AlllOl NBS NATL INST OF STANDARDS & TECH R.I.C.

PUBLICATiONS
All 101909396
/NBS monograph
QC100 .USSe \/l72;1983 C.2 NBS-PUB-C 1959

i/t

NBS MONOGRAPH 172

U.S. DEPARTMENT OF COMMERCE / National Bureau of Standards

Liquefied Natural Gas

Densities: Summary of
Research Program at the
National Bureau of Standards

I
 NATIONAL BUREAU OF STANDARDS

The National Bureau of Standards' was established by an act of Congress on March 3, 1901.
The Bureau's overall goal is to strengthen and advance the Nation's science and technology
and facilitate their effective application for public benefit. To this end, the Bureau conducts
research and provides: (1) a basis for the Nation's physical measurement system, (2) scientific
and technological services for industry and government, (3) a technical basis for equity in
trade, and (4) technical services to promote public safety. The Bureau's technical work is per-
formed by the National Measurement Laboratory, the National Engineering Laboratory, and
the Institute for Computer Sciences and Technology.

THE NATIONAL MEASUREMENT LABORATORY provides the national system of

physical and chemical and materials measurement; coordinates the system with measurement
systems of other nations and furnishes essential services leading to accurate and uniform
physical and chemical measurement throughout the Nation's scientific community, industry,
and commerce; conducts materials research leading to improved methods of measurement,
standards, and data on the properties of materials needed by industry, commerce, educational
institutions, and Government; provides advisory and research services to other Government
agencies; develops, produces, and distributes Standard Reference Materials; and provides
calibration services. The Laboratory consists of the following centers:

Absolute Physical Quantities^ — Radiation Research — Chemical Physics —

Analytical Chemistry — Materials Science

THE NATIONAL ENGINEERING LABORATORY provides technology and technical ser-

vices to the public and private sectors to address national needs and to solve national
problems; conducts research in engineering and applied science in support of these efforts;
builds and maintains competence in the necessary disciplines required to carry out this
research and technical service; develops engineering data and measurement capabilities;
provides engineering measurement traceability services; develops test methods and proposes
engineering standards and code changes; develops and proposes new engineering practices;
and develops and improves mechanisms to transfer results of its research to the ultimate user.
The Laboratory consists of the following centers:

Applied Mathematics — Electronics and Electrical Engineering^ — Manufacturing

Engineering — Building Technology — Fire Research — Chemical Engineering^
THE INSTITUTE FOR COMPUTER SCIENCES AND TECHNOLOGY conducts
research and provides scientific and technical services to aid Federal agencies in the selection,
acquisition, application, and use of computer technology to improve effectiveness and
economy in Government operations in accordance with Public Law 89-306 (40 U.S.C. 759),
relevant Executive Orders, and other directives; carries out this mission by managing the
Federal Information Processing Standards Program, developing Federal ADP standards
guidelines,and managing Federal participation in ADP voluntary standardization activities;
and technological advisory services and assistance to Federal agencies; and
-•rovides scientific
jvides the technical foundation for computer-related policies of the Federal Government,
fhe Institute consists of the following centers:

Programming Science and Technology — Computer Systems Engineering.

'Headquarters and Laboratories at Gaithersburg, MD, unless otherwise noted;

mailing address Washington, DC 20234.
'Some divisions within the center are located at Boulder, CO 80303.
Liquefied Natural Gas Densities:
Summary of Research Program at
the National Bureau of Standards

W. M. Haynes
R. D. McCarty
M. J. Hiza

Chemical Engineering Science Division

National Engineering Laboratory
National Bureau of Standards
Boulder, CO 80303

U.S. DEPARTMENT OF COMMERCE, Malcolm Baldrige, Secretary

NATIONAL BUREAU OF STANDARDS, Ernest Annbler, Director

Issued October 1 983

 Library of Congress Catalog Card Number: 83-600608

National Bureau of Standards Monograph 172

Natl. Bur. Stand. (U.S.), Monogr. 172, 241 pages (Oct. 1983)
CODEN: NBSMA6

U.S. GOVERNMENT PRINTING OFFICE

WASHINGTON: 1983

For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402
Price $
(Add 25 percent for other than U.S. mailing.)
 Contents
Page
1. Introduction 2

2. Major Accompl ishments 4

2.1 Experimental Apparatus 4

2.2 Experimental Measurements 6

2.2.1 Pure Fluid Jata 6

2.2.2 Binary Mixture Jata 6

2.2.3 Mul ti component Mixture uata 7

2.3 Mathematical Models 8

3. LiMG Density Research at Uther Laboratories y

4. Acknowledgments 11

b. Keferences 13

6. Major Publications of LNG Density Project 19

6.1 Reference [5] - Magnetic Suspension Densimeter for

Measurements on Fluids of Cryogenic Interest 19

6.2 Reference [9] - Simplified Magnetic Suspension Densimeter

for Absolute Density Measurements 33

6.3 Reference [6] - Apparatus for Density and Dielectric

Constant Measurements to 35 MPa on Fluids of Cryogenic
Interest 37

6.4 Reference [13] - Measurements of the Orthobaric Liquid

Densities of Methane, Ethane, Propane, isobutane, and i\lormal

Butane 49

6.5 Reference [14] - Orthobaric Liquid Densities of Normal

Butane from 135 to 3U0 K as Determined with a Magnetic
Suspension Densimeter 59
6.6 Reference [16] - Orthobaric Liquid Densities and Excess
volumes for Binary Mixtures of low Molar-Mass Alkanes and
Nitrogen Between 105 and 140 K 65

6.7 Reference [17] - Orthobaric Liquid Densities ana Dielectric

Constants of (Methane + 2-Methyl propane) and (Methane +
("•J-Butane) at low Temperatures 89

ii 1
 Page
6.8 Reference [18] - Liquid Mixture Excess Volumes and Total
Vapor Pressures Usiny a Magnetic Suspension Densimeter with
Compositions Determinea by Chromatographic Analysis:
Methane Plus Ethane 99

6.9 Reference [20] - Orthobaric Liquid Densities and Excess

Volumes for Mul ticomponent Mixtures of Low Molar-Mass
Alkanes and Nitrogen Between 105 and 125 K 107
6.10 Reference [21] - Measurements of Orthobaric-Liquid Densities
of Mul ticomponent Mixtures of LN6 Components (N2, CH^, C2^q,
C3H3, CH3CH(CH3)CH3, C4H1Q, CH3CH(CH3)C2H5 , and CgHj^g^
Between 110 and 130 K 117

6.11 Reference [34] - Four Mathematical Models for the Prediction

of LiMG Densities 127
6.12 Reference [35] - Mathematical Models for the Prediction of
Liquefied-i\latural-Gas Densities 205
6.13 Reference [25] - An Empirical Excess Volume Model for

Estimating Liquefied Natural Gas Densities 223

6.14 Reference [22] - Prediction of Liquefied-Watural-Gas (LNG)
Densities from New Experimental Dielectric Constant Data . 235
 LIQUEFieu fJATUKAL GAS DENSITItS: SUMMARY OF RESEARCH
PROGRAjVI at the MATIOiNlAL BUREAU OF STAf^lUARDS

W. M. haynes, R. u. McCarty and M. J. Hiza

Chemical Engineering Science ui vision
National Engineering Laboratory
i\lational bureau of Standards
Boulder, Colorado 80303

This report summarizes the results of a project concerning the

densities of liquefied natural gas (LiMO) and its components. This
project, initiated in the Properties of Fluids Section of the
Cryogenics Division of the relational Bureau of standards in July 1972!,
was carried out under the sponsorship of a consortiu.n of eighteen
energy companies .

The experimental part of this project has included the following

accomplishments: (a) development of a magnetic suspension densimeter
for absolute density measurements on liquids, including liquid mixtures
in equilibrium with their vapor, at temperatures from 90 to 300 K;
(b) orthobaric liquid density measurements on the major components of
LNG, which include nitrogen (95-12!0 l<), methane (105-160 K), ethane
(100-270 K), propane (100-288.7 K), isobutane (115-300 K), and normal
butane (135-300 K); (c) orthobaric liquid density measurements on
approximately thirty-five binary mixtures of the above components for
all combinations except nitrogen + butane systems, primarily in the
temperature range of 105 to 130 K; ana (d) ortnobaric liquid density
measurements on twenty-seven multi component mixtures (105-120 K),
including several LNG-like mixtures with up to eight components. The
total uncertainty of a single density measurement is approximately 0.1
percent at low temperatures and decreases to approximately 0.06 percent
at room temperature. The estimated standard deviation of a single
density measurement is less than 0.02 percent.

Tne density data have been used to optirnize, test, and compare
several mathematical models as to their suitability for the calculation
of LNG densities for custody transfer. Hodels selected for optimiza-
tion and testing included an extended corresponding states method, a
hard sphere model, a cell model, and an empirical model due to Klosek
and McKinley. The ultimate goal of this project was to produce one or
more mathematical models that could be used to predict the density of
any Li\lG mixture to within an uncertainty of 0.1 percent from an input
of pressure, temperature, and composition. After revisions based on
the new experimental data from this project, each of the models

* British Gas Corp., Chicago Bridge and Iron Co., Columbia Gas Service Corp.,
Oistrigas Corp., Easco Gas LWG, Inc., El Paso Natural Gas, baz de France,
I'^arathon Oil Co., Mobil Oil Corp., Natural Gas Pipeline Co., Phillips Petroleum
Co., Shell International Gas, Ltd., Sonatrach, Southern California Gas Co.,
Tennessee Gas Pipeline Co., Texas Eastern Transmission Co., Tokyo Gas Co., Ltd.,
and Transcontinental Gas Pipe Line Corp.

1
 investigated here satisfy this goal for typical LNG compositions. The
limitations and ranges of validity of the various models are discussed.
Also presented are techniques for predicting Li\iG densities from
dielectric constant measurements and from excess volume calculations.

The last section of this report consists of publications that

provide a complete and detailed account of the results of this
project.

Key words: binary mixtures; density; experimental data; liquefied

natural gas; magnetic suspension densimeter; mul ticomponent mixtures;

prediction methods; pure fluids.

1. Introduction
Uuring the past decade liquefied natural gas has become an increasingly
important commodity on the world energy market. This trend is expected to
continue into the foreseeable future. In the buying ana selling of LNG, the

basis of custody transfer is its heating value. The determination of the heating
value of LMG requires a knowledge of its density, which in turn is dependent upon
its composition, temperature, and pressure. Since, for example, an error of
one percent in density can result in an inequity of between :i>luO,uuO and !>200,000
(at 1983 prices) per 125, OUU m-^ shipload of LNG, the accuracy to which the
density of the liquid can be determined is extremely important in transactions
involving LNG. A one percent error in density was not uncommon at the time this
project was initiated.
There are, at least, two means for determining the density of a large volume
of LNG. One isiethod is by direct field measurements using commercially available
densimeters. A second method consists of the use of a mathematical model or
correlation to calculate a density based on direct measurements of the liquid
composition and temperature.
This report describes a project that has been concerned with the determina-
tion of the density by the second method. In actual transfer situations it is
likely that both methods will be used to complement each other. By either
method, the accuracy to which the density can be determined is limited by both
practical and state-of-the-art considerations. Tnis has resulted in setting a
goal of 0.1 percent for the total uncertainty in the determination of the density
of L.MG.

2
 In July 1972, a project was initiated at this laboratory with an ultimate
goal of providing one or more mathematical models that could be used to calculate

the density of any LNG mixture with a total uncertainty of less than 0.1 percent
based on a knowledge of the pressure, temperature, and composition of the liquid
mixture. The mathematical models would be developed and/or optimized using
density data for the major components of LnG and for binary mixtures of these
components. Density data for multi component mixtures of LNG components would be
used to evaluate and compare the performance of the mathematical models. Some of

the test mixtures would be selected to simulate commercial Li\iG compositions. LNG

consists primarily of methane (concentration level typically greater than 75

percent) with lesser amounts of ethane (up to i5 percent), propane (up to
b percent), butanes (up to 2 percent), nitrogen (up to 2 percent), and pentanes
plus heavier hydrocarbons (up to 0.5 percent).
The accuracy of a calculational technique based on experimental data can be
no better than the input data from which the technique is developed. At the time
this project was initiated, comparisons of saturated (orthobaric) liquid density
data for the pure components of LNG exhibited differences as large as 0.5
percent, while each investigator was generally claiming inaccuracies of 0.1
percent or less. Furthermore, there were only a few sets of published data for
liquid mixtures containing LNG components, especially at low temperatures. These
data were generally limited in scope or had claimed inaccuracies larger than the
desired 0.1 percent. (The pure fluid and mixture data for LMb components from

other investigations are sumi.iarized in Section 3 of this report.) Thus, a major

task of the LNG density project at this laboratory has been to provide an
accurate and internally consistent set of density data for the major components
of LiMG and for mixtures of these components. (These data could also serve as a

basis for calibration or development of gauging and metering methods for Li^b and
for process design and operation of LNG facilities.)
This project was sponsored by a consortium of eighteen international energy
companies, five of which were from countries outside the United States. The
sponsors represented both buyers and sellers of Lf^iG. In carrying out this

research project, the National Bureau of Standards is serving in its traditional

role as an independent third party. All results of this project would be
published in the open literature and hopefully gain wide acceptance throughout
the Ll'^G industry, since qualified sponsor representatives from all parts of the

3
international LNG market would have closely scrutinized the progress of this
project throughout its entirety.
The major purpose of this report is to incorporate under a single cover a

complete account of the LNG density project. The major accomplishments of the
project are summarized in Section while the detailed results are presented in
the papers that comprise Section 6. Most of these papers are reprints of publi-
cations, while a few are presented in their existing, prepublished form. During
the course of this research, several papers [1-4] have been published that
presented a summary or status report of the project. These papers are not
included in Section 6. The present report serves as the final report to the
sponsors of this project.

2. Major Accompl ishments

2.1 Experimental Apparatus

The experimental technique selected for performing the density measurements'
for this project had to satisfy relatively stringent criteria. It must be

capable of absolute density measurements of high accuracy and precision on a

liquia, including mixtures, in equilibrium with vapor at cryogenic temperatures
and at pressures to approximately 4 MPa using an optical cell. A magnetic
suspension densimeter, based on an application of Archimedes' principle, was
developed for this purpose.
Two apparatus [5,6], each incorporating a magnetic suspension technique in
the density determination, were constructed for the measurements on liquids,
including liquid mixtures, associated with this project. For both apparatus, the
total uncertainty of a single density measurement, which is taken as three times
the standard deviation plus the systematic error, is approximately 0.1 percent at
low temperatures and decreases to approximately U.06 percent at room temperature.
The estimated standard deviation of a single density measurement is less than
0.02 percent.
A brief description of the method for determining densities with either

apparatus is as follows. A small magnetic buoy is freely suspended in the fluid

whose density is to be determined. The density of the buoy is significantly
greater I by approximately an order of magnitude) than che density of the fluids

investigatea here. I'he force necessary to lift the buoy is supplied oy one or
more air-core solenoids. The buoy is siiaintained at a stable (vertical) position

4
 .

through the automatic regulation of an electronic servocircuit containing the

lift coil and either a differential transfoniier [5j or a differential
capacitor [7] that aetects the motion of tiie buoy. Ihe horizontal position of
the Duoy is maintained by the axially symmetrical, diverging fiela of the lift
coil

When the buoy is supported at the sanie position in vacuum and in the fluiu
of interest, the upward magnetic force on the buoy supplied by the air-core
solenoids gives a nieans for measuring the density of the fluid since tnis is the

force that must be added to the buoyant force to ualance the downward gravita-
tional force. Thus, the density is deduced from measurements of the currents in
the support coils necessary to support the buoy in vacuum and in the fluia of
interest at the same temperature and at the same vertical position (determined
with a 1^5X microscope).
The mass and volume of the buoy must be determined from independent
measurements. The mass was determined using an analytical microbal ance. The
volume of the buoy at room temperature was determined by using distilled water as
a reference fluid of known density. Thermal expansion data [8] for barium
ferrite, tne material from which the buoy was fabricated, were obtained so that
the volume of the buoy at low temperatures could be calculated.
The first apparatus [5] could be used at temperatures between 90 and 30u K
and at pressures to 5 MPa. After initial tests with a three-coil arrangement to
support the buoy, the densimeter was simplified considerably by reducing to a

system employing only one coil [9J. The second apparatus LoJ, which employed
exactly the same technique for deteraiining density, was significantly different
from, and wiore versatile than, the first apparatus. Tne second apparatus con-
tained a concentric cylinaer capacitor that was used to make dielectric constant
measurements on the same liquid samples for which density data were obtained.
The major reason for the construction of the second apparatus was a need in other
research projects (e.g., PVT and dielectric constant measurements on liquid
propane [10], isobutane [11], and normal butane [12]) for a higher pressure
capability (35 MPa) for the magnetic suspension densimeter. The extensiuon to
higher pressures was not needed in the LHG density project.
It should be noted that the second apparatus gave no improvement over the
first in the accuracy to wnich the density could De determined. The consistency
of the density data from both apparatus should be equivalent to that for either

5
apparatus alone. To ensure that this was the case, measurements on liquid
methane were used as a control throughout the project with both apparatus.

1.1 Experimental heasureinents

Pure Fluid Uata

The first experimental measurements for the Li\iG density project consisted of
the acquisition of orthobaric liquid density data for the major components of
lnG. Comprehensive results were obtained for nitrogen (95-12U K) [b], methane
(105-160 K) [5,13], ethane (100-270 K) [13], propane (100-288.7 K) [13],
isobutane (115-300 K) Li3], and normal butane (135-300 K) [13,14]. Oetailed com-
parisons between the data from this project and independent results demonstrated
the need for an accurate and internally consistent set of data for the major com-
ponents of LimO. Differences as large as 0.5 percent between che density data of
independent investigators were comiiion. At the time this project was initiated,
there were no published data for normal butane at low temperatures. (In the

course of evaluating the performance of the magnetic suspension densimeter,

density data were also obtained for saturated liquid argon (luu-12o K) and
ethylene (105-200 n) Lit>].)

2.2.2 Binary Mixture Data

Next, orthobaric liquid density measurements were carried out on thirty-five
binary mixtures l1d-18] containing the major components of limG. Most of the
binary mixture data was taken in the temperature range of 105 to 140 K. All

binary combinations of the six major components of Li^G were investigated in this
project, with the exception of the nitrogen + isobutane and nitrogen + normal
butane systems.
Prior to the LiMG density project, no low temperature, liquid density data
could be found in the literature for the following systems: nitrogen + ethane,
nitrogen + propane, ethane + isobutane, propane + isobutane, propane + normal
butane, and isobutane + normal butane, all of which have been investigated here.
All of the binary mixture measurements were carried out on liquid samples

condensed into the cell from gravimetrical ly prepared gas mixtures. This was
considered to be the most accurate method to fix the compositions of tne liquid
mixtures. For all mixtures containing nitrogen and/or methane, total vapor
pressures have also been measured. For the methane-rich binary mixtures
containing either isobutane or normal butane, dielectric constant data [17] were
obtained simultaneously with the density results.

6
 txtensive comparisons of the binary mixture data from the present study have
been made with available literature data. In general, differences were less than
0.1 percent, except that the density data here for the ethane + propane system
exhibited approximately an 0.8 percent discrepancy when compared with the data of
Shana'a and Canfield [19]. This was inconsistent with comparisons with the same
authors for the methane + ethane and methane + propane systems as well as for
pure fluid results, where the agreement was typically better than 0.1 percent.

2.Z.3 Mul ti component Mixture Uata

In order to evaluate and test the mathematical models that have been
developed and optimized using the binary mixture and pure component data,
orthobaric liquid density measurements have been performed on twenty-seven
mul ticomponent mixtures [20,21] of Li^u components, primarily in the temperature
range of 110 to 120 K. The mul ticomponent mixtures investigated in this project
ranged from ternary mixtures (methane and binary combinations of nitrogen,
ethane, propane, and normal butane) to four- to eight-component methane-rich
(75-90 mole percent) mixtures containing up to 5 mole percent nitrogen, 15 mole
percent ethane, 7 mole percent propane, 5 mole percent butanes, and 0.44 mole
percent pentanes.
The compositions of some of the six- to eight-component mixtures were
selected to simulate commercial LNG mixtures. The compositions of other
mul ticomponent mixtures were chosen to provide severe tests of the mathematical

models and to complement the use of binary mixture data in the optimization of
the models, txcept for three of the mul ticomponent mixtures, the compositions
were determined from the preparation of the gas mixtures by gravimetric means.
For the other three mixtures, the compositions were determined by gas chromato-
graphic analysis using a thermal conductivity detector. The chromatograph was
calibrated with a gravimetrically prepared mixture. Vapor pressure data have
also been obtained for the mul ticomponent mixtures. For those mul ticomponent
mixtures investigated with the secona apparatus, dielectric constant
measurements iZt:} were performed simultaneously with the density measurements.
Although the pure fluid and binary mixture measurements from this project
have not included data for pentanes, five mul ticomponent mixtures containing
pentanes [21] have been investigated. It was thought that densities of these
systems could be predicted with the mathematical models using existing pure fluid
data for the pentanes from other sources [23,24] and predicting the interaction

7
parameters for binary mixtures containing pentanes from the behavior of the
binary mixtures studied in the LNb density project. (Expressions for represent-
ing the pentane data of Orrit and Laupretre [24] are reported by Hiza [25].)
High accuracy in these predictions is not required since LNG mixtures contain
relatively small fractions of pentanes.

2.3 Mathematical Models

AS mentioned earlier, one method of determining the density of LNG is to
predict that density with a mathematical model of the equation of state, given
the pressure, temperature, and composition of the Lt>i(a. The ultimate goal of this

study was to produce such a model (or models) that would be accurate to within

0.1 percent of the true density. The scope of this study was limited to

saturated liquids over a temperature range of 105 to 140 K with pressures to

2 MPa. The components of LisiG were assumed to be CH^, C2H5, C3H^, iC4Hj^Q,

nC4hj^Q, iC^Hj^^' "^5^12*

Four models were chosen to fit to the experimental data, tach of the
selected models represents a different approach to modeling the equation of state
of a fluid. These are an extended corresponding states model [26-27], a hard
sphere model [28], a cell model [^.9-31], and a graphical model due to Klosek and

McKinley [32]. During the course of optimizing the four models using the
experimental data, it became necessary to revise the functional form of the model
proposed by Klosek and McKinley. No revisions of the functional form of the
other three models were necessary, and only the adjustable parameters were
changed to achieve the desired fit of the experimental aata. The final result
was that all four of the models originally chosen can be used to predict the
density of LNG to within U.l percent of the true density, given the temperature,
pressure, and composition of the Li^jb. uetails of the fitting procedures,
comparisons to experimental data, and computer program listings are given in
references [33-35].
Also during the course of this study, an excess volume model was developed
by Hiza [25]. This model also achieves the 0.1 percent criteria outlined above
under the same restrictions as the other models. Lastly, a model has been
developed by Haynes and ^lcCarty [22] that does not require either the temperature
or pressure as input information (all of the other models require a minimum of

temperature as input and most require both temperature and pressure). This model

8
requires an input of dielectric constant and composition of the li^iu and predicts
the density to within 0.15 percent of the experimental value.

All of the models have restrictions on pressure, temperature, and composi-

tion ranges, which must be defined for each model. The models are of widely
ranging computational complexity. The selection of a particular model is

dependent on the needs and objectives of the individual user. However, on the
oasis of the accuracy of the calculated density of a commercial linIu iiiixture, each
of the models appears to be equally acceptable.
One final comment about accuracy should be emphasized. No mathematical

model of the equation of state can be more accurate than the experimental data to

which it has been fit. Therefore, all of the accuracy claims are dependent upon
the accuracy of the experimental data from this study, which were used almost
exclusively to optimize and test the models.

3. LN6 Density Research at Other Laboratories

Due to the widespread interest in reliable data and the prediction of
properties of mul ti component liquefied natural gas mixtures, in particular
saturated liquid (buDble point) densities, a number of independent research
programs were conducted from whict> a significant amount of density results have
been published in the open literature. These include the studies at baz de
France (Morlet [oo]). Air Products and Cnemicals, Inc. (KloseK and
McKinley [32]), Institute of bas Technology (uonzalez, et al. iy/^^bj, Huebler,
et al . [39]), University of Kansas (Huang, et al . [40], Jensen and Kurata [41]),
University of Oklahoma (bhana'a and Canfield [19]), Shell Research, Ltd. (Boyle
and Reece [42], McClune [23]), university of Wyoming (i^iller, et al. [28,43 - 46]),
•k
and Societe hiationale Elf Aquitane (Roche, et al . [47,48j, Orrit and Laupretre
[24,49], and urrit [50]). There is considerable variation in the precision and
accuracy of measurement and in the completeness of the data published from tnese
studies. Therefore, the contributions of these programs to the precise knowledge
of the density behavior of Lnu mixtures are also quite variable.
From comparisons with these independent results, it appears that the data
from the University of Wyoming are the most precise and are uniformly consistent

* Certain companies are identified in this paper for the purpose of clarity
only. Such identification does not imply any type of endorsement by the National
Bureau of Standards.

9
with the results of the NBS program. Direct comparisons of measurement precision
were also made between the NbS densimeter and the Wyoming apparatus. Measure-
ments were made at the university of Wyoming on one binary mixture and four
mul ti component mixtures, prepared at NBS, with the apparatus calibrated with pure
metnane densities determined with the MBS densimeter [46]. Densities for the
binary mixture and three of the mul ticomponent mixtures were also measured with
the nBS densimeter. Cosnparisons were made through the extended corresponding

states model when measured temperatures were not exactly the same. The maximum
difference found was 0.10 percent for one mul ticomponent mixture point; the
remaining differences were less than 0.03 percent and random. Though the Wyoming
program was carefully planned, measurements for some of the important binary
mixtures (e.g., nitrogen + ethane, nitrogen + propane, ethane + propane, etc.)
were not included, and the accuracy of the molar volume results (though not that
of derived excess volumes) depends directly on the accuracy of the density data
of tne low temperature liquid used to calibrate the apparatus.
The most extensive experimental measurements are those from the Elf Aquitane
program. Their precision of measurement does not appear zo oe quite as good as
the measurements of tne University of Wyoming, and there is a systematic bias of
about -0.1 percent in a sample of mul ticomponent mixtures data [49] compared to
the extended corresponding states model optimized to the iMBi) results. Though not
all of the important binary mixtures were includea in their study, it is notable

that measurements were made on a nearly equimolar mixture of ethane +

propane [bO]. txcess volumes derived from these data and their pure fluid data
between 105 and 140 K are in excellent agreement with those of the NBS program.
Most of their V*^ values are between -0.01 and -0.04 cm"^/mol compared to -0.03
to -O.Ob cm^/mol for a comparable mixture from the present study and about
-0.49 cm^/mol from the University of Oklahoma data at 108.15 K for a mixture
containing 58.5k: mole percent ethane.
With the exception of this large discrepancy in the University of uklahoma
data for the ethane + propane mixture, the remainaer of the data from the
University of Oklahoma, as well as those from Shell Research, Ltd., are of
comparable precision as those from the University of Wyoming and from £lf
Aquitane. From the Shell Research, Ltd. program, only the pure component density
data have been published. i>lone of the remaining programs have provided data

approaching the precision ana accuracy goals desired for custody transfer.

10
 The proyrarns of Llf Aquitane and Shell kesearch, Ltd. provide the only low
temperature experimental densities for pure isopentane and normal pentane, and
these data sets are in good agreement, i.e., within about 0.1 percent. With the
small concentrations of pentanes normally encountered in Ll\lb mixtures, and the
fact that the densities from these investigations are generally within about 0.1
percent of the NBS values for the lower molar mass alkanes, it was felt that
additional measurements on the pure pentanes were unnecessary. In the NBS

program, the pentanes were included only in multi component mixtures, with
compositions much like Li\IG mixtures that would be encountered in practice. These
measurements were made to provide proof that the mathematical models developed
could properly account for the presence of pentanes.
Compared to the LUii research programs noted above, the NBS program is unique
in that a single set of data is provided of uniformly high precision and accuracy

for all of the pure components (i.e., nitrogen and the lower mass alkanes through

the butanes), all the possible binary combinations where liquid phase separation

does not occur, the important ternary combinations, and mul ticomponent mixtures
that contain the highest probable amounts of nitrogen, butanes, and pentanes.
The measurement method employed is a sophisticated state-of-the-art method for

which a detailed analysis of the measuresnent uncertainties have been made and
published, and statistical control of the experiments was assured throughout the
program by repeating measurements of the density of liquid methane with each
experimental run. The mathematical models optimized to the data are representa-
tive of existing theoretical and empirical models in wide use, from the most
simple to the most complex. The concurrent experimental and mathematical
modeling efforts also served to guarantee that sufficient data were obtained to
optimize and to identify the inherent limitations of each.

4. Acknowledgments
The contributions of the members of the LrJG Density Project Steering
Coiiiiiiittee are gratefully acknowledged. The Steering Committee, comprised of
representatives from each of the sponsoring companies, met quarterly from May 11,
1972 to February 15, 1979. The critique of research progress and numerous
suggestions offered by the members were extremely valuable. Special thanks are
due i-lartin k. Cines of the Phillips Petroleum Company who served as Chairman of

the LNG Density Project steering Committee during that time, and who played a
major role in obtaining sponsors for this project. Thanks are also due Louis A.

11
Sarkes of the American Gas Association for his general support of LNG properties
research and for his contriDutions to this project. The AGA coordinated the
funding from the sponsoring companies for this project. L. James Kemp of
i>outhern California Gas Company assumed Chairmanship of the Steering Committee

following the retirement of Martin R. Cines from Phillips Petroleum Company. The
authors woulo like to express their appreciation to Karen A. Bowie and
Lawrence 0. Liberman for their assistance in the preparation of the manuscript.

12
5. References
[I] riaynes, W. h.; hiza, M. J.; hcCarty, R. U. Densities of Lub for custody
transfer. Proceedings of Fifth International Conference on LiMb, vol. 2;

1977 August ii9-Septeniber 1; Uusseldorf, Germany. Chicago, 111.: Institute

of Gas Technology; i>ection III, Paper 11. 30 p.

IZ'j Haynes, W. iM.; McCarty, K. 0. The density of liquefied natural gas.

Proceeding of the Fifty-Eighth annual (GPA) Convention of the Gas

Processors Association; 1979 March 19-^11; Denver, Colo.: Gas Processors

Association, Tulsa, ukla. lu-ib.

[3 J Diller, u. t. L(\ib density determination, hydrocarbon Processing 56(4):

142-144; 1977 April.

[4] McCarty, R. u. LnG densities for custody transfer. Proceedings of the

Fifty-Sixth International School of Hydrocarbon measurements; Ib/Bl

April 14-16; Norman, Okla. 515-517.

[5j riaynes, W. H.; Hiza, H. J.; Frederick, V. Magnetic suspension

densimeter for measurements on fluids of cryogenic interest. Rev. Sci.
Instrum. 47(10): 1237-iii50; 1976 October.

[6] Haynes, W. M. Apparatus for density and dielectric constant measurements

to 35 MPa on fluids of cryogenic interest, d. Res. f^at. Bur. Stand.

(U.S.) b8(4): 241-252; 1983 July-August.

[7] Frederick, N. V.; Haynes, W. IM. Differential capacitance sensor as

position detector for a magnetic suspension densimeter, kev. Sci. Instrum.
50(y): 1154-1155; 1979 September,
[bj Clark, A. F.; Haynes, w. M.; Deason, V. A.; Trapani, R. d.; Low

temperature thermal expansion of barium ferrite. Cryogenics ib(j):

267-270; 1976 May.

[9] Haynes, W. M. Simplified magnetic suspension cieiisimeter for absolute

density measurements. Rev. Sci. Instrum. 48(1): 39-41; 1977 January.
[10] haynes, W. M. Measurements of densities and dielectric constants of liquid
propane from 90 to 300 k at pressures to 35 MPa. J. Chem. Thennodynamics
15(5): 419-424; 1983 May.
[II] Haynes, W. M. Measurements of densities and dielectric constants of liquia
isobutane from 120 to 300 K at pressures to 35 MPa. J. Chem. Eng. Data, to
be published.
[12] riaynes, W. M. Measurements of densities and dielectric constants of liquid
normal butane from 140 to 300 K at pressures to 35 MPa. J. Chem.
Thermodynamics 15(9): 801-805; 1983 September.
13
[13] Haynes, A. M.; riiza, i4. J. Measurements of the orthobaric liquid densities
of methane, ethane, propane, isobutane, and normal Dutane. J. Chem.
Thermodynaiiiics 9(Z): i79-id7; 1977 February.

[14 j riaynes, w. H.; liiza, H. J. Orthobaric liquid densities of normal butane

from 13b to 3U0 K as detennined mth a magnetic suspension densimeter.
Advances in Cryogenic Engineering, Vol Zl, K. U. Timmerhaus and u. H.

Weitzel, ed. ^^ew YorK, f\iY: Plenum Press; 1976. 516-62!1.

[lb] riaynes, w. M. Measurements of the orthobaric liquid densities of argon
(100-120 K) and ethylene (105-200 K). Cryogenics 18(10): 621-623; 1978
October.
[16] iiiza, M. vJ.; riaynes, w. M,; Parrish, vL R. Orthobaric liquid densities and
excess volumes for binary mixtures of low molar-mass alkanes and nitrogen
between 10b and 140 K. J. Chem. Thermodynamics 9(9): c573-896; 1977
September.
[17] Haynes, W. tA. Orthobaric liquid densities and dielectric constants of
(methane + 2-methyl propane) and (methane + n-butane) at low temperatures.
J. Chem. Thermodynainics 15(10): 903-911; 1983 October.
[18] Hiza, H. J.; Haynes, W. M. Liquid mixture excess volumes and total vapor
pressures using a magnetic suspension densimeter with compositions
determined by chromatographic analysis: methane plus ethane. Advances in
Cryogenic Engineering, Vol 23. K. D. Timmerhaus, ed. hiew York, wY: Plenum
Press; 1976. 594-601.
[19] Shana'a, M. Y.; Canfield, F. B. Liquid density and excess volume of light
hydrocarbon mixtures at -165°C. Trans. Faraday Soc. 64(549): 2281-2286;
1968 September.

[20] Hiza, M. J.; naynes, W. M. Orthobaric liquid densities and excess volumes
for mul ti component mixtures of low molar-mass alkanes and nitrogen between
105 and 125 K. J. Chem. Thermodynamics 12(1): 1-10; 1980 January.

[21] haynes, W. M. Measurements of orthobaric-1 iquid densities of

mul ticomponent mixtures of Li^b components (i\i2, CH^, ^2*^6* ^3*^8'
Cri3CH(CH3)CH3, C^H^q, CH3Ch(CH3)C2H5, and C^ri^^) between 110 and 130 K.
J. Chem. Thermodynamics 14(7): 603-612; 198*1 July.

[22] Haynes, W. ; McCarty, K. D. Prediction of liquefied-natural-gas (LNG)

densities from new experimental dielectric constant data. Cryogenics
23(8): 421-426; 1983 August.

14
[23] McClune, C. k. Measurement of the densities of liquefied hydrocarbons from
-100 to -ISO'^C (173 to 93 K). Cryogenics 16(5): 289-295; 1976 May.

[24] Orrit, J.; Laupretre, J. N. Density of liquefied natural gas constituents.

Advances in Cryogenic Engineering, Vol. 23. K. u. Timmerhaus, ed. New

York, NY; Plenum Press; 1978. 573-579.
[25] Hiza, M. J. An empirical excess volume model for estimating liquefied
natural gas densities. Fluid Phase Equilibria 2(2): 27-38; 1978 August.

[26] kowlinson, J. S.; watson, I. u. The prediction of the thermodynamic

properties of fluids and fluid mixtures - I. The principle of
corresponding states and its extensions. Chem. Eng. Sci . 24(10):
1565-1574; 1969 October.

[27] Mollerup, o.; Rowlinson, J. S. The prediction of the densities of

liquefied natural gas and of lower molecular weight hydrocarbons. Chem.
Eng. Sci. 29(b): 1373-1382; 1974 June.

l28] Rodosevich, J. B.; Miller, R. C. Calculation of Li^G excess volumes by a

modified hard-sphere model. Advances in Cryogenic engineering, Vol. 19.
K. D. Timmerhaus, ed. New York, i\IY: Plenum Press; 1974. 339-345.
[29] Renon, ri.; Eckert, C. A.; Prausnitz, J. M. Molecular thermodynamics of
simple liquids. Ind. Eng. Chem. Fundam. 6(1): 52-5b; 1967 February.

[30] Eckert, C. A.; Renon, h.; Prausnitz, u. M. Molecular thermodynamics of

simple liquids. Ind. Eng. Chem. Fundam. 6(1): 58-67; 1967 February.
[31] Aloright, M. A. A model for the precise calculation of liquefied natural
gas densities. NbPA Tech. Pud. TP-3. 1973.

[32] Klosek, J.; McKinley, C. Densities of liquefied natural gas and of low
molecular weight hydrocarbons. White, J. W.; Neumann, A. E. S., ed.

Proceeding of the First International Conference on LNG; 1968 April 7-12;

Chicago, 111. Chicago, 111.: Institute of Gas Technology; Session 5,

Paper 22. 15 p.
[33] McCarty, R. D. A comparison of mathematical models for the prediction of
LivlG densities. iMat. Bur. Stand. (U.S.) NBSIR 77-867; 1977 October. 60 p.

[34] McCarty, k. u. Four mathematical models for tne prediction of Lng

densities. Nat. Bur. Stand. (u.S.) Tech. note i030; 1980 uecemoer. 76 p.
[35] McCarty, R. u. Mathematical models for the prediction of liquefied-
natural-gas densities. J. Chem. Thermodynamics 14(9): S37-854; i98^
September.

15
[36] Morlet, J. The density and miscibility of liquefied hydrocarbon gases at
low temperatures. Rev. Inst. Francais Petrole 18: 127-143; 1963 January.
[37] laonzalez, M. H.; Subrainanian, T. K.; Kao, R. L.; Lee, A. L. Physical
properties of natural gases at cryogenic conditions. White, J.W.; Neumann,
A. £. S., ed. Proceedings of the First International Conference on LMG;
1968 April 7-12; Chicago, 111. Chicago, 111.: Institute of Gas Technology;
Session 5, Paper 21. 38 p.

[38] Gonzalez, H. H,; Lee, A. L. Qew and bubble points of simulated natural
gases. 0. Chem. Eng. ilata 13(2): 172-176; 1968 April.

[39] Huebler, J.; takin, B. E.; Lee, a. L. Physical properties of natural gas
at cryogenic conditions. Proceedings of the Second International
Conference on Li\ib, Vol. I; 197U October 19-23; Paris, France. Chicago,
111.: Institute of Gas Technology; Session 2, Paper 9. 10 p.

[40] Huang, E. T. S.; Swift, G. W.; Kurata, F. Viscosities and densities of

methane-propane mixtures at low temperatures and high pressures. AIChE J.-

13(5): 846-850; 1967 September.

[41] Jensen, R. H.; Kurata, F. Jensity of liquefied natural gas. J. Petrol.

Tech. 21: 683-691; 1959 June.

[42] Boyle, G. J.; Keece, D. bulk measurement of Li^lG. Proceedings of the Second
International Conference on LNG, Vol. I; 1970 October 19-23; Paris,
France. Chicago, 111.: Institute of Gas Technology; Session 2, Paper 5.
14 p.
[43] Rodosevich, J. B.; Miller, R. C. Experimental liquid mixture densities for
testing and improving correlations for liquefied natural gas. AIChE J.
19(4): 729-735; 1973 july.

[44] Miller, R. C. Estimating the densities of natural gas mixtures. Chem. Eng.
81(23): 134-135; 1974 October.
[45] Pan, W. P.; hady, M. H.; Miller, R. C. Dielectric constants and

Clausius-Mossotti functions for simple liquid mixtures: systems containing

nitrogen, argon and light hydrocarbons.. AIChE J. 21(2): 283-289; 1975
March.
[46] Miller, R. C; Hiza, M. J. Experimental molar volumes for some Li\ib-rel ated
saturated liquid mixtures. Fluid Phase Equilibria 2(1): 49-57; i978
August.

16
[47] Rocne, Y.; Perret, J.; Sochi rol, l. uensi ty of L^lb and of its
constituents. Proceedings of the Second International Conference on LUG,
Vol. I; 1970 Uctooer 19-i;3; Paris, France. Chicago, 111.: Institute of Gas
Technology; Session 2, Paper b. 8 p.

l48] koche, Y.; Perret, J.; Bochirol, L.; Brauns, P. Density of Li\lG and of its
constituents. White, J. W., ed. Proceedings of the Third International
Conference on LUG; 1972 September 24-28; Washington, D.C. Chicago, 111.:
Institute of Gas Technology; Session 11, Paper l6. li p.

[49] Orrit, J.; Laupretre, J. M. A cal culational method for obtaining the
density of a liquefied natural gas. Advances in Cryogenic Engineering,
Vol. 23. K. u. Tiinmerhaus, ed. i^^ew York, ImY: Plenum Press; 197b. 56b-b79.
[bO] Orrit, J. t. Densities of liquefied natural gas components mixtures,
{submitted to Fluid Phase tquilibria) 1983 January.

17
6. Major Publications of LiMb Density Project
6.1 Reference [5]

Magnetic suspension densimeter for measurements on fluids

of cryogenic interest*
W. M. Haynes, M. J. Hiza, and N. V. Frederick

National Bureau of Standards, Institute for Basic Standards, Cryogenics Division, Boulder, Colorado 80302
(Received 24 June 1976)

An apparatus incorporating a magnetic suspension technique has been

developed for density measurements on Hquids and liquid mixtures, particularly,
at saturation, at temperatures between 90 and 300 K and at pressures to 5 MPa
(approximately 50 atm). The feasibility of adapting this method, previously used
at room temperature, for low temperature use had been demonstrated in an
earlier study with a density measurement on saturated liquid nitrogen near its
normal boiling point. The present apparatus, which is significantly improved,
and in most respects different from the earlier model, is described in detail. It

includes a cryostat for continuous wide-range temperature control, a windowed

equilibrium cell particularly suited for studies of liquid mixtures, and a new
electronic servocircuit with a linear differential transformer for position control
of the magnetic buoy. Extensive tests and density measurements have been
carried out to evaluate the performance of this apparatus. Densities of saturated
liquid nitrogen between 95 and 120 K and saturated liquid methane between 105
and 160 K are reported. The estimated standard deviation of a single density
measurement is less than 0.02%. The total systematic error in the measurement
process from known sources is approximately 0.05%. The total uncertainty of a
single density measurement, which is taken as three times the standard deviation
plus the systematic error, is approximately 0.1%. Comprehensive comparisons of
the present results with previous experimental data are presented.

1. INTRODUCTION feasibility of adapting the magnetic suspension tech-

nique for density measurements at low temperature was
With the advent of large-scale liquefaction and global demonstrated with a measurement on saturated liquid
transport of natural gas, accurate equilibrium liquid- nitrogen near its normal boiling point. The present
phase densities have become important for custody apparatus (see Fig. for a scaled assembly drawing),
1

transfer. The density data for liquefied natural gas mix- which is a much improved version of the earlier
tures and their components have been generally char- model, incorporates a three-coil support system,^ a
acterized by the existence of large discrepancies (ap- cryostat for temperature control at any temperature
proximately 0.5%) between different sets of data for the between 90 and 300 K, a windowed equilibrium cell
pure components and by a limited amount of mixture suitable for projected studies on liquid mixtures at
data. Hence, an apparatus has been developed to pro- pressures to 5 MPa, and a new electronic servo-
vide accurate, consistent, and comprehensive density circuit with a linear differential transformer that serves
data for liquefied natural gas mixtures and their pre- as a position sensor for a barium ferrite magnetic buoy.
dominant pure components at saturation conditions. The present work has included an extensive evalua-
These measurements will serve as a data base for test- tion of the use of a magnetic suspension densimeter
ing and optimizing selected mathematical models (cor- for absolute density measurements. (Here the term
relations) useful in predicting the densities of liquefied "absolute density" used to indicate that cryogenic
is

natural gas. In the present paper the apparatus and fluids were not used to calibrate the instrument.)
experimental procedures are described in detail and Results of performance tests along with densities
representative measurements for two components of for saturated liquid nitrogen between 95 and 120 K and
liquefied natural gas are presented. saturated liquid methane between 105 and 160 K are
A magnetic suspension densimeter, which is based presented in this paper to demonstrate the low temper-
on an application of Archimedes' principle, has been ature performance of this new apparatus. The estimated
selected for the present measurements primarily be- standard deviation of a single density measurement is
cause this instrument is suitable for an independent, less than 0.02%. The total uncertainty of a single
direct determination of the density of a saturated measurement, which is taken as the sum of the syste-
liquid over wide ranges of temperature and pressure matic error and three times the standard deviation,
within the precision and accuracy required for antici- is approximately 0.1%.

pated technical applications. In an earlier study, ^ the In Sec. II the general principles of a magnetic

1237 Rev. Scl. Instrum., Vol. 47, No. 10, October 1976 Copyright © 1976 American Institute of Physics 1237

19
suspension densimeter are outlined. Section III is mental procedures and measurements is given in Sec.
devoted to a detailed description of the experimental IV. In Sec. V a summary of performance tests carried
apparatus. A comprehensive discussion of the experi- out in evaluating the instrument is presented. This sec-

1238 Rev. Sci. Instrum., Vol. 47, No. 10, October 1976 Oenshneter 1238
tion also deals with the uncertainties in the density ported approximately midway between the gradient
measurements. Finally, the experimental data for coils. These coils are connected in series such that their
saturated liquid nitrogen and methane, along with magnetic field intensity contributions cancel at the float
comparisons with independent measurements, are re- position. (For a Helmholtz coil arrangement the mag-
ported in Sec. VI. netic fields would add at the midpoint.) Thus, the
magnetic field intensity at the float position is due
II. PRINCIPLES OF OPERATION solely to the main coil contribution. Similariy, the mag-
netic field gradients of the gradient coils add at the float
A piece of magnetic material is magnetically sus-
position.
pended in a fluid at a stable equilibrium position
For this three-coil arrangement and with a constant
through the use of a closed-loop servosystem. When
main coil current (/i,) (and subsequently the magnetic
the float (or buoy) is in free support, there are three
moment of the float does not change), the density of the
forces acting on it. In the present system the mag-
fluid (p) is related to the gradient coil current (/c) by the
netic float is more dense than the fluid in which it is
following relation:
suspended; thus, an upward magnetic force is added to
the buoyancy force to balance the downward gravita- p = A + BIc (5)
tional force.
where A and B are constants to be determined by
In the present work a three-coil (sur-core) system^ is
calibration with fluids of known density. This relation-
used to supply the magnetic force. The three coils
ship is valid whether the float is magnetically hard or
are aligned coaxially.Each coil is symmetric about an
soft. Knowledge of the mass and volume of the float is
axis in common with the cylindrical axis of the mag-
not required to obtain relative density results using Eq.
netic buoy. The three coils consist of a main coil,
(5).However, since the magnetic moment of the float
which supplies the major part of the field necessary
istemperature dependent, the constants A and B would
to lift the float, and a pair of gradient coils. The
have to be determined at each temperature of interest.
horizontal position of the float, which is in the shape
At low temperatures at least two fluids of known
of a right circular cylinder (length/diameter > 1)
density, one of which may be "vacuum," are needed
magnetized along its cylindrical axis, is maintained by
for calibrating the instrument for use in performing rela-
the axially symmetrical, diverging field of the main coil.
tive density measurements. However, the densities of
The magnctx force on the ferromagnetic buoy is given
low temperature fluids are not known to sufficient
by
accuracy to permit them to be used as calibration

Fmag
' =M^
aZ
' (1)
fluids and, in fact, this had been one of the major
motivations for developing aii independent technique for
density measurements on cryogenic fluids.
where M is the total magnetic moment of the buoy In the previous work by Haynes and Stewart.' a
and dH/dZ is the variation of the external field
new method was demonstrated for determining absolute
intensity along the common cylindrical axis of the coils. densities using the same three-coil arrangement, but
For the present system the float is a permanent which is dependent upon the magnetic properties of
magnet for which the total magnetic moment (excluding the buoy material. If one uses a permanent magnetic
temperature dependence) can be represented by material for the float and its induced moment is zero

M = .Wn + MiiH). (2) or. at least, very small compared to the permanent
moment, then the magnetic force on the float is given
where Mn is the permanent moment, and M,{H) is the by the relation.
induced moment resulting from an applied field.
At a fixed position along the cylindrical axis of an air- (6)

core solenoid. where C

and D' are constants depending on the rela-
H = kl (3) tive position of the coilsand float, the magnetic moment
of the float, and the dimensions and number of turns
and of the support coils. One should note that for Eq. (6)
dH^ the main coil current (/.„) is not held constant. Now
(4) the density of the fluid is given by the expression.
dZ
p = (m - C/.V - DIgW, (7)
where / is the solenoid current and k and k' are
constants dependent upon the number of turns and where m and V are the mass and volume of the float.

dimensions of the coil. The acceleration of gravity has been included in the
As mentioned earlier, the float is supported at a posi- constants C
and D. (The mass and volume of the float
tion relative to the main coil such that its horizontal must be determined by independent measurements.)
position is fixed. This results in the float being sup- The validity of this equation can be tested experi-
ported at a distance below the main coil slightly larger mentally. For measurements in a vacuum, Eq. (7) re-
than the inside radius of this coil. The float is sup- duces to the following relation:

1239 Rev. Sci. Instrum., Vol. 47, No. 10, Octotwr 1976 Denalcneter 1239

Zl
 .

m = Cht + Die- (8) Table I. Parameters of support coils.

It should be emphasized that the critical assumption Main coil Gradient coils
in deriving Eq. no induced moment
(7) is that there is
Length 6.0 cm 6.2 cm
in the buoy resulting from changes in the magnetic
Inside diameter 7.4 cm 25.4 cm
field intensity over the range of magnetic fields needed Outside diameter 20.24 cm 38.6 cm
for the density measurements. (For the present Number of turns 5000 5000

measurements this includes fields between 0.006 and

0.016 T.)
The current in the gradient coils has been measured can be maintained; (c) the magnetic buoy and the liquid
as a function of the current in the main coil while level are visible within a windowed cell capable of
the floatwas suspended at a constant height in a given withstanding working pressures up to 5 MPa; (d) the
fluid, vacuum, at constant temperature. The
usually volume occupied by the liquid sample within the cell
currents were fitted by the method of least squares to is sufficiently large to render the unavoidable vapor
either Eq. (6) or Eq. (8). The quality of this fit volume of the access tubing, etc.. insignificant within
indicated the validity of the assumption concerning the accuracy goals of experiments involving liquid
the dependence of the magnetic moment upon the mixtures; (e) temperature gradients along the length
magnetic field. (Results will be presented in a later of the working space of the cell are monitored and
section.) For the present measurements the main coil reduced to a practical minimum, i.e., less than a total
current, and thus the magnetic field, could be varied of 10 mK; (f the rate of refrigeration of the equilibrium
)

over a range amounting to 25% of its maximum value. cell is adjustable and continuous; and (g) the tempera-

The minimum ratio of Istllc was determined by the ture of the ceil can be controlled at any temperature
loss of horizontal stability of the float. between 90 and 300 K.
From the above discussion it should be apparent A concrete block structure provides a stable sup-
that relative density measurements using Eq. (5) port for the cryostat, the coils, and the microscope.
Us, constant) can be carried out to complement the The cryostat assembly is suspended from an aluminum
results of absolute measurements using Eq. (7) with the plate that spans the top of the concrete structure.
same three-coil arrangement. In performing measure- The stainless steel central support tube of the cryo-
ments using Eq. (7) for which the main coil current stat is attached to this aluminum plate with a brass
is varied over a considerable range, it is experi- collar (not shown in Fig. 1). The lower end of the
mentally difficult to fix or determine the position of stainless steel tube is soldered to a copper tube to
the main coil from vacuum to liquid measurements which a heat exchanger is soldered.
within the desired precision. This problem does not Refrigeration is provided by continuous transfer of
arise when making relative density measurements for liquid nitrogenfrom the cryostat reservoir through the
which the main coil current is held constant. In heat exchanger. Regulation of the refrigeration rate is
principle, both procedures should give identical density accomplished by controlling the nitrogen vent rate to the
ratios for the same fluids at constant temperature. atmosphere..
measurements can be
In practice, the relative density Refrigeration is supplied to the bottom of the
carried outdeduce the systematic error in the
to equilibrium cell through the copper radiation shield and
absolute density measurements that depends on the to the top of the cell through two copper braided
magnitude of the main coil current. These tests have straps. The radiation shield, which contains a slit
been performed in the present work and will be dis- aperture in a position corresponding to the cell window,
cussed in a later section. is attached to a copper plate at the top of the shield
with Wood's metal to facilitate removal of the shield
III. APPARATUS and alignment of the shield within the glass tail section.
The glass tail is connected to the stainless steel vacuum
A. General considerations and cryostat jacket through a stainless steel-to-glass transition joint.

An assembly drawing (approximately to scale) of the

B. Equilibrium cell
major components of the apparatus is shown in Fig. 1
The general dimensions and configuration of the cryostat The equilibrium cell, made of electrolytic tough
and equilibrium cell were determined by the dimensions pitch copper, has an overall length of approximately
and configuration of the main coil and the two gradient 30 cm, including the copper-plated bellows at the bot-
coils. (See Table I for coil dimensions.) tom. The cell closure is a compression fitting with a
Criteria considered essential, and which have been silver-plated, solid copper 0-ring. The closure plug,
satisfied, in the design of the apparatus are as follows: threaded backing ring, and compression screws are
(a) all materials of construction in close proximity to stainless steel. The top of the cell is held in align-
the magnetic suspension assembly are nonmagnetic; (b) ment by a stainless steel tube, brazed to the which is
the entire assembly is rigidly supported so that the rela- closure plug extends through the uppermost
and
tive position of the buoy, microscope, and support coils cryostat plate. The upper end of the cell support tube

1240 Rev. Sci. Inttnim., Vol. 47, No. 10, October 1976 Densimeter 1240
is threaded through a nut and bearing assembly to al- along its cylindrical axis. Its length and diameter are
low small adjustments in the vertical position of the cell approximately 0.64 and 0.51 cm, respectively.
and thus the vertical position of the magnetic buoy. Barium ferrite is a magnetically hard, ceramic ma-
The bottom of the cell is thermally anchored to the terial with a density of approximately 5 x 10^ kg/m^. It
radiation shield by mechanical attachment of a flanged is somewhat porous and has a high electrical resistivity.

bellows section. A thin coat of commercially avail- The buoy used in the present work has been perma-
able conducting grease is used to enhance thermal nently magnetized in a saturation field of T. After
1

contact between the bellows flange and the shield. initial cycling over maximum ranges of temperature

Two thermal links of copper braid are also mechanically (90- 300 K) and magnetic field intensity (0-1.3 x 10^
attached to the top of the cell from the top plate of A/m) the barium ferrite magnet exhibited no hysteresis
the radiation shield using thin coats of conducting over the ranges of currents and temperatures needed
grease between the contact surfaces. in the density measurements. The magnet is operating
The outside diameter of the main part of the cell is in a very small segment (5-13 x l(P A/m) of a broad
4.13 cm and that of the closure section is 7.60 cm. hysteresis loop. The slope of the B-vs-H curve is small
The internal working space is 1.99 cm in diameter and constant. The change in the magnetic moment of the
and approximately 20 cm long. More than half of the float with temperature varies from 0.2%/K at 300 K to
internal volume, however, is occupied by the mag- 0. 1%/K at 100 K. as determined from the present meas-

netic buoy- sensor coils assembly (~10 cm^) and the urements in vacuum. This variation in the magnetic
vapor bulb (~35 cm^). The vapor bulb is a slip fit in the moment with temperature results in a random error in
top section of the cell and is attached directly to the the density of approximately 0.002% at 100 K, increasing
closure plug. The internal free volume of the cell is to 0.004% at 300 K. This estimate is based on a
approximately 20.5 cm^. Anticipating measurements on reproducibility in the temperature of 2 mK from vacuum
mixtures, the volume (—0.3 cm^) in the annulus be- to liquid measurements.
tween the vapor bulb and the cell wall has been made as Since the barium ferrite magnet is porous, it must
small as possible, since this is normally part of the be coated with a material which does not allow fluid to
vapor volume. penetrate into the buoy. Copper was found suitable
The electrical lead-throughs for the sensor coils con- for this purpose. After a piece of barium ferrite
sist of three No. 32 coated copper wires encapsulated was ground into the shape of a right circular cylinder
with an epoxy adhesive in a copper capillary tube. >^ith chamfered edges (0.25 mm deep), a conducting
This tube extends through the closure plug to the bot- layer of copper was uniformly plated onto the ceramic
tom of the vapor bulb. The epoxy adhesive, an magnet by chemical reduction. On top of this conduct-
alumina-filled resin with elevated temperature curing ing layer, copper was electroplated to a thickness of
agent, was selected based on an earlier study^ of approximately 0.25-0.50 mm: a much thicker coating
low temperature properties of epoxy adhesives. By resulted at the edges. Then most of the copper was
pressurizing with helium gas while the adhesive was removed with a diamond tool until the Copper-
hot. the entire length of the capillary tube was filled with plated barium ferrite magnet was in the shape of a right
adhesive to minimize the available vapor volume in the circular cylinder without chamfered edges. The final
equilibrium cell. buoy had a minimum of 0.06 mm of copper over the
The cell window was designedto allow viewing the cylindrical surface and the end faces and at least 0.25 mm
liquid sample from the base of the vapor bulb down on the edges where the magnet had been chamfered.
to 2 cm from the bottom of the liquid space. The The edges were up to a thicker coating since
built
viewing slit is 0.6 cm wide x 7.5 cm long. The window the plating strengthwould be weakest at these
consists of a piece of tempered Pyrex"* glass of 0.32-cm locations. (For density measurements it is not neces-
thickness with semicircular ends and chamfered edges. sary that the plating be symmetric about the center of
The seed between the glass and the cell is made with mass of the buoy.) Gold was flashed (10"^ mm) over
indium wire compressed in a racetrack groove ma- the entire copper surface as a protective coating.
chined in a flat surface on the cell. The glass is
compressed against the indium seal with a hardened D. Electromagnetic support coiis
beryllium copper plate secured by stainless steel screws.
Each of the coils is composed of two separate
A rubber asbestos pad is placed between the beryl-
coils of 2500 turns of epoxy-coated aluminum foil
lium copper plate and the glass to relieve thermal and
of approximately 0.025-mm thickness. The foil for the
mechanical strains. The maximum working pressure of
main coil has a width of 2.5 cm while that for the
the equilibrium cell is limited by the maximum working
gradient coils is 3.0 cm. Selected coil parameters
pressure of the window.
are given in Table I.

Each of the gradient coils is rigidly supported by three

C. Magnetic buoy
aluminum rods resting on the concrete structure. The
The buoy is a barium ferrite (BaFei20,9) magnet in gradient coils have a separation distance of approxi-
the shape of a right circular cylinder magnetized mately 23 cm. No effort has been made to control the

1241 R«v. Sci. Inttrum., Vol. 47, No. 10, October 1976 Dontimator 1241

Z6
 12V 12V 12V 12V

J MAIN
J COIL

POWER SUPPLY
( 100 VOLTS)

1.35V KELVIN
Hg CELL VARLEY
(TEMP i DIVIDER
LOAD
(CURRENT
STABIL-
CONTROL)
IZED)

-vw
DIFFERENTIAL

VOLTMETER

^THERMAL LAGGING
ENCLOSURE

Fjg. 2. Current control circuit for main coil.

position of the gradient coils except through the con- functions as follows: The main coil current flows through
trol of the laboratory temperature. The dH/dZ of the a stable resistor Rjtd (01 i^' in the emitter of Qo (the
gradient coils changes very slowly (2.5 x IO~''/cm) current-controlling transistor). The voltage developed
from a maximum at the float position. Since the change across Rstd 'S compared with a reference voltage from
in dH jdZ with position for the main coil is two orders of a Kelvin-Varley voltage divider by amplifier A,. The
magnitude larger than that for the gradient coils, output of A, drives the base of Qj and controls the cur-
special care has been taken to minimize changes in rent through Rstd and the main coil. Since the loop
the position of the main coil. gain of this circuit is exceedingly high (>I0*) and the
First, the main coil is supported at its midpiane so that thermal drifts are very low, the long-term stability of
thermal expansion effects inside the coil are minimized. the main coil current is of the order of 0.0019?^. This
A circular copper plate has been soldered to the brass current controller has a temperature coefficient of the
tube on which the main coil is wound. A coil of aluminum order of 5 ppm/K and a sensitivity to power sup-
foil (2500 turns) is clamped tightly to each side of the ply variations of less than 2 x 10~® A/V. A slow drift of
copper plate with a thin layer of vacuum grease be- the coil current of 2.5 x 10~* A/'h has negligible ef-
tween the aluminum foil and copper for maximum fect on the density measurements since the main coil
thermal contact. The copper plate (and thus the coil) current is measured each time that the gradient coil

is supported by quartz rods extending to the concrete current is determined. (It should be noted that the
structure. Thus, any changes in the position of the mid- precision and accuracy of the density results depend on
plane of the coil due to room temperature changes the short-term stability and not the absolute accuracy of
would be negligible. Copper plates with cooling water the standard resistor.)
have been placed on the outside faces of each sec- The gradient coils are included in a control circuit
tion of the main coil such that the average temper- (Figs. 3 and 4) that senses and maintains the vertical
ature of the main coil is below room temperature position of the buoy. The requirements for stability
for maximum current (1 A). However, for absolute of a magnetic support or levitation system are dis-
density measurements, the temperature of the main coil cussed in detail in Refs. 5 and 6 and will not be repeated
changes from vacuum to liquid measurements (1.2 K here. Suffice it to say that the characteristic equation
for nitrogen at 100 K) since a larger current is required system has a root with a positive
for a magnetic levitation
to support the float in vacuum than in the liquid for real and for system stability the plot of the
part,
a given gradient coil current. frequency response of the characteristic equation (the
Nyquist plot) must encircle the origin once in the
E. Support system electronics counterclockwise sense. This is accomplished by the
duaJ-lead network composed of R,, R,, C,, and Cj in
The current in the main coil, which contributes the Fig. 4. A dual-lead network is required for this system
major portion of the magnetic force necessary to lift the since minimization of the steady state offset or buoy
buoy, is provided by the circuit of Fig. 2. The circuit position change requires at least a first-order integrator

1242 R*v. Sci. Instrum., Vol. 47, No. 10, October 1976 0«nBlinet«r 1242
in the control loop. This integrator consists of amplifier in Fig. 5 to complement the detailed circuit diagrams in

A, and R, and C,. The additional resistor R3 is re- Figs. 2-4. The circuit diagrams are presented in

quired to stabilize the combined integrator and dual-lead detail since this circuit is considerably different from
network. A block diagram of the electronics is presented any found in the literature. It should be noted that the

12V i20pf
HC1550G ^-^^^
+12V

PHASE SHIFTER

TO THE CURRENT
CONTROL FOR THE
GRADIENT COILS

COMPENSATION CIRCUIT

Fig. 4. Position sensor electronics and compensation circuit.

1243 Rev. Scl. Inttrum., Vol. 47, No. 10, Octob«r 1976 Densimeter 1243
current control amplifier for the main-coil constant- Ihisiriii

current source is the same as that for the gradient CIrciits

coils.
r-WJ— VtAifi Ctitrillii
frtmi
«Mtl|l
Perhaps the most interesting part of the support ttniit Snrti
Cntril
system is the position sensor. In the past, position
sensing has been accomplished by either optical'-*
(ligjit beam-photomuitiplier tube arrangement) or 3, Hciliitir

electronic' *"'^ (inductance pickup coil or differential

transformer) means. The requirements of small size, low ifl Friiiiicf litiiriiir t
power dissipation, high sensitivity, the ability to with- Trusliriir ClBMKItlM
stand high pressure and low temperatures, and excep- Circaits

tional dimensional integrity resulted in the use of a linear

differential transformer (LDT) wound on a machinable
glass ceramic form. The LDT consists of a pair of coils
tiltiii Ciitrillil
equally spaced about a transverse rectangular opening Citfiit SMrti
in the ceramic form inside which the buoy
suspended. is

Each coil is wound with 34 turns of double silk-

Crciits
wrapped 50/44 Litz wire in circumferential grooves of
2.8-mm width and 6. 1-mm depth. The outside diameter Fig. 5. Block diagram of a magnetic suspension densimeter:
of the coils and the ceramic holder is 17.3 mm. The
G,, G;— gradient coils; M — mam coil; S,. Sj — sensing coils; B —
magnetic buoy.
complete transformer assembly has been potted with a
low-viscosity epoxy resin to prevent intrusion of the
pressurized fluid into the windings. has been checked against vapor pressure measurements
The sensitivity of the LDT as measured at the phase- on liquid methane and nitrogen.
sensitive detector (synchronous detector) is of the order The uncertainty of the calibration is approximately
of 1 V/mm. The noise voltage at the sensor amplifier 0.002 K. Due to the specifications of the potentiometric
input is less than 10~* V/Hz"^. The operational measuring system, uncertainties temperature in the
amplifiers in the compensation circuit have very low. off- amount to a maximum of O.OiO K at
K, increasing to 1 10

set drifts and are thermally- lagged to a large copper 0.030 K at 300 K. The temperature of the sample
block in the amplifier chassis. Consequently, the vertical holder is controlled to better than 0.005 K, approxi-
movement of the buoy due to operational amplifier mately the same as the reproducibility of the temper-
drifts and rf amplifier noise is less than 10"' mm. ature measurements.
The LDT is the major component of a ratio-trans- A current of mA for the thermometer is supplied
1

former-type bridge. Pan of the bridge is located at the by an electronic constant current source. The voltage
top of the cryostat and is connected to the LDT through drop across a 100 H calibrated standard resistor in series
long coaxial lines. The bridge is constructed so as to re- with the thermometer is monitored continuously with a
move proximity effects of the outer conductors of the six-digit differential voltmeter. With this setup the un-

coaxial lines. The neutral position of the buoy is con- certainty and repeatability of the current is approx-
trolled by a small variable capacitor between the center imately 0.002^. The voltmeter is checked periodically
lead and one side of the LDT. Conductance balance against a microvolt potentiometer using a calibrated
is achieved with a resistor connected across the capac- standard cell.

itor. Unbalance in the LDT due to the inductance The total uncertainty in the reported temperatures in

of the long leads is minimized by tuning the two halves Sec. VI is believed to be less than 0.03 K. (At room
of the LDT to resonance with matched capacitors temperature it would be less than 0.04 K.) The 0.03-K
(as large as possible for'the 0.5-MHz exciting signal). total uncenainty consists of approximately 0.025-K.

These capacitors are connected near the LDT outside systematic error and 0.005-K random error. The syste-
the sample cell but inside the vacuum space. matic error was determined from the uncenainties in
the calibration of the thermometer and in the potentio-
metric measuring system.
IV. EXPERIMENTAL MEASUREMENTS AND
PROCEDURES The temperature of the cell is regulated by balancing
coarsely variable cooling with precisely controlled
A. Temperature and pressure electric heating. A control heater near the bottom of the
cell is connected to a dc power regulator, which is part
The techniques and instrumentation for the measure- of a measuring/regulation system that also includes a
ment and control of the temperature and pressure are six-dial potentiometer and a microvolt
microvolt
standard. The primary temperature sensor, calibrated amplifier. A
second heater at the top of the cell is con-
on the IPTS 1968, is a platinum resistance thermometer nected to a manual power supply. Independent heaters
secured in a well at the base of the equilibrium cell at the top and bottom of the cell are used to minimize
with Wood's metal. The calibration of the thermometer temperature gradients along the length of the cell.

1244 Rev. Scl. Inttrum.. Vol. 47, No. 10, October 1976 Denaimeter 1244

2b
Temperature gradients are detected by monitoring the float observed under two experimental conditions,
is

vapor pressure of the liquid in the vapor bulb. i.e., first immersed in gas and then im-
with the float
Pressures below 7 bars are measured with a 0- to mersed in liquid. Both methane and nitrogen were used
7-bar spiral quartz Bourdon gauge. This gauge has a to carry out this procedure. During this procedure
resolution of less than 3.5 x iO""* bar. It has been cali- the microscope tilt is adjusted so that the "apparent"
brated against an air dead-weight gauge giving an uncer- position of the float does not depend on the refractive
tainty of approximately 0.1% over the full range. For index of the fluid inside the cell within the resolution
pressures above 7 bars, a double-revolution 0- to 20-bar of the microscope. The temperature of the cell is
Bourdon gauge is used. It has also been calibrated controlled to better than 0.01 K during these observa-
against the air dead-weight gauge and can be read to the tions. Care is taken to ascertain that the float is ob-
nearest 0.005 bar. served through the same parts of the windows as during
Vapor pressure measurements have been used in the the density measurements. Necessarily a different part
present work primarily as a means of monitoring of the float is observed. This presents no special problem
temperature gradients. They have also been used as a since there are diamond tool marks the entire length of
check against the calibration of the platinum resistance the float that can be used as reference lines. This
thermometer. In general, the vapor pressures for procedure has been repeated frequently during the
methane have been consistently lower than those of course of the reported measurements. It has been found
Prydz and Goodwin.'^ This pressure difference cor- that a readjustment of the microscope tilt is necessary
responds to a temperature difference of 0.01-0.02 K. only if the apparatus is perturbed by disassembly.
The discrepancies between the nitrogen vapor pressures
of the present work and those of Wagner''' correspond C. Volume of buoy
to an average temperature difference of less than 0.015
K. with no particular trend observed. Absolute density measurements [Eq. (7)] with a mag-
netic suspension densimeter require a determination of

B. Position of buoy the volume of the magnetic buoy. In the past this
has been accomplished either by using distilled water as
The accuracy and precision of the density measure- a reference fluid of known
density'^ or by making direct
ments depend on being able to suspend the float at the length and diameter measurements on a uniformly con-
same position relative to the support coils in vacuum structed float.' Here the volume has been determined
and in the test liquid at the same temperature. A by the first method.
125 X filar micrometer microscope is used to determine The volume (0.13485 cm' at 300 K) of the magnetic
the position of the float. It has a resolution of ap- buoy was determined within 0.02%. The distilled water
proximately 5 X 10"" mm. The maximum error in the was vacuum distilled to remove air. Any problems with
position determination is 2 x 10~^ mm. which cor- bubbles inside the cell were rectified by pressurizing the
responds to an error in density of less than 0.03%. water with helium gas. The volume of the buoy has
This error includes approximately equal contributions been measured three times at 300 K and once at 290 K.
from the alignment of the microscope, so that the "ap- These four measurements gave a standard deviation of
parent" position of the float is independent of the
' 0.005% using thermal expansion data for barium fer-
index of refraction of the fluid inside the sample ceil, rite to calculate the volume change from 290 to 300 K.

and from the repositioning of the float for vacuum and This resulted in a 99% confidence interval of rO.015%.
liquid measurements. The first contribution would be a The systematic error in the volume determination de-
systematic error while the second is random. The preci- pends on the uncertainty in the density of water.
sion of the density measurements depends primarily on This should be less than 0.005%. '«
the reproducibility of the position of the buoy from The present paper repons density data for cryogenic
vacuum to liquid measurements. Thermal expansion data for barium ferrite were
fluids.
As mentioned above, the microscope must be aligned used to calculate the volume of the float at iou
so that the "apparent" position of the float does net temperatures. Recently the linear thermal expansion
depend on the index of refraction of the fluid inside the of polycrystalline barium ferrite was measured at this
sample cell. In viewing the float inside the cell one must laboratory at temperatures from 76 to 293 K with a
look through the glass tail of the cryostat and the quartz tube dilatometer." The volume of the float at 100
sample holder window. The "apparent" position of the K is approximately 0.4% less than the room temper-
float would be dependent on the refractive index of the ature value.
fluid inside the cell unless the light rays are perpendicular Barium ferrite is somewhat anisotropic and its thermal
to the interface between the test liquid and the inside expansion was measured both parallel and perpendicular
surface of the sample holder window. to the magnetization direction. An anisotropy of 15-20%
The procedure for adjusting or aligning the micro- was observed. The estimated overall uncertainty of the
scope is as follows. First, the float is positioned at thermal expansion measurements would correspond to
rest on the ceramic holder at the same horizontal an error of 6% in the adjustment to the volume.
position as when in suppon. Then a fiducial mark on the This would produce an uncertainty of approximately

1245 R«v. Sci. Instrum., Vol. 47. No. 10, Octobvr 1976 Oonalmotar 1245

<Ll
0.02% in the volume at 120 K. The uncertainty in random and well approximated by a normal probability
the adjustment to the volume due to the contraction of distribution. These results demonstrated that within the
the copper plating is negligible. precision of the current measurements the magnetic
The on the volume of the float
effect of pressure moment of the barium ferrite float is independent of the
is encountered in the
negligible at the highest pressures magnetic field intensity over the range of fields (0.006-
present work. The bulk modulus for barium ferrite 0.016 T) considered in the present work.
should be less than that for copper, which is 1.35 Other coil configurations have been tried that gave re-
X 10" Pa.'* This value would correspond to a volume sults that substantiated the observations given above.
correction of approximately 0.001% at 2.5 MPa, the First the bottom gradient coil was disconnected re-
vapor pressure of nitrogen at 120 K. sulting in a two-coil system. The values of the constant
(O in Eq. (7) for the two- and three-coil arrange-
D. Mass of buoy ments under otherwise identical experimental conditions
were in agreement within the experimental uncertainty.
The mass of the magnetic buoy was determined with
In other words, the magnetic moment of the float (con-
a 10-g capacity equal-arm microbaiance. A calibrated
tained in C) remains constant for significantly differ-
class M
weight set was used in the weighings. Pre-
ent magnetic field intensities.
cautions were taken to assure that the copper plating
Next the pair of approximately identical (gradient)
and gold protective coating provided an impervious
coils were connected to the constant current source in a
The buoy was rapidly
barrier to liquids under pressure.
Helmholtz arrangement. The main coil was connected
cycled between room temperature and 76 K. It was then
into the servosystem. The float was first supported at
immersed in liquid ethanol at 5-atm pressure and room
a constant height at fixed temperature by the force
temperature for 4 h. The buoy was weighed twice after
supplied by only the main coil. The float was located
removal from the ethanol bath. The average of these
at the midpoint of the Helmholtz coils while the cur-
weighings was 9 ftg higher than the mean of the initial
rent in the Helmholtz coils was increased to 0.6 A. There
four weighings. A significantly larger increase in the
was an insignificant change in the main coil current
mass was expected if the plating on the float contained a
(lO"* A) for this relatively enormous change in the
pinhole that allowed ethanol to penetrate into the porous
magnetic field intensity (about an order of magnitude
ceramic. Thus, at room temperature, the metallic
more than for the typical data point). The results of these
coating appeared to be impervious after thermal cycling.
tests demonstrated that this technique could be used for
The reproducibility of liquid methane density measure-
absolute density measurements.
ments (see experimental results) after many cycles be-
In carrying out density measurements with this tech-
tween room temperature and 100- 140 K is further
nique, the position of the main coil must remain fixed
evidence of the resistance of the metallic coatings to
over a wide range of currents for the vacuum and liquid
damage from thermal cycling.
measurements. This means that the "constant" k' in
The final mass of the magnetic buoy (0.73706 g)
Eq. (4) and the constants. C and D, in Eq. (7) must
was taken as the mean of all weighings, including the
be independent of the current in the coil. It was hoped
weighings after immersion in ethanol. corrected for air
that themeasures exercised in controHing the tempera-
buoyancy. A 99% confidence interval for the six weigh-
ture and position of the main coil (see Sec. Ill D)
ings, based on a standard deviation of 6 ^g. was ± 10
would result in such behavior. It was experimentally
fig. The systematic error in the mass determination,
impracticable to determine if k' was a constant by
resulting primarily from the weight set cahbration, is
monitoring the position of the coil. However, the effect
believed to be less than 5 /ig.
of a change in the current in the main coil on the
density measurements could be obtained in another man-
V.EVALUATION OF PERFORMANCE OF ner. As shown in Sec. II the fluid density is linearly
MAGNETIC SUSPENSION DENSIMETER proportional to the current in the gradient coils if the
main coil current is held constant. Not only is the posi-
A. Performance tests
tion of the main coil fixed, but also the magnetic field
The use of Eq. (7) for obtaining absolute densities intensity at the midpoint of the gradient coils (float
has again been verified experimentally'. The gradient position) is constant. Relative measurements [Eq. (5)]
coil current has been measured as a function of the carried out vacuum, nitrogen, methane, and
with
main coil current over a range of approximately 20% propane at 110 K were not consistent with the density
of the maximum main coil current. These measurements ratios computed from absolute measurements [Eq. (7)]
were carried out while supporting the float in a given for the same fluids. These results showed that the
fluid (includingvacuum) at constant height at a fixed "constants" {k'. C, and D) vary slightly with the main
temperature. The various pairs of and Iq were coil current. This variation produces a density error
fitted by the method of least squares to Eq. (7). The relative to the difference between main coil currents
residual standard deviation, with the gradient coil cur- for vacuum and liquid measurements (constant Ig) of
rent as the dependent variable, was less than 10"' A. approximately 1.4%/A. This corresponds to a density
Examination of the residuals showed that they were adjustment of 0. l-0.2%/(g/cm^) to the results from Eq.

1246 Rev. Sci. Instrum., Vol. 47, No. 10, October 1976 Densimeter 1246

Z8
 1

Table 11. Estimated uncertainties in density measurements.

Percent error in density

Metliane Nitrogen Normal butane

Systematic errors 105 K 160 K 95 K 120 K 135 K 300 K

Mass ot tioat (oec. iv u) ±U.UU^

Volume of float at 300 K (Sec. IV C) ±0.02 ±0.02 ±0.02 ±0.02 ±0.02 ±0.02
Thermaiexpansioncoefficientofbariumferrite(Sec. IV C) ±0.022 ±0.017 ±0.023 ±0.021 ±0.020 ±0.001
Position of float (Sec. IV B) ±0.015 ±0.018 ±0.010 ±0.012 ±0.009 ±0.009
Position of main coil, determined from relative
measurements. Eq. (5) (Sec. V A) -0.02 ±0.02 ±0.02 ±0.02 ±0.02 ±0.02
Temperature of sample fluid (Sec. IV A) ±0.010 ±0.021 ±0.023 ±0.076 ±0.004 ±0.008

Three times standard deviation ±0.045 ±0.054 ±0.030 ±0.036 ±0.027 ±0.027

Total uncenainty" ±0.09 ±0.10 ±0.07 ±0.12 ±0.06 ±0.06

" The total uncertainty was determined from the square root of the sum of the squares of the systematic errors added to three times the
standard deviation.

(7). Since all measurements are based on the volume de- relativemeasurements [Eq. (5)]. The density difference
termination using distilled water as a reference fluid of between the relative and absolute measurements, ex-
known density, the adjustment must be applied relative pressed in terms of the difference in main coil currents
to this point. between the vacuum and liquid points il, - If) for a
A further check on this density-dependent adjust- constant gradient coil current, is independent of temper-
ment was accomplished by measuring the densities of ature. The temperature dependence of the magnetic
two other fluids of known density at room temperature. moment of the float must be included if the difference
The densities of liquid samples of normal pentane is expressed in terms of density.
and ethanol were determined by the Mass and Volume The relative measurements were carried out using the
Section of the Mechanics Division of the National same experimental setup and parameters (mass, volume,
Bureau of Standards in Washington, DC, to an estimated and position) as the absolute measurements. Thus the
inaccuracy of less than 0.01%. Measurements on these total uncertainty in the adjustment determination should
fluids at 300 K
with the present apparatus using Eq. only include the random error in the measurement
(7), combined with the adjustment based on the relative process. Thus an estimated uncertainty of ±0.02% in the
density measurements at low temperature, gave adjustment is based on the comparisons with the refer-
densities that agreed with the reference values to better ence fluids since this density difference was larger than a
than 0.02%. Thus, the densities presented in this paper 99% confidence interval from a fit of the differences
based on absolute measurements [Eq. (7)] have been between relative and absolute measurements.
adjusted slightly to be consistent with the results of

B. Error analysis
Table Experimental results
III. for saturated liquid nitrogen
(molecular weight = 28.0134).
Evaluation of the experimental parameters involved in

the present work showed that the uncertainty in the

T lOCKpexp - ptalc)
(K) (mol/n (mol/l) density measurements depends primarily on the un-
certainties in the determination of the volume of the
95.000 25.6731 25.6755 -0.009
float, the relative position of the float and the main
100,000 24.6400 24.6391 0.004
105.000 23.4998 23.4988 0.005 coil, and the temperature of the sample fluid. The
110.000 22.2079 22.2071 0.004 precision of the density measurements is inversely
115.000 20.6719 20.6748 -0.014
proportional to the difference between the density of the
120.000 18.6818 18.6856 -0.020
fluid and the density of the float. This variation with
100.000 24.6361 24.6391 -0.012
density cannot be observed for the measurements re-
105.000 23.4%3 23.4988 -0.01
110.000 22.2044 22.2071 -0.012 ported in this paper because of the density range covered.
115.000 20.6787 20.6748 0.019 The precision also varies with temperature, correspond-
120.010 18.6800 18.6808 -0.004
ing to the change in the magnetic moment of the float
95.075 25.6624 25.6605 0.007 with temperature.
100.075 24.6207 24.6229 -0.009
Table II summarizes the uncertainties in the density
100.075 24.6279 24.6229 0.020
105.075 23.4841 23.4807 0.015 measurements for the nitrogen and methane results
110.075 22.1840 22.1862 -0.010 presented in this paper. Normal butane is also included
115.075 20.6490 20.6492 -0.001
in this table, primarily to show the performance of the
117.575 19.7356 19.7326 0.015
120.075 18.6525 18.6497 0.015 instrument at higher temperatures. Normal butane
densities, determined with this instrument at tempera-

1247 Rev. Scl. Instrum., Vol. 47, No. 10, October 1976 Densimeter 1247

29

I
tures between 135 and 300 K, will be presented in a Table IV. Parameters of Eq. (9) for nitrogen.

forthcoming paper.
a = 19.39217=' (0.11)" b3 = 23.32977 (4.8) Pr = 11.21 mol/l
The standard deviation given in the table for methane b, = 26.01408 (2.5) 7", = 126.20 K a = 0.016%
at 105 K was based on twelve measurements. The t,„ = -39.497.S9 (6.5)

other standard deviations were computed relative to the

°These coefficients (a. 6,. fco, i>3) and those in Table VI were obtained
density and temperature of this point. Standard devia-
from a least-squares program in which the experimental densities in
tions at the other temperatures and densities could not g/cm' to five digits were convened to molar densities within the
be estimated from statistical analysis since sufficient program.
" Standard errors of coefficients rn parentheses.
repetitivemeasurements were not taken for these points.
Measurements on more dense fluids, such as argon and
krypton, have demonstrated the expected variation of
pc and Tr are selected values of the critical density
the precision with density. Measurements near room and temperature and a. b, are coefficients determined by
temperature, especially those on distilled water, have least squares. Equation (9) consists of a scaling law
exhibited the expected improvement in the precision as modification'* to a generalized Guggenheim equation.^"
the temperature is increased. The parameters of Eq. (9) for nitrogen, along with the
The total uncertainty of a single density measure- residual standard deviation (cr) of the fit and the standard
ment was taken as the square root of the sum of the errors of the coefficients, are given in Table IV. The
squares of the systematic errors plus an allowance of standard errors of the calculated densities in Table III
three times the standard deviation for random error. are not presented since they are significantly smaller
than the standard deviation. The main purpose of the
Vi. EXPERIMENTAL RESULTS above expression is to facilitate comparisons with in-
dependent measurements. The choice of the critical
A. Nitrogen
point parameters is arbitrary in the sense that the critical

The experimental saturated liquid densities for

point density can be varied by as much as 5% and the
critical point temperature by 0.2% without changing the
nitrogen are presented as a function of temperature
fit of the data within the estimated precision of the
(IPTS 1968) in Table III. The explanation for some of
present measurements. For nitrogen the critical point
the nitrogen points not being at integral temperatures
involves the presence of a temperature gradient along
values were obtained from the PVT data compilation
of Jacobscn^'.
the length of the cell in some of the initial measure-
In Fig. 6 independent experimental data available
ments with this apparatus. Each of the data points was
taken from a new charge of nitrogen. For each
in the literature are compared with the expression

datum point the vacuum measurements were carried out representing the present nitrogen data over the temper-

immediately before or after the liquid measurements. ature range of the present measurements. The present
data are also shown. The experimental temperatures
The same procedure was used for methane. The three
data sets for nitrogen over the temperature range of
95- or 100-120 K were obtained at approximately
0.6p
six-month intervals. Of the two data sets at integral
temperatures, one was taken with methane, and the 0.5 -

other with nitrogen, in the vapor bulb. The third set 0.4 -
was taken with no liquid in the vapor bulb. (For the
0 3 -
120.010-K point a slight temperature gradient was ap-
plied to remove bubbles from the sample liquid inside the 0.2 -

cell.) -
0.1
The nitrogen samples were taken from commercially
available, research grade gas. A purity specification 1 0 -
a.
of 99.99 mol% was
given by the supplier. The gas has -0.1 -

been analyzed chromatographically with a thermal con-

-0.2 -
ductivity detector and found to be within the specified
purity.The nitrogen gas was passed through a room -0.3 -

temperature molecular sieve trap, primarily for re- -0.4 -

moval of water.
-0.5 -
The saturated liquid densities (p) of nitrogen have
been fitted as a function of temperature (7") by the -0.5 —
100 105 110
method of least squares to an equation of the following TEMPERATURE. K
form:
Fig. 6. Deviation plot of experimental densities of saturated liquid
nitrogen compared with values calculated from Eq. (9) using
T \'+''-l"3
J \0.35

p- p, = a |1 - -j +
^
I
^. (l -_ ) . (9)
parameters from Table IV: O 3
present data. Brauns.^^
A Rodosevich and Miller.^' • Goldman and Scrzise.^'
Weber." C
Terry et al."
I 7-c
O Streett and Staveley," Mathias ei at}*

1248 Rev. Sci. Instrum., Vol. 47, No. 10, October 1976 Densimeter 1248

3U
 Table V. Results for saturated liquid methane (molecular weight The liquid samples were obtained from either of two
= 16.04303).
cylinders of different batches of commercial, research
grade methane. The purity for each cylinder as specified
Standard
Temperature Mean density deviation Number of by the supplier is 99.99 mol% minimum. Analyses in this
(K) (mol/l) (%) data points laboratory substantiated these specifications. As with
nitrogen the methane gas was passed through a room
105.000 26.9458 0.015 12
110.000 26.4985 0.019 11 temperature molecular sieve trap to remove moisture
115.000 26.0443 0.014 12 and any heavy contaminants not detected by analysis.
120.000 25.5721 0.016 17
125.000 25.0845 0.016 18
The 123 experimental data points have been fitted by
130.000 24.5775 0.017 16 the method of least squares (9). Only three
to Eq.
135.000 24.0540 0.016 23 coefficientswere needed to fit methane data within
the
140.000 23.5067 0.018 5
the precision of the measurements. The fourth term in
145.000 22.9312 0.014 5
150.000 22.3218 2 the equation was not statistically significant. The co-
160.000 20.9876 2 efficients of Eq. (9), the standard errors of the co-
efficients, the residual standard deviation (cr) of the fit,

and a selected critical density^** and temperature^" are

of other measurements have not been adjusted to the given in Table VI for methane. Equation (9), with the
temperature scale of the present results. coefficients for methane, has been used to compare the
The data of Brauns," Weber,^^ and Rodosevich and present results with experimental data from independent
Miller.-''each obtained with a technique different from investigators. The methane data from the present work
that used here, fall within 0.15% of the present results. are not shown on the deviation plot because of the large
The densities of Goldman and Scrase^^ and Terry er number of points. All of the experimental densities
al.-^ are consistently lower than the present results by of methane obtained in the present work will be pre-
approximately 0.3-0.4%. This pattern is also followed sented in tabular form in a future publication.
for methane by Terry et al.. and preliminary analysis For saturated liquid methane there have been con-
of recent results for argon and krypton obtained with siderably more measurements than for saturated liquid
the magnetic suspension densimeter indicate compar- nitrogen. All of the other measurements were obtained
isons with Terry et al., for these fluids exhibit the same with techniques different from that of the present work.
behavior. Some of the older, less precise data have been included
in Fig. 7 for the sake of completeness. It is satisfying
B. Methane and Olives,^'
that the relatively recent data of Orrit
McClune,^^ Goodwin and Prydz,^^ and Shana a and
Methane has been used as a control fluid through-
out the entirety of a project that has included
density measurements on six pure fluids and, at least,
thirty mixtures of these components. As a result, well l.2l-

over one hundred data points have been accumulated

1.0^
for saturated liquid methane, primarily in the temper-
0.81-
ature range 105- 150 K. Although these experimental
data are not presented in this paper, the results of an 0.6 l-

analysis of the methane data are helpfiil in evaluating j

0.4^
the performance of the magnetic suspension densimeter
0 2
used in the present work. More than ten data points
were obtained at each 5-K interval between 105 and s
°

135 K. The mean of the experimental densities, along -0.2

with the standard deviation and the number of points 6>
-0.4 |-D
at each temperature, are given in Table V. (The
standard deviations are not given at 150 and 160 K since -0.6

there only two data points at each of these

are -0.8
temperatures.) Each of the data points represents a new
methane scunple. The vapor bulb contained liquid
-1.21-L
methane for all points. no 120 130 140 ISO 160
TEMPERATURE. K

Table VI. Parameters of Eq. (9) for methane. Fig. 7. Deviation plot of experimental densities of saturated liquid
methane compared with values calculated from Eq. (9) usmg
parameters from Table VI: Terry et al..-^ Q Orrit and Olives."
a = 18.65812 (0.035)" 190.555 K A McClune.^ C Goodwin and Prydz." ® Shana a and Canfield.^
6, = 6.712030 (0.21) Pc 10.16 mol/l
T Vennix,'* ® Gngor.^" V Davenport et al.." Klosek and
bi = -0.9472020 (0.20) 0.016% McKiniev.'* © Sinor and Kurata,^" A Jensen and Kurata." A Bloomer
and Parent,-" O Fuks et al.," • Keyes et al..*' Moran." O Van
" Standard errors of coefficients in parentheses. Itterbeek ei al."

1249 Rev. Scl. Instrum., Vol. 47, No. 10, Octobsr 1976 Densimeter 1249

I
Canfield^'' agree with the present results within 0.1%. an endorsement or recommendation by the National
signify
Bureau of Standards.
However, as mentioned earlier, the densities of Terry ' J, B. Breazeale. Ph.D. thesis. University of Virginia (1955),
lower than the present results
et al.^^ are systematically ' H. S,Monon. Jr.. Ph.D. thesis. University of Virginia (1953).
' i. P. Senter. Rev. Sci. Instrum. 40. 334 (1%9).
by 0.3-0.4%. The extrapolated densities of Vennix^^ at « J. W. Beams, C. W. Hulburt, W. E, Lotz, Jr.. and R. M,

approximately 150 and 160 K are 0.4% larger than the Montague. Jr.. Rev. Sci. instrum. 26. 1181 (1955),
present results although the discrepancy at 140 K is ap- ' S, P. Almeida and T. H, Crouch, Rev, Sci. Instrum,

42. 1344 (1971).

proximately 0.01%. '° M,
G. Hodgins and J, W, Beams, Rev, Sci, Instrum,
42, 1455 (1971).
" J. L, Hales, J, Phys, E 3, 855 (1970),
C. Other fluids " S, C, Greer, M, R, Moldover. and R. Hocken. Rev„ Sci,
Instrum, 45, 1462 (1974).
To demonstrate the performance of the magnetic " R, Prydz and R, D, Goodwin. J, Chem, Thermodyn, 4, 127
suspension densimeter at low temperature, this paper (1972).
'•
W, Wagner, Cryogenics 13, 470 (1973),
presents only a representative cross section of the data
J,W. Beams and A. M, Clarke, Rev. Sci, Instrum, 33,
that have been taken in the present work. Data have 750 (l%2),
recently been obtained for other pure fluids (ethane, M, Menache and G. Girard. Metrologia 9, 62 (1973).
" A. F. Clark. W, M, Haynes, V, A. Deason. and R. J,
propane, isobutane. normal butane, argon, and krypton) Trapani. Cryogenics 16, 267 (1976),
over a density range extending to 2.3 g/cm^ (krypton) G. Simmons and H. Wang. Single Crystal Elastic Constants
and Calculated Aggregate Properties: A Handbook (MIT
and over a temperature range extending to 300 K (iso- Press. Cambridge, MA, 1971), p, 181,
butane and normal butane). These data will be published " R. D. McCarty, Natl. Bur, Stand, (U,S,) Intern, Rept, 74-357
(1974),
in the near future. " Y. C. Hou and J. J. Martin. AIChE J. 5, 125 (1959).
R. T. Jacobsen. Ph.D. thesis, Washington State University
ACKNOWLEDGMENTS ^
(1972).
P. Brauns (private communication, 1973).

The authors would like to express their appreciation " L. A. Weber. J. Chem. Thermodyn. 2, 839 (1970).
" J. B. Rodosevich and R. C. Miller. AIChE J. 19, 729 (1973).
to the following individuals who have made significant " K, Goldman and N. G. Scrase, Physica 44, 555 (1%9).
contributions to the development of this apparatus. M. J. Terry, J. T. Lynch, M. Bunclark, K. R. Mancell,
Chairman of the and L. A, K. Staveley, J, Chem, Thermodyn, 1, 413 (1969),
M. R. Cines of Phillips Petroleum Co., W. B. Streelt and L. A, K, Staveley, Adv. Cryog, Eng, 13,
LNG Density Project Steering Committee, was instru- 363 (1968),
mental in the development of support for this project. ^ E. Mathias. H, K. Onnes, and C, A, Crommelin, Commun.
Leiden 145C, 19 (1914).
W. G. Layne, D. L. Smith, and A. N DiSalvo as- " J. D. Olson, J, Chem, Phys. S3, 474 (1975).

sisted in the design and fabrication of the apparatus. R, D. Goodwin, Natl. Bur. Stand. (U.S.) Tech, Note No.
653 (1974),
J. E. Cruz designed some of the electronics associated " J. Orrit and J. Olives, distributed at 4th International Conference
with the refrigeration and temperature control systems. on Liquefied Natural Gas, Algeria 1974), (

Whetstone determined the densities of liquid ^^C. R. McClune, Cryogenics 16, 289 (1976),
J. R.
" R, D. Goodwin and R. Prvdz. J. Res. Natl. Bur. Std.
samples of normal pentane and ethanol. which were used (U.S.) 76A, 81 (1972),
as reference fluids. P. V. Tryon assisted with the " M, Y. Shana a and F. B. Canfield, Trans. Faradav Soc, 64,
2281 (1968).
statistical analysis of the data.
A, J, Vennix, Ph.D. thesis. Rice University (1%5).
A. F, Grigor, Ph.D. thesis, Pennsylvania State University (1966).
A, J, Davenport, J. S. Rowlinson, and G. Saville, Trans,
* This work was carried out at the National Bureau of Standards Faraday Soc. 62, 322 il%6).
under the sponsorship of British Gas Corp., Chicago Bridge and J. Klosek and C. McKinley, Proceedings of the First International
Iron Co.. Columbia Gas Service Corp.. Distrigas Corp., Conference on LNG, Chicago (1968), Paper 22,
Easco Gas LNG. Inc., El Paso Natural Gas. Gaz de France, " J. E. Sinor and F. Kurata, J. Chem. Eng. Data 11, 1%6), 1 (

Marathon Oil Co. Mobil R and D Corp.. Natural Gas Pipeline Co..
. R. H. Jensen and F. Kurata, J. Petrol. Technol. 21, 683 (1969),
PhillipsPetroleum Co.. Shell International Gas, Ltd.. Sonatrach, ^'
O, T, Bloomer and J. D. Parent, Chem. Eng, Progr, Symp,
Southern California Gas Co.. Tennessee Gas Pipeline. Texas Ser. 49, 1 1 (1952).
Eastern Transmission Co.. Tokyo Gas Co., Ltd.. and Trans- "2 S.Fuks, J. C. Legros, and A. Bellemans. Phvsica 31, 606
continental Gas Pipe Line Corp.. through a grant administered by the (1%5).
.\merican Gas Association. Inc. " F, G. Keyes, R, S. Tavlor. and L. B. Smith. J. Math. Phvs,
'
W. .M. Havnes and J. W. Stewart, Rev. Sci. Instrum. 42, 1, 21 1 ( 1922).
1 142 (1971). " D. W. Moran. Ph.D. thesis. Imperial College, University of
-
J. W. Beams. Rev. Sci. Instrum. 40, 167 (l%9). London (1959).
R. M. McClintock and M. J. Hiza. Mod. Plast. 35, 172 (1958). " A. Van Itterbeek, O, Verbeke, and K. Staes, Physica 29,
' The use of a trade name for clarity and completeness does not 742 (1%3),

1250 Rev. Sci. Instrum., Vol. 47, No. 10. October 1976 Densimeter 1250

6Z
 .

6.2 Reference [9]

Simplified magnetic suspension densimeter for absolute

density measurements*
W. M. Haynes
National Bureau of Standards, Institute for Basic Standards, Boulder, Colorado 80302

(Received 24 September 1976)

A magnetic suspension densimeter, incorporating three support coils, has been reduced to
a system using only one coil. This simplifies considerably the design of the apparatus and

the procedures involved in the measurements. This instrument can be used for absolute
density measurements; i.e., it does not have to be calibrated with reference fluids of known
density.

During the past two decades the magnetic suspension to support the buoy, and a pair of gradient coils con-
densimeter'"'^ has evolved into a versatile research nected in such a way that at the buoy position their
instrument that has been used routinely for liquid density magnetic fields cancel and their magnetic field gradients
measurements to a precision of ten parts per million. add. The magnetic moment of the buoy is held constant;
(Reference 14 presents a detailed review of the princi- itdepends only on the stability of the constant current
ples, applications, and development of the magnetic source that supplies the main coil current. In the first
densimeter.) In the original design (Fig. 1) of the instru- applications of the three-coil artangement (with a mag-
ment a piece of soft ferromagnetic material is suspended netically soft buoy) the instrument still had to be cali-
in a fluid of unknown density by the magnetic force brated but now the density range for a given buoy was
produced by a single solenoid. The magnetic force on increased significantly over that for a single-coil system.
the buoy is given by the relation Also the calibration was simplified since the density was
linearly proportional to the curtent in the gradient coils
dH for a constant main coil current. The buoy may be
(1)
Hz either magnetically soft or hard for the three-coil system

where M is the magnetic moment of the buoy, H is the Recently, in the adaptation of the magnetic suspension

axial magnetic field intensity of the air core solenoid, densimeter for use at low temperatures,"''^ the same
three-coil arrangement (Fig. 2) was employed to deter-
and Z is the distance along the vertical. The density
of the fluid is related to the current (/) in the support mine the force on a magnetic material in a nonuniform
(p)
field; i.e., it was used to determine the variation of the
coil by
magnetic susceptibility of the buoy material with over H
p = A + BP (2) the range of magnetic needed for the den-
fields

where A and B are constants to be determined by cali- sity measurements. It was found that the magnetic

bration of the instrument with reference fluids of known force on a barium ferrite buoy could be represented by
density. The above relation is based on the assumption the expression

that the magnetic moment of the buoy is linearly pro- dH„ ^ dHc
(3)
portional to the magnetic field intensity. dZ dZ
It is not unusual for the magnetic susceptibility of the
where the subscripts denote the main and gradient coils.
magnetically soft material of the buoy to vary slowly
with H. Thus a relatively large number of reference
fluids are required to calibrate a one-coil instrument to SUPPORT
maintain a given precision over its entire operating COIL^

range. The precision is inversely proportional to the

difference between the density of the fluid and the
density of the buoy. To cover an extended density range I HmicroscopeTj 4^
it is necessary to construct and calibrate buoys of dif- SENSOR
ferent densities. The calibration of the instrument must COILS

be checked periodicEilly to test the buoy material for

hysteresis.
These procedural tasks were reduced considerably
with the development of a three-coil support system^
(Fig. 2) which allows variations mdHldZ while is held

constant. The three-coil arrangement consists of a main Fig. 1. Schematic diagram of a magnetic suspension densimeter with
coil, which supplies the major part of the force necessary one support coil.

39 Rev. Scl. Instrum., Vol. 48, No. 1, January 1977 Copyright © 1977 American Institute of Ptiyslcs 39

33
This result demonstrated that, over the range of mag-
CONSTANT
netic fields considered (0.006-0.016 T), there was no in- CURRENT
duced moment in the barium ferrite buoy (a magnetically SOURCE

hard materieil with a permanent moment). Therefore, it .

was possible to determine fluid densities without the use

of calibration fluids. The instrument constants [C, in D MAIN
Eq. (3)] could be determined from measurements of the 7 COIL

gradient coil current as a function of the main coil current GRADIENT

COILS
while the buoy was supported at a fixed position at con- MAGNETIC
L qMICROSCOPEl=l
BUOY
stant temperature in vacuum. The mass and volume of
i

\ SENSOR
the buoy must be determined from independent meas- \ COILS

urements.
Now that it had been determined with the three-coil
arrangement (Fig. 2) that the magnetic moment of the
barium ferrite buoy was independent of the magnetic
field strength over the range of magnetic fields needed
CONTROL
for the desired density measurements, the instrument CIRCUIT

could be simplified considerably by reducing to a one-

coil system (Fig. 1). The magnetic force on the buoy de-
pended only on the magnetic field gradient (dH/dZ) and Fig. 2. Schematic diagram of a magnetic suspension densimeter with
not on the magnetic field intensity (H). Thus the single three support coils.

(main) coil was connected into the servosystem and the

gradient coils were eliminated.
vacuum points for a single-coil system require only a
It should be noted that a one-coil system can be used
measurement, while those for a three-coil
single current
for absolute density measurements as long as the varia-
system require a series of measurements of l^, and Iq
tion in the magnetic moment of the buoy with magnetic while holding the buoy at a fixed position. Then the
field intensity is known. This information could be ob- currents must be fitted by the method of least squares to
tained by techniques other than that used in the present to obtain the instrument constants. Also the
. Eq. (3)
work, e.g., a vibrating sample magnetometer. This is error analysis is less complicated with a one-coil
mentioned to emphasize that, with a knowledge of M(H)
s'ystem.
of the magnetic buoy material, the initial apparatus
There is one inherent disadvantage in the method for
design could be simplified significantly with the use of
obtaining absolute densities with either the one-coil or
only one coil.
three-coil systems where the magnetic force depends
To carry out density measurements with the one-coil
dH/dZ
only on dH/dZ. The change in with position is
systerfl with a barium ferrite buoy, first the current
significantly larger than the change in the product of
necessary to support the buoy in vacuum /, ) at a given (
H and dHjdZ with position. Thus a very precise deter-
position and temperature must be determined. Then the
mination ( lO"'' mm) of the separation distance of the main
current necessary to support the buoy in a liquid (If)
coiland buoy was required in the present work and has
of unknown density at the same position and tempera-
been described in Ref. 17.
ture is measured. The density of the fluid is related to
Experimental densities have been determined for
this current ratio by
saturated liquid argon and methane at temperatures be-
tween 100 and 120 K with both the three-coil and one-coil
(4) systems. The results agreed within the imprecision of
the measurements, which was approximately 0.015%
where m and V are the mass and volume of the buoy. for methane and 0.005% for argon. Thus the simpler
It should be emphasized that no reference fluids are one-coil system is now utilized for all density measure-

required for calibration of the one-coil system (or the ments with the present instrument.
three-coil system) used in the present work. However, The servosystem that is used with the single coil was
the volume of the float may best be determined near originally connected to the gradient coils in the three-
room temperature using distilled water as a reference coil system. No modifications to the servosystem as
of known density and, in fact, this procedure was
fluid described in Ref. 17 were required for its use with the
used in the present work.'' single coil.
There are several obvious advantages of the one- The present series of density measurements in this
coil system over one using three coils; most are in- laboratory with the magnetic suspension densimeter
volved with the experimental design and procedures. represented a nontypical use of the instrument. In
With elimination of the two gradient coils and the align- the past the instrument has been used for highly precise
ment problems associated with them, the design and (usually less than 0.001%) measurements over a short
construction of the apparatus are much simpler. The density range at a single temperature. Here a relatively
constant current source is no longer needed. The modest precision (less than 0.02%) was the result of

40 Rev. Scl. Instnim., Vol. 48, No. 1, January 1977 Magnetic suspension densimeter 40

34
measurements over large density (0.3-2.3 g/cm^) and vacuum measurements for each datum point. Thus, for
temperature (95-300 K) ranges with a single buoy. density data which the highest precision and accuracy
in

The total uncertainty of the present measurements was are not required, the use of a one-coil system and a
approximately 0.1%. barium ferrite buoy would result in a relatively simple
In order to obtain the highest precision and accuracy and efficient measurement process that requires few
with the present instrument it was necessary to perform vacuum measurements.
vacuum measurements immediately before or after the
liquid measurements. The possible use of the instrument
without repeating the vacuum measurements for each * This work was carried out at the National Bureau of Standards
datum point has been investigated. In other words the under the sponsorship of British Gas Corp., Chicago Bridge and
Iron Co., Columbia Gas Service Corp., Distrigas Corp.. Easco
precision and accuracy of the instrument has been
Gas LNG, Inc.. El Paso Natural Gas, Gaz de France. Marathon
vacuum measure-
estimated, based on a single series of Oil Co., Mobil R and D Corp., Natural Gas Pipeline Co., Phillips

ments as a function of temperature at the beginning Petroleum Co., Shell International Gas. Ltd., Sonatrach, Southern
California Gas Co., Tennessee Gas Pipeline, Texas Eastern
(and/or end) of a set of density measurements without Transmission Co.. Tokyo Gas Co., Ltd., and Transcontinental
regard to the temperature and magnetic field cycling of Gas Pipe Line Corp., through a grant administered by the
American Gas Association, Inc.
the buoy. The separation distance of the coil and the '
J.W. Beams. C. W. Hulburt, W. E. Lotz, Jr., and R. M. Montague,
buoy must be maintained throughout the set of measure- Rev. Sci. Instrum. 26, 1181 (1955).
Jr.,

ments.
' J. W. Beams andA. M. Clarke, Rev. Sci. Instrum. 33. 750 (1962).
^ A. M. Clarke, D. W. Kupke. and J. W. Beams, J. Phys. Chem. 67.

In the present setup the position of the main coil was 929 (1%3).
* D. V. Ulrich. D. W. Kupke. and J. W. Beams, Proc. Nat. Acad.
fixed as described in Ref. 17. The position of the float
Sci. U.S.A. 52, 349 (1964).
was determined with a 125x microscope and the buoy ' J. W. Beams. Rev. Sci. Instrum. 40, 167 (1%9).

could be returned to a desired position either elec- « J. P. Senter, Rev. Sci. Instrum. 40, 334 (1969).

' P. F. Fahev, D. W. Kupke, and J. W. Beams, Proc. Nat. Acad. Sci.

tronically or mechanically. Over a period of several
U.S.A. 63, 548 (1969).
months during the course of density measurements at * A. M. Wims, D. Mclntvre, and F. Hynne, J. Chem. Phys. 50,

low temperatures on several different fluids, the sample 616 (1%9).

» R. Goodrich, D. F. Swinehart, M. J. Kellv, and F. J. Reithel,
cell was cycled between approximately 100 to 130 K Anal. Biochem. 28, 25 (l%9).
and 300 K more than twenty times. The standard devia- J. L. Hales. J. Phys. E 3, 855 (1970).
W. M. Haynes and J. W. Stewart, Rev. Sci. Instrum. 42, 142 (1971).
tion of the vacuum measurements at low temperatures
1

" S. P. Almeida and T. H. Crouch, Rev. Sci. Instrum. 42, 1344 (1971).
was of the order of 0.05% with no systematic trend M.G. Hodginsand J. W. Beams, Rev. Sci. Instrum. 42. 1455(1971).
" D. W. Kupke and J. W. Beams, Methods Enzymol. 26. 74 (1973).
observed. This was a measure of the thermal hysteresis
" S. C. Greer. M. R. Moldover, and R. Hocken, Rev. Sci. Instrum.
of the magnetic moment of the barium ferrite buoy over 45, 1462 ( 1974).
rather large temperature cycles. This result demon- '» Greer. T. E. Block, and C. M. Knobler, Phys. Rev. Lett.
S. C.
34, 250 (1975).
strated that densities with a total uncertainty of approxi- " W. M. Haynes, M. J. Hiza, and N. V. Frederick, Rev. Sci. Instrum.
mately 0.2% could be expected without repeating 47, 1237 (1976).

41 Rev. Sci. Instrum., Vol. 48, No. 1, January 1977 Magnetic suspension densimeter 41
 6.3 Reference [b]

JOURNAL OF RESEARCH of the National Bureau of Standards

Vol. 88, No. 4, July- August 1983

Apparatus for Density and Dielectric Constant Measurements

to 35 MPa on Fluids of Cryogenic Interest

W. M. Haynes and N. V. Frederick

National Bureau of Standards, Boulder, CO 80303

Accepted: March 11, 1983

An apparatus has been develojjed for simultaneous measurements of fluid densities and dielectric constants
at temperatures from 70 to 320 K and at pressures to 35 MPa. A magnetic suspension technique, based on an
application of Archimedes' principle, is employed in the density determination, while a concentric cylinder
capacitor is used for obtaining the dielectric constant data. The apparatus can be used not only for
determining densities and dielectric constants of compressed gases and liquids (including mixtures), but for
saturated liquid and vapor properties as well. Also included is the capability for acquiring liquid-vap>or
equilibrium data for mixtures. The total uncertainty of a single density measurement is estimated to be
approximately 0.1% for densities as low as 50 kg/m'; at lower densities, the uncertainty increases. The
imprecision of the density data is typically less than 0.02%. The total uncertainty in the dielectric constants is

approximately 0.01%. Experimental data for a 0.85 CH4 + O.15 CjH^ mixture are given here to demonstrate
the performance of the apparatus.

Key words: Clausius-Mossotti function; compressed fluid; concentric cylinder capacitor; density; dielectric
constant; excess volume; magnetic suspension densimeter; methane-ethane mixture; saturated liquid; vapor
pressure.

density, temperature, and pressure.

Introduction 2) Calibration fluids are not required.
1.
3) The technique can be used not only to measure
A magnetic suspension densimeter [1,2]' was used in compressed fluids, but also to measure
densities of
program to measure the orthobaric liquid
a large-scale liquid and vapor densities along the coexistence
major components [1,3,4] of liquefied
densities of the boundary.
natural gas (LNG) and mixtures [5-9] of these
components. This technique, based on an application The measurements for the LNG density project
of Archimedes' principle, was selected for the LNG were concentrated in the temperature range from
density project for several reasons: 100-140 K at pressures typically less than 0.2 MPa.
The apparatus was designed for a maximum pressure
1) It is capable of absolute density measurements of of 5 MPa. Near the end of the LNG density project,
high accuracy and precision over wide ranges of the pressure range of the densimeter was expanded to
at least 35 MPa. The expansion was because of a need
.About the Authors, Paper: W, M. Ha\nes is with
to map the PVT surfaces of fluids with critical points
NBS' Chemical Engineering Science Dixision. and
significantly above ambient temperature. (A gas
N. V, Frederick with its Electromagnetic Technology
expansion technique [10-13] used at this laboratory for
Dixision. The work reported on was carried out at
NBS under the sponsorship of the American Gas
PVT measurements on cryogenic fluids could not be
used for this application.) The expansion of the
.Association. Inc. and the National Aeronautics and
densimeter's pressure capability resulted in a new
Space Administration.
instrument, described here, significantly different from
'
Figures in brackets indicate literature references at the end of and more versatile than the previous one [1] developed
this paper. specifically for the LNG density project.

241

37
 Although many of the components of the new measurements for the LNG density project [7,9].
densimeter changed, the technique used is the same. Extensive tests were made to ensure that the density
To detect the position of the magnetic buoy, a Hnear results obtained with the new apparatus were in
differential capacitance sensor [12], compatible with agreement with those from the previous apparatus.
the higher pressure environment, has been developed The consistency of the density data was one of the
to replace inductance sensor. Although the
an most important considerations in the development of
properties of the buoy had been well characterized in mathematical models [15-18] for predictions of
the previous work [1], it was necessary to determine mixture (LNG) densities. The apparatus was then used
the effect of pressure on its volume. A new support to measure the densities and dielectric constants of
coil for lifting the buoy and a new microscope lens liquid propane [19-20], isobutane [19,21], and normal
combination for determining the position of the buoy butane [19,22] over temperature ranges from their
were also required in adapting the technique to higher triple points to 300 K at pressures to 35 MPa.
pressures.
Since the dielectric constant of a fluid is closely
2. Measurement Methods
related to its density through the Clausius-Mossotti
function, a concentric cylinder capacitor was added 2.1 Density
inside sample cell to enable simultaneous
the
In the magnetic suspension densimeter used, a piece
measurements of dielectric constant and density on the
same of magnetic material (barium ferrite magnetic buoy in
fluid samples. Dielectric constant measurements
the shape of a right circular cylinder magnetized along
can serve as simple and reliable substitutes for density
measurements. The addition of the capacitor was its suspended freely by the force
cylindrical axis) is

motivated to some extent by the fact that some produced from the magnetic field of a single air-
axial

commercial densimeters being developed for custody core solenoid. The vertical motion of the magnetic
transfer applications in LNG transactions include
buoy is controlled by the automatic regulation of the
devices based on capacitance measurements. solenoid current with a closed-loop servocircuit that

The new apparatus includes a differential capacitance sensor to detect the

incorporates a cryostat design
different from that employed with the previous
position of the buoy. (The horizontal position of the
densimeter, but similar to ones used with other
buoy is maintained by the axially symmetrical,
diverging field of the solenoid.) In the present system,
instruments [10-13] in this laboratory. The cryostat is

suitable for continuous temperature control between

the magnetic buoy is more dense than the fluids in

70 and 320 K. A new high-pressure window design which it is suspended. Thus, an upward magnetic
developed for the equilibrium cell allows the position force is added to the buoyant force to balance the
of the buoy to be determined by optical means. With downward gravitational force.

the new cell and cryostat, it was not possible to In earlier work
[2] with a densimeter that utilized a

change the position of the buoy by mechanical means; three-solenoid arrangement to supply the magnetic
this is now accomplished electronically.
force, it was found that the magnetic moment of a

The apparatus can also be used for liquid-vapor barium ferrite buoy was independent of magnetic field
intensity over the range of fields (0.006-0.016 T)
equilibrium measurements on mixtures. Means have
been provided for mixing (recirculation of vapor needed to support the buoy. Barium ferrite is a
through liquid), sampling of vapor and liquid, and magnetically hard material with a permanent moment.

subsequent composition analysis. Compared to the This meant that a one-coil system could be used to
previous densimeter [1], improvements have been determine fluid densities without the need of
made calibration fluids. To carry out density measurements
the characterization of the temperature,
in
pressure, and composition of the fluid samples. With with a one-coil system and a barium ferrite buoy, first
the previous apparatus [1], it was possible to observe the current (7^,) necessary to support the buoy in
the liquid-vapor interface through a window that vacuum at a given position (buoy-coil separation

extended most of the length of the sample space. In distance) and temperature is measured. Then the

adapting the magnetic suspension densimeter for current (If) necessary to support the buoy in a fluid of

pressures to 35 MPa over a wide temperature range, it unknown density at the same position and temperature
was not practical to retain this feature. This feature, is determined. The
density (p) of the fluid is related to

although convenient, was not essential for any of the these currents by the relation,

measurements performed with the new apparatus.

The new apparatus was first used to complete p=7<i-f). (1)

242
where m and V are the mass and volume of the buoy. determined by the dimensions^ of the high-pressure cell
Measurement procedures for using this equation to and the support coil. The outside diameter of the
determine fluid densities for the instrument developed stainless steel vacuum jacket is 20.3 cm while the
in the present work are discussed later. Also presented outside diameter of the glass of the cryostat is 9.0
tail

is a detailed description of the

magnetic suspension cm. There is 0.4-cm clearance on the diameter
system. between the cryostat tail and the support coil. The

2.2 Dielectric Constant

Access Tubes for Capillaries.
Coaxial Cables and Wires
A stable concentric cylinder capacitor was used for Vacuum
dielectric constant measurements. First, the capaci- LNj Fill

tance (Q with the fluid (of unknown dielectric

constant) between the cylindrical electrodes is
measured. Then, at the same temperature, the vacuum
capacitance (CJ is determined. The dielectric constant
(f) is calculated from the relation,
Metallic Bellows
€ = C/Q. (2)

3. Apparatus
LN2 Reservoir

3.1 Cryostat

The major features of the cryostat are shown to

scale in figure 1, an assembly drawing of the appara-
tus. This cryostat is similar to some used previously at
this laboratory [10-13]. Those modifications necessary
to adapt the cryostat for use with a magnetic suspen- Reflux Tube
sion densimeter will be emphasized.
The cryostat was supported by a 1.3-cm-thick ^ — Wood's Metal Joint
aluminum plate suspended from concrete block Cold Ring
columns at a height 2.3 m above floor level. The
aluminum plate was reinforced with 10-cm-wide
Guard Ring
channel beams so that the position of the
stainless steel
cryostat was independent of the amount of liquid
Vapor Pressure Bulb
nitrogen in the reservoir. Apparent changes in the Vacuum Jacket
Copper
buoy-coil separation distance resulted if the position of
Sample Holder Cold Wall
the cryostatwas not maintained during the course of
Sample Fluid Shield
measurements. The aluminum plate from which the
cryostat was suspended could be leveled and clamped Capacitor

in place using four bolts in contact with the concrete

Support Coil Cooling Coils
block columns.
The cryostat was fabricated from nonmagnetic Platinum Resistance
materials. The inner cylinders or cans (liquid nitrogen Thermometer ^ Vacuum Window
reservoir,cold ring, guard ring, shield, cold wall) Differential la
Microscope
were copper; the outer vacuum jacket and access Capacitance Sensor
High Pressure Window
tubes were primarily stainless steel; and the flanges, Magnetic Buoy
support plates, and fittings were mostly brass. The Vapor Pressure Bulb
Quartz Leg
various cylindrical portions of the cryostat and the
central support (reflux) be aligned
tube had to Glass Tail
(concentric and vertical) could be
so the buoy
suspended symmetrically about the cylindrical axis of Figure 1-Assembly drawing (approximately to scale) of an apparatus
the differential capacitance sensor. Many of the for density and dielectric constant measurements to 35 MPa on

dimensions of the cryostat comp>onents were cryogenic fluids.

243

39
overall length of the cryostat is approximately 94 cm.
The liquid nitrogen reservoir has a of
capacity
3.2 Equilibrium Sample Cell
approximately 7 x lO' cm' and needs refillingunder
normal conditions about every 16 h. Metallic bellows
Figure 2 is a detailed cross sectional diagram of the
were placed in the fill and' vent lines of the reservoir copper equilibrium cell. The overall dimensions of the
to eliminate relative movement, as a function of cell used here are roughly the same as for the cell used
temperature, between the cell and the outer stainless with the earlier version [1] of the magnetic suspension
steel jacket of the cryostat. This step was necessary to densimeter, except for an increase in the wall thickness
maintain alignment of the capacitance sensor and buoy required for use at higher pressures. The new cell has
located inside the cell with the support coil located an overall length of approximately 29.5 cm, primarily
outside the cryostat, independent of the experimental determined by the position at which the buoy must be
conditions. Five access tubes, which passed through suspended relative to the diverging field of the support
the liquid nitrogen reservoir, were available for
of the buoy. The outside
coil for horizontal stability
introducing capillaries, coaxial cables, and wires into diameter of the main part of the cell is 4.44 cm, while
the vacuum space and cell inside the cryostat.
The cryostat was comprised of one large vacuum
space; holes (3 cm diameter) were cut into the
Capillaries Reflux Tube
electroformed copper cylinders attached to the liquid
nitrogen reservoir andguard ring for visual
the Coaxial Cables

observation of the buoy. With this arrangement it was

Vapor Pressure
not feasible to use exchange gas for fast-cooling the Bulb
cell. The exchange gas would have been in direct

contact with the outer stainless steel can which

included a glass tail as its lower section. The glass tail
was connected to the outer jacket through a stainless
steel-to-glass transition joint; to ensure integrity of this
seal, the joint should be kept at a temperature near
Copper 0-Ring

ambient.
Outer Electrode
Fast-cooling of the cell was accomplished using the
refiuxing action of nitrogen in the central support Inner Electrode

tube.The maximum cooling rate was approximately

Capacitor
50 K/h. The reflux tube is evacuated during
measurements when the cell is controlled at a desired
temperature. For additional cooling, three flexible
copper braided straps (not shown in fig. 1), each with Grounded Guard
a cross sectional area of 5 mm^ per strap, have been
Copper Sample Holder
connected symmetrically from the top of the cold
shield to the cell.
Support Tube
The cryostat was designed to facilitate ease in
assembly and disassembly and to minimize alignment
Sapphire Window
problems of critical components. All demountable
joints had enough latitude to make alignment Platinum Resistance
reasonably straightforward. The outer vacuum jacket Thermometer

was attached to the top of the cryostat through a Magnetic Buoy

rubber o-ring seal using a split-ring assembly not Indium Gasket
Differential
shown in the drawing. The inner cans were attached Capacitance
with either Wood's metal or screws. Once the joints Sensor
were secured, the and no
entire structure was rigid,
problems with alignment were encountered after Capillary
Vapor Pressure
numerous temperature cycles. A nut and bolt i

(fiberglass) arrangement was installed between the

cold shield and the bottom of the cell to eliminate
pendulous vibrations of the cell. Figure 2-The equilibrium cell.

244

4U
 the top closure section has a diameter of 8.10 cm. The a groove cut lengthwise in the cell wall. Vapor
dimensions of the internal sample space are 1.90 cm pressure bulbs fabricated from copp>er (each with an
diameter by approximately 23 cm length. About one internal volume of approximately 2 cm^) have been
third of the internal volume is occupied by the installed at the ends of the cell thermal in close
concentric cylinder capacitor and the magnetic buoy- contact (soft-solder) with the outside surface. These
capacitance sensor assembly. This results in an internal bulbs, along with differential thermocouples (chromel-
free volume of approximately 43 cm\ as determined constantan) secured at the same positions, are used to
by filling the cell with water. monitor temperature differences between the ends of
To facilitate easy access to the assemblies inside the the cell. Four independent heaters of 110 to 160 Cl
cell, a flange-gasketed seal was used to close the cell. each for temperature control have been wound
The was effected by compressing a silver-plated,
seal bifilarly at different positions along the length of the
solid copper O-ring between a stainless steel plug and cell.

a smooth, flat ledge of the copper cell. This assembly A total of six coaxial cables enter through the lid of
was similar to that used earlier [1]. A thin- walled the cell into the sample space in pressure-tight
stainless steel tube, used to support the cell, was assemblies. Three of the cables are for the capacitance
soldered into the cell lid (the stainless steel closure sensor, two for the cylindrical capacitor, and one is a
plug). This support tube was also used for reflux gas as spare. Each coaxial cable consists of an outer stainless
described earlier. steel sheath of 0.51 mm
diameter insulated from a 0.13
For visual observation of the magnetic buoy mm diameter inner conductor with polytetrafluoro-
position, a new window assembly for pressures to 35 ethylene. These cables have been found to be leak-
MPa has been developed. The assembly consists of a tight along short lengths at gas pressures to 70 MPa at

sapphire disk (1.90 cm diameter by 0.63 cm thickness) room temperature. Cables approximately 1 meter in
with chamfered edges, indium gasket, asbestos-rubber length have been inserted into stainless steel capillaries
backing ring to relieve thermal and mechanical strains, (1.07 mm outside diameter X 0.66 mm inside diameter)
and stainless steel plug and ring with four stainless that extend from inside the sample cell to a position
steel set screws to provide a pressure- and vacuum- above the top of the cryostat where the coaxial cables
tight seal. It has a 0.63 cm diameter viewing area. An are soldered into the capillaries. This means that con-
effort was made to minimize the size of the high tinuous lengths of the coaxial cables extend from
pressure window assembly, especially in the horizontal inside the sample cell to a region in which the cables
direction. This effort resulted from an attempt to are leak-tight. At low temperatures, the cables do not
minimize the inner diameter of the support coil so seal because of the relatively large difference in the
current and power in the coil were not excessively thermal expansion coefficients of stainless steel and
large. polytetrafluoroethylene. The stainless steel capillaries
A total of four capillaries has been introduced into are soldered into collars in the lid of the cell and at the
the sample space. Two capillaries are soldered into top of the cryostat.
collars in the cell lid; the other two are soldered into
ends terminating at the bottom
the cell wall with their
of the sample space. This makes it possible to fill the 3.3 Magnetic Suspension System
cell from either the top or bottom, or from both
The magnetic suspension system, employed here in
simultaneously. procedures can be critical
Filling
the density determination, uses some components from
when condensing liquid mixtures into the cell. Under
the previous densimeter [1]. The magnetic buoy, with
normal conditions, a pair of capillaries, one at the top
its well-characterized properties deduced from the
and one at the bottom, is available for recirculating
earlier work, has been described in detail. The buoy is
the vapor through the liquid for mixing; the other cap- a barium ferrite magnet in the shape of a right circular
illary at is for pressure measurements while the
the top cylinder (0.51 cm diameter X 0.64 cm length) magne-
free capillary at the bottom can be used for liquid
tized along its cylindrical axis. Barium ferrite is a mag-
samphng. By having capillaries at both ends of the netically hard, ceramic material with a density of ap-
cell, it is easy to remove low vapor pressure liquids proximately 5x10^ kg/m^ Since barium ferrite is
from the cell.
I

porous, the buoy was plated with copper to a

At approximately the same height at which the thickness of approximately 0.06 mm. A thin (10"' mm)
||
magnetic buoy is suspended, a platinum resistance protective coating of gold was flashed over the
1
thermometer has been soldered with indium into a copper. No problems were encountered with fluids
closed-end copper tube that has been soft-soldered into penetrating the copper layer at pressures to 35 MPa.
Ij

245

I
-

41
Barium ferrite was selected as the buoy material since position of the buoy can be made insensitive to the

its magnetic properties are consistent with absolute dielectric constant of the sample fluid. For a detailed
density measurements [2]. Over the range of magnetic description of the capacitance sensor, see reference
fields needed to support the buoy, the magnetic mo- [13].

ment of the barium ferrite buoy had been found to be The previous densimeter [1] included mechanical
independent of the magnetic field intensity. means for fine adjustment of the position of the buoy
The densimeter described here included a single in going from vacuum-to-liquid measurements. The
solenoid that supplied the force required to lift the new system does not incorporate this feature. Small
buoy. The simplification in the densimeter design in changes in the buoy position are accomplished
going from a three-coil to a one-coil system, which electronically in the new system by adjusting the
resulted from a determination of the magnetic offset voltage of the integrator amplifier in the
properties of barium ferrite, has been discussed in servocircuit. A calibrated 0.5-fl standard resistor,

detail in an earlier paper [2]. (Although no further use placed in series coil, can handle the
with the support
of a three-coil arrangement with gradient coils was relatively large currents needed to lift the buoy in the
anticipated, the outside diameter of the new support present work. The measurement of the voltage drop
coil was made less than the inside diameter of the across this resistor, which is immersed in an oil bath,
gradient coils used in the previous work [1] in the enters directly into the density determination.
eventuality that a need for tests on new buoy materials A 125x filar micrometer microscope is used to
ensued.) determine the position of the buoy. A new lens
The new support coil is composed of two separate combination, which includes an objective lens (38 mm)
coils of 2275 turns each of epoxy-coated aluminum foil with a larger working distance (less power) and a
of approximately 0.025 mm thickness and 2.5 cm higher-power (14x) eyepiece, gives approximately the
width. Each of the two coils, epoxied to a central same magnification as with the earlier apparatus [1].
quartz tube at a separation distance of 1.3 cm, has an The microscope is rigidly mounted on an aluminum
inside diameter of 10.16 cm and
an outside diameter of support table, adjustable in three dimensions. The
roughly 22.2 cm. To be compatible with the new entire assembly is secured firmly with bolts to the
higher-pressure cell, the inside diameter of the new massive concrete block structure.
coil is somewhat larger than that used previously [1].

The quartz tube is attached to a fiberglass plate

3.4 Concentric Cylinder Capacitor
supported by three quartz rods (2.54 cm diameter) that
extend to another fiberglass plate resting on a part of The capacitor for dielectric constant measurements
the concrete block structure, used for rigid support of was located in the top portion of the sample space, as
the entire Water-cooled copper plates,
apparatus. seen in figure 2. The central support mandrel of the
insulated with thin mylar sheets, have been placed in capacitor was connected rigidly to the sample cell lid
close thermal contact with all faces of the two coils. using a slotted brass tube, crimped at its ends, that
No problems have been encountered with this clamped tightly over mating surfaces on the mandrel
arrangement for mounting and cooling the coils for and the lid. The same type of brass tube was used to
use at currents to 1.5 A, which corresponds to 270 W connect the differential capacitance sensor to a nut at
heat dissipation in the coils. the bottom of the capacitor. The use of the brass tubes
The servocircuit is essentially the same as that used for support of the components inside the cell resulted
before [1] except for a new type of sensor for in an extremely rigid assembly that provided means
detecting the position of the buoy.Some problems had for independent rotational orientation of each
been encountered with the earlier densimeter [1] with component. The coaxial leads to the differential
pressurized fluid slowly penetrating into the coil capacitance sensor located below the capacitor passed
windings of an inductance sensor, resulting in a slow through the brass tubes and a slot in the central
change in the position of the suspended buoy. Thus, a support mandrel of the capacitor.
differential capacitance sensor [13], a solid monolithic The concentric cylinder capacitor design used in the
structure, has beendeveloped for position detection in present work was based on that developed by
the present work that entails measurements in fluids at Younglove and Straty [23]. Two slightly different
pressures to 35 MPa. The sensitivity of the capacitors used with the present apparatus,
are
capacitance sensor was at least equivalent to that of depending on
whether the measurements are for
the linear differential transformer used previously. By mixtures or for pure fluids. For mixture measurements,
balancing a bridge of which the sensor is a part, the slots are cut into the cylinders parallel to the

246
cylindrical axes, similar to the ring and bar design of Vapor pressure measurements on liquid nitrogen,
Pan et al. [24]. This modification was made to methane, ethane, and propane at temperatures from
minimize the chances for composition gradients in the 100-290 K have been used to check the calibration of
cell by allowing free passage of fluid mixtures between the thermometer. Temperatures from vapor pressure
the electrodes. The vacuum capacitance of the measurements, using selected data from the literature,
capacitor with slotted cylinders was approximately 20 generally agreed with measured temperatures
pF, while that for the one with solid cylinders was (platinum resistance thermometer) to better than 0.02
about 33 pF. K. Further details on temperature measurements are
The dimensions of the capacitors used here were presented in reference [1].

proportionally the same as those of Younglove and Temperature differences between the ends of the
Straty [23]; however, the overall size was significantly cell are monitored with vapor pressure bulbs located

smaller. The overall external dimensions of each at the ends of the cell. The bulbs have been filled with

capacitor were 6.4 cm length X 1.77 cm diameter. For the fluids mentioned above, the selected fluid
the capacitor with slots, the widths of the five slots in depending on the temperature range needed. With the
each cylinder were 0.32 cm, while the slot lengths in reflux tube evacuated, overall temperature differences
the outer cylinder were 4. 1 cm and those in the inner were typically less than 0.01 K, or within the precision
cylinder cm. The outside diameter of the
were 1.9 of vapor pressure measurements. Thus, a
the
inner cylinder was 1.06 cm. The outer cylinder had a differentialthermocouple, with junctions placed near
thickness of 0.32 cm, while that of the inner cylinder the vapor pressure bulbs, was not needed as part of a
was 0.24 cm. There was a separation distance of 0.38 control loop to regulate the temperature distribution
mm between the cylindrical electrodes. Small pieces along the length of the cell.
of Kapton^ film (0.05 mm thickness) were used to A third vapor pressure bulb was soldered in close
insulate the electrodes from the support assembly. thermal contact with the guard ring. This vapor
Both of the capacitors were fabricated from copper. A pressure bulb was used to check a differential
thin protective coating of gold was flashed over the thermocouple between the guard ring and the cell.
surfaces. This thermocouple was part of a control circuit to
maintain the temperature of the guard ring
approximately equal to that of the sample cell. The
guard ring was connected to a copper radiation shield
4. Measurements that surrounded the cell to provide an approximately

4.1 Temperature and Pressure isothermal environment for the sample space.
The techniques and instrumentation for control of
The primary temperature on the
sensor, calibrated the temperature
of the cell are standard. Four
IPTS-68, is a platinum resistance thermometer, which independent heaters along the length of the cell are
had been used with the previous densimeter [1]. The available, if needed, to minimize temperature
uncertainty of the calibration is approximately 0.002 gradients. Only the middle two heaters have been used
K. The potentiometric system for temperature in the control circuit.
It should also be noted that
measurements gives uncertainties that range from temperature gradients along the length of the cell can
approximately 0.010 K at 100 K to 0.030 K at 300 K. be monitored by observing the vapor pressure of the
The temperature of the sample space is regulated liquid inside the cell as a function of the liquid level.
within a few mK, approximately the same as the
Pressures of the fluid under test are usually
reproducibility of the temperature measurements. A measured with a dual-range, precision oil dead-weight
current of 1 mA
for the thermometer, supplied by an
gauge. Its sensitivity ranges from 2x10^ MPa at 3
electronic constant current source, is determined to an
MPa to 2X10"^ MPa at 35 MPa. The overall
uncertainty of approximately 0.002%. The total uncertainty in pressure is approximately 0.01%,
uncertainty in the measured temperatures estimated
is increasing somewhat at lower pressure. With the high
to be less than 0.03 K. range piston, this gauge cannot be used for pressures
less than 0.2 MPa; with the low range piston, the
' In order to describe wterials and experimental procedures lowest pressure is approximately 0.04 MPa.
adequately, it is occasionally necessary to identify commercial
products by manufacturers' or trade names. In no instance does such
A Bourdon-tube gauge with a range of
spiral quartz

identification imply endorsement by the National Bureau of

0-1.38 MPa
normally used for vapor pressure
is

Standards, nor does it imply that the particular product is

measurements. It has been calibrated against an air
necessarily the best available for that purpose. dead-weight gauge; maximum uncertainty in the

247

43
 calibration was 70 Pa. The resolution of this gauge is coil in vacuum and in the test fluid at the same temper-
better than 20 Pa. ature.Although the microscope lens combination (sec.
Both pressure gauges are characterized by relatively 3.3) used here is different from that of the previous
small free volumes, approximately 0.5 cm' for each. densimeter, the present arrangement results in position
This is an important consideration when performing measurements of the same quality. The maximum
X 10'' mm,
mixture (phase equilibria) measurements with this error in the position determination is 2
apparatus, e.g., to minimize vapor space corrections. which corresponds to an error in density of less than
0.03% for a density of 5 X 10^ kg/m\ or larger.
The position of the buoy is observed through a 0.63
4.2 Density
cm thick sapphire window in the cell. It has been ob-
served that the apparent position of the buoy changes
The accuracy and precision of density measure-
slowly with the pressure inside the cell; as the cell
ments with the magnetic suspension technique used
volume expands, the angle of the sapphire window
here depend on a knowledge of the mass and volume
changes. The magnitude of has been deter- this effect
of the buoy, along with the capability to determine the
mined microscope is adjusted so
as follows. First, the
buoy with high resolution independent
position of the
that the position of the buoy is independent of the
of the medium in which the buoy is suspended. The
index of refraction of the fluid inside the cell. This
same barium ferrite buoy as used with the previous
procedure is accomplished by observing the buoy
densimeter was employed here; its properties have
resting at a stationary position on the capacitance
been well characterized and are discussed elsewhere
sensor assembly as the cell is alternately filled with gas
[1]. The mass of the buoy is 0.73706 g and its volume
and liquid at a pressure slightly greater than 0.1 MPa;
at 300 K is 0.13485 cm\ Thermal expansion data [25]
e.g., methane at 120 K. The temperature of the cell is
for barium ferrite were previously obtained to calcu-
maintained during these tests. Then, the cell is filled
volume of the buoy at low temperatures. The
late the
with liquid at a pressure less than 0.1 MPa; either
change in the volume of the buoy for a temperature
propane, isobutane, or normal butane have proved
change from 100-300 K is approximately 0.4%.
ideal for these tests at temperatures between 100 and
The earlier densimeter was used at relatively low
300 K. Then, the pressure inside the cell is increased
pressures compared to those used now. Based on the to greater than 35 MPa while observing the buoy at
properties of similar materials,it was estimated that
rest on the sensor assembly. The change in the index
the effect of pressure on the volume of the barium
of refraction for a 35-MPa change in pressure along an
ferrite buoy was negligible (<0.001%) for the highest
isotherm for liquid propane, isobutane, or normal
pressures (2.5 MPa) encountered with the previous butane is negligible compared to the change in
densimeter. Since the new apparatus is used at
refractive index in going from vapor-to-liquid for
pressures up to 35 MPa, bulk modulus measurements
methane at 120 K. The change in positionwith pres-
[26] have been carried out on a sample of barium
sure corresponds to a change in density with pressure
ferrite, the buoy material. Bulk modulus (Bj) data (in
of less than 5x 10"^ kg/(m'-MPa) for liquid methane at
units of MPa) from 75-295 K are represented as a
120 K.
function of temperature {T in units of K) by the
The currents in the support coil needed to suspend
expression,
the buoy in vacuum and in the fluid of interest (see eq
(1)) are determined by measuring the voltage drops
across a 0.5-fl standard resistor in series with the
0.0206014
support coil. The voltages were routinely measured to
B r= 6295^.01 629- 175.15
r-75
-1 (3) 5x10 * V with a high resolution differential-type
) voltmeter, which corresponds to a change in density
of less than 0.01% for a density of 5 X 10^ kg/m\

where B2„= 1.3030 x 10' MPa, the bulk modulus at

4.3 Dielectric Constant
295 K. Coefficients were determined from nonlinear
least squares. The correction to the volume of the
The dielectric constant is determined from a
buoy at room temperature for a pressure of 35 MPa measurement of the ratio of the capacitance of the
amounts to 0.027%. concentric cylinder capacitor with fluid between the
For density measurements, it is necessary to suspend electrodes to the capacitance under vacuum. The
the buoy at the same position relative to the support capacitances are measured with a three-terminal ac

248

44
 bridge operated at an oscillator frequency of 5 kHz. constant data for liquid methane exhibited differences
Measurements to a resolution of 10"* can be obtained of <0.01% when compared with the data of Straty
with little difficulty with this bridge. and Goodwin [28].

Both capacitors used in this work yielded equal Next, data were obtained for a binary mixture of
results within the precision of the measurements in methane and ethane, a system that had been
testson liquid methane. Measurements on mixtures are extensively investigated with the earlier densimeter.
inherently more difficult and are generally character- The data for three mixtures of methane and ethane
ized by greater scatter because of the difficulties associ- had been used to optimize several mathematical
ated with obtaining homogeneous liquid mixtures in models [15-18] developed for prediction of LNG
the sample space. Examples of dielectric constant mea- densities. The most accurate and versatile of these
surements on pure fluids and mixtures with the appara- models was the extended corresponding states method
tus described here are given in references [7,19-22,27]. [15-17]. The new data for a methane + ethane mixture
Measurements of the vacuum capacitance are would be compared with predictions from this model.
normally obtained just before or just after fluid mea- The mixture, for which data are reported here, had
surements are performed. Since vacuum measurements also been used to cross-check results from this laborato-
are required for each run in the density measuring ry obtained with the magnetic suspension densimeter
technique used here, therewas no reason vacuum ca- with those from another laboratory that employed an
pacitances should not be recorded at the same time. entirely different technique for determining density
Thus, there was no reason to acquire a calibration [29]. The experimental density of Miller and Hiza [29]
curve representing the vacuum capacitance as a func- for this methane + ethane mixture at 110.08 K differed
tion Vacuum measurements can be
of temperature. by 0.01% from that calculated from the extended cor-
made to a precision of 10^ pF and are usually stable responding states model [16].
within 0.0002 pF when the sample cell is cycled The experimental orthobaric liquid densities, vapor
between low temperature and room temperature. pressures, and dielectric constants of a 0.85147 CH4 +
(Vacuum capacitances have been observed to be stable 0.14853 C2H6 mixture are presented as a function of
to better than 10"^ pF for more than 20 temperature temperature in table 1. Excess volumes and values for
cycles over a period of several months.) The total the Clausius-Mossotti (CM) function, as well as values
change vacuum capacitance from 300-100 K is
in the for the excess function, are also given in table 1. The
approximately 0.3%. Based on tests by Younglove and excess volume (V^) is defined by the relation,
Straty [23] on a similar cylindrical capacitor, the effect
of pressure on the capacitor was sufficiently small to
neglect. It is estimated that, based on the resolution of V^=V- J^x.V.ll+P.iPrP)]^ (4)
the capacitance measurements and the stability of the
capacitor design, the total uncertainty in the dielectric
constant measurements is approximately 0.01%. where V is the molar volume of the mixture at a given
temperature at saturation pressure p, V, is the molar
volume of component / at the same temperature at sat-
5. Performance of Apparatus
uration pressure p, x, is the mole fraction of compo-
,

5.1 Results and Discussion nent /, and /3, is the isothermal compressibility of com-
ponent /. The CM function is defined by the
Before completing measurements for the LNG expression,
density project,it was necessary to ensure that the
new densimeter yielded results consistent with those
from the previous densimeter. First, density data were
obtained for several components of LNG (e.g.,
methane, ethane, etc.); the data obtained with the new
apparatus agreed to better than 0.02% with the results
where p is the density and € is the dielectric constant.
1 1,3,4] determined at the onset of the LNG density
Then the excess Clausius-Mossotti function (CM^) for
project with the earlier version [1] of the magnetic
a liquid mixture is defined, analogous to V^, by the re-
suspension densimeter. As mentioned earlier,
lation,
measurements on saturated liquid methane were used
||

as a check on the measurement process during the

CM^ = CM-2;x,CM,, (6)
I
entirety of the LNG density project. Dielectric

249

4b
 Table 1. Orthobaric liquid densities (p) and dielectric constants (£) of 0.85147 CH4 + 0.14853 C^H,, mixture
(molecular weight = 18.1265 g mol ') as a function of temperature (D and pressure (P). PcsM- density calculated
from extended corresponding states model; V^, excess volume; CM, Clausius-Mossotti function; CM^, excess
Clausius-Mossotti function.

__
T P P CM CM^
K MPa mol-dm"' PCSM cm'-mol' cm^-mol"' cm'-mol'

115.00 0.116 25.3618 -0.059 -0.410 1.67297 7.2243 0.005

120.00 0.166 24.9911 -0.013 -0.493 1.66070 7.2220 0.000
125.00 0.232 24.5983 -0.045 -0.567 1.64828 7.2239 -0.001
130.00 0.316 24.1950 -0.036 -0.655 1.63561 7.2258 -0.003

where CM refers to the Clausius-Mossotti function of increasing temperature.

the mixture at a given temperature at the saturation After the new apparatus had been thoroughly
pressure of the mixture, and CM, is the Clausius- tested, it was used to complete the measurements for
Mossotti function of pure component / at the same the LNG density project [7,9]. Next, the apparatus
temperature and pressure as the mixture. Adjustments was used to obtain data for propane [19-20], isobutane
of the pure component CM,'s to the saturation [19,21],and normal butane [19,22] at pressures up to 35
pressure of the mixture are sufficiently small to MPa. The performance of the apparatus at high
neglect. pressures is demonstrated by these pure fluid data.

In the calculation of and CM, the pure compo- ,

nent molar volumes of methane and ethane were calcu- 5.2 Error Analysis
lated from equations in references [1,3] obtained from
fitting experimental orthobaric liquid densities deter- Detailed discussions of the systematic and random,
mined with the earlier version of the magnetic suspen- errors involved in measurements with the magnetic
sion densimeter used in the LNG density project. suspension densimeter used in the present work have
Vapor pressures for methane and ethane were taken been presented elsewhere [1,3-9, 19-22]. The
from Goodwin [30] and from Goodwin at al. [31], re- uncertainty in the density measurements depends
spectively. Isothermal compressibilities for methane primarily on the uncertainties in the determination of
were taken from Rowlinson [32]; those for ethane the volume of the buoy, of the relative position of the
from Miller [33]. The dielectric constants of Straty and buoy and the support coil, and of the temperature of
Goodwin [28] were used for methane, while those of the sample fluid. With the new apparatus, designed for
Weber [34] were used for ethane. higher pressures than the previous one, the effect of
Also presented in table 1 are comparisons between pressure on the apparent position of the buoy,

the experimental densities (p„pt) from this work and resulting from slight movement of the cell window,
densities calculated from the extended must now be included. Maximum uncertainty in the
(Pcsm)
corresponding states model [15-17]. The average density determination resulting from this effect is

absolute deviation is 0.038%. This result, combined 0.02%.

with the pure fluid comparisons, demonstrates that The effect of pressure on the volume of the buoy,
data obtained with the new densimeter are consistent which is extremely small at maximum design pressure
with data taken with the previous instrument. The of the cell, must also be considered. As discussed in
total uncertainty of a single density measurement for section 3.3, the reduction in the buoy volume for a
these binary mixture data is estimated to be ±0.1%. pressure of 35 MPa is 0.027%. The uncertainty
The present results are also consistent within involved in this adjustment is negligible.
experimental error with the data point of Miller and As discussed [1], the total uncertainty of a single
Hiza [29] for the same mixture. density measurement is taken as the square root of the
The calculated excess Clausius-Mossotti values were sum of the squares of the systematic errors plus an
less than 0.07% of the mixture CM values for the allowance of three times the standard deviation for
methane 4- ethane mixture data presented here. This random error. The imprecision of measurement (or
result was expected based on dielectric constant and standard deviation) is typically less than 0.02%. Both
density measurements on other mixtures containing the precision and accuracy of density measurements
nonpolar constituents [7,9,24]. It appears that the with a magnetic suspension densimeter depend on the
excess CM values show a slow decrease with difference between the density of the buoy and the

250

4b
 .

density of the fluid. For the present arrangement,

where the density of the buoy is significantly larger 6. References
(typically an order of magnitude) than the density of
the fluids, the precision and total uncertainty change [1] Haynes, W. M.; Hiza, M. J.; Frederick, N. V. Magnetic
slowly with fluid density. The total systematic error in suspyension densimeter for measurements on fluids of
the measurement process from known sources is cryogenic interest. Rev. Sci. Instrum. 47(10); 1237-1250;
low temperatures, decreasing 1976 October.
approximately 0.05% at
|2) Haynes, W. M. Simplified magnetic suspension densimeter for
to approximately 0.03% at room temperature. This
absolute density measurements. Rev. Sci. Instrum 48(1):
results in an estimated total uncertainty in the density 39^1; 1977 January.
of approximately 0.1% at low temperatures and 0.06% [3] Haynes, W. M.; Hiza, M. Measurements of the orthobaric
J.

at room temperature. (Of course, the uncertainty in liquid densities of methane, ethane, propane, isobutane, and
normal butane, J. Chem. Thermodyn. 9(2): 179-187. 1977
the density determination also depends on the values
February.
of the derivatives, (dp/dP)j- and (dp/dT)p, for the (4) Haynes, W. M.; Hiza, M. J. Orthobaric liquid densities of
particular region of the PVT surface for the fluid normal butane from 135 to 300 K as determined with a
under investigation.) magnetic suspension densimeter. Advances in Cryogenic
Engineering, Vol. 21. Timmerhaus, K. D.; Weitzel, D. H.,
The uncertainties in the dielectric constant determi-
ed. New York, NY: Plenum Press; 1976 516-521.
nation have been discussed in detail (sec. 4.3 and refs. Hiza, M. Haynes, W. M.; Parrish, W. R. Orthobaric liquid
J.;
[5]
[7,9,19-22]). The total uncertainty in the dielectric densitiesand excess volumes for binary mixtures of low
constant measurement is estimated to be approximate- molar-mass alkanes and nitrogen between 105 and 140 K. J.

ly 0.01%. Chem. Thermodyn. 9(9): 873-896; 1977 September.

[6] Hiza, M. J.; Haynes, W. M. Liquid mixture excess volumes
The estimates of the uncertainties in the density and
and vapor pressures using a magnetic suspension
total
dielectric measurements can be tested to
constant densimeter with compositions determined by
some degree by making comparisons with reliable data chromatographic analysis: methane plus ethane. Advances
from independent sources. Such comparisons have in Cryogenic Engineering, Vol. 23. Timmerhaus, K. D., ed.

been made for many fluids over wide ranges of experi- New York, NY: Plenum Press; 1976; 594-601.

mental parameters (such as temperature, pressure, den- 17] Haynes, W. M Orthobaric liquid densities and dielectric
sity, etc.) [7,9,19-22]. In general, these comparisons constants of (methane + isobutane) and (methane + normal
butane) at low temperatures. J. Chem. Thermodyn. to be
have confirmed the estimates of the uncertainty levels.
published.
There has been little information presented in this
[8] Hiza, M. J.; Haynes, W. M. Orthobaric liquid densities and
paper concerning the problems associated with excess volumes for multicomponent mixtures of low molar-
mixture measurements compared to pure fluid mass alkanes and nitrogen between 105 and 125 K. J.

measurements. The uncertainties involved in the Chem. Thermodyn. 12(1): 1-10; 1980 January.
determination of the composition of mixtures have [9] Haynes, W. M. Measurements of orthobaric-liquid densities of

been discussed in detail in previous papers [1,5-9].

multicompKDnent mixtures of LNG components (Nj, CH,.
CjHe, CjHg, CHjCH(CH3)CH3, C^H.o, CH3CH(CHj)C,H5,
and CjH.j) between 110 and 130 K. J. Chem. Thermodyn.
14(7): 603-612; 1982 July.

(10) Goodwin, R. D. Apparatus for determination of pressure-

density-temperature relations and spyecific heats of
hydrogen to 350 atmospheres at temperatures above 14 K.
J. Res. Natl. Bur. Stand. (U.S.) 65C(4): 231-243; 1961
Oc tobe r- Decem be r
[11] Prydz, R.; Straty, G. C. The thermodynamic properties of
compressed gaseous and liquid fluorine. Natl. Bur. Stand.
(U.S.) Tech. Note 392; 1970 October; 182 p.
(12] Straty, G. C; Prydz, R. Fluorine compatible apparatus for
accurate PVT measurements. Rev. Sci. Instrum. 41(8):
1223-1227; 1970 August.
The authors are grateful to the following individuals
(13] Straty, G. C. (p.V,T) of compressed fluid ethene. J. Chem.
who made significant contributions in the develop- Thermodyn. 12(8): 709-716; 1980 August.
ment of this apparatus. W. R. Bjorklund, A. N. (14] Frederick, N. V.; Haynes, W. M. Differential capacitance
DiSalvo, W. G. Layne (deceased), and D. L. Smith sensor as p>osition detector for a magnetic suspension
densimeter. Rev. Sci. Instrum. 50(9): 1154-1155; 1979
assisted in the fabrication of the apparatus. M. J. Hiza
September.
participated in many fruitful discussions. H. M.
McCarty, R. D. A comparison of mathematical models for the
(15]
Ledbetter provided prepubhshed bulk modulus data prediction of LNG densities. Natl. Bur. Stand. (U.S.)
for barium ferrite. NBSIR 77-867; 1977 October. 60 p.

251

47
1 16] McCarty, R. D. Four mathematical models for the prediction 125] Clark, A. F.; Haynes, W. M.; Deason, V. A.; Trapani, R. J.
of LNG densities. Natl. Bur. Stand. (U.S.) Tech. Note Low temperature thermal expansion of barium ferrite.
1030; 1980 December. 76 p. Cryogenics 16(3): 267-270; 1976 May.
1 17] McCarty, R. D. Mathematical models for the prediction of [26] Ledbetter, H. M. Fracture and Deformation Division,
liquefied naturai gas densities. J. Chem. Thermodyn. 14(9): National Engineering Laboratory, Bureau of
National
837-854; 1982 September. Standards, Boulder, CO 80303. private communication.
|18j Hiza, M. J. An empirical excess volume model for estimating 127] Haynes, W. M.; McCarty, R. D. Prediction of liquefied-
liquefied natural gas densities. Fluid Phase Equilibria 2(2): natural-gas (LNG) densities from dielectric constant
27-38; 1978 August. measurements. Cryogenics, to be published.
1 19] Haynes, W. M.; Younglove, B. A. Dielectric constants of 128] Straty, G. C; Goodwin, R. D. Dielectric constant and
saturated liquid propane, isobutane, and normal butane. polarizability of saturated and compressed fluid methane.
Advances in Cryogenic Engineering, Vol. 27. Fast, R. W., ed. Cryogenics 13(12); 712-715; 1973 December.
New York, NY: Plenum Press; 1982. 883-891. 129) Miller, R. C; Hiza, M. J. Experimental molar volumes for

Haynes, W. M. Measurements of densities and dielectric some LNG-related saturated liquid mixtures. Fluid Phase
120]
K Equilibria 2(1): 49-57; 1978 August.
constants of liquid propane from 90 to 3(X) at pressures
to 35 MPa. J. Chem. TTiermodyn. to be published. 130] Goodwin, R. D. TTie thermophysical properties of methane,

Haynes, W. M. Measurements of densities and dielectric from 90 to 500 K at pressures to 700 bar. Natl. Bur. Stand.
[21]
isobutane from 120 to 300 K at (U.S.) Tech. Note 653; 1974 April; 274 p.
constants of liquid
pressures to 35 MPa. J. Chem. Eng. Data, to be published. 131] Goodwin, R. D.; Roder, H. M.; Straty, G. C. Thermophysical

[22] Haynes, W. M. Measurements of densities and dielectric properties of ethane, from 90 to 600 K at pressures to 700

constants of liquid normal butane from 140 to 3(X) K at

bar. Natl. Bur. Stand. (U.S.) Tech. Note 684; 1976 August.
35 MPa. Chem. Thermodyn. to be 319 p.
pressures to J.

published. 132] Rowlinson, J. S. Liquid and Liquid Mixtures. 2nd Ed. London:
Butterworths; 1969. 51.
[23] Younglove, B. A.; Straty, G. C. A capacitor for accurate wide
[33] Miller, R. C. Estimating the densities of natural gas mixtures.
range dielectric constant measurements on compressed
fluids. Rev. Sci. Instrum. 41(7): 1087-1089; 1970 July. Chem. Eng. 81(23): 134-135; 1974 October.
134] Weber, L. A. Dielectric constant data and the derived
124] Pan, W. P.; Mady, M. H.; Miller, R. C. Dielectric constants
Clausius-Mossotti function for compressed gaseous and
and Clausius-Mossotti functions for simple liquid mixtures:
liquid ethane. J. Chem. Phys. 65(1): 446-449; 1976 July.
systems containing nitrogen, argon and light hydrocarbons.
AlChE J. 21(2): 283-289; 1975 March.

252

4d
 '

6.4 Reference [13]

M-700
J, Chem. Thermodyn. Vol 9, No. 2, 179-87 (Feb 1977)

Measurements of the orthobaric liquid

densities of methane, ethane, propane,
isobutane, and normal butane

W. M. HAYNES and M. J. HIZA

Cryogenics Division, National Bureau of Standards,
Colorado 80302, U.S.A.
Institute for Basic Standards, Boulder,

{Received 9 August 1976)

The orthobaric liquid densities of the major components of natural gas have been deter-
mined with a magnetic suspension densimeter. This paper reports results for methane
(105 to 160 K), ethane (100 to 270 K), propane (100 to 288 K), isobutane (115 to 300 K),
and normal butane (135 to 3(X) K). The imprecision of the measured densities is approxi-
mately 0.015 per cent the estimated overall uncertainty is 0.1 per cent at low temperatures
;

and decreases to 0.06 per cent at 300 K. A simple expression has been used to represent
the densities as a function of temperature. Comprehensive comparisons with the experi-
mental results of other investigators are presented.

1. Introduction

Liquefied natural gas (lng) is expected to become an increasingly important com-

modity on the world energy market. The basis for sale of lng is its total heating value,
which requires a knowledge of both density and composition. A project was initiated
at this laboratory to provide orthobaric (saturated) liquid densities for the major
components of lng, and for mixtures of these components. The densities will be
used to develop a mathematical model or correlation that predicts the density of
LNG type mixtures with an inaccuracy of 0.1 per cent, given a knowledge of the
composition and temperature of the liquid. In the development of an accurate
mathematical model (correlation), it is important to have both an accurate and an
internally consistent set of density data.
Before this project was started there were significant temperature ranges for
which saturated (orthobaric) liquid density data did not exist for some of the major
components of lng. For nitrogen and methane there were discrepancies as large
as 0.5 per cent between different sets of data. Not only was it important to fill in

" This work was carried out at the National Bureau of Standards under the sponsorship of
BritishGas Corp., Chicago Bridge and Iron Co., Columbia Gas Service Corp., Distrigas Corp.,
Easco Gas LNG, Inc., El Paso Natural Gas, Gaz de France, Marathon Oil Co., Mobil R&D
Corp.,
Natural Gas Pipeline Co., Phillips Petroleum Co., Shell International Gas, Ltd., Sonatrach,
Southern California Gas Co., Tennessee Gas Pipeline, Texas Eastern Transmission Co., Tokyo
Gas Co., Ltd., and Transcontinental Gas Pipe Line Corp., through a grant administered by the
American Gas Association, Inc.

49
180 W. M. HAYNES AND M. J. HIZA
gaps, but also to provide new independent measurements of sufficient accuracy to
help resolve inconsistencies.
In this paper orthobaric liquid densities for methane, ethane, propane, isobutane,
and normal butane are reported. Results for nitrogen were presented in an earlier
paper.*" Major emphasis has been placed on the low temperature region of 105 to
140 K; however, measurements have been carried out to 160 K for methane, to
270 K for ethane, to 288.7 K for propane, and to 300 K for isobutane and normal
butane. The densities have been represented as a function of temperature with an
expression that is used to facilitate comparisons with other measurements.
The present measurements were carried out with a magnetic suspension densi-
meter.'" In this method a magnetic buoy is freely suspended in the liquid of interest
by the force generated from the axial oiagnetic fields of air-core solenoids. The
motion of the buoy is controlled by the automatic regulation of a servo-circuit.
The magnetic force necessary to maintain the buoy at a given position is inversely
proportional to the buoyant force on the buoy. Thus, using Archimedes' principle,
along with measurements of the mass and volume of the buoy, the density of the
liquid is obtained.

2. Experimental

The experimental apparatus and its operation have been described in detail else-

where.'" At low temp)eratures the experimental procedures for the measurements on

the hydrocarbons other than methane differed significantly from those for nitrogen
and methane. The density of a given fluid is determined from measurements of the
magnetic force necessary to support a barium ferrite buoy in a vacuum and in the
fluid at the same position and temperature. For nitrogen and methane the vacuum
measurements were performed imipediately before or after the liquid measurements.
At low temperatures it has been found impossible to evacuate the sample cell within
a reasonable time aifter it has been filled with one of the heavier hydrocarbons. Most
of the liquid can be removed by pressurizing with helium gas; however, a liquid
film is left on the surfaces, including those of the buoy, inside the cell. The buoy
cannot be brought into support until the film is removed. Thus, for the heavy hydro-
carbons at low temperatures, the vacuum points must be obtained before the liquid
measurements.
So that more than one point could be obtained in a given day for a heavy hydro-
carbon at low temperatures, vacuum points were obtained at two temperatures
separated by 5 K before liquid was condensed into the cell. Then the liquid measure-
ments were performed at each of these temperatures. Performance tests had demon-
strated that the barium ferrite buoy does not exhibit any detectable hysteresis at
low temperatures within the precision of the current measurements for a temperature
range of, at least, 20 K. After the vacuum and liquid measurements were performed for
one of the heavy hydrocarbons at low temperatures the cell was warmed to a tem-
perature above the normal boiling temperature of the test fluid for evacuation.
Methane was used as a control fluid during the heavy-hydrocarbon measurements
at low temperatures. Each day a new methane point was taken to insure that the

bO
 LIQUID DENSITIES OF CH«, QHo, C3H,, i-CMio, AND w-C^Hjo 181

warm-up and cool-down of the apparatus did not affect the apparent position of the
buoy from liquid-to-vacuum measurements. The position of the buoy was deter-
mined with a high-powered microscope that had been adjusted initially so that the
jipparcnt position of the buoy did not >Jepend on the index of refraction of the fluid
msidc the cell. It was found that the temperature cycling of the cell had no detectable
effect on the apparent buoy position.
Some of the results presented in this paper were taken with a one-coil system
instead of the three-coil arrangement described in the apparatus paper.*** The evolu-
tion to the use of only one coil is discussed in reference 2.

All of the gases were of research grade quality. The minimum purities as specified
by the suppliers were 99.9 moles per cent for isobutane and normal butane and
99.99 moles per cent for methane, ethane, and propane. The gases were analyzed
chromatographically with a thermal-conductivity detector and found to be within
the specified purities except for isobutane. It was found that the isobutane contained
approximately 0. 1 5 per cent of normal butane. This relatively large amount of normal-
effect (< 0.01 per cent) on the density results.
butane impurity has a negligible
The methane gas was passed through a room-temperature molecular-sieve trap
to remove moisture and any heavy contaminants not detected by analysis. The
other hydrocarbon gases were normally not passed through a molecular-sieve trap.

3. Results

The experimental orthobaric liquid densities of methane, ethane, propane, isobutane,

and normal butane are presented as a function of temperature (IPTS-68) in tables
1 to 5. The relatively large number of points for methane at any given temperature

resulted from the use of methane as a control fluid throughout the project. Although
the mean experimental densities of methane have been given in an earlier paper***
they are presented again here, along with other information (calculated densities,
etc.), so that the orthobaric liquid densities of all the low molecular-weight alkanes

TABLE I. Orthobaric liquid densities of methane, where F is the temperature (IPTS-68), /7„pt is
the mean experimental density for n observations at a given temperature, />c»ic is the density
calculated from equation (1) and is the largest value of (pe^pt Poaic)IPct\c —

r/K /'«pt/mol dm"' n Pc.ic/moI dm"^ lOMn...

105.000 26.9458 12 26.9456 0.030

110.000 26.4985 11 26.5005 0.035
115.000 26.0443 12 26.0429 0.028
120.000 25.5721 17 25.5712 0.036
125.000 25.0845 18 25.0839 0.037
130.000 24.5775 16 24.5790 0.036
135.000 24.0540 23 24.0540 0.031
140.000 23.5067 5 23.5060 0.024
145.000 22.9312 5 22.9311 0.024
150.000 22.3218 2 22.3243 0.014
160.000 20.9876 2 20.9857 0.037

5i
182 W. M. HAYNES AND M. J. HIZA

TABLE 2. Orlhobaric liquid densities of ethane, whcrv T is the temperature (llTS-bS), is

the experimental density, /i„.i„ is the density calculated from equation (I), and A is the value
of (/'.«Pt - /'o»lo)//'c»lo

T /'•Kpt T
10^.1
K mol dm"^ mol dm"^ K mol dm"^ mol dm " ^
100.000 21.3408 21.3388 0.009 170.000 18.6867 18.6869 -0.001
105.000 21.1585 21.1568 0.008 180.000 18.2793 18.2787 0.003
110.000 20.9746 20.9742 0.002 190.000 17.8612 17.8586 0.015
115.000 20.7927 20.7907 0.010 200.000 17.4289 17.4240 0.028
120.000 20.6022 20.6063 -0.020 210.000 16.9713 16.9720 -0.004
125.000 20.4186 20.4208 -0.011 220.000 16.4988 16.4989 -0.001
130:000 20.2317 20.2343 -0.013 230.000 15.9973 15.9994 -0.013
135.000 20.0461 20.0466 -0.002 240.000 15.4642 15.4670 -0.018
140.000 19.8566 19.8575 -0.005 250.000 14.8899 14.8919 -0.013
150.000 19.4751 19.4748 0.002 260.000 14.2610 14.2598 0.008
160.000 19.0857 19.0850 0.004 270.000 13.5493 13.5477 0.012

TABLE 3. Orthobaric liquid densities of propane, where T is the temperature (lPTS-68), pt^^x. is
the experimental density, Pc&ic is the density calculated from equation (1), and A is the value
of (/7sxpt — Pcalc)IPc&\c

T PaxpC Poalo T Pcalc

10^ A lO^J
K mol dm"^ mol dm"^ K mo! dm~^ mol dm"^

100.075 16.3065 16.3048 0.011 140.075 15.3751 15.3755 -0.002

105.075 16.1872 16.1885 -0.008 145.075 15.2588 15.2590 -0.002
110.075 16.0718 16.0723 -0.003 150.075 15.1400 15.1424 -0.016
115.075 15.9557 15.9562 -0.003 200.000 13.9560 13.9524 0.026
120.075 15.8411 15.8401 0.006 240.000 12.9271 12.9285 -0.010
125.075 15.7250 15.7241 0.006 270.000 12.0733 12.0742 -0.008
130.075 15.6085 15.6080 0.003 280.000 11.7622 11.7631 -0.008
135.075 15.4910 15.4918 -0.005 288.706 11.4790 11.4775 0.013

TABLE 4. Orthobaric liquid densities of isobutane, where T is the temperature (IPTS-68), pexpi is
the experimental density, Pc«io is the density calculated from equation (1), and A is the value
of (/>„pt —
Pc»lc)//'calc

T Pnpt T
K mol dm"' mol dm"^
WA K
Poalc

mol dm"' mol dm"'

10^ J

115.075 12.7305 12.7313 -0.006 145.075 12.2353 12.2372 -0.015

120.075 12.6489 12.6491 -0.001 150.075 12.1534 12.1544 -0.008
125.075 12.5687 12.5669 0.015 228.000 10.8273 10.8263 0.009
130.075 12.4850 12.4846 0.003 288.706 9.6676 9.6687 -0.012
135.075 12.4015 12.4022 -0.005 290.000 9.6411 9.6417 -0.007
140.075 12.3215 12.3197 0.014 300.000 9.4300 9.4287 0.014
 LIQUID DENSITIES OF CH4. CaHe, CaHe, /-QHio. AND n-QHio 183
TABLE 5. Orthobaric liquid densities of normal butane, where T is the temperature (IPTS-68),
/^xpt is the experimental density, />c»io is the density calculated from equation (1), and A is the value
of (Pexpt — Poalo)//'o»lo

T T
10= J
K mol dm'^ mol dm"^ K mol dm"' mol dm"'

135.075 12.6517 12.6524 -0.005 165.075 12.1634 12.1659 -0.020

140.075 12.5706 12.5714 -0.006 170.075 12.0839 12.0846 -0.005
145.075 12.4920 12.4904 0.013 230.000 11.0911 11.0905 0.005
150.075 12.4089 12.4093 -0.003 288.706 10.0325 10.0324 0.001
155.075 12.3299 12.3283 0.014 290.000 10.0067 10.0073 -0.007
160.075 12.2484 12.2471 0.010 300.000 9.8103 9.8099 0.004

investigated in the present work are included in a single paper. All of the experimental
points for methane will be presented in a future report/^' Each methane point was
taken from a new filling of the cell. For the other hydrocarbons no more than two
points were taken from a single filling.
The experimental densities p have been fitted as a function of temperature T to
the expression:

(p-p,)/mol dm-3 = a(l-T/TJ°-^' + ^ fe/l-T/r,)"+<'-^>/3', (1)

i= 1

which incorporates a scaling-law modification^*^ to a generalized Guggenheim

equation.*'* The coefficients o, f); determined by least squares, and selected values
of the critical teinperature 7^ and density for each fluid are given in table 6.*^"*^'
Only three coefficients were needed to fit the results for methane, which covered a
relatively small temperature range compared with that for the other fluids.

The residual standard deviations of the fit io equation (1) for each fluid are given
in table 6. These values substantiate the estimate of the imprecision of the density
measurements, which is approximately 0.015 per cent. The estimated inaccuracy in

TABLE 6. Parameters of equation (1),

(p - /)„)/moi dm"' = fl(i - r/re)"'" +i bi\ - 7/r„)<>

where Tc and Po are the critical temperature and density. The coeflRcients a, bi, fta, were obtained
from a least-squares program in which the experimental mass densities to five digits were converted
to molar densities within the program. The standard deviations a and molar masses are also given M
Methane Ethane Propane Isobutane Norma) butane

a 18.65812 12.55205 8.684459 7.657535 7.286063

h, 6.712030 13.43284 18.04086 8.145251 11.96308
b, -0.9472020 -19.00461 -29.46261 -13.10582 -19.87592
11.07716 16.43559 8.145894 11.60211
TJK 190.555'"' 305.33 <»' 369.82 <«' 408.13<"> 425.16<>»
dm"' 10.16<''> 6.86<»> 5.00"«' 3.80"°> 392(11)
Pc/mo]
0.013 0.012
10V<P> 0.016 0.012 0.011
M/g mol " *
16.04303 30.07012 44.09721 58.1243 58.1243

53
184 W. M. HAYNES AND M. J. HIZA
the densities is O.I per cent at low temperatures and decreases to 0.06 per cent at

300 K. The total uncertainty in the reported temperatures is estimated to be less than
30 mK at 100 K and less than 40 mK
at 300 K. These uncertainty limits in the
temperature correspond to a maximum uncertainty of 0.02 per cent in the density
for the results reported in this paper. A detailed error analysis of the magnetic
suspension densimeter used in the present work has been given elsewhere.'"
Equation (1), along with the parameters given in table 6, has been used for com-
'^-25) Deviation
parisons of the present results with independent experimental data.'*'
plots for ethane, propane, isobutane, and normal butane are presented in figures i

0.8 1
r
e
0.6 V e

e e
ciT 0.4

f 0.2

-0.2

-0.4
100 140 180 220 260
T/K

FIGURE 1. Deviation plot of experimental orthobaric liquid densities of ethane compared with
values calculated from equation (1) using parameters from table 6. O, present results; •, Shana'a
and Canfield;"^' O, Chui and Canfield;'"' ©, Orrit and Olives;'"' A Rodosevich and Miller;'^^',
, McClune;'^*' ,
Klosek and McKinley;*^''' A, Douslin and Harrison;''*' 0, Maass and
Wright ;"»' V, Jensen and Kurata;'^''* T, Kahre.'">

through 4. The deviation plot for methane was presented in an earlier paper
thus, it is not included here.
In comparing the results of other investigators with equation (1) some general
trends are observed. Below 140 K
the densities of Shana'a and Canfield,"^' Chui
and Canfield,"^> Orrit and 01ives,'^*> Rodosevich and Miller," McClune,"''' and
Klosek and McKinley'*'^ are generally lower than the present results by 0.05 to
0.1 per cent. Exceptions to this trend are: for isobutane the densities of Rodosevich
and Miller"*' between 114 and 120 K are larger (maximum of 0.1 per cent) than
the present results and exhibit a significantly different temperature dependence;
and for ethane the change in density with temperature reported by Klosek and
McKinley"'' is appreciably larger than that observed in the present work.

54
 LIOUID DLNSirilS Ol- CH,. C.Ho. CM.,. /-C,H,„. AND //-t .H,., IS5

0.4V t
0.4
-e-T" T
0.3

0.2 -
•• • V
^0.1

5 0 0 0 n ^ o-o —
Pn
o-®^
°
~0 JT

; * " o 0
a "a® A
-0.1 ^0 0° DO

-0.2

100 140 180 220 260

T/K
FIGURE 2. Deviation plot of experimental orthobaric liquid densities of propane compared
with values calculated from equation (1) using parameters from table 6. O, present results;
*, Siiwinski;'8' •, Shana'a and Canfield;''^' (D, Orrit and 01ives;<"* A, Rodosevich and Miller;""'

n. McClune;"«' ,
KJosek and McKinley;'"' 0, Maass and Wright;"^' V, Jensen and Kurata;""'
, Kahre;'"' Tomlinson;'"' ^, Seeman and Urban;'"' ®, NGAA;'"'; C, Van der Vet.'"'

0.3

-
J 0.2
c «

5i: 0 .1 -
• •
• •
• •
e
-O O-j,
o
g 5^
® d»a
-0.1 -

120 160 200 240 280

T/K
FIGURE 3. Deviation plot of experimental orthobaric liquid densities of isobutane compared
with values calculated from equation (1) using parameters from table 6. O, present results;
•, Sliwinski;'*' ®, Orrit and Olives;'"' A, Rodosevich and Miller;'"' Mcaune;'"' T, Kahrc;"**' ,
®, NGAA;'"' €), Van der Vet."""

55
 186 W. M. HAYNES AND M. J. HIZA

0.3 1
r

0.2

O.i -

-0.1

-0 2' I \ I I I \ i : \ \ i

120 160 200 240 280

r/K
'
FIGURE 4. Deviation plot of experimental orthobaric liquid densities of normal butane compared
with values calculated from equation (1) using parameters from table 6. O, present results;
•, Sliwinski;'^' ®, Orrit and Olives;"*' ,
McClune;''«' Kahre;'^!' ®, NGAA;'"' C, Van
der Vet.<"'

Above 140 K and Canfield/^^^ Orrit and Olives/'*^ and

Ihe data of Chui
McClune^'^' from the present results by less than 0.05 per cent. At higher
differ
temperatures (above 280 K) the data of Sliwinski^^* for propane, isobutane, and
normal butane generally differ from the present results by less than 0.05 per cent.
The orthobaric liquid densities reported by Douslin and Harrison*'^* for ethane at
temperatures between 248 and 263 K were systematically larger than the present
results by 0.3 to 0.35 per cent. Some of the cider less precise (but frequently used)
data have been included on the deviation plots for the sake of completeness.
Ahhough Klosek and McKinley^'^' give densities for isobutane and normal
butane at temperatures between 105 and 133 K, these densities are not experimental,
and therefore, are not plotted on figures 3 and 4. A few comments on the reliability
of the omitted values are appropriate. Their densities were obtained from the Francis
equation. ^^^^ For normal butane at temperatures below its triple-point temperature,
their results were systematically higher by less than 0.2 per cent than the densities
obtained through extrapolation of the present results. ^'^^^ However, their isobutane
densities at temperatures between 116 and 133 K were 1.5 per cent higher than those
of the present work.

4. Summary
This research has provided accurate and self-consistent measurements of the ortho-
baric liquid densities of methane, ethane, propane, isobutane, and normal butane
at temperatures down to 100 K. Most of the measurements recently reported by
other workers differ from the present results by less than 0.1 per cent. In subsequent
papers, density measurements on liquefied mixtures of the major components of
liquefied natural gas will be reported.

66
 LIQUID DENSITIES OF CH4, CzHe. Cjhi„, /-C,H,o, AND /;-C,H,„ 187

The authors would like to acknowledge R. D. McCarty and M. J. Brown for assistance
with the correlation and reduction of the data.

REFERENCES
1. Havncs, W. M.; Mi/a, M. J.; Frederick, N. V. Rev. Sci. Instnini. 1976, 47, 1237.
2. Hayncs, W. M. Rev. Sci. Instrtiin. In press.
3. Mcihaiie data arc available from authors upon request and will be published in a Nat. Bur.
Stand. (U.S.) report.
4. McCarty, R. U. Nui. Bur. Stand. (U.S.), Internal Rept. 1974, 74-357.
5. Hou, Y. C; Martin, J. J. A.I.CIi J. 1959, 5, 125.
6. Goodwin. R. D. Nat. Bur. Siaiul. (U.S.), Tech. Note 653, 1974.
7. Olson, J. D.
C/iem. Pliys. 1975, 63, 474.
J.
8. Sliwinski, P. Z. Phys.Chem. (Frankfurt) 1969, 63, 263.
9. Das, T. R.; Eubank, P. T. Advances in Cryogenic Engineering Vol. 18. Timmerhaus, K. D.:
editor. Plenum Press: New York. 1973, p, 208.
10. Das, T. R.; Reed, Jr., C. O.; Ei.-bank, P. T. J. Chem. Eng. Data 1973, 18, 253.
11. Das, T. R.; Reed, Jr., C. O.; Eubank, P. T. J. Chem. Eng. Data 1973, 18, 244,
12. Shana'a, M. Y.; Canfield, F. B. Trans. Faraday Soc. 1968, 64, 2281.
13. Chui, C. H.; Canfield, F. B. Trans. Faraday Soc. 1971, 67, 2933.
14. Orrit, J.; Olives, J. distributed at 4th International Conference on Liquefied Natural Gas.
Algeria. 1974.
15. Rodosevich, J. B.; Miller, R. C. A.I.Ch..E.J. 1973, 19, 729.
16. McClune, C. R. Cryogenics 1976, 16, 289.
17. Klosek, J.; McKinky, C. Proc. 1st Int. Conf. on LNG, Paper 22, Chicago, 1968.
18. Douslin, D. R.; Harrison, R. H.J. Chem. Thermodynamics 1973, 5, 491.
19. Maass, O.; Wright, C. H. J. Am. Chem. Soc. 1921, 43, 1098,
20. Jensen, R. H.; Kurata, F. J. Petrol. Technol. 1969, 21, 683.
21. Kahre, L. C. J. Chem. Eng. Data 1973, 18, 267.
22. Tomlinson, J. R. Natural Gas Processors Assoc. Tech. Publ. TP-I, Tulsa, Oklahoma, 1971.
23. Seeman, F.-W.; Urban, M. Erddl und Kohle Erdgas Petrochem. 1963, 16, 117.
24. Tech. Conmt., Natl. Gas. Assoc. Am. Ind. Eng. Chem. 1942, 34, 1240.
25. Van der Vet, A. P. Congress Mondial du Petrol., Paris, Vol. II, 1937, p. 515.
26. Francis, A. W. Ind. Eng. Chem. 1957, 49. 1779.
27. Haynes, W. M.; Hiza, M. J. Advances in Cryogenic Engineering Vol. 21. Timmerhaus, K. D.;
Weitzel, D. H.; editors. Plenum Press: New York. 1976, p. 516.

57
6.5 Reference [14]

516-21^(1976)''^°^^"^''
Enpineerinp .
Vol 21. Plenum Press. New York.

ORTHOBARIC LIQUID DENSITIES OF

NORMAL BUTANE FROM 135 TO 300 K AS
DETERMINED WITH A MAGNETIC
SUSPENSION DENSIMETER*
W. M. Haynes and M. J. Hiza

Cryogenics Division, Institute for Basic Standards, Boulder, Colorado

INTRODUCTION
Of
the principal constituents of natural gas, normal butane is the first in the
series of paraffin hydrocarbons that has a triple-point temperature significantly
higher than the normal boilingpoint (1 12 to 1 15 K) of methane-rich liquefied natural
gas (LNG). Unlike isobutane, with a triple-point temperature (113.6 K) near the
normal boiling point of methane-rich LNG, normal butane freezes at the relatively
high temperature of 134.8 K. Thus any estimate of the contribution of n -butane
content to the molar volume (or molar density) of LNG requires a relatively long
extrapolation into the subcooled liquid region of n-butane. It is also known that
r-butane is the first in the series of paraffin hydrocarbons exhibiting geometrical
isomerism, with nearly instantaneous equilibrium, which contributes to the tempera-
ture dependence of the molar density ['"^]. It follows that an analytical expression
that provides the most reliable means of extrapolating n-butane densities into the
subcooled liquid region cannot be based on correspondence alone, but must be based
on extensive and accurate data above the triple-point temperature.
Other than references [^'^] that report results of saturated liquid density
measurements near room temperature, no references were found that report experi-
mental orthobaric liquid densities of n-butane below 283 K. (The results given by
Klosek and McKinley at temperatures below the triple-point temperature are
probably based on their own data at somewhat higher temperatures.) The purposes
of the present study were to obtain orthobaric liquid densities for n-butane that
cover a relatively large temperature range at low temperatures, as well as some
measurements up to ambient temperature, and to provide an analytical representa-
tion of these data that is used for extrapolation into the subcooled liquid region to
compare with other results [%
A
relatively new technique for determining absolute densities of cryogenic fluids
['] was used in the present work. It is simple in principle and is based on a

straightforward application of Archimedes' principle. Using a magnetic suspension

* This work w;is c;irrii'il out Ht the N.ition;il Bureau of Sl.indards under the sponsorship of the I.NCi
FX-nsity I'rojoit .Sli-«-iin>' ( omniiitcc, ihroiij'.h a I'.rant adniini'itercd by llic Anicricin (i ts Associ.ilinn.
Inc

516

69
 Orthobaric Liquid Densities of Normal Butane from 135 to 300 K 517

technique, a cylinder of magnetic material of known mass and volume is freely

suspended by air-core solenoids in a stable configuration in the fluid whose density is
to be determined. At such a stable equilibrium position, the upward magnetic force is
added to the buoyancy force to balance the gravitational force. Densities of the fluid
can be determined directly from the measured currents in the coils necessary to
support the magnetic "float" at the same position in vacuum and in the fluid at
constant temperature. The vertical position of the float is controlled by the automatic
regulation of a servo-circuit, which includes a linear differential transformer for
sensing inductively the position of the float. The horizontal position of the float is

determined and maintained by the divergence of the magnetic field intensity.

EXPERIMENTAL
A complete description of the apparatus and a detailed discussion of the
performance characteristics are beyond the scope of this paper and will be published
elsewhere ['°]. Only those essentials of the experiment necessary for clarity are
included here.
The feasibility of using a magnetic suspension densimeter for measurements on
cryogenic fluids had been demonstrated earlier with an absolute density measure-
ment on liquid nitrogen at the normal boiling point The development of the
present apparatus, shown in Fig. 1, was based on experience with that apparatus and
incorporates a cryostat with continuous wide-range temperature control, a win-
dowed equilibrium cell suitable for studies of the liquids and liquid mixtures of
interest here, a new servo-circuit for vertical position control of the magnetic buoy,
and a high-powered (125 x ) microscope for monitoring the position of the magnetic
buoy.
The magnetic buoy is a barium ferrite (BaFe]20i9) ceramic magnet in the shape
of a right circular cylinder, approximately 5 mm
in diameter and 6 mm
in length. It
has been plated with a thickness of approximately 0.07 mm
of copper over its entire
surface to prevent fluid from penetrating into the porous ceramic. Gold was flashed
over the copper surface. The mass of the buoy was determined to better than 0.002%
with an analytical balance. The volume of the buoy was determined within 0.02% at
room temperature by using distilled water as a reference fluid of known density. At
temperatures below ambient, the change in volume was calculated using recent
thermal expansion data obtained for barium ferrite ["].
Evaluation of the experimental parameters involved in this method and exten-
sive performance tests indicate that the uncertainty in the density measurements
depends primarily on the uncertainty in the determination of the position of the
buoy, relative to the main coil, from vacuum to liquid measurements. The position of
the float was determined to less than lO""* mm
with a 125 x microscope. The position
of the main coil was maintained from vacuum to liquid measurements by supporting
the coil at its midplane with quartz rods and by controlling the temperature of the coil
with water.
The total uncertainty in the density measurements was estimated to be less than
0.1% at the lower temperatures; near room temperature the absolute error should
be less than 0.05%. The imprecision of the density measurements was less than
0.02% over the complete temperature range.
A platinum resistance thermometer calibrated on the International Practical
Temperature Scale (1968) was used for the temperature measurements. The uncer-
tainty of the calibration was approximately 0.002 K. Due to the specifications of the

bu
SIS W. W. Haj-nes and M. J. Hiza
 Orthobuic Liquid Densities of Normal Butane from 135 to 300 K 519

potentiometric measuring system, uncertainties in the temperature amounted to a

maximum of 0.0135 K at 135 K, increasing to 0.030 K at 300 K. The temperature
of the sample holder was controlled to better than 0.005 K, approximately the same
as the reproducibility of the temperature measurements.
The purity of the n-butane used in these experiments was determined by
chromatographic analysis with thermal conductivity detection. 7.5-m liquid A
partition column was used, capable of separating air (or nitrogen) from methane in
small concentrations and the paraffinic from olefinic hydrocarbons. The only detect-
able impurity in this sample was air (or nitrogen) at approximately 25 ppm. The limits
of detectability of the analyzer were approximately 20 ppm of air (or nitrogen) and
methane and approximately 10 ppm of ethane and the higher hydrocarbons.

RESULTS
The experin:ental densities for n -butane are given in Table I with the corre-
sponding experimental temperatures. These data were fitted by the method of least
squares to an equation of the form

(1)

where p is the density in moles/liter, is a characteristic density in moles/liter-

(approximately the critical density), T<. is a characteristic temperature in kelvin
(approximately the critical temperature), n is defined as l+5(/ — 1), and a, b, are
coefficients determined by least squares. The parameters of (1) also are given in
Table I, as well as the standard deviation and the molecular weight.
The deviations of various experimental results [*'^] and several frequently used
[^^"'•*]
literature values from the values calculated from (1) are plotted as a function
of temperature in Fig. 2. The accuracy with which equation (1), with parameters as

Table I. Densities of Saturated Liquid Norma] Butane and

Parameters of Equation (1)

Peak. 100(p„p- Pcalc)

moles/liter T, K moles/liter

12.6517 135.075 12.6524 -0.005

12.5706 140.075 12.5714 -0.006
12.4920 145.075 12.4904 0.013
12.4039 150.075 12.4093 -0.003
12.3299 155.075 12.3283 0.014
12.2484 160.075 12.2471 0.010
12.1634 165.075 12.1659 -0.020
12.0839 170.075 12.0846 -0.005
11.0911 230.000 11.0905 0.005
10.0067 290.000 10.0073 -0.006
9.8103 300.000 9.8099 0.005

Standard deviation 0.01 3% T^ = 425.0 K

a - 7.4673 W332 = 3.876 moles/liter
- 1I.25V63703 Molecular weight 58.12430
hi - -18.03664709
/j-, - 11 22771351

6Z
520 \V. Nf. Haynes and IM. J. Hiza

o It l llllIll|ltlN
—mu r
^
I

• il'IOlll O flllCII

• Jr

111 ];i Ill i?i

ntmnm. KitriaiiK. I

Fig. 2. Deviation plot of densities of Fig. 3. Deviation plot of

saturated liquid normul butane com- densities of liquidnormal
pared with values calculated from butane Klosek
of and
equation (1 ); data from this work and McKinley C] below the
the (Note that
Iiterature[*''-'^'''*]. triple-point temperature
results from references 12 through 14 compared with the extrapo-
are not experimental data.) lated values calculated from
equation (1) for two cases:
(a) a least squares fit of the
data from the present work
from 135 to 170 K using
two coefficients and (b) a
least squares fit of the data
from the present work from
135 to 300 K using four
coefficients. (Note that the
baseline represents two sets
of extrapolated results
obtained from the present
data and that both curves
with designated points rep-
resent a single set of
values from Klosek and
McKinley ['].)

given in Table I, can be extrapolated above 300 K has not been investigated, but
from the comparisons with the work of Sliwinski [^] and Kahre C] it appears that the
equation gives reasonably accurate values up to 370 K.
Two different methods of extrapolation to temperatures below 135 K are
presented in Fig. 3. First, the present data from 135 to 170 K were fitted to equation
(1) using two coefficients (a = 7.318406322 and f>i = 3.479543073) and then
densities were calculated for temperatures below the triple point and compared with
the values of Klosek and McKinley Next, the present data from 135 to 300 K
were fitted, using equation (1) with four coefficients, and again comparisons were
made with the values of Klosek and McKinley. The graph demonstrates the
significant differences obtained for the two extrapolation methods and also provides
comparisons with the only other published density values for normal butane at
temperatures below 283 K. It should be noted that no effort has been expended in
the present work to produce a "best" method for extrapolation into the subcooled
liquid region.

CONCLUSIONS
The results of this study provide the first set of experimental density data for the
orthobaric liquid phase of n-butane at temperatures between 135 and 300 K. The

b3
 Orthobaric Liquid Densities of Normal Butane from 135 to 300 K 521

absolute error is estimated to be less than ±0.1%. The analytical expression selected
represents these data with a standard deviation less than the estimated imprecision of
measurement, i.e., ±0.02%. It is felt that the analytical expression given here

provides a reasonable means of extrapolation of n-butane densities into the sub-

cooled region for liquefied natural gas applications.

ACKNOWLEDGMENTS
The authors wish to acknowledge the contributions of the LNG Density Project Steering Committee
and the financial support of the companies they represent.

REFERENCES
1. R. A. Bonham and L. S. Bartell, J. Am. Chem. Soc. 81:3491 (1959).
2. R. A. Bonham, L. S. Bartell, and D. A. Kohl, J. Am. Chem. Soc. 81:4765 (1959).
3. A. L. Verma, W. F. Murphy, and H. J. Bernatein, /. Chem. Phys. 60:1540 (1974).
4. M. R. Cines, Phillips Petroleum Co., private communication (January 1974).
5. T. W. Schmidt, Phillips Petroleum Co., private communication (May 1974).
6. P. Sliwinski. Z. Phys. Oiem. (Frankfurt) 63:263 (1969).
7. L. C. Kahre, / Chem. Eng. Data 18:267 (1973).
8. J. Klosek and C. McKinley, in: Proceedings of First Intern. Conference on LNG, Chicago, Illinois
(1968), Paper 22, Session 5.
9. W. M. Haynes and J. W. Stewart, Rev. Sci. Instr. 42:1142 (1971).
10. W. M. Haynes, N. V. Frederick, and M. J. Hiza, unpublished dat:i.
11. A. F. Clark, \V. M. Haynes, V. A. Deason, and R. J. Trapani, Cryogenics (to be published).
12. A. W. Francis, Ind. Eng. Chem. 49:1779 (1957).
13. Natural Gas Processors Supplies Association, Engineering Data Book, Ninth Edition, NGPA, Tulsa,
Oklahoma (1972).
14. American Petroleum Institute, Project 44, Selected Values of Physical and Tliermodynamic Properties
of Hydrocarbonsand Related Compounds, Carnegie Press, Pittsburgh, Pennsylvania (1953).

DISCUSSION
Question by R. C. Hendricks, NASA Lewis Research Center: Have you \ erified the results from your
primary platinum temperature sensor by comparison with a second platinum temperature sensor? If not,
how did you assure yourself that the temperature values obtained in your study were accurate?
Answer by author: No. Vapor pressures of pure methane were measured in the vapor bulb at the top
of the cell and in the equilibrium space to determine the reliability of temperature measurement and
control, as well as departures from isothermal conditions in the equilibrium cell.

04
6.6 Reference [16]

M-754
/. Chem. Thermodynamics 1977, 9, 873-S96

Orihobaric liquid densities and excess

volumes for binary mixtures of low molar-
mass alkanes and nitrogen between
105 and 140 K*

M. J. mZA, W. M. HAYNES, and W. R. PARRISH

Cryogenics Division, Institute for Basic Standards,
National Bureau of Standards, Boulder, Colorado 80302, U.S^.

{Received 21 February 1977)

A magnetic suspension densimeter has been used to determine ortbobaric liquid densities
of gravimetrically prepared binary mixtures of the major components of liquefied natural
gas (LNG) i.e. nitrogen, methane, ethane, propane, i-butane, and n-butane, generally
between 105 and 140 K. All binary combinations were included in this study, with the
exception of nitrogen +i-butane and nitrogen +
/r-butane. Uncertainties in the reported
liquid-mixture densities are discussed in detail. Comparisons are made between excess
volumes computed from the present results and coirnarabie values from the literature.
It was found that the volumetric properties of binary liquid mixtures of the heavy hydro-
carbons (those mixtures not containing nitrogen or methane) are closely approximated
by ideal mixing. Some observations are included on the use of excess volumes of the heavy
hydrocarbon systems to determine effective molar volumes of n-butane in liquid mixtures
below its triple-point temperature. For mixtures containing nitrogen or methane, approxi-
mate total vapor pressures are given.

1. Litroduction

Custody transfer of liquefied natural gas (LNG) on the world energy market has
imposed one of the most stringent requirements for accuracy in the prediction of
orthobaric (saturated) liquid mixture densities. Since heating value is directly related
to the density and composition of the be able to predict LNG
liquid, it is desirable to
densities within a small known uncertainty, preferably within ±0.1 per cent, for any
condition encountered in commerce.
The most promising theoretical methods for predicting the properties of fluid
mixtures require that the pure-component characteristic parameters be combined to
give the mixture characteristic parameters of the included binary pairs. It is well
known that small adjustments to these parameters can significantly affect the difference
" This woric was carried out at the National Bureau of Standards under the sponsorship of British
Gas Corp., Oiicago Bridge and Iron Co., Columbia Gas Service Corp., Distrigas Corp., Easco Gas
LNG, Inc., El Paso Natural Gas, Gaz de France, Marathon Oil Co., MobQ Oil Corp., Natural Gas
Pipelkie Co., Phillips Petroleum Co., Shell International Gas, Ltd., Sonatrach, Southern California
Gas Co., Tennessee Gas Pipeline Co., Texas Eastern Transmission Co., Tokyo Gas Co., Ltd., and
Transcontinental Gas Pipe Line Corp., through a grant administered by the American Gas
Association, Inc.
874 M. J. HIZA, W. M. HAYNES, AND W. R. PARRISH

between predicted and actual mixture properties. Though reasonable estimates of

deviations from the combining rules can be made in special cases/^^ the best values
for a given application and model must still be derived from accurate and consistent
experimental results at several temperatures for the binary mixtures and pure com-
ponents of interest
For the development of a model to predict LNG densities reliably, for example,
one needs suflBciently accurate densities of uniformly high precision for the pure
components and the possible binary combinations, as well as for selected multi-
component mixtures, to test and optimize the prediction method. To provide this
data base, a comprehensive study was initiated in our laboratory to obtain orthobaric
liquid densities of the desired accuracy for the major components of LNG and for
their mixtures.
Previously, orthobaric liquid densities were reported for nitrogen^ and the low
molecular-weight alkanes —methane,^^'^^ ethane, propane, /-butane, and n-butane^''^
—as determined with a magnetic suspension densimeter.
The present study was conducted with the same apparatus to obtain orthobaric
liquid densities for gravimetrically prepared binary mixtures of these components,
generally between 105 and 140 K. All of the possible binary systems were studied,
with the exception of nitrogen -f- and nitrogen + «-butane. Based on the
/-butane
phase equilibria for nitrogen + and nitrogen + propane^'^ it was estimated
ethane^*^
that the limits of miscibility for the nitrogen + butane systems would preclude the
possibility of obtaining experimental densities for these systems in the temperature
range of interest Where possible, total vapor pressures were also determined, but these
are considered approximate since the main focus was to assure the reliability of the
density measurements.
Excess volimies, computed for each of the binary mixtures studied, show that
mixtures of the heavy hydrocarbons exhibit nearly ideal mixing. Effective liquid-
phase molar volumes for «-butane below its triple-point temperature were obtained
from the ethane 4- n-butane liquid-mixture densities by assuming that the small
excess volume remains constant in the temperature range studied.
Only a few sets of orthobaric liquid-mixture densities in the literature^*^ provide
data applicable to the LNG problem discussed above. These data are generally
Umited in either the temperatures or compositions covered, or have admitted in-
accuracies significantly larger than the desired ±0.1 per cent. Comparisons are made
between excess volumes from the present study and comparable values taken from
the literature. Prior to the present study, there were no liquid densities in the tempera-
ture range of interest for nitrogen + ethane, nitrogen + propane, ethane -H i-butane,
propane + z-butane, propane + n-butane, or i-butane -I- w-butane.

2. Experimental
The magnetic suspension densimeter used in this study was discussed in detail else-
where.^^* All of the results reported here were obtained with the one solenoid
arrangement^'^ With this arrangement a barium ferrite magnetic buoy of known
mass and volume is freely suspended, in vacuum and in the liqmd, by the force

66
 UQUm DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 875
generated from the axial magnetic field of a single air-core solenoid. The position

of the buoy controlled by automatic regulation of the solenoid current with a

is

radio-frequency servo-circuit Since the magnetic force necessary to support the buoy
at a given position is dependent on the buoyant force, measurement of the solenoid
current needed to support the buoy in the liquid relative to vacuum at the same
position and temperature yields the liquid density directly.
The relation used to compute the density of the liquid with the one solenoid
system is

Pi-(mjn){l-(/i//r)}. (1)

where pi is the mass density of the Kquid, is the mass of the buoy, V^, is the volume

of the buoy, and /, and /, are the solenoid currents required to support the buoy at
the same position in the liquid and vacuum, respectively.
The procedures used in this study were basically the same as those used in measuring
liquid densities of the pure fluids.^^*^*^^ Vacuum measiu:ements were made im-
mediately before or after those for nitrogen 4- methane liquid-mixture points, and
each measurement was made with a separate fill. Since the temperature of the equi-
librium cell had to be increased nearly to room temperature for effective removal
of the heavy hydrocarbons, it was necessary to make vacuum measurements before
introducing mixtures containing the heavy hydrocarbons. Measurements for these
mixtures generally were made in pairs at 5 K
increments with a single fill. With few
exceptions a liquid-methane density measurement was made each day as a control
on the measurement parameters. Liquid methane was used as the vapor-bulb fluid
to provide the criteria for adjusting the control heat to the top and bottom of the
cell to minimize temperature gradients. The heater currents were adjusted so that the

methane vapor pressure was consistent with the temperature (IPTS-68) determined
with a platinum resistance thermometer mounted near the bottom of the cell. Vapor
bulb and equilibrium cell pressures were measured simultaneously with quartz
Bourdon-tube pressure gages (0 to 0.69 MPa) calibrated against an air dead-weight
gage. The estimated maximum uncertainty in calibration of these gages is about
±70 Pa.
Since the imcertainty in the measured density of a binary liquid mixture is inherently
larger than for the pure fluids due to the added uncertainty in composition, pre-
cautions were taken to assure that the uncertainty in composition was minimized and
that the composition in the equilibrium cell was homogeneous.
Basically, there are three options available to fix the composition of a liquid
mixture to be studied experimentally. These are (1) liquid-phase sampling and analysis
relative to calibration mixtures, (2) introducing known quantities of each pure com-
ponent and mixing within the experimental chamber, and (3) introducing the desired
mixture, prepared under carefully controlled conditions, into the equilibrixmi chamber
in the mixed state. Each method, or the various combinations, requires an exact
accounting of the amount of each component at some point in the process. The
third method was considered to be the most desirable and potentially the most
accurate for this study.
The mixtures were prepared gravimetrically in thoroughly cleaned and dried metal

t)7
876 M. J. mZA, W. M. HAY>fES, AND W. R. PARRISH
cylinders, each with a freevolume of about 3 dm'' and a tare mass of about 4 kg.
In a few cases, cylinders with about 7 dm^ free volume and a tare mass of about
12 kg were used. AH gases were research grade, and were analyzed chromatographi-
cally before use. In addition, nitrogen and methane were passed through room-
temperature molecular-sieve adsorption columns to remove moisture and any heavy
contaminants not detected by analysis. The amount of each component added to a
cylinder was determined by difference weighings using a Class S weight set and a
precision equal-arm balance with a capacity of 25 kg. An identical cylinder filled
with nitrogen gas at atmospheric pressure was used as ballast on the opposite pan.
The standard deviation of 10 repetitive weighings with the balance was determined
by the manufacturer to be 0.51 mg. The standard deviation of 1 1 repetitive weighings
as used in the preparation of mixtures in the laboratory was determined to be 0.67 mg.
Since the uncertainty in the Class S weights is ten times lower than the standard
deviation in the weighings, the uncertainty in the amount of substance of each com-
ponent in the mixture prepared is dependent only on the random error in the weighing-

TABLE 1. Uncertainty 5;c in mole fraction of each component in the prepared mixtures ba^ed on
the total amount of substance n prepared

Component n = 1 raol rt = 5 mol

±0.00014 ±0.00003
CH« ±0.00025 ±0.00005
CaHg ±0.00013 ±0.00003
C3H9 ±0.00009 ±0.00002
C«Hio ±0.00007 ±0.00001

The estimated error in mole fraction of each component is given in table 1 for both
1and 5 mol of mixture prepared. Since two weighings are required to determine the
amount of substance of each component added, the estimated errors are based on
6 times the standard deviation determined in our laboratory.
The composition, molar mass, and the total amount of substance prepared are
given in table 2 for each of the binary mixtures included in this study. A comparison
of the uncertainties in component composition given in table 1 with the total amount
of substance of mixture prepared, given in table 2, is a direct indication of the un-
certainty in composition of each mixture. Three of the mixtures given in table 2
(mixtures with x « 0.5 of methane -f- ethane, methane -i- propane, and ethane +
propane) were obtained from the U.S. Bureau of Mines in Amarillo, Texas. The
estimated uncertainties in component composition of these gravimetrically prepared
mixtures are roughly the same as those given in table 1.^'^
A schematic diagram of the apparatus as used to measure liquid mixture densities
is given in figure 1. This arrangement incorporates the same capabilities as the closed
loop vapor-recirculation apparatus used in liquid-vapor equilibrium measure-
ments^'' except for vapor sampling. A window in the equilibrium cell allows visual
observation of the liquid sample from about 2 cm above the bottom of the cell cavity

66
 "

LIQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 877

TABLE 2. Prepared binary mixtures : M denotes molar mass and n the total amount of substance
prepared

Mixture Af/g mol /i/mol

0.04752Na + 0^5248CH4 16.6119 6.30

0.30349Na + 0.6965 ICH* 19.6759 4.37
0.49242Na + O.SOTSSCH* 21.9375 6.00
0.05933Na + 0.94067CaH« 29.9481 5.42
0!02014Na + 0.97986C3Ha 43!7733 IJO
0.03794N2 + O.962O6C3H9 43.4870 3.73
0.06740Na + O.9326OC3H8 43.0132 123

O.35457CH4 + 0.64543CaH« 25.0%5 5.04

O.49325CH4 + 0.50675CaH, • 23.1513 25.56
O.68OO6CH4 + 0.31994CaH. 20.5309 4.87
0.29538CH« + 0.70462C3H8 35.8106 1.27
0.49637CH4 + OJO363C3H8 » 30.1720 2.19
O.7492OCH4 + O.25O8OC3H8 23.0790 3.54
0.85796CH4 + 0.14204C3Ha 20.0279 4.79
0.48687CH4 + OJI313/-C4H10 37.6362 1.63
0.58828CH4 + 0.41172n-C4Hio 33.3687 1.92
0.91674CH4 + 0.08326n-C4Hio 19.5467 3.85

0.50105CaH« + 0.49895C3Ha » 37.0689 2.23

0.67287CaH, + 0.32713C3H8 34.6588 3.04
0.68939CaH9 + 0.31 061 /-C:4Hio 38.7840 1.21
0.72436CaH« + 0.27564/-C4Hio 37.8030 1.78
0.65343CaH8 + 0.34657n-C4Hio 39.7929 0.83
0.6711 7CaH8 + 0.32883«-C4Hio 39.2952 0.96
O.4903OC3H8 + 0.50970/-C4Hio 51.2468 0.74
O.50326C3H8 + 0.49674j-C4Hio 51.0650 0.75
O.58692C3H8 + 0.41308/i-C4Hio 49.8915 0.53
O.6O65OC3H8 + O.3935O/1-C4H10 49.6169 0.54
O.6O949C3H8 + 0.39051/i-C4Hio 49.5749 0.58
O.47039/-C4H10 + 0.52961n-C4Hio 58.1243 1.03

• Obtained from the U.S. Bureau of Mines, Helium Operations, Amarillo, Texas.

up of the vapor bulb. The vapor bulb, attached directly to the closure
to the base
plug, is a top section of the cell. The vapor leaves the top of cell through
slip fit in the

a capillary tube passing through the vapor bulb. The density equilibrium system
was designed so that the vapor volume, excluding the free volume (65 cm^) in the
recirculation pump, is extremely small relative to the liquid volume. The vapor
volume includes approximately 0.42 cm^ in the cell, 0.43 cm^ in the access tubing
within the cryostat, and 3.3 cm^ in the pressure gage and tubing outside the cryostat
The volume occupied by the liquid is approximately 20.5 cm'. The vapor volumes
were calculated from known dimensions, the liquid volume was determined by filling
the cell with water, and the pump volume was determined by gas expansion. Without
the pump volume, the effect of the vapor volimies on composition and density of the
liquid is quite small and becomes important only at the higher mixture vapor pressures
(0.2 to 0.3 MPa).
60

69
878 M. J. HIZA, W. M. HAYNES, AND W. R. PARRISH

I
Recirculation

Liquid Magnetic
mixture buoy

i J

FIGURE 1. Schematic diagram of experimental apparatus.

The prepared gas mixtures were condensed directly and continuously into the
equilibrium cell through the capillary vapor inlet tube at the bottom of the cell

until the lastvapor bubble disappeared below the vapor bulb. This filling method
was used to provide continuous mixing of the liquid during the fill, analogous to
vapor recirculation. For mixtures containing only the heavy hydrocarbons, the vapor
pressures were extremely low, and the amount of vapor space left in the cell was
imimportant. Nevertheless, the cell was filled with liquid as discussed above. There
were no visible bubbles when filling with these mixtures.
For methane + heavy hydrocarbon mixtures, comparison of the measured densities
before and after vapor recirculation provided a simple test of homogeneity of the
liquid mixture. For these mixtiires, the vapor phase is essentially pure methane at
the temperatures included in this study. Ethane, the most volatile of the heavy com-
ponents, has a vapor pressure of only 0.00383 MPa at 140 K.^^^^ Initially, the re-
circulation pump was filled with pure methane gas at a pressure equivalent to the
vapor pressure of the mixture for the second experimental temperature, and isolated
from the system. The solenoid currents for a pure liquid methane point and two
vacuum points, separated by 5 K, were taken first. The prepared binary gas mixture
to be studied was then condensed into the equilibrium cell in the manner discussed
above. Solenoid currents for the two liquid-mixture points were measured at tempera-

7U
 LIQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 879

tures corresponding to those of the vacuum points, the lower-temperature point

generally being taken last Subsequently, the system was opened to the pump, and
vapor was recirculated through the liquid at one to three bubbles per second (probably
less than 1 cm^ min"^) for several minutes. A
low recirculation rate was used to
avoid entraiimient The pump was then turned off, and the solenoid current for the
liquid mixture point was remeasured. No change in density could be detected outside
the precision of the measurement for any of the reported binary-mixture points thus
tested.
For mixtures containing heavy hydrocarbons only, the most direct means to
determine that the liquid sample was homogeneous and of the same composition as
the prepared mixture was by liquid sampling and analysis. Ethane + propane was
selected for this test Liquid samples were withdrawn under constant hehum-gas
pressure (about 0.3 MPa) after the density measurements were completed. During
sample withdrawal, the solenoid current was monitored to detect any variations in
density as a different part of the liquid sample surrounded the magnetic buoy. Due
to the extremely low vapor pressures, the compositions of these liquid mixtures are
not subject to change by preferential vaporization during sampling. Each liquid
sample was analyzed chromatographically using the prepared gas mixture for cali-
bration. For these liquid mixtures, no change in solenoid current could be detected
during sample withdrawal, and the composition of the liquid was found to be con-
sistent with that of the prepared mixture within the precision of the gas analysis,
i.e. within a few hundredths of 1 mole per cent

Liquid samples of methane + ethane and methane + propane were also with-
drawn and analyzed subsequent to the density measurements and recirculation of
the vapor. However, it was extremely difficult to obtain samples of consistent com-
position due to preferential vaporization. At the higher temperatures (^135K)
samples with compositions consistent with the prepared mixtxires could be obtained
by using the recirculation-pump volume as pressure ballast during withdrawal. At
lower temperatures it was necessary to pressurize the cell with helium gas to obtain
samples with compositions equivalent to those of the prepared mixtures. In general,
however, analyses of liquid samples of the methane + hydrocarbon mixtures were
mainly useful to determine that there were no gross discrepancies (greater than a
few tenths of 1 mole per cent) in composition and that the liquid mixtures were not
inadvertently contaminated. No attempt was made to recirculate the vapor or to
sample and analyze the liquid mixtures containing nitrogen.

3. Results and discussion

EXPERIMENTAL RESULTS
The experimental orthobaric liquid-mixture amount-of-substance densities are given
in table 3 as a function of temperature. The excess volumes F^, also given in table 3,
were computed from the expression:

(2)

where is the molar volume, is the saturation pressure, x is mole fraction, and P

71
880 M. J. mZA, W. M. HAYNES, AND W. R. PARRISH

TABLE 3. Ortiiobaric liquid amount-of-subsiance densities p of binary mixtures of low molar-mass

alkanes and nitrogen. T, temperature (IPTS-68); ^ouc. P calculated from equation (4); p, pressure;
V^, excess volume

T P 1 r\1 P'm,
103
Pc*.lo P
Mixture
K mol dm"' MPa cm' mol"*

0.04752N2 + 0.95248CH4 105.00 26.8476 -0.006 0.138 -0.144

110.00 26.4052 0.010 0.199 -0.258
115.00 25.9374 -0.002 0.263 -0.442
120.00 25.4522 -0.006 0.350 -0.889
125.00 24.9496 0.005 0.460
130.00 24.4210 0.007 0.583
135.00 23.8600 -0.017 0.730
140.00 23.2809 0.009 0.920

0.30349N2 + 0.69651CH4 100.00 26.8735 —0.010 0.345 -0.584

105.00 26.3393 0.023 0.466 -0.885
110.00 25.7686 -0.005 0.618 -1.353
115.00 25.1790 -0.017 0.801 -2.253
1 1A AA 24.5737 0.010 1.011
A CIA
—4.510

0.49242N2 + 0.50758CH4 95.00 27.0801 0.006 0.330 -0.541

1AA AA
100.00 26.4 Jos 0.003 0.465 —0.777
105.00 25.8106 —0.012 0.637 — 1.154
110.00 25.1387 -0.016 0.844 -L803
115.00 24.4431 0.006 1.097 -3.056
120.00 23.7096 0.017 1.398 —6.110
125.00 22.S315 0.010 1.731
130.00 22.1005 —0.015 2.110

0.05933Na -I- 0.94067C2H8 105.00 21.4718 0.385 -0.423

110.00 21.2912 0.463 -0.590
115.00 21.0845 0.547 -0.804
120.00 20.8998 0.638 — 1.385
0.02014Na + 0.97986C3Hs 110.00 16.2131 0.357 -0.208
115.00 16.0931 0.471 -0.289
0.037941^2 + O.yoZOoCaHs 105.00 16.4638 0.495 —0.310
110.00 16.3410 0.671 —0.392
0.06740N2 + O.9326OC3H8 100.00 16.8055 0.631 —0.427
105.00 16.7084 0.880 -0.628
0.35457CH4 + 0.64543CaH8 105.00 23.1032 0.002 0.0256 -0.382
110.00 22.8777 0.009 0.0392 -0.442
115.00 22.6478 0.007 0.0580 -0.505
120.00 22.4035 -0.050 0.0826 -0.554
125.00 22.1872 0.030 0.115 -0.673
130.00 21.9441 0.002 0.155 -0.798
0.49325CH4 + 0.50675CaH, 105.00 23.9619 0.031 0.0325 -0.525
110.00 23.6937 -0.004 0.0503 -0.569
115.00 23.4328 0.003 0.0749 -0.639
120.00 23.1559 -0.047 0.108 -0.697
125.00 218933 -0.024 0.150 -0.801
130.00 22.6290 0.007 0.205 -0.927
135.00 22.3581 0.024 0.272 -1.068
140.00 22.0765 0.009 0.355 -1.223

7Z
 —

UQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 881

TABLE 3 continued

T yz
iViiAiurc
P
L \J — ^ P
K mol dm"^ Pcala MPa cxn^ mol ^

0.68006CH« + 0.31994CaHa 105.00 25.1027 0.002 0.0416 -0.524

24 7802 0 004 0 Ofi45 —0 562
115.00 24.4612 0.003 0.0963 -0.621
120.00 24.1402 -0.001 0.138 -0.698
125.00 23.8212 0.002 0.193 -0.802
130.00 23.5007 -0.001 0.262 -0.933

O.29538CH4 + 0.70462C3Hi 105.00 18.5132 0.0270 —0.469

11000 18 3^24 0 0409 —0 524
0.49637CH4 + 0.50363C3H8 105.00 20.4909 0.005 0.0384 —0.727
110.00 20.3046 -0.002 0.0591 -0.814
115.00 20.1180 —0.006 0.0874 —0.915
120.00 19.9311 -0.007 0.125 -1.033
125.00 19.7471 0.014 0.173 — 1.179
iin on —0 004 0 2'?2 — 329 1

O.7492OCH4 + O^OSOCsHg 105.00 23.4768 0.005 0.0478 —0.700

110.00 23.2064 —0.009 0.0738 —0.783
115.00 22,9364 —0.004 0.110 —0.887
150 no 0 017 0 158 — 1 014
125.00 22.3818 -0.010 0.222 -1.141
130.00 22.1019 0.002 0.303 -1.310

O.85796CH4 + O.I42O4C3H8 105.00 —0.007 0.0516 -0.553

1 10 00 24 fi%'%\ 0 009 0 0801 —0.617
115.00 24.2941 0.006 0.120 -0.684
120.00 23.9491 0.002 0.173 -0.764
125.00 23.5941 -0.021 0.242 -0.855
130.00 23.2461 0.012 0.330 -0.992

0.48687CH4 + 0.51313i-C4Hio 110.00 17.3575 0.0629 —0.803

115.00 17.2076 0.0938 —0.884
120.00 17.0639 0.1361 —1.002
10.71v0 0 1852

0.58828CH4 + 0.4I172n-C4Hio 120.00 18.6495 0.1636 —1.312

125.00 18.4772 0.2281 — 1.466
125.00 18.4853 0.2281 -1.489
130.00 18.3058 0.3183 -1.646

0.91674CH4 + 0.08326/i-C4Hio 105.00 25.1536 0.009 -0.599

110.00 24.7960 -0.021 -0.638
115.00 24.4512 0.010 -0.719
120.00 24.0910 -0.011 -0.797
125.00 23.7370 0.006 -0.916
130.00 23.3789 0.018 -1.063
135.00 23.0110 0.003 -1.234
140.00 22.6391 -0.015 -1.451

0.50105CaH, + 0.49895C3H, 105.00 18.3618 0.021 -0.040

110.00 18.2169 0.001 -0.036
115.00 18.0726 -0.012 -0.034
120.00 17.9282 -0.020 -0.033
125.00 17.7880 0.000 -0.047

/J
 —

882 M. J. mZA, W. M. HAYNES, AND W. R. PARRISH

TABLE 3 continued

T yz
Mixture
P 10*" ——————
K mol dm'^ />o*io MPa cm' mol"

in on
1 jU.UU
1
1/.041i
A A1
— U.Ull 1
— U.U41
135.00 17.4988 0.009 -0.051
140.00 17.3526 0.013 -0.051

0.67287C2H8 + 0.3271 3C3H, 125.00 18.6192 -0.044

130.00 18.4646 -0.053
A nc-y
14U.UU — U.UOJ
A ruze

A £.QQ1Q/^ ZJ -LA ^lAi^l .*

LT 1
1
7/ .J171 1^
/ lo
1 A Al 7
1 AA 1 7 HAA +1 A A1 C
U.U1 J

O.72436C2H8 + 0.27564/-C4Hio 105.00 17.9779 +0.013

1 1 A AA 17
1 /.04U1
1 A AAO
1 AA — AU.UIU
Al A
1 "JA AA

1 1 < AA 71 94
17/.Zlo4
1 — U.Ujj
120.00 17.0824 -0.045

U.o/il/Cang 1 1 A An 17 </M7 A /Vie

-r U.jZo9j/i>C4riio
115.00 17.3706 -0.000 -0.050
125.00 17.1031 0.009 -0.060
1 lA AA 1 if Qi^7X A A7n A Ai4'5

11^ AA 1/S 979< A AA^

— U.UUO A A>1<

140.00 16.6947 0.014 -0.054

lAC
lUj.UUAA lA m9n — u.uuz
1 t A AA A AA7
110.00 14.ZI Jo + AU.U7CS
1 A70
115.00 14.1219 0.011 +0.056
120.00 14.0257 0.001 +0.057
— AO.UUZ
AA*^ _1_A AC/4

130.00 13.8342 -0.002 +0.052

A CAI'^X/^ XT _L A >4AiC7/1.' ^ Tjr 1 ^< AA 1 "3 Q71 Q A Ay17
+U.U4/
1

1 '3A AA A A</f
130.00 13.0737 +0.054
1

1 1 A AA 1/1 A9'}Q — U.Uj4

135.00 14.1748 +0.021
140.00 14.0786 +0.006

0,60650C3H8 + 0.39350/:-C4Hio i4o:oo 14.1343 +0.010

145.00 14.0333 +0.015
150.00 13.9346 +0.009

O.6O949C3H, + 0.3905I/»-C«Hio 115.00 14.6487 -0.017

. 120.00 14.5521 -0.036

0.47039/-C4Hio + O.5296I/J-C4H10 125.00 12.6943 +0.016

130.00 12.6133 +0.013
135.00 12.5271 +0.048
140.00 12.4447 +0.053

74
 :

UQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 883

is the isothermal compressibility. Subscripts m

and i refer to the mixture and to the
pure components, respectively. The pure-component molar volumes were calculated
from the fit of the experimental results/^* Only those for nitrogen, methane, and
ethane were adjusted to the mixture pressure where appropriate. Isothermal com-
and methane were taken from Rowlinson^"^ and those for
pressibilities for nitrogen
ethane were taken from Miller^^^^ assuming linear temperature dependence. Excess
volumes for a few representative methane +
ethane points were also computed
using isothermal compressibilities from the recent methane data of
derived
Goodwin.^**^ The excess volumes using Goodwin's values of P were less than 0.1
per cent different from the excess volumes computed using those of Rowlinson.
Rowlinson's values were used here only because excess volumes for methane and
nitrogen systems reported in the literature are based on his values. Thus, a direct
comparison of excess volumes can be made without assessing the contribution of
different compressibility values. The ethane compressibility values of Miller were
chosen for the same reason. Vapor pressures for nitrogen, methane, and ethane were
taken from Strobridge,^^^^ Goodwin,^^*^ and Goodwin, Roder, and Straty,^^^^
respectively.
molar volumes for n-butane below its triple-point temperature (134.8 K)
Effective
were obtained from equation (2) by assuming that the small negative excess volume
of ethane -H «-butane was constant and equal to the average value of the excess
volumes above the triple-point temperature of n-butane. These subcooled liquid
molar volumes were fitted as a linear function of temperature to obtain the equation

F(n-C4) = 65.63936 -H0.0992160(r/K). (3)

The molar volumes calculated from this expression at 135 and 140 K agree with the
experimental values^'^ within 0.01 per cent. At 105 K, the molar volume of n-butane
calculated from the equation given in reference 3 fitted to the experimental liquid
by equation (3). Equation (3)
densities is 0.056 per cent larger than the value given
was used compute the excess volumes given in table 3 for all of the /i-butane
to
mixtures below 135 K. For i-butane, the liquid-phase equation given in reference 3
was used to obtain subcooled liquid molar volumes below the triple-point temperature
(113.6 K).
In examining the excess volumes for propane 4- n-butane and /-butane -f- n-butane
using n-butane molar volumes from equation (3), an interesting anomaly was noted.
The excess volumes for these systems, as shown in figure 2, appear to have a dis-
continuity at about the triple-point temperature of n-butane. Molar volumes of
n-butane determined from these systems in the same manner as those determined
from ethane + n-butane are in good agreement with each other. However, the
molar volumes calculated from a linear fit of the values from propane + n-butane
and /-butane -I- n-butane are 0.14 and 0.09 per cent lower than those from equation (3)
at 110 and 130 K, respectively, and the extrapolated volumes are about 0.075 per cent
lower than the experimental liquid molar volumes at 135 and 140 K.
Though the differences noted are not very large, the imprecision of the density
measurements for these heavy hydrocarbons and their mixtures is small enough (see
884 M. J. HIZA, W. M. HAYNES, AND W. R. PARRISH

the ethane + w-butane and propane + /-butane mixtures in table 3) to suggest that
this behavior is real and would be worth investigating in more detail. This behavior
might also be important in the development of precise methods to predict densities
of mixtures containing n-butane.
The pressures listed in table 3 are considered only approximate mixture vapor
pressures, and not all of these were directly measured in the present experiment.
Pressures are given only for mixtures where sufficient measurements were made to
allow either interpolation in temperature or composition for a given system or
comparison with existing phase equilibria. For nitrogen + methane, several random
pressure measxirements were made which were consistent with those interpolated
from previous phase-equilibrium measurements in our laboratory. As a result,
all of the pressures listed for nitrogen + methane were obtained from graphical

0.10 r

K„ -
0-04 per cent in
'o
0.05
;_rH +n-CH
E
_^ h ,
A Jo .04 per cent

-0.05
110 120 130 140 150 160
r/K
FIGURE 2. Excess volumes of propane +
n-butane and /-butane -f- /i-butane from the present
study. O, O.58692C3H9 4- 0.41308n-C4Hio; 0. O.6O65OC3H, -i- 0.39350/i-C*Hio: O, 0.60949
CjHa -i- 0.3905 In-CiHio; ,
O.47039/-C4H10 -f 0.52961/i-C4Hio.

interpolations of the phase-equilibrium measurements. Pressures for nitrogen +

ethane and nitrogen + propane were
measure with any consistency since
difficult to

they are very strong functions of composition. Most of these pressures were estimated
by comparison of the present measurements with the phase-equilibrium data of Lu
and his colleagues.^*' For methane -I- ethane, the pressure corresponding to the
second point from one filling of the equilibrium cell was generally consistent with
phase-equilibrium data,^^''^°^ while the first point was usually high by a few per
cent Pressures for these mixtures were smoothed to the pressures of the second point
by comparison with the data of Miller and Staveley^*'^ and of Miller, Kidnay, and
Hiza.^^'' For methane + propane, discrepancies in the pressures between points
taken with the same fill were not as apparent, and the values given in table 3 were
not smoothed. Comparisons between the ratios of measured to Raoult's law pressures
from the present study with those from the measurements of Stoeckli and Staveley,^-^*
Cutler and Morrison,^^^^ and Calado, Garcia, and Staveley^^^' are shown in figure 3.
Pressures given for methane + /-butane and methane + ;i-butane, though no phase

7b
 UQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 885

1 I ! I r

0.8580CH4 + 0.1420CjHg
LOl — 1

90 100
1 i

110
1

120
1

130
1

140
r/K
FIGURE 3. Comparison of ratios of actual to ideal total vapor pressures for methane + propane
at two compositions. O, present stxidy; g, Stoecidi and Staveley;'*" A, Cutler and Morrison
, Calado, Garda, and Staveley.**"

equilibrium measiu^ments are available for comparison, would probably exhibit

departures similar to those shown in figure 3.

ANALYSIS OF ERRORS IN LIQUID-MDCTURE DENSITIES

Where density measurements were made at five or more temperatures, amount-of-
substance densities p at constant composition were fitted as a function of temperature
to the expression previously used in representing orthobaric amount-of-substance
densities of the pure components

p-p. = a{i-(r/Tj}''-"+ i &.{i-(r/r,)}(^^<'-^)/'>. (4)

The least-squares coefficients a and Z>f, and the values of the critical temperature
and critical amount-of-substance density Pj for each mixture thus treated are given
in table 4. The fourth coefficient was not statistically significant in the fit of any
of these results. Since experimental critical parameters are rarely available for mixtxires
at the desired composition, those used here were obtained from the correlation of
Chueh and Prausnitz.^^*^ The standard deviations from the fit of the experimental

densities to equation (4) are also given in table 4 as a percentage of p. The largest
standard deviation obtained was 0.029 per cent in p for (0.35457CH4 -I-

O.64543C2H6).
The overall uncertainty in the measured densities of the binary mixtures is slightly
larger than for the pure fluids due to the added uncertainty in composition and is
dependent on the type of mixture. Estimated uncertainties in the density measure-
—
ments for three representative mixtures nitrogen + methane, methane -f- ethane,
—
and ethane + n-butane are given in table 5. The estimated errors due to mixture
preparation were determined from the uncertainty in mole fraction of each com-
ponent in the mixture for total amount of substance of mixture prepared (tables 1
and 2). The errors due to vapor-volume correction were estimated by assuming a
volume imcertainty equivalent to that portion of the vapor volume in the equilibrimn
cell at the experimental temperature. The sources of the other systematic errors arc

77
886 M. J. HIZA, W. M. HAYNES, AND W. R. PARRISH

TABLE 4. Parameters of equation (4) where To and are the critical temperature and critical
amount-of-substance density of the mixture. The coefficients a, bu aud 63 were obtained from a
least-squares program in wliich the experimental mass densities to five digits were convened to
amount-of-substance densities within the program; c is the standard deviation

a bi To Pc 10= X
Mixture
mol dm"' mol dm"' mol dm"^ K mol dm"'

0.04752N2 -f- O.95248CH4 16.41033 18.47257 -11.57943 188.04 10.258 0.011

O.30349N2 -h 0.6965 ICH4 14.92272 20.19078 -10.85599 173.20 10.747 0.023
0.49242N2 + O.50758CH4 14.54530 20.60585 -10.07112 161.48 11.018 0.014
O.35457CH4 -1- O.64543C2HS 13.02060 6.275239 276.40 8.196 0.029
O.49325CH4 -1- 0.50675C2Ha 12.99333 7.219530 262.24 8.762 0.028
O.68OO6CH4 -i- O.31994C2H, 18.86369 -16.08578 19.91557 239.93 9.482 0.003
0.49637CH4 + 0.50363C3H, 9.168092 7.647523 320.33 7.372 0.009
O.7492OCH4 -f O.25O8OC3H8 6.367272 24.40468 -11.53932 271.15 9.163 0.013
0.85796CH4 + O.I4204C3H8 8.371599 23.49229 -10.69945 241.28 9.836 0.016
0.91674CH4 -f- 0.083 26/J-C4H10 8.975024 12.09284 238.92 11.044 0.015
0.50105CsHa + 0.49895CjH8 9.598348 5.440200 344.42 6.125 0.015
0.671 17CaH« + 0.32882n-C^Hio 8.334169 5.886011 359.88 6.082 0.013
0.49030C3Ha + O.S0970i-C^Hio 7.647038 4.065571 39163 4.472 0.007

TABLE 5. Estimated uncertainties in density measurements

m5p\/p
Source of error 0.49242Na 0.49325CH4 0.671 nC^H,
+O.50758CH4 -t-0.50675C2H9 -h0.32883/r-C4H;

Systematic errors
llOK 130K llOK 130 K llOK 130 K
Mass of float 0.002 0.002 0.002 0.002 0.002 0.002
Volume of float at 300 K 0.020 0.020 0.020 0.020 0.020 0.020
Thermal expansion coefficient
of barium ferrite 0.022 0.020 0.022 0.020 0.022 0.020
Position of float 0.012 0.013 0.012 0.012 0.009 0.009
Position of main coil.
determined from relative
measurements 0.020 0.020 0.020 0.020 0.020 0.020
Temperature uncertainty of
0.030 K 0.016 0.023 0.007 0.007 0.005 0.005
Mixture preparation 0.001 0.002 0.001 0.001 0.002 0.002
Vapor volume correction 0.006 0.020 0.001 0.005 0.000 0.000

Total systematic error •

0.042 0.048 0.039 0.038 0.037 0.036
Three times standard
deviation " 0.042 0.048 0.081 0.084 0.039 0.039

"
Total uncertainty 0.08 0.10 0.12 0.12 0.08 0.08

" The total uncertainty was determined from the square root of the sum of the squares of the

systematic errors added to three times the standard deviation from the fit to equation (4).

76
 UQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 887

the same as those discussed earlier^^^ for pure-fluid measurements. The total un-
certaintywas taken as the square root of the sum of the squares of the systematic
errors plus three times the standard deviation for random error. The standard
deviations are those obtained from the fit of the results to equation (4) taken as a
percentage of the mixture amount-of-substance density at the specified temperature.
For the estimate of random errors for pure-fluid measurements given in reference 2,
the standard deviation of a single density measurement was computed relative to the
density and temperature of methane at 105 K for which repetitive measurements
had been obtained. This method of assessing random error could not be used here.
Since the standard deviation for methane obtained from equation (4)^^^ was in good
agreement with that computed from repetitive measurements, standard deviations for
the binary mixtures obtained with equation (4) are considered a reasonable basis for
estimating the random error.
It is probable that the overall uncertainties in the densities for some of the mixtures
investigated in this study are larger than those shown in table 5. Uncertainties in the
results for nitrogen + + propane could exceed the maximum
ethane and nitrogen
total uncertainty shown in table 5 by a few hundredths of a per cent Because of the
dew-point limitations on the pressure of the gas mixtures prepared and the high
liquid-mixture vapor pressures, it was difficult to maintain adequate driving force
during the fill to provide continuous condensation and mixing. Of these two systems,
results for nitrogen + propane are considered the least reliable. Also, during the
course of this study, difficulties were encountered in attempting to obtain consistent
results for methane + /-butane where xiCH^) > 0.9. This difficulty could have been
due to a problem in the filling procedure or to a dew-point related problem in the
prepared gas mixture. Though this problem was not encountered in measurements
of the (0.91674CH4 -f- 0.08326«-C4Hio) densities, it is quite possible that the overall
uncertainties in these results are also larger by a few hundredths of a per cent than
the maximum given in table 5.

Fitting the results for mixtiu-es to equation (4) does not allow a test of their con-
sistency as a function of composition. Where densities are available for a given binary
system at several temperatures and at three or more compositions, preferably at mole
fractions of approximately 0.3, 0.5, and 0.7, excess volumes can be examined for
consistency using a temperature-dependent Redlich-Kister expansion. In this study,
compositions for methane -i- ethane and methane + propane were selected to allow
this treatment As shown in table 2, each of these systems also includes one mixture
obtained from a laboratory with many years of experience in preparing gravimetric
standards. Excess volumes for these systems were fitted to a Redlich-Kister expansion
of the form:

F^/cm^ mol-^ = XiX2[{ao + ai(T/K)+a2(T/K)^}

+ {bo + biiTIK) + b:,{T{KY} (2;c, - 1) -h {co + Ci(T/K) + c^(TlYif}Clx, - 1). (5)

where is the mole fraction of methane. The least-squares coefficients a„ and c,,
and the standard deviation, in per cent of the average mixture molar volume, for
each system are given in table 6. The C| coefficients were not statistically significant

7y
888 M. J. mZA, W. M. HAYNES, AND W. R. PARRISH
TABLE 6.Parameters of equation (5) for methane -r ethane and for methane + propane; ,Vi is the
methane mole fraction and a the standard deviation

methane + ethane methane + propane

Co -1 4.22 J 84 -11.96725
0.2649993 0.236748
fla -0.001419196 -0.001429237
6o -18.87392 16.59465
0.3108472 -0.3212125
b2 -0.001349755 0.001398221
Co -52.93613
Ci 0.9887147
-0.004667488
l(yff<F-i> 0.030 0.040

for methane + ethane. The excess volumes calculated from equation (5) are com-
pared in figure 4 with the experimental values at 110 from the present study. K
The standard deviation for methane + ethane obtained with equation (5) is in
very good agreement with that obtained with equation (4) for the methane + ethane
mixture in table However, the standard deviation for methane + propane obtained
5.

with equation (5) more than a factor of two higher than that obtained with equation
is

(4) for any of the individual mixtures in the set The higher value from equation (5)
is considered to be more indicative of the random error in the results for methane +

propane.
The standard deviation given in table 6 for methane + propane was the highest
value obtained in the analysis. With a total systematic error of about ±0.04 per cent
added to three times this standard deviation (±0.12 per cent), the overall uncertainty
for methane + propane would be approximately ±0.1 6 per cent. With the possible
exceptions noted above (i.e. nitrogen + ethane, nitrogen + propane, and methane +
/i-butane), this is believed to be a suitable estimate of the maximum overall uncertainty
in the densities of binary liquid mixtures obtained in this study.

0 V 1 1 1 1 i

1
I 1
I 1 1
I I

0 0.2 0.4 0.6 0.8 1

.x(CH^

FIGURE 4. Excess volumes for methane +

ethane and methane + propane from the present
study at 110 K. O, experimental; calculated from fit to equation (5).
,

8u
 UQUm DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 889
COMPARISONS OF EXCESS VOLUMES
Equation (5) has also been used to compare the present results for methane + ethane
and methane + propane with those of other investigators at 108.15 K, a temperature
approximately common to all of their measurements. The excess volumes were taken
directly from the literature where reported, or otherwise were calculated from the
pure-fluid molar volumes in their data set Though some of the results were 0.1 to
0.3 K below 108.15 K, the differences in temperature do not affect the comparisons
of excess volumes appreciably. From equation (5), a change in temperature of 0. 1 K
results in a change in excess volume of 0.001 cm^ mol"* for methane + ethane and
about 0.002 cm^ mol"* for methane + propane.

In figure 5, methane + ethane calculated from equation (5) at

excess volumes for
108.15 K are compared with those from the experimental data of Shana'a and
Canfield/^^> Rodosevich and Miller/^*** Pan, Mady, and Miller,^^"^' Klosek and
McKinley,^^'^ and Jensen and Kurata,^^'^ The data from references 26 and 27 are
at about 108 K and those from reference 28 are at about 107.9 K. The excess volumes
from Shana'a and Canfield and from Miller and his colleagues are in excellent agree-
ment with the curve calculated from the present results. With the exception of one
point each from Shana'a and Canfield and from Miller and his colleagues, the
differences are equivalent to much less than 0.1 per cent in the mixture molar volume.
In contrast, the excess volumes from Klosek and McKinley and from Jensen and
Kurata are much less consistent with the present results. The differences from the
calculated curve are equivalent to 0.15 to 0.3 per cent in the mixture molar volume.
In figure 6, excess volumes for methane + propane calculated from equation (5)

81
890 M. J. mZA, W. M. HAYNES, AND W. R. PARRISH

I 1 —J I \ 1

0 0.2 0.4 0.6 0.8 1

FIGURE 6. Comparison of excess volumes for methane + propane at 108.15 K. Calculated,

from fit of present results to equation (5); ,

Shana'a and Canfield;'*" A, Rodosevich and Miller-/^*'
A, Pan, Mady, and MUler;'"> O, Jensen and Kurata.'29>

at 108.15 K are compared with those from the experimental data of Shana'a and
Canfield/^'^ Rodosevich and Miller/^^^ Pan, Mady, and Miller,^^''^ and Jensen and
Kurata.^^^^ For this system the excess volumes from Shana'a and Canfield and from
Miller and his colleagues are also in excellent agreement with the curve from the
present results with a maximum difference of about 0.06 per cent in the mixture
molar volume. The excess volume from Jensen and Kurata again is not consistent
with the present results. The difference is about 0.5 per cent in the average mixture
molar volume, in the opposite direction from that found for methane + ethane.
Though there are only a few methane + /i-butane points reported in the literature, a
comparison can be made at 108.15 K in the methane-rich region between the excess
volume interpolated from the present results for the (0.91674CH4 + O.O8326/J-C4H10)
mixture and excess volumes from the experimental data reported by Shana'a and
Canfield^^^' and by Miller.^^^' This comparison is given in figure 7. The excess volume
from Shana'a and Canfield was computed using their pure methane data and the

_n 0 1 I
1 ^
'

0.90 0.95 1

x(CH^
FIGURE 7. Comparison of excess volumes of methane + n-butane at 108.15 K. O, Interpolated
from the present study; ,
Shana'a and Canfield A. Rodosevich and Miller.'"'

82
 UQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 891

molar volume of «-butane obtained from equation (3). The excess volumes of Miller
are those reported in his paper. The «-butane molar volumes used by Miller are
within 0.02 per cent of the values given by equation (3). The difference between the
excess volume interpolated from the present results and the curve drawn through the
excess volumes of Shana'a and Canfield and of Miller is about 0.14 per cent of the
mixture molar volume. Since the data from Shana'a and Canfield and from Miller
appear to be in as good agreement for methane + n-butane as for methane + ethane
and methane + propane, it is possible that at least part of the 0.14 per cent difference
is due to an unaccounted error in our results for the mixture, as suggested earlier in

this paper. Unfortunately, data are not available for comparison with the present
results for (0.58828CH4 + 0.41172n-C4Hio), the accuracy of which we have no
reason to question. Measurements for this mixture could not be made at the lower
temperatures since the n-butane content is the solid solubility limit at about 1 16 K.^^°^

-0.5 'I 5 1

2 -1.0 —
e
B

-1.5

-1
1 ._. 1
1
1
I
1
I l_
1

100 110 120 130 140

T/K
FIGURE 8. Comparison of excess volumes from the present study for methane + heavy hydro-
carbon mixtures having x x 0.5, O, 0.49325CH4 + 0.50675C3H«; ,
0.49637CH4 + 0.50363C9H«;
A, 0.48687CH« + 0.51313/-C«Hio; O. 0.58828CH* + 0.41172rt-C«Hio.

The only data for methane + /-butane in this temperature range are those of
Rodosevich and Miller^^^^ for two mixtures with mole fractions 0.9152 and 0.9462
of CH4. The excess volumes they report for these mixtures at 108 K are about half-
way between their results for methane + propane and for methane + n-butane at
the same methane mole fraction. The excess volumes for the mixture of methane -f-
/-butane with x « 0.5 from the present study are equivalent to those obtained for
methane propane at 110
-I- K
and the same composition. Excess volumes from the
present study are shown in figure 8 for mixtures with x « 0.5 of methane -i- ethane,
-f" propane, and + /-butane. Excess volumes are also included for (O.58828CH4 +
0.41172«-C4Hio) for comparison. It is clear that more extensive measurements of
high precision are needed to describe accurately the composition dependence of the
excess volumes for both methane + butane systems.
The small excess volumes given in table 3 for binary mixtures containing the
—
higher molar-mass alkanes—ethane, propane, /-butane, and /x-butane show that at

83
892 M. J. HIZA, W. M. HAYNES, AND W. R. PARRISH
cryogenic temperatures the molar volumes for binary mixtures of these components
are closely approximated by the assumption of ideal mixing. The only other investi-

gators who report data for binary mixtures of these alkanes at cryogenic temperatures
are Shana'a and Canfield/^^^ They report densities for several mixtures of ethane +
propane and for one mixture of ethane + /i-butane.
For ethane + propane, the excess volume reported by Shana'a and Canfieid at
108.15 K for the mixture (0.5852C2H6 + O.4148C3H8) is compared in figure 9 with
those determined from the present results at two similar compositions. If the excess
volumes for ethane + propane exhibit an asymmetry similar to those for methane +
ethane (figure 5), the excess volumes for the two ethane + propane mixtures from
the present study should be nearly the same, with the excess volume at the higher
mole fraction of CjHg shghdy larger. Shana'a and Canfieid report a maximum

-0.1 h

-0.2 h
I

-0.3

-0.4
0.1 per cent in
1

-0.5
105 110 115 120 125 130 135

T/K

+
FIGURE 9. Comparison of excess volumes for ethane propane. O, O, Present study at
xiCilie) =- 0.50105 and 0.67287, respectively; ,
Shana'a and Canfieid'"' at .r(QH,) =- 0.5852.

excess volume for this system of — 0.485 cm^ mol~^ for a mixture containing
(O.7O36C2H5 + 0.2964C3Hg). The difference in excess volumes from the two sets of
measurements is surprisingly large, equivalent to approximately 0.8 per cent in the
mixture molar volume. Since the pure-fluid molar volumes for ethane and propane
from Shana'a and Canfieid are only higher than those determined with the present
apparatus^^^ by approximately 0.06 and 0.12 per cent, respectively, it is logical to
suspect composition error or incomplete mixing as the source of this disagreement
To assess on the excess volume, it was assumed
the efiect of composition uncertainty
that the mixture mass density from the present study was exact and that the actual
mole fraction of propane was 0.01 larger than the composition of the prepared
mixture with x » 0.5. This results in an error of — 0.062 cm^ mol"^ in the excess
volume. If the excess volume reported by Shana'a and Canfieid were assumed to be
correct, it would be necessary for the actual mole fraction of ethane in the liquid in
the present study to be larger by 0.064 than that of the prepared mixture.

84
 —

UQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 893

As discussed in the experimental section of this paper, chromatographic com-

parisons of the compositions of liquid samples for ethane +
propane were in agree-
ment with the composition of the prepared mixture within a few hundredths of 1 mole
per cent The fact that no change in density of the mixture could be detected during
sample withdrawal is considered evidence that the liquid mixture was homogeneous.
The fact that the two mixtures studied were prepared independently in two separate
laboratories is also confirmation that there was no significant error in the mixture
preparation.
Recent enthalpy of mixing measurements in binary mixtures of methane, ethane,
and propane reported by Miller and Stavele/"^ should also be referred to in regard
to the ethane + propane results discussed above. These investigators report that the
enthalpy of mixing at 112 K
about 129 Jmol"^ for methane + propane with
is

X = 0.5, about 69 J mol~^ for methane + ethane with x = 0.5, and probably less
than 5 Jmol"^ for ethane + propane throughout the composition range. A com-
parison of ratios of the enthalpies of mixing with ratios of the excess volmnes for these
systems also tends to support the more ideal excess volumes obtained in the present
investigation.
Shana'a and Canfield include a discussion on the invalidity of the principle of
congruence based on a comparison of their excess volumes for ethane + propane
and methane + ethane with those for methane + propane as a function of the
equivalent carbon number. When the excess volumes from the present study are
used, methane -I- ethane shows the same departure as expected, but ethane 4- propane
isin excellent agreement with methane + propane. For a detailed discussion of the
method, the reader is referred to Shana'a and Canfield's paper.
In figure 10, the excess volume computed from the molar volume of the (O.SSSSCzHg
+ 0.1167«-C4Hio) mixture reported by Shana'a and Canfield is compared with the
excess volimies of this system computed from our results. The excess volume of
Shana'a and Canfield was computed using their density for pure ethane and the molar

0 I
1 1 1 1
i
1 1

s
l_
o
-0.1

0.1 per cent in V„

-0.2'
110 115 120 125 130 135 140

FIGURE 10. Comparison of excess volumes for ethane +

n-butane. O, -o-. Present study at
xiCJcU) = 0,67117 and 0.65343, respectively; ,
Shana'a and Canfield""" at xiCiiU) 0.8833. -
61

bb
894 M. J. HIZA, W. M. HAYNES, AND W. R. PARRISH

volume of n-butane from equation (3). For this system the difference in excess volumes

iseqmvalent to about 0.25 per cent in the mixture molar volume. Since the molar
volumes for n-butane in the subcooled liquid region given by equation (3) were
obtained from the results for this system the lack of temperature dependence is an
artificial constraint For this system, if the actual mole fraction of n-butane in the

liquid mixture at 110 Kwas 0.01 larger than the composition of the mixtures prepared
for this study, an error of —0.119 cm^ mol~^ in the excess volume would result.

0 0.1 0.2 0.3 0.4 0.5

x(N2)

FIGURE 11. Comparison of excess volumes for nitrogen -i- methane. O, Present study;
, Rodosevich and MiUer;"" A, Liu and Miller;"" V, Massengill and Millcr.<^='

Of the data available for liquid mixtures of nitrogen + methane,^*^ the excess
volumes reported by Rodosevich and Miller, Liu and Miller,^^^^ and Massengill
and Milier^^^' were selected for comparison with excess volumes from the present
results. These comparisons are shown in figure 11. The excess volumes of Massengill
and Miller are from the data of Liu and Miller adjusted with additional nitrogen
measurements. All of the excess volumes of Miller and his colleagues appear to be
in agreement with excess volumes from the present results within 0. 1 per cent of the
mixture molar volumes. The height of the symbols is equivalent to about 0.1 per cent
of the mixture molar volumes.
The fact that the nitrogen + methane excess volumes are in good agreement does
not mean that the mixture molar volumes are in agreement. It is worth noting that
the data reported by Liu and Miller and by Massengill and Miller were obtained with

bb
 UQUID DENSITIES FOR BINARY MIXTURES OF LNG COMPONENTS 895
a calibration based on the molar volumes of saturated liquid argon reported by
'^'^^
Terry et alP^^ while all of the other data of Miller and his colleagues^^^* referred
to here were obtained with a calibration based on molar volumes of saturated liquid
methane taken from Goodwin and Prydz.^^*^ From the discussion given earlier/^^
the nitrogen + methane molar volumes of Rodosevich and Miller should be higher
but consistent with the present results within about 0.1 per cent, while those of Liu
and Miller and of Massengill and Miller should be higher than the present results
by about 0.4 per cent

4. Smninaiy
The measurements of the present study provide a set of orthobaric liquid densities
for binary mixtures of the major components of LNG which are generally consistent
with those for the pure components within the precision of the measurements. The
maximum random error in the densities of the binary mixtures is believed to be about
±0.12 per cent, roughly twice that for the pure fluid densities. The known systematic
errors in the densities for both the pure fluids and the binary mixtures were estimated
to befrom 0.03 to 0.05 per cent As noted, the uncertainties in the densities for
methane + /j-butane with ^(CHJ > 0.9, and for nitrogen + ethane and nitrogen +
propane, could exceed these estimates.
The volume found was —6.11 cm^ mol"^ for nitrogen + methane
largest excess
with X = K. This is equivalent to 13 per cent of the ideal mixture molar
0.5 at 120
volume. In contrast, the present results show that molar volumes of mixtures of the
alkanes, excluding methane, are closely approximated by ideal mixing in the
temperature range studied. The most significant disagreement between the present
results and published data is the 0.8 per cent difference in density for ethane +
propane from that of Shana'a and Canfield.
Based on the excess volumes of the alkane mixtures, excluding methane, from the
present study, it was shown also that effective molar volumes of subcooled n-butane
in a binary liquid mixture may differ by a small but significant amount dependent
on the other components present It is suggested that this behavior would be worth
investigating in more detail for n-butane and other fluids such as /j-pentane. There is
also a definite need for additional measurements on the binary mixtures containing
both n-butane and /-butane with methane.
In subsequent papers, the results of measurements of orthobaric liquid densities
of multicomponent mixtures and the ability of theoretical models to predict the
measured densities will be reported.

The authors acknowledge the contributions of M. J. Brown, R. C. Miller, and R. D.

McCarty to the correlation of the results.

REFERENCES
1. Robinson, R. L, Jr.; Hiza, M. J. In Advances in Cryogenic Engineering, VoL 20. Timmerhaus,
K. D.: editor. Plenum Press: New York. 1975. p. 218.
2. Haynes, W. M.; Hiza, M. J.; Frederick, N. V. Rev. Sci. Instrunu 1976, 47, 1237.
3. Haynes, W. M.; Hiza, M. J. /. Chem, Thermodynamics 1977, 9, 179.

d7
896 M. J. HIZA, W. M. HAYNES, Aim W. R. PARRISH
4. Yu, P.; Elshayal, I. M.; Lu, B. C.-Y. Can. J. Chem. Eng. 1969, 47, 495.
5. Poon, D. P. L.; Lu, B. C.-Y. In Advances in Cryogenic Engineering, Vol. 19. Timmerhaus, K. D.:
editor. Plenum Press: New York. 1974, p. 292,
6. Hiza, M- J. ; Kidnay, A. J. ; Miller, R. C. Equilibrium Properties of Fluid Mixtures, A Bibliography
of Data on Fluids of Cryogenic Interest. NSRDS Bibliographic Series. IFI Plenum; New Yorlc.
1975.
7. Haynes, W. M. Rev. Sci. Instrum. 1977, 48, 39.
8. Tully, P. C; Emerson, D. Personal communications. November 1974, August 1976.
9. Duncan, A. G., Hiza, M. J. In Advances in Cryogenic Engineering Vol, 15. Timmerhaus, K. D.:
editor. Plenum Press: New York. 1970. p, 42,
10. Hiza, M, J.; Duncan, A. G. Rev. Sci. Instrum. 1969, 40, 513.
11. Goodwin, R, D.; Roder, H, M.; Straty, G. C. Nat. Bur. Stand. U.S., Tech. Note 684, 1976.
12. Rowlinson, J. S. Liquids and Liquid Mixtures. 2nd Ed. Butterworths: London, 1969.
13. Miller, R. C. Chem. Eng. 1974, 81. 134,
14. Goodwin, R, D. Nat. Bur. Stand. U.S., Tech. Note 653, 1974.
15. Strobridge, T. R- Nat. Bur. Stand. U.S., Tech. Note 129. 1962.
16. Parrish, W. R.; Hiza, M. J. In Advances in Cryogenic Engineering Vol. 19. Timmerhaus, K- D,:
editor. Plenum Press: New York. 1974, p. 300.
17. Kidnay, A, J.; Miller, R, C; Parrish. W. R.; Hiza. M. J. Cryogenics 1975, 15, 531.
18. Miller, R- C; Staveley, L. A. K, In Advances in Cryogenic Engingeering, Vol. 21, Timmerhaus,
K- D. and Weitzei. D, H.: editors. Plenum Press: New York, 1976, p. 493,
19. Miller, R. C; Kidnay, A, J.; Hiza, M. J, /. Chem. Thermodynamics 1977, 9, 167.
20. Wichterle, L; Kobayashi, R, J. Chem. Eng. Data 1972, 17, 9.
21. Stoeckli, H, F.; Staveley, L. A. K, Helv. Chim. Acta 1970, 53, 1961.
22. Cutler, A, J. B.; Morrison, J. A, Trans. Faraday Soc. 1965, 61, 429.
23. Calado, J. C. G.; Garcia, G.; Staveley, L. A. K. /. Chem. Soc, Faraday Trans. 1, 1974, 70, 1445.
24. Chuefa, P. L.; Prausnitz, J. M. AIChEJ. 1967, 13, 1107.
25. Shana'a, M. Y.; Canfield, F, B. Trans. Faraday Soc. 1968, 64, 2281.
26. Rodosevich, J, B.; MiUer, R. C, AIChEJ. 1973, 19. 729.
27. Pan, W. P.; Mady. M, H.; Miller. R. C, AIChEJ. 1975, 21, 283,
28. Klosek, J,; McKinley, C, In Proceedings of the First International Conference on LNC
White, J. W, and Neumann, A. E, S.: editors. Institute of Gas Technology, Chicago, Illinois.
1968. Paper No. 22.
29. Jensen, R. H.; Kurata, F. /. Petrol. Technol. 1969, 21, 683.
30. Kuebler, G. P.; McKinley, C. In Advances in Cryogenic Engineering Vol, 21, Timmerhatis, K. D,;
Weitzei, D. H.: editors. Plenum Press: New York. 1976, p. 509.
31. Liu, Y.-P.; Miller, R. C. /. Chem. Tfiermodynamics 1972, 4, 85.
32. Massengill, D, R. ; Miller, R. C. /. Chem. Tliermodynamics 1973, 5, 207,
33. Terry, M, J.; Lynch, J. F.; Bunclark, M.; ManseU, K. R.; Staveley, L. A. K./, Chenu Thermo-
dynamics 1969, 1, 413.
34. Goodwin, R. D.; Prydz, R. /, Res. Nat. Bur. Stand. US. 1972, 76A, 81,

88
Reference [17j

0-37
J. Chem. Thermodynamics 1983, 15, 903-911

Orthobaric liquid densities and dielectric

constants of (methane + 2-methyl-
propane) and (methane + /i-butane)
at low temperatures^

W. M. HAYNES
Thermophysical Properties Division, National Engineering Laboratory,
National Bureau of Standards, Boulder, Colorado 80303, U.S.A.

( Received 12 January 1983)

Measurements of the orthobaric liquid densities and dielectric constants of methane-rich

(methane + 2-methylpropane) and (methane + n-butane) have been obtained at temperatures
between 110 and 140 K. Densities were determined with a magnetic-suspension densimeter,
while a concentric-cylinder capacitor was used for simultaneous measurements of dielectric
constant. These measurements were part of an experimental program that has provided a
consistent and comprehensive set of densities for the major components of liquefied natural gas
(LNG) and their mixtures, which was used to develop mathematical models for the calculation
or prediction of LNG densities. Along with the (methane -(- 2-methylpropane) and
(methane + n-butane) experimental densities and dielectric constants, are presented
experimental vapor pressures, as well as excess volumes, Clausius-Mossotti functions, and
excess Clausius-Mossotti functions, derived from the densities and dielectric constants.
Comparisons are shown between the excess volumes of the present work and those from
independent measurement using an extended corresponding-states model that had been
optimized to the results from this work. The total uncertainty of a single density measurement
is approximately O.I per cent with a precision of a few parts in 10*. Dielectric constants are

estimated to be accurate to approximately 0.05 per cent. A brief description of the apparatus,
experimental method, and procedures is also presented.

1. Introduction

An experimental program*'"*' has been carried out at this laboratory to measure

with an uncertainty of less than 0.1 per cent the orthobaric liquid densities of the
major components" of liquefied natural gas (LNG) and mixtures**- of these
components. The pure fluid and binary mixture results have been incorporated into
the development of several mathematical models*'"'^* that can be used to predict
within an uncertainty of less than 0.1 per cent the density of any LNG mixture at

" This work was carried out at the National Bureau of Standards under the sponsorship of the British
Gas Corporation, Chicago Bridge and Iron Co., Columbia Gas Service Corp., Distrigas Corp., Easco Gas
LNG, Inc., El Paso Natural Gas, Gaz de France, Marathon Oil Co., Mobil Oil Corp., Natural Gas
Pipeline Co., Phillips Petroleum Co., Shell International Gas, Ltd., Sonatrach, Southern California Gas
Co., Tennessee Gas Pipeline Co., Texas Eastern Transmission Co., Tokyo Gas Co., Ltd., and
Transcontinental Gas Pipe Line Corp., through a grant administered by the American Gas Association,
Inc.

56

89
904 W. M. HAYNES

saturation from an input of the temperature, composition, and pressure of the liquid
mixture. Multicomponent-mixture results have been used to test the models.
Orthobaric liquid densities from this project have been previously reported for
nitrogen,"' methane,*''-" ethane,*^' propane,*^* 2-methylpropane,''' and n-butane;*^-'*'

all binary combinations*** of those components except (nitrogen + 2-methylpropane)

and (nitrogen + n-butane); and selected multicomponent mixtures'^' ^' with up to
eightcomponents that included 2-methylbutane and n-pentane in addition to those
mentioned above. In an earlier paper**' describing binary-mixture measurements,
the need for additional measurements on binary mixtures of methane with either
2-methylpropane or n-butane was cited. Experimental problems had been
encountered in the earlier study in obtaining consistent results for methane-rich
binary mixtures containing 2-methylpropane or n-butane. It was conjectured that

these difficulties were associated with liquid methane condensing in the narrow
annulus around a vapor-pressure bulb that fitted tightly inside the top part of the
cell. Thus, the present work was carried out to resolve these inconsistencies and to
provide values for these mixtures that would complement and extend the earlier
results. The earlier work provided results for {xCH4-F(l — xXCH3)2CHCH3} and
{xCH4-l-(l — x)C4H,o} with x « 0.5 while the present study includes results for
methane-rich mixtures (x from 0.78 to 0.93).
The density results from the present study were obtained using a magnetic-
suspension densimeter*'-'' that was part of a new apparatus,*'-^* which differed
significantly from that employed for the previous measurements.**' (The new
apparatus*^' was developed because a magnetic-suspension densimeter of higher-
pressure capability, 35 MPa, was desired for other projects.) Although the new
apparatus exploited the same technique for density measurements as that used in the
earlier work and gave results that agreed, within experimental precision, with those
obtained with the previous apparatus, there were significant new and improved
features of the apparatus that were important to the present work. These will be
discussed in the next section.
The new apparatus contained a capacitor that was used for obtaining dielectric
constants along with the densities. Through the use of the Clausius-Mossotti
function, dielectric constant measurements can serve as a simple and reliable
substitute for density measurements. In the custody transfer of LNG, densities are
determined by direct measurements or indirectly using a calculational technique
based on measurements of composition and temperature. One technique for
commercially available densimeters is predicated on a density determination
through the application of dielectric-constant measurements.Thus both the
dielectric constants and densities from this study can contribute to equitable custody
transfer of such a valuable commodity as LNG.

2. Experimental

The experimental method, apparatus, and procedures for density measurements

have been discussed in detail in other papers.*'"®' Thus, this section will focus on a
description of the technique and procedures for the dielectric-constant

90
 DENSITIES OF {xCH^ + (l -x)(CH,)jCHCH3} AND {xCH^ + (l -x)QH,o} 905

measurements, along with a discussion of those features of the apparatus and

procedures that are specific to the present study.
The dielectric-constant measurements were performed with a concentric-cylinder
capacitor of a design similar to that of Younglove and Straty."^* There is one major
difTerence between their design and that used here. Instead of solid cylinders, the
capacitor in the present work consisted of cylinders with slots cut parallel to the
Pan et a/.'*** This modification
cylindrical axes, similar to the ring-and-bar design of
was carried out for the same reason Pan et ai: to reduce as far as
as that given by
possible the chances for composition gradients in the cell. The copper capacitor was
located in the top portion of the cell, while the magnetic buoy and position sensor
for the magnetic-suspension densimeter were contained in the lower portion.
Assuming the absence of composition and temperature gradients, simultaneous
measurements of density and dielectric constant were made on the same liquid
samples.
The dielectric constant was determined from measurements of the ratio of the
capacitance with liquid between the electrodes to the capacitance in vacuo. The
capacitance vacuo was approximately 20 pF. The overall external dimensions of
in

the capacitor were 6.4 cm in length by 1.77 cm in diameter. The widths of the five
slots in each cylinder were 0.32 cm, while the slot lengths in the outer cylinder were
4.1cm and those in the inner cylinder cm. Kaptonf film insulators of
were 1.9

0.05 mm thickness were used between the cylinders and the support assembly. The
entire capacitor assembly was fabricated from copper and subsequently plated with
gold. Capacitances were measured with a three-terminal a.c. bridge operated at an
oscillator frequency of 5 kHz.
Although the resolution of the measured capacitances was approximately
10"* pF, the overall uncertainty in the dielectric-constant measurements for the
mixtures investigated here was approximately 0.05 per cent. The limiting factor was
the reproducibility of the measurements, which was based on measurements from
several fillings taken from the same cylinder of prepared mixture. The dielectric-
constant determination was the most sensitive monitor inside the sample space to
indicate the reproducibility of liquid compositions each time the cell was recharged.
In vacuo capacitances, measured to a precision of approximately 10~*pF, were
observed before each run and were stable to better than 10" pF over the duration
of this project.
The densities were determined using a one-solenoid magnetic-suspension
densimeter''* with a differential-capacitance sensor" used to detect the position of
the magnetic buoy. The total uncertainty" ** in the density determination was
estimated to be 0.12 per cent, which includes an allowance of three times the
standard deviation for random error. As demonstrated in a following section, the
precision of measurement was better than 3 x 10"*.
Procedures for preparing mixtures of LNG components gravimetrically have been

and experimental procedures adequately, it is occasionally necessary to

t In order to describe materials
identifycommercial products by manufacturer's or trade names. In no instance does such identification
imply endorsement by the National Bureau of Standards, nor does it imply that the particular product is
necessarily the best available for that purpose.

91
906 W. M. HAYNES

discussed previously.**' For the mixtures investigated here, at least 3.8 mol was
prepared for each composition. This corresponds to less than 0.0001 uncertainty in
the mole fraction of CH4 and considerably less uncertainty for either
(CHajzCHCHa or C4H,o.
Vapor pressures were measured with a spiral quartz Bourdon-tube gauge with a
range of 0 to 1.38 MPa that had been calibrated against an air dead-weight gauge.
Maximum uncertainty in the calibration was 70 Pa. Vapor-pressure measurements
were generally reproducible to within 3 x lO"* MPa for different fillings of the cell
with the same mixture.
Prior to the filling of the cell with each new mixture sample, a density
measurement was performed on liquid methane to serve as a check on the density-
measurement process. The densities were obtained to ensure that the ahgnment of
the cryostat and cell windows had not changed as a result of temperature cycling of
the apparatus. The windows had been aligned, along with the microscope, so that
the buoy-position determination was independent of the refractive index of the fluid
inside the cell. The cell had to be warmed to room temperature to effect complete
removal of mixtures containing butanes. Measurements in vacuo needed for both
density and dielectric-constant determinations were obtained before each new fiUing
of the cell. A maximum of four points at 5 K increments was obtained for each
filling.

The mixtures were condensed and continuously into the cell at a pressure
directly
greater than the vapor pressure of methane either through two capillaries at the
bottom of the cell or simultaneously through capillaries at both the top and bottom
of the cell. Based on the consistency of the results, it was found that, within the
precision of the measurements, the density and dielectric-constant results were
independent of the filling routines used.
All of the measurements in this project were made on liquid mixtures in
equilibrium with vapor. To minimize composition or density errors, it was of the
utmost importance to reduce as much as possible the low-temperature vapor
volume in the sample space. The total internal volume of the sample cell was 43 cm-',
while the low-temperature vapor volume was less than 0.2 cm-'. Assuming that the
vapor composition was pure methane for the mixtures studied here, the maximum
adjustment (decrease) to the measured densities because of differences in
composition between the prepared mixtures and the mixtures condensed into the
cell was 0.045 per cent, which was at 140 K for the mixtures with x % 0.78. This

corresponded to a correction of 0.03 per cent in the dielectric constant.

3. Results and discussion

The experimental orthobaric liquid densities, vapor pressures, and dielectric

constants for (methane -I- 2-methylpropane) and (methane -I- n-butane) are presented
as a function of temperature (IPTS-68) in table 1. Excess molar volumes and
Clausius-Mossotti functions, as well as excess Clausius-Mossotti functions, are also
given in table 1. The excess molar volume is defined by the relation:

Kn' = K„-L^iK*{i+'Ci(pr-P)}. (1)

92
 1 1 '

DENSITIES OF {xCH^ + (l -xKCHj^CHCHj} AND {xCH^ + (I -xjQH.o} 907

TABLE Orthobaric liquid densities and dielectric constants of {xCH« + (l -xKCHjljCHCH,} and
1.

{xCHt + (l —xJC^Hio}. M, molar mass; T, temperature (IPTS-68); p pressure; p„ experimental amount-

of-substance density; p„p„ experimental density; p^,,^, density calculated from equation (4); V^, excess
molar volume; e, dielectric constant; ^q^,, Clausius-Mossotti function; <^cm> excess Clausius-Mossolti
function

T P <^CM 4>CM
e
K MPa mol dm '"

p„,t cm' mol'' cm' -mol"' cm' mol"'

0.92044CH4+0.07956(CHj)jCHCH3 (M = 19.3910gmor')
1 15.00 0.1254 24.3633 -0.016 -0.547 1.69665 7.7352 0.024
120.00 0.1805 24.0029 -0.006 -0.624 1.68372 7.7326 0.021
125.00 0.2521 23.6403 0.011 -0.726 1.67103 7.7322 0.020
130.00 0.3434 23.2752 0.037 -0.859 1.65838 7.7321 0.018
135.00 0.4567 22.8920 0.004 -1.001 1.64527 7.7326 0.017
22.5037 -0.030 -1.186 1.63196 7.7320 0.015

0.78329CH4 + 0.21671(CH3)jCHCHj (M = 25.1625 g-moP')

110.00 0.0782 21.9144 0.005 -0.837 1.81689 9.7662 0.050
115.00 0.1164 21.6652 0.001 -0.941 1.80487 9.7639 0.051
V. i o/ 21.4136 -0.011 -1.065 1.79271 9.7605 0.050
125.00 0.2329 21.1668 0.006 -1.229 1.78094 9.7581 0.049
130.00 0.3170 20.9125 -0.008 -1.412 1.76906 9.7571 0.049
135.00 0.4208 20.6629 0.006 -1.648 1.75729 9.7543 0.047
140.00 0.5474 20.4082 0.001 -1.925 1.74568 9.7548 0.047

0.92788CH4 + 0.072 nQH.o (M = 19.0779 g mol -

190 on
1 ZVJ.VA/
n 11
U. R^n
OZU 24.2615 -0.006 -0.661 1.67289 7.5511 0.007
125.00 0.2547 23.8868 0.005 -0.756 1.66033 7.5524 0.005
130.00 0.3470 23.5047 0.000 -0.872 1.64764 7.5538 0.003
135.00 0.4616 23.1225 0.012 -1.031 1.63477 7.5528 -0.001
140.00 0.6023 22.7284 -0.011 -1.219 1.62190 7.5547 -0.002

0.92780CH* +0.07220C4H,o (M = 19.0813 g - mol

115.00 0.1270 24.6285 -0.586 1.68465 7.5445 0.002
120.00 0.1824 24.2783 -0.693 1.67243 7.5418 -0.003
125.00 0.2549 23.8999 -0.782 1.65994 7.5446 -0.003

0.77982CH4 + 0.220 18C4H,o (M = 25.3085 g- mol -')

120.00 0.1702 21.6066 0.003 - 1.288 1.78385 9.5877 0.006
125.00 0.2369 21.3549 -0.002 -1.444 1.77286 9.5925 0.007
130.00 0.3228 21.1020 -0.010 -1.633 1.76202 9.5989 0.009
135.00 0.4291 20.8555 0.016 -1.879 1.75113 9.6013 0.008
140.00 0.5576 20.5978 -0.007 -2.151 1.73990 9.6049 0.007

0.77762CH4 +0.22238C4H,o (M = 25.4011 gmol -')

115.00 0.1179 21.8054 -0.027 -1.135 1.79554 9.6122 0.005
120.00 0.1699 21.5601 -0.001 - 1.273 1.78485 9.6180 0.007
125.00 0.2372 21.3164 0.037 -1.444 1.77382 9.6193 0.004
130.00 0.3230 21.0605 0.021 -1.623 1.76272 9.6249 0.005
135.00 0.4291 20.801 -0.009 -1.836 1.75139 9.6291 0.005
140.00 0.5579 20.5448 -0.021 -2.107 1.74020 9.6328 0.004

93
908 W. M. HAYNES

where V„ is volume of the mixture at a given temperature at saturation

the molar
pressure p, V*
molar volume of pure component at the same temperature at
is the i

saturation pressure pf, x, is the mole fraction of component i, and is the

isothermal compressibility of component i. The Clausius-Mossotti function 0cm is

defined by the expression:

<^cM = (£-l)/(£ + 2)p„, (2)

where p„ is the amount-of-substance density and e is the dielectric constant. Then

the excess Clausius-Mossotti function </>cm for a liquid mixture is defined,
analogously to V^, by the relation:

0CM = 0CM-L^i<^CM.P (3)

where </)cm refers to the Clausius-Mossotti function of the mixture at a given

temperature at saturation pressure and </»cm.b is the Clausius-Mossotti function of
pure component i at the same temperature and pressure as the mixture. Along an
isotherm, the Clausius-Mossotti function is a very slowly -varying function;
therefore, adjustments of saturated-liquid pure-component Clausius-Mossotti
values to the saturation pressure of the mixture are sufficiently small to neglect.
In the calculation of and 0cm, i>
the pure-component molar volumes of CH4,
(CH3)2CHCH3, and C4H10 were calculated from equations*^' used to fit the
experimental orthobaric liquid densities obtained with the first version of the
magnetic-suspension densimeter used in the LNG density project. For 2-methyl-
propane and n-butane, the equations were used for extrapolation below their triple-
point tempieratures. The molar volumes of CH4 were adjusted to the mixture
pressure using vapor pressures from Goodwin*'^' and isothermal compressibilities
from Rowlinson.'^^' The compressibility corrections for 2-methylpropane and n-
butane were sufficiently small to disregard. The dielectric constants of the pure
components, used in the calculation of 0cm. ^^^e taken from Straty and i'

Goodwin'*^' for CH4 and from Haynes and Younglove<^^' for (CH3)2CHCH3 and
C4H10.
The binary-mixture results presented in this paf>er, along with the previous pure-
fluid and binary-mixture results from the LNG density project, have been used to
optimize several mathematical models'^" for the prediction of LNG densities. The
most accurate and versatile of the models is the extended corresponding-states
model,'^""' which will be utilized here to demonstrate the consistency of the results
for (methane -I- 2-methylpropane) and (methane -f- n-butane) from the LNG density
project at this laboratory and to make comparisons with values for these mixtures
from other sources.
In figure 1 the experimental results for {xCH4 + (l -xXCH3)2CHCH3} from the
LNG density project and from Rodosevich and
compared with values
Miller*^°' are
calculated from the extended corresponding-states method. The differences are less
than 0.1 per cent for all points, including the values for x % 0.5 obtained earlier**' in
the LNG density project. Figure 2 shows comparisons of experimental results for
{xCH4-|-(l — x)C4Hio} from the LNG density project, from Miller,*^" and from
Shana'a and Canfield,*^^* with values predicted from the extended corresponding-

94
 _

OF {xCH^+(l -x){CH,)jCHCH,}
DENSITIES ^. AND {xCH^+d -x)C4»to} 909

1 1 1

-
a

e O o
Z »
e

100 110 120 130 140

T/K
FIGURE Comparison of experimental densities of {xCH4+(l -xXCHjjjCHCHj} from this work,
1.

from Hiza al.,^** and from Rodosevich and Miller/^"" with values calculated from the extended
et
corresponding-states model."""* O, This work, x = 0.92044; A, this work, x = 0.78329; j^c, Hiza et al,
X = 0.48687; V, Rodosevich and Miller, x = 0.9462; T, Rodosevich and Miller, x = 0.9152.

States model. In general, the agreement is good (within ±0.1 per cent) except for the
results for (0.92CH4 + 0.08C4Hioy*' obtained with the first version of the magnetic-
suspension densimeter used in the LNG density project. As reported earlier in this
paper, experimental difficulties were encountered with that apparatus in performing
measurements on methane-rich mixtures containing heavy hydrocarbons. Therefore,
the results for (O.92CH4-I-O.O8C4H10) were not included in the development of the
LNG mathematical models.***'
The excess molar volume is plotted as a function of x in figure 3 for isotherms

at 115 K for {xCH4 + (l -xXCH3)2CHCH3} and {xCH4-H(l -x)C4Hio}. The

smoothed curves represent values calculated from the extended corresponding-states
model. The results from the LNG density project are shown on the figure. The
maximum excess molar volume for {xCH4-t-(l — x)C4Hio} at 115 K,
— 1.24cm-'-mol~* at x = 0.65, was approximately 20 per cent larger than that
observed for {xCH4-l-(l — xXCH3)2CHCH3} at the same temperature.
For those compositions with points at five or more temperatures, the
experimental amount-of-substance densities p„ have been fitted as a function of
temperature T to the relation:

(Pn-Pn.c)/(niol-dm-^) = a(l-r/rj°-^^+ f - TO* (4)

i= 1

T/K
FIGURE Comparison of experimental densities of {xCH4-(-(l -x)C4H,o} from this work, from
2.

Hiza et a/.,**' from Miller,'^" and from Shana'a and Canfield,'^^' with values calculated from the extended
corresponding-states model.*'"'" O, This work, x = 0.92788; •, this work, x = 0.92780; A, this work,
X = 0.77982; A, this work, x = 0.77762; , Hiza et ai. x = 0.91674; Hiza, et ai. x = 0.58828; V.
Miller, x = 0.9506; T, Miller, x= 0.9206; 3^, Shana'a and Canfield, x = 0.8843.

9b
 910 W. M. HAYNES

0 0.2 0.4 0.6 0.8 1

X
FIGURE 3. Excess molar volumes of (a), {xCH^ + (l -xXCHjj^CHCHj} and (b),

{xCH4 + (l — x)C4H,o} as a function of x. Smoothed curves represent values calculated from the extended
corresponding-states model.""'" Points from the LNG density project are shown: O, This work; ,
Hiza et al.'*>

where and p„ are the critical temperature and the critical amount-of-substance
,,

density; a and 6, are coefficients determined by least squares. (This equation had
been used previously*^' to represent the orthobaric liquid densities of the pure
components*'" of LNG and For the
selected binary mixtures*** of these components.)
results presented here, onlytwo coefficients were statistically significant in the fits.
The critical temperatures and densities were obtained from Chueh and Prausnitz*^^'
using the pure-component critical-point values given in refei-ence 3. These fits were
relatively insensitive to the selected critical-point values. The results of the fits are
given in table 2. The maximum standard deviation was 0.028 per cent in the density,
which agrees with that observed in the other binary mixture measurements reported
earlier**' for the LNG density project. Deviations of the experimental densities from
values calculated from equation (4) are given in table 1.

The systematic errors involved in density measurements on mixtures with the

magnetic-suspension technique used here have been discussed in detail
elsewhere.*' *' The total uncertainty is taken as the square root of the sum of the
squares of the systematic errors plus an allowance of three times the standard

TABLE 2. Parameters of equation (4) where and ^ are the critical temperature and critical amount-of-
substance density of the mixture"

7; P-.c
Mixture a
K moldm"^

0.92044CH4 + 0.07956(CH3)2CHCH3 11.21295 10.64118 232.19 10.17 0.026

0.78329CH4 -l-0.21671(CH3)jCHCHj 6.265006 11.14928 285.80 9.77 0.007
0.92788CH4 + 0.072 2C4H o 1
,
9.448745 12.23371 233.14 10.99 0.010
O.77982CH4 + 0.2201 8C4H 0 ,
4.436119 12.72402 297.59 10.31 0.012
0.77762CH4-i-0.22238C4H,o 4.238446 12.93083 298.37 10.29 0.028

"The coefficients (a, /»,) were obtained from a least-squares program in which the experimental mass
densities to five digits were converted to amount-of-substance densities within the program.

96
 DENSITIES OF {xCH, +(1 -x)(CH,)jCHCH,} AND {xCH,+(I -xjQH.o} 91 1

deviation for random error. The justification for using the standard deviation from
equation (4) as the basis for estimating the random error was discussed in detail in
references 3 and 4. Combining the maximum standard deviation of 0.028 per cent
from the fits to equation (4) with the systematic errors from references 1 and 4, the
total uncertainty of a single density measurement for the mixtures studied here was
estimated to be ±0.12 per cent.
As noted, computed values of the excess Clausius-Mossotti function, as defined in
equations (2) and (3), are given in table 1. A distinct difference was observed in the

behavior of (/)cm for the two mixtures investigated here. For {xCH4 + (l — x)C4Hio},
both nonpolar species, the calculated values were less than 0. 1 per cent of the
mixture (pcM values. Within the measurements
precision for of the
{xCH4 + (l — x)C4Hio}, there was little, if any, evidence of a relation between the
<j!)cM
values and the composition. There was some indication that the (/>cm values
decreased with temperature for the mixture with x = 0.93.
For {xCH4 + (l — xXCH3)2CHCH3}, in which 2-methylpropane has a small dipole
moment, the (^cm values were more than 0.2 per cent of 0cm for x = 0.92 and as
large as 0.5 per cent of 0cm for ^ = 0-78. The large 0cm values observed for binary
mixtures containing a species with a small dipole moment compared with mixtures
of nonpolar species exemplified the behavior observed by Pan et a/.*^*' in dielectric-
constant measurements on mixtures of LNG components, including their results for
{0.91CH4 + 0.09(CH3)2CHCH3} and {O.95CH4 + O.O5C4H10}.

The contribution of M. J. Hiza in the preparation of mixtures is gratefully

acknowledged.

REFERENCES
1. Haynes, W. M.; Hiza, M. J.; Frederick, N. V. Rev. Sci. Instrum. 1976, 47, 1237.
2. Haynes, W. M.; Hiza, M. J. Advances in Cryogenic Engineering, Vol. 21. Timmerhaus, K. D.;
Weitzel, D. H.: editors. Plenum Press: New York. 1976, p. 516.
3. Haynes, W. M.; Hiza, M. J. J. Chem. Thermodynamics 1911, 9, 179.
4. Hiza, M. J.; Haynes, W. M.; Parrish, W. R. J. Chem. Thermodynamics 1977, 9, 873.
5. Haynes, W. M. Rev. Sci. Instrum. 1977, 48, 39,
6. Hiza, M. J.; Haynes, W. M. 7. Chem. Thermodynamics 1980, 12, 1.
7. Haynes, W. M. J. Chem. Thermodynamics 1982, 14, 603.
8. Haynes, W. M. J. Res. Nat. Bur. Stand. (U.S.). in press.
9. McCarty, R. D. Nat. Bur. Stand. (U.S.). Interagency Report NBSIR 77-867. 1977.
10. McCarty, R. D. Nat. Bur. Stand. (U.S.). Tech. Note 1030. 1980.
11. McCarty, R. D. J. Chem. Thermodynamics 1982, 14, 837.
12. Hiza, M. J. Fluid Pha.<ie Equilibria 1978, 2, 27.
13. Younglove, B. A.; Straty, G. C. Rev. Sci. Instrum. 1970, 41, 1087.
14. Pan, W. P.; Mady, M, H.; Miller, R. C. AIChEJ. 1975, 21, 283.
15. Frederick, N. V.; Haynes, W. M. Rev. Sci. Instrum. 1979, 50, 1 154.
16. Goodwin, R. D. Nat. Bur. Stand. (U.S.). Tech. Note 653. 1974.
17. Rowlinson, J. S. Liquids and Liquid Mixtures. 2nd edition. Butterworths: London, 1969.

18. Straty,G. C; Goodwin, R. D. Cryogenics 1973, 13, 712.

19. Haynes, W. M.; Younglove, B. A. Advances in Cryogenic Engineering Vol. 27. Fast, R. W.: editor.
Plenum Press: New York. 1982, p. 883.
20. Rodosevich, J. B.; Miller, R. C. AIChEJ. 1973, 19, 729.
21. Miller, R. C. Chem. Eng. 1974, 81, 134.
22. Shana'a, M. Y.; Canfield. F. B. Trans. Faraday Soc. 1968, 64, 2281.
23. Chueh, P. L.; Prausnitz, J. M. AIChEJ. 1973, 19, 729,

97
Reference [lb]

ADVANCES IN CRYOGENIC ENGINEERING, Proc. Cryogenic Engineering

Conf. (Boulder, Colo., Aug 2-5, 1977) Vol 23, K. D. Timmerhaus,
Editor. Plenum Press, New York (1978)

M—6
LIQUID MIXTURE EXCESS VOLUMES
AND TOTAL VAPOR PRESSURES USING
A MAGNETIC SUSPENSION DENSIMETER WITH
COMPOSITIONS DETERMINED BY
CHROMATOGRAPHIC ANALYSIS: METHANE
PLUS ETHANE*
M. J. Hiza and W. M. Haynes
National Bureau of Standards, Boulder, Colorado

INTRODUCTION
The measurements discussed in this paper are part of a comprehensive
experimental program to obtain orthobaric liquid densities for the major
components of LNG and their mixtures using a magnetic suspension densimeter.
The present measurements on methane plus ethane were made to evaluate vapor
recirculation as a method of liquid mixing and the consistency of liquid-phase
compositions determined by analysis of liquid samples. The results demonstrate the
feasibility of using a magnetic suspension densimeter to obtain accurate isothermal
phase equilibria and liquid density data simultaneously, with compositions deter-
mined by chromatographic analysis.
The magnetic suspension densimeter, developed for absolute measurements of
cryogenic liquid and liquid mixture densities, f''^] is a sophisticated but straightfor-
ward application of Archimedes' principle. Owing to the experimental procedures
involved in the determination of temperature-dependent parameters, this instru-
ment is particularly suited for efficient use in isothermal density measurements as a
function of cither pressure or composition.
In measurements of orthobaric liquid densities reported elsewhere for the
major components of LNG l''^'*) and their binary mixtures ['], the potential for
elHcient use of the densimeter at constant temperature was not utilized. In the
binary mixtures study ['] for example, measurements were made on gravimetrically
prepared mixtures as a function of temperature to minimize the uncertainty in
composition of the liquid mixture. Vapor recirculation was used only as a test for
nonhomogeneity of the liquid mixture. In that study, measurements were made on

* Work carried out at the National Bureau of Standards under the sponsorship of British Gas Corp.;
Chicago Bridge and Iron Co.; Columbia Gas Service Corp.; Distrigas Corp.; Easco Gas LNG, Inc.; EI
Paso Natural Gas; Ga^ dc France; Marathon Oil Co.; Mobil Oil Corp.; Natural Gas Pipeline Co.;
Phillips Petroleum Co.; Shell International Gas, Ltd.; Sonatrach; Southern California Gas Co.;
Tennessee Gas Pipeline Co.; Texas Eastern Transmission Co.; Tokyo Gas Co., Ltd.; and Trans-
continental Gas Pipe Line Corp., through a grant administered by the American Gas Association. Inc.
594

99
 Liquid Mixture Excess Volumes and Total Vapor Pressures 595

methane plus ethane mixtures at about 35 and 68 mole % methane from 105 to
130 K and at about 49 mole % methane from 105 to 140 K. Times involved in the
steps of the procedure restricted measurements to two data points per day. A
vacuum point at each experimental temperature was taken prior to measurements
on the liquid mixture, since the equilibrium cell had to be warmed nearly to room
temperature for effective removal of the heavy component from surfaces within the
cell. Thermal hysteresis of the magnetic moment of the buoy, although very
small, dictates that, for the most precise density measurement [ ] large thermal
cycles (e.g., from 100 to 300 K) should be avoided between the vacuum and liquid
points. Since the position of the buoy (which must be the same within about
0.002 mm for the vacuum and liquid points) is determined with a 1 25 x microscope,
ambient temperature must remain constant during these measurements. If a series
of density measurements can be made (at least within a single day) without
changing the experimental temperature, only one vacuum point would be required
for the series, and the experimental time per data point would be reduced
significantly.
The results of isothermal measurements of liquid mixture densities and the
corresponding total vapor pressures are reported for the methane plus ethane
system at 125 and 135 K with liquid-phase compositions determined by chromato-
graphic analysis. Excess volumes, computed for the mixtures studied, are compared
with those obtained from the densities of gravimetrically prepared mixtures and
with values calculated from a temperature-dependent Redlich-Kister equation
fitted to all the methane plus ethane data obtained with this apparatus. Excess
pressures, derived from the total vapor pressures, are compared with values from
the prepared mixtures and from phase equilibria data in the literature. In this
discussion, total vapor pressure is used interchangeably with mixture vapor pres-
sure and equilibrium pressure.

EXPERIMENTAL
A
schematic diagram of the apparatus as used in the present study is given in
Fig. 1. It incorporatesan equilibrium system patterned after the closed-loop vapor
recirculation apparatus used for liquid-vapKir equilibria measurements [^'^] and the
simplified one-coil magnetic suspension densimeter [^]. A window in the equili-
brium cell provides visual observation of the magnetic buoy and the liquid sample.
With the simplified version of the densimeter, a barium ferrite magnetic buoy
of known density is freely suspended by the force generated from the axial field of a
single air core solenoid. Measurement of the solenoid current needed to support
the buoy in liquid relative to that in vacuum at the same position and temperature
yields the liquid density directly. The relationship used to compute the liquid
density is

P/=P/.[l-(/y/.)l (1)

where pi and are the mass densities of the liquid and the magnetic buoy, and //
and are the solenoid currents required to support the buoy in the liquid and
vacuum at constant temperature and at the same position relative to the support
coil.
The procedures used in this study were basically the same as those followed in
measuring the liquid densities of gravimetrically prepared binary mixtures [^].
Solenoid currents were measured first with pure liquid methane in the cell and then

100
596 M. J. Hiza and W. M. Maynes

L J

Fig. 1. Schematic of experimental apparatus.

with the cell interest. (A methane point was taken

evacuated at the temperature of
each day as a controlon the measurement process, and these were used in a
statistical analysis of the data.) Then the mixture to be studied was introduced into
the cell, and the desired density and pressure measurements were made.
For measurements on the prepared binary mixtures [*], an effort was made to
reduce the vapor volume to a minimum. A 0 to 100 psi (0 to 0.69 MPa) quartz
Bourdon tube gauge with a free volume of approximately 0.5 cm^ was used for
equilibrium pressure measurements. The recirculation pump, with a free volume of
approximately 65 cm', was isolated during the time the cell was being completely
filled and during the density measurements. For binary mixtures containing
methane plus ethane, the pump was filled with pure methane gas at the mixture
vapor pressure so that the liquid sample could be mixed by vapor recirculation,
without significant pressure change, while monitoring the density. This procedure
was used as a test for nonhomogeneity of the liquid sample.
In contrast, for measurements on mixtures with compositions determined from
liquid sample analysis, a large vapor volume is desirable. A
large vapor volume
minimizes pressure changes in the equilibrium cell during liquid sample withdrawal,
and thus reduces changes in the liquid composition due to preferential vaporiza-

101
 Liquid Mixture Exceu Volumes and Total Vapor Pressures 597

tion. In the present study, equilibrium pressures were measured with a 0 to 100 psi
(0 to 0.69 MPa) quartz Bourdon tube gauge with an cstinnated free volume of
80 cm'. In addition, the recirculation pump was in communication with the equili-
brium system at all times, and only enough liquid was added to the cell to raise the
liquid level at least 1 cm above the top of the magnetic buoy when in suspension,
which allowed about 6 cm' of vapor space in the cell.
At 135 K, data were obtained at four compositions between 35 and 75 mole %
methane by starting with pure liquid ethane in the cell and adding incremental
amounts of pure liquid methane. At 125 K, data for an approximately cquimolar
mixture were obtained by adding pure liquid methane to the cell subsequent to
measurements on a gravimetrically prepared mixture at 35 mole %
methane. A
sufficient amount of the prepared mixture was removed through the liquid sample
line so that the liquid level, after adding methane, was a centimeter or so above the
suspended buoy. In each case the liquid sample was mixed by vapor recirculation.
The liquid sample line consists of a 25-cm length of 0.028-cm-ID tubing
exiting the bottom of the equilibrium cell and about 2 m of 0.048-cm-ID tubing
leading out of the cryostat to a small volume valve. The line to the chromatograph
consists of about 5 m of 0.159-cm-ID tubing. The sample receiver shown in Fig. 1
was sized to hold the gas equivalent of 1 cm' of liquid sample. This receiver
provides a sufficient volume of sample for several repetitive analyses and a method
of averaging composition variations which might occur during liquid sample with-
drawal. If vapor samples had been analyzed in this study, these would have been
withdrawn directly from the recirculation pump. Additional vapor volume probably
would have been necessary to provide sufficient sample for purging and analysis.
Different liquid sampling and analysis procedures were used at the two
temperatures of this study. At 135 K, before a sample was taken for analysis, the
line from the densimeter to the chromatograph and the sample valve were evacu-
ated and then purged with sample continuously for 6 min at a flow rate of approx-
imately 120 stp cm'/min. For the first data point, with sample line initially contain-
ing nearly pure ethane, it was necessary to repeat the density and pressure
measurement and sampling procedure to flush the residual ethane from the lines.
For these analyses, the sample receiver was not used, and the chromatograph, with
a 9-m liquid partition column, was calibrated with an equimolar methane plus
ethane standard. At 125 K, the same purging procedure was used, but the sample
was then introduced into the sample receiver which had been maintained under
vacuum. A recirculation pump ['] attached to the receiver was used to stir the
sample before analysis. Two analyses were made which were in agreement within
0.02 mole % of the composition of each component. In this case, the chromato-
graph with a 4-m silica gel column was calibrated with a 68 mole %
methane plus
ethane standard.
Tht: imprecision in measurement of component concentration was determined
to be about ±0.0.'^ mole % with either separation column. The uncertainty in
component concentration due only to errors in analyzer calibration is estimated to
be about ±0.05 mole %. Considering all other known sources of systematic error
in the density measurements (' *'] foi methane plus ethane mixtures of the
compositions studied here, an uncertainty of ±0.05 mole %
in component concen-
tration combined with the other systematic errors would result in an estimated total
systematic error of about ±0.05 % in the mixture density. This was determined
from the square root of the sum of the squares of the systematic errors. Similarly,
the estimated tot;il systematic error in the mixture vapor pressure measurement

102
598 M. J. Hiu and W. M. Haynn

due component concentration, gauge calibration, and temperature

to errors in
uncertainty approximately ±0.01 x lO' Pa.
is

Owing to the small number of data points obtained, it is not possible to make
an estimate of the random error in these measurements. In attempting to establish
reliable liquid sampling procedures for these measurements, it was quite apparent
that obtaining representative samples is not easily accomplished for systems of this
type, particularly at lower temperatures. Problems were encountered when there
was inadequate vapor volume to prevent preferential vaporization, low sample
driving force, and insufficient purging of the sampl^ lines to the analyzer.

RESULTS
The methane
plus ethane total vapor pressures, P„, and molar volumes, V„,
obtained study are given in Table I. Also given are the corresponding excess
in this
values, and V^, determined from the difference between the actual mixture
value and the mole fraction average of the pure component values at the conditions
of the rnixture. The pure component properties used by Hiza et al. ['] were also
used here.
The excess volumes in Table I and those obtained from measurements on the
gravi metrically prepared mixtures [ ] were fitted to a Redlich-Kister expansion of

the form

= XxX2[{ao + a,T + a2T^)+(bo + biT + b2T^)i2x:-\)

+ ico+c,T + C2T^)(2x,-lf]
(2)

where and X2 are the mole fractions of methane and ethane, respectively. 7'he
jci

least-squares coefficients (a, and b,) are given in Table II. The c, coefficients were
not statistically significant in representing excess volumes for this system. The
standard deviation, given in Table II in percent of the average mixture molar
volume (0.035%), is slightly higher than the comparable value, 0.030%, obtained
in fitting excess volumes from the data on prepared mixtures only The
differences between excess volumes obtained in the present study and values
calculated with the coefficients from Table II, in percent of the molar volumes, are
also given in Table I.

Table I. Orthobaric Liquid Molar Volumes, Total Vapor Pressures, and Cor-
responding Excess Functions for Methane Plus Ethane with Compositions Deter-
mined by Chromatographic Analysis
[ Vfaic
from equation (2) with coefficients given in Table II]

T, X, X,
"'^xlOO
K ClU QH, PaxlO'*' PaxlO"'* cni'/mole cm''/mole v„

135.00 0.3563 0.6437 2.043 + 0.280 46.056 -0.876 -0.020

0.4447 0.5553 2.505 +0.310 45.171 -1.026 -0.018
0.5934 0.4066 3.150 + 0.229 43.836 -1.126 +0.021
0.7415 0.2585 3.743 +0.099 42.690 -1.040 -0.066
125.00 0.5014 0.4986 1.515 +0.162 43.573 -O.S35 -0.032

* IxlO* Pa = 1 bar.

103
 Liquid Mixture Excess Volumes and Total Vapor IVessures 599

Table II. Coefficients for Equation (2)

from a Least^Squares Fit of all Excess
Volumes for Methane Pius Etiiane
Obtained from Measurements with the
Magnetic Suspension Densimeter

«0 -14.89469
Ol 0.2770596
az -0.001473116
bo -37.98245
0.6427122
-0.002783724
0.035%

In Fig. 2, the excess volumes are compared with those, obtained from
measurements on prepared mixtures at 125 and 135 K and curves showing the
composition dependence of excess volumes calculated with the coefficients in Table
II. At 135 K, the two prepared mixture points shown at about 35 and 68 mole %
methane were obtained by extrapolation from a least-squares fit of the densities of
these mixtures as a function of temperature [']. All others are experimental data.
From the several sources of methane j)lus ethane liquid-vapor equilibria data
listed in a recent survey of the literature [ ], excess pressures were taken from the
measurements of Miller and Staveley ['] at 115.77 K and of Wichterle and Kobay-
ashi at 130.37 K for comparison with excess pressures given here. This
comparison is given in Fig. 3. Miller and Staveley measured total vapor pressures of
prepared mixtures, while Wichterle and Kobayashi analyzed the liquid and vapor

METHANE HOLE FRACTION

0 0.2 0.4 O.e 0.8 1.0

Fij;. 2. Comparison of excess volumes for methane plus ethane from the present study with those from
measuremrnts on prepared mixtures anil with curves calculated from cq. (2).

104
600 M. J. Hiza and W. M. H«ynw

1 1 1 1

Preient Study
o Pr»p*rK) MUtur»»
• CompoiltloM by C4i Antlysli
D
O.JxlO - miler t St*vtl»y
—
Ulch(«rl» & Kob^yiihi. 1)0.3; K
135 Kjr^ >v

0.2x10
125 IC>,^0 O

0.1x10
//y y

1
1 1
0
0 0.2 0.4 0.6 O.B 1.0

HETHANE MOLE FRACTION

Fig. 3. Comparisons of excess pressures for methane plus ethane from the present study with those from
measurements on prepared mixtures and with values obtained from the hteraiure.

Compositions at the equilibrium pressure of the mixture. Two points from vapor-
pressure measurements at 125 K on prepared mixtures in the densimeter are also
shown. There appears to be a discrepancy between tHe present data and those of
Wichterle and Kobayashi. To be consistent, excess pressures at 130.37 K should
fall between the values at 125 and 135 K, while those of Wichterle and Kobayashi

are predominantly higher than the present values at 135 K. However, a data point
of the latter authors at about 60 mole %
methane, when plotted with these new
values and that of Miller and Staveley at the same composition, produces a smooth
curve of excess pressure as a function of temperature.

SUMMARY
The experimentalresults given here demonstrate the feasibility of efficiently
utilizing the magnetic suspension densimeter to obtain both isothermal phase
equilibria and liquid mixture density data, with compositions of the equilibrium
phases determined by chromatographic analysis. A more complete evaluation of
sampling procedures is needed to determine the best procedure to obtain
representative liquid mixture samples reliably. Although vapor-phase compositions
were not analyzed in the present study, obtaining representative samples of the
equilibrium vapor would not present a difficult problem. Using excess volumes
from liquid density data on accurately prepared mixtures as a reference, measure-
ment of the liquid mixture densities along with liquid-vapor equilibria measure-
ments provides an excellent means of assessing the reliability of liquid-phase
compositions obtained in the phase equilibria measurements.

REFERENCES
1. \V. M. Hayncs. M. J. Hiza, and N. V. Frederick. Rev. Sri. hisir. 47:12M (1976).
2. W. M. Waynes. Rev Sci. Instr. 48:39 (1977)
3. W. M. Haynes anil M, J. Hiza, / Chern. Vifrmodynamics 9:179 (1977).

105
 Liquid Mixture Excess Volumes and Total Vapor Pressures 601

4. W. M. Haynes and M. J. Hi/a, in: Adunnces in Cryogenic Engineering, Vol. 21, Plenum Press, New
York (1976), p. 516.
5. M. J. Hiza, W. M. Haynes. and W. R. Parriih, / Chem. Thermodynamics 9;873 (1977).
6. A. G. Duncan and M. J. Hiza, in: Advances in Cryogenic Engineering. Vol. 15, Plenum Press, New
York (1970), p. 42.
7. M. J. Hiza and A. G. Duncan. Rev. Sci. Instr. 40:513 (1969). ,

8. A. J. Kidnay. M. J. Hiza, and R. C. Miller. Cryogenics 13:575 (1973),

9. R. C. Miller and L. A. K. Staveley, in: Advances in Cryogenic Engineering, Vol. 21, Plenum Press,
New York (1976), p. 493.
10. I. Wichterle and R. Kobayashi, / Chem. Eng. Data 17:9 (1972).

DISCUSSION
Question by R. Kobayashi, Rice University: Are there any attempts to apply the maxnetic
densimeter to measure gas densities as well as liquids?
Answer by author: The precision and accuracy of a magnetic suspension densimeter, which is a
straightforward application of Archimedes' principle, depends on the difference between the density of
the float and the density of the fluid Thus, it is very difficult to make precise density measurements on a
low-density gas. It is one of our long-range goals. (We have tried some gas measurements in selected
regions.) For the extremely low gas densities you need a low-density float as well as very precise
measurements of the position of the float and of the force (current in electromagnetic coils) necessary to
support the float.

106
6.9 Reference [2U]

A- 145
J. Chem. Thermodynamics 1980, 12, 1-10

Orthobaric liquid densities and excess

volumes for multicomponent mixtures of
low molar-mass alkanes and nitrogen
between 105 and 125
M. J. HIZA and W. M. HAYNES
Thermophysical Properties Division,
Natiorml Engineering Laboratory,
National Bureau of Standards,
Boulder, Colorado 80303, U.S.A.

{Received 15 March 1979)

A magnetic suspension densimeter has been used to determine orthobaric liquid densities of
gravimetrically prepared multicomponent mixtures containing the major components of
liquefied natural gas, i.e. nitrogen, methane, ethane, propane, isobutane, and normal butane,
between 105 and 125 K. These results were obtained to provide a test of the capability of
mathematical models to predict the densities of liquefied natural-gas mixtures. Combinations of
the subject components were chosen to provide the most severe test of the models and the
possibility of using the measured densities to optimize parameters of the models. Deviations are
given between the experimental densities for each mixture and values predicted with an extended
corresponding-states model optimized to binary-mixture and pure-component orthobaric liquid
densities obtained with the same apparatus. Uncertainties of the present results are discussed in
relation to the experimental technique, the knowledge of the compositions of the liquid
mixtures, and the comparisons between the experimental and predicted results. Approximate
total vapor pressures are also given for each mixture at the temperatures studied.

1. Introduction

The present study is part of an experimental program to obtain orthobaric liquid

densities for the major components of liquefied natural gas (lng) and their mixtures,
primarily in the temperature range from 105 to 140 K. The overall purpose of the
program is to provide sufficiently accurate densities (±0.1 per cent) of uniformly high
precision ( + 0.02 per cent) for the pure components, the possible binary
combinations, and selected multicomponent mixtures, for optimization and testing of
prediction methods.

" This work was carried out at the National Bureau of Standards under the sp)onsorship of British Gas

Corp., Chicago Bridge and Iron Co., Columbia Gas Service Corp., Distrigas Corp., Easco Gas LNG, Inc.,
El Paso Natural Gas, Gaz de France, Marathon Oil Co., Mobil Oil Corp., Natural Gas Pipehne Co..
Phillips Petroleum Co., Shell International Gas Ltd., Sonatrach, Southern California Gas Co., Tennessee
Gas Pipeline Co., Texas Eastern Transmission Co., Tokyo Gas Co. Ltd., and Transcontinental Gas Pipe
Line Corp., through a grant administered by the American Gas Association, Inc.

0021-9614/80/010001 -I- 10 $01.00/0 © 1980 Academic Press Inc. (London) Ltd.

1

107
2 M. J. HIZA AND W. M. HAYNES

Previously, orthobaric liquid densities were reported for nitrogen,"* the low molar
mass alkanes: methane," '^*
ethane,'-^' propane,*^* isobutane,*^* and normal
butane and their binary mixtures,**' as determined with a magnetic suspension
densimeter.
In this paper, orthobaric liquid densities between 105 and 125 K, determined with
the same apparatus, are reported for 10 gravimetrically prepared multicomponent
mixtures containing methane. Total vapor pressures are also reported, but these are
considered approximate since the main focus was to assure the reliability of the
density measurements. As an indication of the consistency of the densities of the
multicomponent mixtures with those for the pure fluids and binary mixtures reported
earlier, deviations are given between the experimental densities for each
multicomponent mixture and values predicted with an extended corresponding states
LNG density model with parameters determined from the binary mixture and pure-
component results only. A brief discussion is also included on the consistency of other
multicomponent liquid densities in the literature with the results presented here.

2. Experimental

The magnetic suspension densimeter used in thisstudy has been discussed in detail
elsewhere.'^' ^' All of the results reported here were obtained with the one solenoid
arrangement'^' using the procedures followed in measuring the densities of the binary
liquid mixtures.*'*' The relation used to compute the density of the liquid with the one
solenoid arrangement is

P. = Pb{l-(V/v)}, (1)

where p, and are the mass densities of the liquid and the buoy, and /, and are the
solenoid currents required to support the buoy at the same temperature and at the
same position, relative to the solenoid, in the liquid and vacuum, respectively.
The multicomponent mixtures were prepared gravimetrically as gas mixtures in
thoroughly cleaned and dried metal cylinders, each with a free volume of about
3.3 dm'', using research-grade gases which were analysed chromatographically for
impurities before use. In addition, nitrogen and methane were passed through room-
temperature molecular-sieve adsorption columns to remove moisture and any heavy
hydrocarbon contaminants not detected by analysis. Estimated errors in component
mole fractions, based on six times the standard deviation of weighing,*"*' are given in
table 1 for both 1 and 5 mol of mixture prepared.
Composition, molar mass, and the total amount prepared are given in table 2 for
each of the multicomponent mixtures included in this study. Compositions given for
mixtures containing only hydrocarbons are the same as the prepared compositions.
Compositions given for mixtures containing nitrogen have been adjusted slightly (by
as much as 0.0005 in the mole fraction of and CH4) to include all or most of the
vapor-phase correction at each temperature studied. In making the composition
adjustments, the mole fraction of each component was rounded to the nearest 0.0001.
These vapor-phase corrections to the compositions were made to reduce the final

1U8
 LIQUID DENSITIES FOR MULTI COMPONENT lng MIXTURES 3

TABLE I. Uncertainty 5.v in mole fraction of each component in the prepared mixtures based on the total
amount of substance n prepared.

Component 5 a-

n-- 1 mol n= 5 mol

±0.00014 ±0.00003
CH4 ±0.00025 ±0.00005
C^H, ±0.00013 ±0.00003
C3H8 ±0.00009 + 0.00002
C4H10 ±0.00007 ±0.00001

TABLE 2. Multicomponent mixtures studied : x, mole fraction M, molar mass

; ; «, the total amount of substance
prepared

Mixture .xfN,) x(CHJ .v(CjH,) a{i-C4H,o) .x(n-C4H,o) M/g mol '

n/mol

A 0.34242 0.31372 0.34386 30.0903 1.78

B 0.80284 0.09902 0.09814 20.1852 4.93
C 0.85443 0.05042 0.04038 0.02577 0.02901 20.1885 1.89
D 0.3380 0.3414 0.3206 24.5861 4.66
E 0.1635 0.6704 0.1661 20.3301 4.66
F 0.0995 0.7977 0.1028 20.1181 4.67
G 0.0484 0.8526 0.0483 0.0507 18.7223 4.41
H 0.0490 0.8060 0.0468 0.0482 0.0500 20.7423 3.00
I 0.0554 0.7909 0.0560 0.0500 0.0477 20.9017 3.16
J 0.0425 0.8130 0.0475 0.0487 0.0241 0.0242 20.6168 3.09

adjustment of the mixture densities from a maximum of about 0.05 per cent to 0.01
per cent or less.

The prepared gas mixtures were condensed directly and continuously into the
windowed equilibrium cell through a capillary tube entering at the bottom until no
vapor space could be observed. This filling method provided continuous mixing of the
liquid while the gas mixture was being condensed into the cell. For one mixture
(mixture C in table 2) a comparison was made between the measured densities at
105 K obtained before and after the vapor phase was recirculated through the liquid
for several min at a bubble rate of 1 to 3 s"^ with an external vapor-recirculation
pump. No change in measured density was detected after recirculation. Subsequently,
liquid samples were withdrawn, and two comparisons were made chromato-
graphically with the composition of the prepared mixture. During withdrawal of
liquid samples, the recirculation pump (with a free volume of about 65 cm^) was kept
open to the vapor space of the equilibrium system to reduce the amount of CH4
vaporized from the liquid in the cell. The liquid samples were, nevertheless, lean in
CH4 by about 0.3 mole per cent, and mole fractions of heavy components in the
liquid were all higher than those in the prepared mixture. The mole fractions of CjH^,
and CjHg were higher by approximately 0.001, while the mole fractions of
(CHjIjCHCHj and C4H10 were higher by approximately 0.0006 and 0.0002,
respectively. Since it would be difficult, with this arrangement, to completely avoid

109
4 M. J. HIZA AND W. M. HAYNES

partial vaporization of methane from the Hquid samples, the main value of the
analysis was comparison of the mole fractions of the heavy components in the
in the
liquid samples to those in the prepared mixtures on a methane-free basis. On this
basis, the variations in the liquid-phase compositions observed for mixture C are not
considered unreasonable.
Attempts to study mixtures comparable to mixture C containing only one of the
butanes were unsuccessful. Differences between measured and predicted densities
were three to four times larger than expected, even after several min of vapor
recirculation, and the mole fractions of the heavy components from analysis of liquid
samples were inconsistent with those in the prepared mixture. Analysis of a liquid
sample for one mixture containing mole fraction approximately 0.05 of
(CH3)2CHCH3, without C4H10, for example, gave mole fractions of CjH^ and C3H8
in the liquid approximately 0.002 higher than those in the prepared mixture, and a
mole fraction of (CH3)2CHCH3 lower than that in the prepared mixture by about
0.0013. This inconsistency in composition could be interpreted as a non-uniform mole
fraction of (CH3)2CHCH3 in the liquid of up to 0.0033. Since the magnetic buoy is

essentially a point sensor relative to the size of the cell, the direction of error in
observed density depends on the composition of liquid actually surrounding the
buoy. From observations made during the course of this study and in the previous
study on binary mixtures,'*' it appeared that mixtures which were most likely to

present problems in filling the equilibrium cell with a homogeneous liquid sample of
the same composition as the prepared gas mixture were those of hydrocarbons only
which contain the butanes and mole fraction 0.85 or more of CH4. Though mixture C
fits these criteria, we did not detect any abnormalities in the liquid mixture during

measurement, as noted above, which we could consider significant.

3. Results and discussion

EXPERIMENTAL RESULTS
The experimental orthobaric liquid-mixture amount-of-substance densities are given
in table 3 as a function of temperature. Pressures listed in table 3 are considered only
approximate mixture vapor pressures, and in a few cases, measurements were not
made and values given were obtained by interpolation. Based on comparisons of
pressures measured for the binary mixtures with reported phase equilibria,*"**
pressures reported here are thought to be either equal to or slightly higher than the
true multicomponent mixture vapor pressures.
Also given in table 3 are the corresponding experimental excess volumes V^,
computed from

l^'=Kn-Kdea,. (2)
where
K,,,i = ^:XX{l+KrAPi-PJ}^ (3)

V is molar volume, p is saturation pressure, x is mole fraction, and k> is isothermal

compressibility. Subscripts m
and i refer to the mixture and pure components.

110
 '

LIQUID DENSITIES FOR MULTICOMPONENT lng MIXTURES 5

TABLE 3.Orthobaric amount-of-substance densities p„ of multicomponent mixtures of low molar-mass

alkanes and nitrogen : 7, temperature (IPTS-68); p, pressure K^, excess volume calculated from equations
;

(2) and (3); ^caic density calculated from the extended corresponding-states model as given in reference 18

'
Mixture T/K p/MPa pJmo\ dm '
V^lcm^ mol 10'(p-Pc.ic)P^'

A 110.00 0.0427 20.5342 -0.573 -0.03

115.00 0.0637 20.3497 -0.646 -0.02
115.00 0.0637 20.3567 -0.663 -hO.Ol
120.00 0.0914 20.1786 -0.766 -h0.06
120.00 0.0914 20.1803 -0.770 -t-0.07
125.00 0.128 19.9858 — 0.848 -1-0.04

B 105.00 0.0477 24.9975 -0.533 -1-0.02

110.00 0.0747 24.6696 -0.587 -0.02

115.00 0.112 24.3515 -0.676 4-0.01
120.00 0.162 24.0335 — 0.790 -1-0.06

C 105.00 0.0515 24.8775 -0.596 -f-0.18

110.00 0.0818 24.5382 -0.638 -f-0.11

115.00 0.119 24.2083 -0.716 -hO.lO

120.00 0.170 23.8859 — 0.832 -1-0.14

D 105.00 0.719 24.9084 -2.108 -fO.07

110.00 0.949 24.5350 — 2.764 -1-0.04

1 1 J.UU 1 7^7
1 .ZJ 1 Z*+.l*f — J .o81 u
1 fi — 0 06
E 105.00 0.369 25.8587 - 1 .069 -0.00
110.00 0.481 25.4598 -1.433 -0.01
115.00 0.618 25.0466 -2.052 -0.05
120.00 0.773 24.6359 -3.496 -0.04
F 105.00 0.282 25.3950 -0.845 -0.01
110,00 0.362 25.0153 -1.091 -0.04
115.00 0.455 24.6465 -1.542 -t-0.01

120.00 0.561 24.2667 -2.525 -1-0.04

G 105.00 0.158 25.8910 -0.513 -0.07

110.00 0.214 25.5081 -0.656 -0.07
115.00 0.283 25.1224 -0.900 -0.06
120.00 0.365 24.7283 -1.425 -0.04

H 105.00 0.204 24.7803 -0.726 -h0,03

110.00 0.255 24.4327 -0.870 -0,03
115.00 0.314 24.1039 -1.154 4-0.00
120.00 0.379 23.7707 -1.730 -1-0.05

I 105.00 0.193 24.8080 -0.786 -0.00

110.00 0.253 24.4664 -0.954 -0.06

J 105.00 0.183 24.8496 -0.717 + 0.04

110.00 0.252 24.5159 -0.870 -^0.03
115.00 0.338 24.1783 -1.109 -f-0,02

120.00 0.443 23.8577 -1.634 + 0.10

111
 :

6 M. J. HIZA AND W. M. HAYNES

respectively. Pure-component molar volumes were calculated from the fit of the
experimental results."'^* Isothermal compressibilities for nitrogen, methane, and
ethane were calculated from the expressions

K7-(N2)/MPa"' = -0.01032 + 0.51368/(126.2- T/K), (4)

Kr(CH4)/MPa-' = -0.001595 + 0.29488/(190.555- T/K), (5)

and
Kj-(C2H6)/MPa-' = {(r/250K) + 0.45} x 10-^ (6)

where T
thermodynamic temperature. These expressions are considered adequate
is

representation of isothermal compressibilities of nitrogen and methane from

Rowlinson,*^' those of methane from Goodwin,'^' and those for ethane from Miller.*®*
Isothermal-compressibility corrections were not made for molar volumes of propane,
isobutane, and normal butane. Vapor pressures for nitrogen, methane, and ethane
were taken from Strobridge,'^* Goodwin,'^* and Goodwin, Roder, and Straty,'*°*
respectively.

UNCERTAINTIES IN MEASURED LIQUID-MIXTURE DENSITIES

The uncertainty in measured density of a liquid mixture is inherently larger than for a
pure due to the added uncertainty in composition. Previously, detailed
fluid
discussions were given on random and systematic errors associated with magnetic
suspension densimeter measurements for pure liquids'" and binary liquid mixtures.***
In each case the total uncertainty was taken as the square root of the sum of the
squares of known systematic errors plus three times the standard deviation for
random error.
For measurements on pure liquid methane, the predominant component in
liquefied natural gas, the total uncertainty in density was estimated to be about ±0.1
per cent, with approximately equal contributions from random and systematic
For measurements on (methane + propane) liquid mixtures, which typify the
errors.*'*

binary liquid mixtures studied, the total uncertainty was estimated to be about +0.16
per cent with approximately the same contribution from known systematic errors as
in pure liquid methane measurements.*'** The random error in the (methane +
propane) density measurements was estimated from the standard deviation obtained
by fitting the excess volumes, for the four different compositions studied, to a
Redlich-Kister expansion as a function of both composition and temperature. Fitting
the results as a function of temperature at constant composition resulted in
significantly smaller standard deviations, comparable to that for pure methane.
However, it was felt that the evaluation based on both temperature and composition
would give a more realistic estimate of the random error.
not possible to obtain a standard deviation for the multicomponent-mixture
It is

results from which an estimate of random error can be made as a function of both
temperature and composition, nor are the results for any mixture sufficient to obtain
a meaningful standard deviation as a function of temperature only. However, there is

nz
 LIQUID DENSITIES FOR MULTICOMPONENT lng MIXTURES 7

no reason to believe that the random error in density measurements for the
multicomponent Hquid mixtures in the present study should be significantly larger
than that noted above for the (methane + propane) binary mixtures. Estimated
systematic errors are also about the same. The increased uncertainty in composition
of prepared mixtures due to the increased number of components present adds very
little to the total systematic error.
There is a potential source of error, suggested by our experimental observations,
for mixtures containing high mole fractions of CH4 with higher molar-mass alkanes
including either or both butanes, but with no nitrogen. These are mixtures for which
the vapor phase is essentially pure methane. Thus, as the mole fraction of CH4 in the
liquid mixture is increased, the difference between the mixture vapor pressure and
that for pure methane becomes quite small, and the probability increases for
condensing pure methane in the vapor space, particularly at the lowest temperatures,

while the prepared mixture is being condensed into the equilibrium cell. If methane
thus condensed were to accumulate in a location where it could not be readily
blended back into the liquid mixture, and if the mixture were homogeneous, analysis
of representative liquid samples would indicate the mixture to be lean in methane
similar to the effect of partial vaporization during liquid sampling. Consequently, this
experimental test alone would not be sufficient to confirm a possible error in
measured density due to such an occurrence.
In the top section of the equilibrium cell, the annulus volume (of approximately

0.3 cm^) between the vapor bulb and the cell wall"* is the only possible location
where methane might condense without being in good communication with the bulk
of the liquid mixture in the cell. If pure liquid methane (rather than methane vapor)
completely filled this annulus when mixture C, for example, was condensed into the
cell, the mole fraction of CH4 in the liquid mixture would be about 0.0024 lower and

the measured mass density would be about 0.2 per cent higher than for the prepared
mixture, with a variation in the density error of about 0.015 per cent between 105 and
120 K. For comparison, if methane were to condense in a 20 cm length of the vapor-
exit capillary with an internal diameter of 0.048 cm, the error introduced in the
densities for mixture C would be about 0.03 per cent, about the same as the
imprecision of measurement. Due to vapor-phase enhancement of the mole fraction
of CH4 in (nitrogen methane) gas mixtures in equilibrium with the liquid,"' ''' the
-I-

presence of nitrogen in the multicomponent liquid mixture, e.g. mixture J as against

mixture C, would reduce the probability of methane condensing out in the vapor
space.

DEVIATIONS FROM PREDICTED DENSITIES

Comparisons between experimental and predicted liquid densities for multi-
component mixtures of lng components primarily provide a test of the predictive (or
correlative) capability of a proposed lng density model. However, when the
parameters of a model are optimized only to the pure-fluid and binary-mixtures
results obtained in the same experimental apparatus, these comparisons also can
provide information on the consistency of the results for the multicomponent

113
8 M. J. HIZA AND W. M. HAYNES

mixtures with those for the pure fluids and binary mixtures, within known Hmitations
imposed development of the model.
in the
Of several models proposed for the prediction of orthobaric liquid densities of lng
mixtures,'*' the extended corresponding-states model'' ^' appears to be the most
general in application. McCarty'^*' has demonstrated that, when optimized to pure-
fluid and binary-mixtures densities obtained with the magnetic suspension densimeter

of the present study, all of the pure-fluid and binary-mixtures results are correlated
generally within their estimated accuracy, and, in most cases, nearly within the
precision of measurement. It was
comparisons of the values for
felt that
multicomponent mixtures with values predicted from this model would be most likely
to identify inconsistencies.
Percent deviations of the experimental orthobaric liquid amount-of-substance
densities from values predicted with the extended corresponding-states model,
without further modification, are also given in table 3. The average deviation or bias
is -1-0.016 per cent, and the average absolute deviation 0.05 per cent. The maximum
is

deviation is -1-0.18 per cent for the 105 K point for mixture C. Only the results for
mixture C show a systematic bias of 0.1 per cent or larger. Excluding the results for
mixture C results in an average deviation or bias of -1-0.003 per cent, an average
absolute deviation of 0.04 per cent, and a maximum deviation of -1-0.1 per cent.
It is also worth noting the significant results of comparisons made with values
predicted from two less complex models, which are more limited in their application.
These are both excess-volume models, the empirical model of Hiza'^'* and the
modified hard-sphere model of Rodosevich and Miller,''^' with parameters optimized
to the same pure-fluid and binary-mixture values. Excluding results outside the
applicable range of each model, deviations of experimental from predicted densities
are in substantial agreement with those of the extended corresponding-states model.
Excluding the results for mixtures D and E and the 120 K point for mixture F, the
maximum deviation from values predicted with the empirical excess-volume model is

-1-0.14 per cent for the 105 K point for mixture C. Excluding the results for mixtures
containing nitrogen at the highest temperatures, the 120 K points for mixtures E
through J and the 115 K point for mixture D, the maximum deviation from values
predicted with the hard-sphere model is' -I- 0.1 8 per cent, also for the 105 K point for
mixture C. In fact, deviations of the densities for mixture C from values predicted with
the hard-sphere model are the same as those given in table 3 for the extended
corresponding-states model.
Though the densities for mixture C are systematically larger than the predicted
values by 0.1 per cent or more, the deviations given in table 3 generally are random
and, for the most part, are significantly smaller than ±0.1 per cent.

CONSISTENCY OF OTHER MULTICOMPONENT lng MIXTURE DENSITIES

Orthobaric liquid densities for multicomponent mixtures of lng components are
open literature, and only some of those available approach the accuracy
scarce in the
goal of the present study. Our primary source of references is a fluid-mixtures
bibliography"^' current to January 1975 and more recent references from a file

114
 LIQUID DLNSn iliS FOR MULTI( 0MP0N|;NT i N(i MIX IUIU-S 9

maintained to update that bibliography. A summary of the results of comparisons of

literature data with values predicted with the extended corresponding-states model
are included here as an indication of the precision and consistency of available values
with those of the present study. The values are divided, for purposes of discussion,
into two groups: first, those with an apparent imprecision of more than ±0.3 per
cent, and second, those with an apparent imprecision of less than ±0.3 per cent.
In the first group are the results reported by Morlet,*^^' Gonzalez and Lee,*^''
Klosek and McKinley,'*^* and Jensen and Kurata.*^^' Morlet'^°* reported results
graphically for several ternary mixtures of (methane + ethane + propane) at 90.2 K.
These results appear reproducible within about ±1.0 per cent, and tabulated
densities for three compositions from a fit of them deviate from predicted values by
— 0.5 to - 1.1 per cent. Gonzalez and Lee*^'* reported bubble-point densities for 10
methane-rich mixtures, simulating natural gas, down to 99.8 K. For the four
mixtures not containing helium or hexanes and higher hydrocarbons, comparisons
were made with predicted values at the lowest temperature reported. For three of the
mixtures at 116.5 K, the densities deviate from predicted values by —0.60 per cent,
+ 0.25 The density for the fourth mixture at 144.3 K
per cent, and —26.15 per cent.
deviates from the predicted value by —12.14 per cent. Klosek and McKinley**''*
reported densities for 13 multicomponent mixtures with three to seven components
from 94.37 to 125.04 K. Though their values deviate from predicted values by ± 1 per
cent, the average deviation or bias is only +0.04 per cent, the average absolute
deviation being 0.36 per cent. Since the values for each mixture generally deviate from
predicted values systematically, the variation may be due largely to an uncertainty in
composition. Jensen and Kurata'^^' reported results for essentially three methane-
rich mixtures containing nitrogen, ethane, propane, and normal butane from 98.15 to
113.15 K. These values appear to have an imprecision of about ±0.5 per cent. From
comparisons made at 108.15 and 113.15 K, they deviate from predicted values by
+ 1.03 to +2.21 per cent.
In the second group are the results reported by Shana'a and Canfield,*^^'
Rodosevich and Miller,*^*' Miller,*^' Orrit and Laupretre,*^*' and Miller and Hiza.*"*
Shana'a and Canfield'^^' reported results for two methane + ethane + propane
mixtures at 108.15 K. These deviate from predicted values by —0.09 per cent and
+ 0.15 per cent. Rodosevich and Miller*^"^' and Miller'^' reported results for three
ternary mixtures and one quaternary mixture at 108.00 and 1 15.00 K. In addition,
results included in reference 26 for one ternary mixture obtained by Rodosevich at
108.00, 110.00, and 115.00 K
were included here. From predicted values, the average
deviation or bias of these is —0.03 per
cent, the average absolute deviation is 0.03 per
cent, and the maximum deviation is —0.06 per cent. Very recently, Orrit and
Laupretre*^^' reported a sample of their results for five methane-rich mixtures of four
to eight components containing nitrogen and hydrocarbons through pentane from
95.50 to 123.61 K. From predicted values, the average deviation or bias of these
values is —0.10 per cent, the average absolute deviation is 0.10 per cent, and the
maximum deviation is —0.175 per cent.
The last investigation in the second group, that of Miller and Hiza,'^^' is not
completely independent of the present study. Measurements were made between 105

115
 ;

10 M. J. mZA AND W. M. IIAYNi:s

and 1 15 K at the University of Wyoming with Miller's apparatus calibrated on pure-

methane values obtained with the magnetic suspension densimeter of the present
study.' Mixtures studied were prepared in our laboratory. These included two
mixtures for which results are reported here (mixtures H and I) and two similar four-
component mixtures without nitrogen. From predicted values, the average deviation
or bias of these values is -1-0.04 per cent, the average absolute deviation is 0.04 per
cent, and the maximum deviation is -f- 0.1 02 per cent (for mixture I at 105 K). Though
there is a slight bias, these deviations are consistent with the accuracy of the
calibration.
Additional measurements are now in progress with a new design of the magnetic
suspension densimeter, significantly different than that used in the present study.
Measurements are being made on both binary and multicomponent mixtures for
which the data are insufficient, and for multicomponent mixtures containing the
pentanes. These new results will be reported in subsequent papers.

The authors acknowledge the contributions of M. J. Brown and R. D. McCarty

during the course of the present study.

REFERENCES
1. Haynes, W. M.; Hiza, M. J.; Frederick, N. V. Rev. Sci. Insrmm. 1976, 47, 1237.
2. Haynes. W. M.; Hiza. M. J. J. Chem. Thermodynamics 1977, 9, 179.
3. Haynes, W. M.; Hiza, M. J. In Advances in Cryogenic Engineering, Vol. 21. Timmerhaus, K. D.;
Weitzel, D. H.: editors. Plenum Press; New York. 1976. p. 516.
4. Hiza, M. J.; Haynes, W. M.; Parrish. W. R. J. Chem. Thermodynamics 1977, 9, 873.
5. Haynes, W. M. Rev. Sci. Instrum. 1977, 48, 39.
6. Rowlinson, J. S. Liquids and Liquid Mixtures. 2nd Ed. Butterworths London. 1969. :

7. Goodwin, R. D. Nat. Bur. Stand. U.S., Technical Note 653, 1974.

8. Miller. R. C. Chem. Eng. 1974, 81, 134.
9. Strobridge, T. R. Nat. Bur. Stand. U.S., Technical Note 129, 1962.
10. Goodwin, R. D.; Roder, H. M.; Straty, G. C. Nat. Bur. Stand. U.S., Technical Note 684, 1976.
11. Miller, R. C; Kidnay, A. J., Hiza, M. J, Am. Inst. Chem. Eng. J. 1973, 19, 145.
12. Kidnay, A, J.; Miller, R. C; Parrish, W. R.; Hiza, M. J. Cryogenics 1975, 15. 531.
13. Klosek, J.; McKinley, C. In Proceedings First International Conference on LNG. White, J. W.
Neumann, A. E. S.: editors. Institute of Gas Technology: Chicago, III. 1968, Paper 22, Session 5.
14. Albright, M. A. Nat. Gas Proc. Assn. Tech. Publ. TP-3, 1973.
15. Mollerup, J.; Rowlinson, J. S. Chem. Eng. Sci. 1973, 29, 1373.
16. Rodosevich, J. B.; Miller, R. C. In Advances in Cryogenic Engineering, Vol. 19. Timmerhaus, K. D.:
editor. Plenum Press: New York. 1974. p. 339.
17. Hiza, M. J. Fluid Phase Equilibria 1978, 2, 27.
18. McCarty, R. D. Nat. Bur. Stand. U.S.. Interagency Rept. NBSIR 77-867, 1977, 60 pp. See also:
Haynes, W. M.; Hiza, M. J.; McCarty, R. D. In: Proceedings Fifth International Conference on
LNG, Vol. 2. White. J. W.; McGrew, W.; Farmer, S.; Hansen, D.; Jacobshagen, H.: editors.
Institute of Gas Technology: Chicago, 111. 1977, Paper 11, Session III.
19. Hiza, M. J.; Kidnay. A. J.; Miller, R. C. Equilibrium Properties of Fluid Mixtures, A Bibliography of
Data on Fluids of Cryogenic Interest. NSRDS Bibliographic Series. IFI Plenum: New York. 1975.
20. Morlet, Rev. Inst. Franc. Petrol. 1963, 18, 127.
J,

21. Gonzalez,M. H.; Lee, A. L.J. Chem. Eng. Data 1968, 13, 172.
22. Jensen, R. H.; Kurata, F, J. Petrol. Technol. 1969, 21, 683.
23. Shana'a, M. Y.; Canfield, F, B. Trans. Faraday Soc. 1968, 64, 2281.
24. Rodosevich, J. B.; Miller, R. C. Am.
Inst. Chem. Eng. J. 1973, 19, 729.
25. Orrit, J. E.; Laupretre. J. M.
Advances in Cryogenic Engineering, Vol.
In 23. Timmerhaus, K. D.:
editor. Plenum Press: New York. 1978, p. 566.
26. MUler, R. C; Hiza, M. J. Fluid Phase Equilibria 1978, 2, 49.

116
 7

b.lu Reference [21]

A-279
J. Chem. Thermodynamics 1982, 14, 603-612

Measurements of orthobaric-liquid
densities of multicomponent
mixtures of lng components
(N2, CH4, C2H6, C3H8, CH3CH(CH3)CH3,
C4Hio» CH3CH(CH3)C2H5, and C5H^2)
between 110 and 130

W. M. HAYNES
Thermophysical Properties Division, National Engineering Laboratory,
National Bureau of Standards. Boulder, Colorado 80303, U.S.A.

{Received 14 September 1981)

A magnetic suspension densimeter has been used to measure the orthobaric-liquid densities of 1
multicomponent mixtures of the major components of liquefied natural gas (lng) at
temperatures from 110 to 130 K. These mixtures ranged from a ternary mixture containmg
nitrogen, methane, and butane to 4-to-8-component methane-rich (74 to 90 moles per cent)
mixtures containmg up to 5 moles per cent of nitrogen, 16 moles percent of ethane, 7 moles
per cent of propane, 5 moles per cent of the butanes, and 0.44 mole per cent of the pentanes.
Some of the compositions were selected to simulate commercial lng mixtures. Results of vapor-
pressure measurements are also presented. The major purpose of this work was to obtain
muliicomfKinent-mixture values that could be used to test mathematical models that have been
developed for the prediction of lng densities. To demonstrate the consistency of the
multicomponent-mixture values, comparisons are presented between experimental densities and
calculated values from an extended correspwnding-states method that was optimized to pure-
fluid and binary-mixture results from the lng density project here. The total uncertainty of a
single density measurement is estimated to be approximately 0.1 percent, which includes an
allowance of three times the standard deviation for random error. The imprecision of
measurement is a few parts in 10*.

1. Introduction

A project has been carried out at this laboratory to provide one or more
mathematical models that can be used to predict the density of a liquefied natural gas
(lng) mixture at saturation within 0.1 per cent of its true value from an input of the

° This work was carried out at the National Bureau of Standards under the sponsorship of British Gas

Corp., Chicago Bridge and Iron Co., Columbia Gas Service Corp., Distrigas Corp., Easco Gas LNG, Inc.,
El Paso Natural Gas, Gaz de France, Marathon Oil Co., Mobil Oil Corp., Natural Gas Pipeline Co.,
Phillips Petroleum Co., Shell International Gas. Ltd., Sonatrach, Southern California Gas Co., Tennessee
Gas Pijjeline Co., Texas Eastern Transmission Co., Tokyo Gas Co., Ltd., and Transcontinental Gas Pipw
Line Corp., through a grant administered by the American Gas Association, Inc.

37

117
604 W. M. HAYNES

temperature, composition, and pressure of the liquid mixture. Such a model would
commodity. In order to develop
serve as a basis for custody transfer of this valuable
and test such models, an accurate and consistent set of densities was needed for the
major components of lng and for mixtures of these components. Results from this
project have been previously reported for (a) nitrogen,*' ' methane,''-^' ethane,'^*
propane,'^* isobutane.*^' and normal butane*^-^' (major components of lng), (b)
binary mixtures'* of these components, including all combinations except
(nitrogen + isobutane) and (nitrogen + normal butane), and (c) selected
multicomponent mixtures*^' of these components with an emphasis on ternary
mixtures containing methane and binary combinations of nitrogen, ethane, and
propane. All of these results were obtained with a magnetic suspension densimeter,*"
an instrument based on an application of Archimedes' principle.
In an earlier paper giving results for multicomponent mixtures,*^' reference was
made to experimental problems that resulted in inconsistencies in densities for
methane-rich (80 moles percent or more) mixtures containing only hydrocarbons
with approximately 5 moles per cent of butanes. The resolution of these
inconsistencies was one of the major objectives of the present series of measurements.
Also, more results were needed for 5- and 6-component LNO-like mixtures containing
less than 5 moles per cent of the butanes and/or less than 5 moles per cent of nitrogen,

and for 7- and 8-component LNO-like mixtures containing up to 0.44 mole per cent of
the pentanes to test and define the limits of the models for prediction of densities for
compositions approximating those of commercial lng. These types of mixtures had
not been investigated in the first series of measurements. (In the mathematical
models,'^"^' results for pentanes were obtained from the literature""' from papers
that also included results for the other components of lng that agreed with results
from the present project. The behavior of binary mixtures containing pentanes was
predicted by extrapolating the results of binary-mixture measurements on other lng
components.)
The results presented here were obtained with a new magnetic suspension
densi^neter*'" that was significantly different from, and more versatile than, the
apparatus*'' that was used for most of the previous measurements of this project.
However, the new apparatus did employ the same technique for determining densities
and gave results for pure fluids and binary mixtures that agreed, within the precision
of the measurements, with values obtained with the previous apparatus. [The new
apparatus was constructed because there was a need for a higher-pressure capability
(35MPa) for the magnetic suspension densimeter in other research projects.]
New or improved features" " of the new apparatus that were pertinent to the study
of multicomponent mixtures here are as follows: (a) improved temperature control
and measurements, (b) improved handling of mixtures, especially the filling of the cell
so as to obtain homogeneous liquid mixtures of known composition, (c) improved
vapor-pressure measurements, and (d) a capacitor for dielectric-constant
measurements. (The dielectric constants for the multicomponent mixtures have not
been analyzed as yet due to a lack of values for some of the pure components.) The
new apparatus was also used for a study of (methane -I- isobutane) and
(methane -h normal butane).***

118
 DENSITIES OF lng MIXTURES 605

2. Experimental

The method of measurement,*' the apparatus," " and the experimental procedures
and measurements" '^ *'^ " '^'
have been described in detail elsewhere. Only those
details that are different or essential for this papjer will be discussed here. A one-
solenoid magnetic suspension system"^' was used to obtain the results reported here.
Measurements on liquid methane were used as a control on the measurement
process during the course of the mixture measurements. A density measurement on
liquid methane was performed before each filling of the cell with a new mixture

sample. This procedure was followed to insure that the warm-up and cool-down
cycles of the apparatus did not affect the apparent position of the buoy from liquid-to-
vacuum measurements. After mixture measurements were completed for each sample,
the cell was warmed to room temperature to facilitate complete removal of the heavy
hydrocarbons. A maximum of four points at 5 K increments was obtained for each
sample.
There were a two at the top and two at
total of four capillaries entering into the cell,
the bottom. A and one at bottom) was available for
pair of these (one at top
recirculation of the vapor through the liquid. The second capillary into the bottom of
the cell was available for liquid sampling. The second line into the top of the cell was
used solely for pressure measurements. This was the only capillary not used in the
filling process.
The mixtures were condensed directly and continuously into the equilibrium cell
either through the two capillaries at the bottom of the cell (analogous to vapor
recirculation) or through the recirculation and liquid-sampling capillaries at the top
and bottom simultaneously. An effort was made to determine if the homogeneity of
the liquid sample in the cell was dependent on the filling procedure. As indicated by
the consistency of the densities, it was found that the measured densities were
independent of the filling routines attempted.
All measurements were on saturated liquids. In the present work it was impossible,
and not essential, to observe the liquid- vapor interface through a small high-pressure
sapphire window positioned near the mid-point of the cell along its length. The cell

was initially filled with each new sample until the mixture was in the compressed
liquid, or at a pressure a few tenths of a MPa above its vapor pressure. Then, small
amounts of liquid were withdrawn from the bottom of the cell until the saturation
pressure was attained, as denoted by reproducibility of the pressure readings. It was
extremely important to have as small vapor volume as possible in the sample space.
As the cell was warmed to the temperature of the next point, the pressure of the
mixture was monitored as a function of temperature. This information could be used
to calculate the amount of vapor volume in the cell for the prior point and to confirm

that the liquid mixture had been at saturation. The vapor volume in the cell was
sufficiently small (0.2 cm^) so that the liquid went into the compressed state for a

temperature change of less than 2 K.

Itwas thought that the most accurate method to determine the compositions of the
liquid mixtures was to prepare them by gravimetric means under carefully controlled
conditions and, subsequently, to introduce them in a mixed state into the equilibrium

119
 —

606 W. M HAYNES
chamber. The preparation of mixtures of lng components, with the exception of
pentanes, has been described in previous papers.**-^' The pentane fractions were
obtained from bottles of commercially available research-grade liquids. The lowest
purities, as specified by the supplier from freezing-temp>erature determinations, were
99.99 and 99.93 moles per cent for isopentane and normal pentane, respectively. The
most probable impurity in normal pentane was isopentane while those for isopentane
were neopentane and normal pentane.
With one exception (mixture C in table 1,3.1 mol), at least 4 mol was prepared for
each mixture investigated here. This corresponds to less than 0.0001 uncertainty in
mole fraction of methane in the prepared mixture and considerably less uncertainty
for each of the other components. Due to insufficient mixing in a base mixture, the
compositions of three of the mixtures (mixtures m, n, and p) were determined by
analyses of samples using a gas chromatograph with a thermal-conductivity
detector.*'^' The uncertainty in the compositions determined in this manner
corresponds to an uncertainty of roughly 0.03 per cent in the density. At least six
samples, with subsequent analyses, were taken for each mixture. Half of these samples
were taken after heating the cylinders, in which the gas mixtures were contained, for
approximately 30 min with an infrared lamp. The compositions for the six samples
agreed within experimental uncertainty and did not exhibit any dependence on the
temperature of the mixture being tested.
The total liquid volume within the cell was approximately 43 cm^, as determined
by filling the cell with water. The total vapor volume of the sample space was
approximately 7.4 cm^ and can be divided into three parts as follows (a) inside cell
:

0.2 cm-*, (b) access tubing within cryostat —

1 cm^, and (c) access tubing and pressure

gauge outside cryostat— 6.2 cm^. The volumes of the tubing and pressure gauge were
computed from known dimensions. The composition of the liquid in the cell should
be slightly different from the prepared mixture composition as a result of the vapor
volume. The magnitude of the adjustment to the measured densities to account for
this composition change, so that the experimental densities can be assigned to the
prepared mixture composition, is discussed in the next section.
For allmeasurements reported here, the cell was always isolated from a
recirculation pump that could be used to mix the liquid inside the cell. Based on the
consistency of the densities for different filling procedures and new mixture samples
and from comparisons of the experimental densities with predicted values, it was not
deemed necessary to recirculate the vapor through the liquid to mix the samples or to
take liquid samples and analyze them. There were other means for gaining
information on the reproducibility of liquid compositions within the cell for different
samples from the same prepared mixture and on the homogeneity of the liquid
sample, as discussed below.
Vapor pressures were measured for each of the mixtures investigated here using a
spiralquartz Bourdon-tube gauge (0 to 1.38 MPa) calibrated against an air dead-
weight gauge. The maximum uncertainty in the gauge calibration was estimated to be
approximately 70 Pa. Vapor-pressure measurements for different samples and filling
procedures generally agreed within 3 x lO"'^ MPa.
A capacitor composed of slotted concentric cylinders located in the top part of the

120
 :

DENSITIES OF lng MIXTURES 607

cell was used for dielectric-constant measurements on the mixtures. The capacitor had
a vacuum capacitance made routinely to within
of 19.8 pF. Readings could be
0.0001 pF. This device was the most sensitive monitor within the sample space to
determine the reproducibility of liquid compositions for different filling procedures
and for different samples from the same mixture.
Although the cell was monitored for temperature gradients along its length using
vapor bulbs at the top and bottom of the cell in only a few runs, a differential
thermocouple with junctions at the same locations as the vapor bulbs was used in all
runs. The vapor-bulb measurements were used as a control on the thermocouple
measurements. The reflux tube was always evacuated while performing
measurements. At temperatures between 110 and 130 K, greatest power input into
the cell to balance the cooling was 0.5 W. The guard ring was controlled at a
temperature somewhat above the cell tempjerature. Under the above conditions, no

problems were encountered with temperature gradients along the length of the cell
within the precision of the vapor-pressure measurements.
The vapor pressures and dielectric constants were obtained for different portions
(different heights) of the liquid sample from that of the density determination.
Assuming an isothermal sample space, information on composition (or density)
gradients along the length of the cell could be gained from the various measurements
at different positions within the sample space.

3. Results

Experimental orthobaric liquid densities and vapor pressures of mixtures of lng

components are presented in table I as a function of temperature (IPTS-68). The
compositions presented were determined gravimetrically except those for
in this table

mixtures m, n, and p. The compositions of these mixtures were determined by

analyses of the prepared mixtures using a gas chromatograph with a thermal
conductivity detector, as discussed in the preceding section.
Small corrections were made to the measured densities to account for differences in
compositions between the prepared mixtures and the liquid mixtures in the
equilibrium which resulted from some of the sample space being occupied with
cell,

vapor. The largest adjustment was less than 0.05 per cent, which was for mixture a at
125 K. All other corrections were less than 0.02 per cent. For all mixtures containing
hydrocarbons only, it was assumed that the vapor composition was 100 per cent
methane. For those mixtures containing nitrogen, the vapor compositions were
estimated from partial-pressure calculations using available phase-equilibria data for
^'
(nitrogen -I- methane)."*' '

Excess volumes defined by the relation

V^=V^-Yx,V,{\+K,(p-pJ\, (1)
i

are also given in table 1. In this expression V„ is the molar volume of mixture at a
given temperature at saturation pressure is the molar volume of pure

I'll
608 W. M HAYNES
TABLE 1. Orthobaric-liquid densities of multicomponent mixtures of lng components. M, Molar mass;

T. temperature (IPTS-68); p, pressure; eip.- experimental amount-of-substance density, /»j,p„

experimental density; V^, excess volume calculated from equation (1 ); density calculated from
the extended corresponding-states method as given in references 7 to 9

T _p_ P..C.P, 10^(P„p.-p„n.)

K MPa mol dm"^ cm' mol"' p„„

(a ) (0.0593 1 N 2 + 0.8907 ICH4 + O.O4998C4 H 1 0)

(M = 18.8562 g mol"')
110.00 0.2400 25.3450 - 0.713 + 0.03
115.00 0.3145 24.9440 -0.983 +0.01
120.00 0.4082 24.5383 -1.595 +0.00
125.00 0.5196 24.1141 -0.04

(b) {0.86040CH4+0.04600C2Hf, + 0.04790C3H8+0.04570CH3CH(CHj)CH3|

(M = 19.9552 g mol"')
115.00 0.1186 24.2654 -0.631 -0.03
120.00 0.1710 23.9371 -0.741 +0.01
125.00 0.2387 23.5868 -0.842 -0.01
130.00 0.3248 23.2331 -0.970 -0.02
135.00 0.4320 22.8637 -1.110 -0.07

(c) {0.85378CH4+0.05178C,Hb+0.04703C3H8+0.04741CH3CH(CH3)CH3l
(M = 20.0838 g mol-')

115.00 0.1191 24.2100 - 0.640 - 0.03

120.00 0.1706 23.8779 -0.739 -0.02
125.00 0.2379 23.5324 -0.843 -0.04
130.00 0.3238 23.1834 -0.975 -0.05

(d) (0.85133CH4 + 0.05759C2H6 + 0.04808C3H8+0.04300C4H,o)

(M = 20.0092 g mor')
.115.00 0.1180 24.3243 -0.675 -0.01
120.00 0.1700 23.9965 -0.785 +0.03
125.00 0.2374 23.6586 -0.907 +0.05
130.00 0.3232 23.3108 -1.045 +0.05
135.00 0.4301 22.9634 -1.226 +0.09

(e) (0.84566CH4+0.07924C2H^+0.05060C3H8 + 0.02450C4H,o)

(M = 19.6051 gmol"')
115.00 0.1167 24.5569 -0.621 -0.01
120.00 0.1683 24.2126 -0.711 -0.01
125.00 0.2350 23.8698 -0.832 +0.03
130.00 0.3201 23.5204 -0.975 +0.07

(f) i0.04801N,+0.8094OCH4+0.04542CjH<, + 0.05050CjH8+0.04667CH3CH(CH3)CH3|

. (M = 20.6355 g- mor")
115.00 0.3005 24.1487 -1.129 - 0.04
120.00 0.3863 23.8075 -1.679 -0.03
125.00 0.4874 23.4518 -0.05
130.00 0.6125 23.0893 - 0.07
 — )

DENSITIES OF LNG MIXTURES 609

TABLE 1 continued

K MPa mol dm"^ cm^ mo!"' p^m

(g) (0.02628N J + 0.8 249CH4 + 0.08484C2 H<, + 0.0493 Cj Hg + O.O27O8C4 H

1 1
, 0)

(M = 20.0706 g mol - '

115.00 0.2214 24.4562 -0.908 -0.02

120.00 0.2874 24.1119 -1.254 -0.01
125.00 0.3768 23.7507 -0.05
130.00 0.4793 23.3954 -0.03

(h) {0.85892CH4+0.11532C2H6 + 0.01341C3He+0.00530CH3CH(CH3)CH3+0.00705C4H,ol

(M = 18.5565 g mol"')

115.00 0.1185 25.0957 -0.480 -0.03

120.00 0.1706 24.7131 -0.534 -0.07
125.00 0.2372 24,3294 -0.613 -0.08
130.00 0.3225 23.9490 -0.730 -0.06

(i) {0.84558CH4 + 0.08153C2H, + 0.O4778C3He+0.01259CH3CH(CH3)CH3+0.01252QH,o:

(M = 19.5838 g mol"')

115.00 0.1166 24.5586 -0.620 +0.01

120.00 0.1680 24.2180 -0.716 +0.03
125.00 0.2348 23.8688 -0.826 +0.04
130.00 0.3188 23.5154 -0.963 +0.07

(j) ;0.00601Nj +0.9061 3CH4+0.06026C,Hfe+0.02154C3H8+0.00300CH3CH(CH3)CH3

+ 0.00306C4H,o! (M = 17.8195 g- mol"')

115.00 0.1478 25.3834 -0.425 -0.01

120.00 0.2043 24.9894 -0.551 +0.02
125.00 0.2785 24.5702 -0.02
130.00 0.3722 24.1578 +0.02

(k) ;0.00973N2 +O.88225CH4 +O.O7259C2H, +O.O256IC3H8 +0.00490CH3CH(CH3)CH3

+ 0.0049 2C4H,ol (M = 18.3094 g- mol-')

115.00 0.1639 25.2023 -0.556 -0.04

120.00 0.2247 24.8047 -0.698 -0.00
125.00 0.3022 24.4022 -0.03

(!) {0.01383Nj + 0.85934CH4+0.08477C:,H6+0.02980C3Hs+0.00519CHjCH(CH3)CH3

+ 0.00707C4H,oi (M = 18.7496 g moi-')
115.00 0.1812 25.0384 -0.662 +0.06
120.00 0.2441 24.6661 -0.874 +0.07
125.00 0.3223 24.2880 +0.09
130.00 0.4222 23.8981 +0.10

(m)0.85341CH4 + 0.07898C2H,+0.04729C3H8+0.00854CH3CH(CH3)CH3 + 0.00992C4H,o

+ 0.00097CH3CH(CH3)CjH5+0.00089C5H,2: (M = 19.3587 g mol" ')
25.0063 -0.517 -0.01
110 00 0.0787
24.6566 -0.581 -0.03
115.00 0.1172
120.00 0.1686 24.3079 - 0.670 - 0.02
23.9525 -0.777 -0.01
125.00 0.2351
23.5883 -0.901 0.00
130.00 0.3210

i/:6
 — )

610 W. M. HAYNES

TABLE 1 continued

K MPa mol dm"^ cm^ mol"' p„m

(n) !0.75442CH,+0.15401C2H6+0.06950CjH9+0.00978CH3CH(CH3)CH3+0.01057C4H,o
+ 0,0O089CH3CH(CH3)C,H5 + 0.0O083C5H,2i (Af = 21 .1060 g • mor '

110.00 0.0723 24.2529 - 0.633 - 0.05

115.00 0.1081 23.9619 - 0.740 + 0.01
120.00 0.1549 23.6535 -0.840 +0.01
125.00 0.2153 23.3351 -0.949 -0.01

(o) {0.00859N2+0.75713CH4 + O.13585C2H(,+O.O6742CjH8+O.01336CH3CH(CH3K:H3

+ 0.01326C4H,o + 0.O0223CH3CH(CHj)C2H5+0.00216C5H,j; (M = 21.3094 g moj- ')

110.00 0.1155 24.1809 - 0.741 +0.03

115.00 0.1595 23.8731 -0.857 +0.02
120.00 0.2155 23.5709 -1.055 +0.05
125.00 0.2873 23.2644 +0.08
130.00 0.3744 22.9514 +0.11

(p) |0.00801N2+0.74275CH4 + 0.16505CjH6 + 0.06547C3H8+0.00843CH3CH(CH3)CH3

+ 0.00893C4H,o + 0.00069CH3CH(CH3)C2H5+0.00067C5H,3; (Af = 21.0976 g mor')

110.00 0.1158 24.3141 -0.676 -0.05
115.00 0.1584 24.0160 -0.807 -0.02
120.00 0.2093 23.6937 -0.965 -0.06
125.00 0.2853 23.3804 -0.05

(q) 10.00599Nj +O.9OO68CH4 +0.06537C,Hfe +O.O22OOC3H8 +0.00291CH3CH(CH3)CH3

+ 0.00284C4H,o + 0.00010CH3CH(CHj)C2H5 + 0.0001 IC5H, 2! (M = 17.9026 g mol"')

115.00 0.1456 25.3600 - 0.449 + 0.02

120.00 0.2024 24.9656 -0.571 +0.03
125.00 0.2762 24.5450 -0.03
130.00 0.3698 24.1289 -0.02

component at the same temperature at saturation pressure p,, Xj is the mole fraction
i

of component i, and k, is the isothermal compressibility of component i.

The pure-fluid molar volumes were obtained from expressions given in references 1,
2, and 16, while the compressibilities for nitrogen, methane, and ethane were obtained

from expressions presented in reference 6. Compressibility corrections for

hydrocarbons beyond ethane were sufficiently small to neglect. For those mixtures
containing nitrogen, excess volumes were not computed for temperatures above
120 K. (The critical temperature of nitrogen is approximately 126.2 K.) Vapor
pressures for nitrogen, methane, and ethane were obtained from references 17 to 19,
respectively.
To demonstrate the consistency of the results presented here, table 1 includes
comparisons between the experimental densities and the values calculated using an
extended corresponding-states method,'^"^' which was optimized to only the pure-
fluid (except for pentanes) and binary-mixture values from the lng density project at
this laboratory. Of the mathematical models developed at this laboratory*^"'' for the
prediction of lng densities, the extended corresponding-states method is probably the
most accurate and certainly the most general since it handles the widest ranges of

124
 DENSITIES OF lng MIXTURES 611

pressure, temperature, and composition. All other models'^"^' developed at this

laboratory were characterized by limitations in the ranges of either temperature or
composition covered for mixtures containing nitrogen and butanes. The maximum
deviation between the experimental and calculated densities using the extended
corresponding-states method was 0.11 per cent, while the average absolute deviation
was 0.04 per cent.
Comparisons of the present results with other mathematical models optimized to
the pure-fluid and binary-mixture values from this project are presented in
reference 8. Comparisons of multicomponent mixture values (lng components) from
other sources with those of the present study have been accomplished indirectly in
earlier papers.'^' ^' Thefrom other work were compared with the calculated
results
values from the extended corresponding-states method that had been optimized to
the pure fluid and binary-mixture values from this project.
Detailed descriptions of the random and systematic errors involved in density
measurements on liquids and liquid mixtures with a magnetic suspension densimeter
have been presented elsewhere. The total systematic error in the measurements
on multicomponent mixtures should be little different from that for binary mixtures.
The added uncertainty in the composition of the prepared mixtures resulting from the
larger number of components is insignificant.
The random error in the multicomponent-mixture measurements was difficult to
evaluate since, for most mixtures, there was not a sufficient amount prepared to make
a statistically significant number of repetitive measurements on different samples from
the same mixture. It is believed that the random error in these multicomponent
mixture measurements should not be significantly different from that for
(methane -I- isobutane) and (methane + normal butane) measurements'^' (maximum
standard deviation of 0.024 percent) obtained with the same apparatus.
Based on the analysis of errors for density measurements on pure fluids and binary
mixtures with a magnetic suspension densimeter, it is believed that the maximum
total uncertainty in the densities for the multicomponent mixtures presented in this

paper is approximately 0.12 per cent. The total uncertainty has been taken as the root
mean square of the known systematic errors added to three times the standard
deviation for random error. This estimate was confirmed, to some extent, in the
comparisons between the experimental results and the predicted values from the
extended corresponding-states method. As reported earlier in this section, the
maximum deviation was 0.11 per cent and the average absolute deviation was 0.04
per cent.

The contributions of M. J. Hiza in the preparation of mixtures, R. J. Richards in the

analysis of mixtures, and R. D. McCarty and R. G. Sargent in the reduction and
analysis of results are gratefully acknowledged.

REFERENCES
1. Haynes, W. M.; Hiza, M. J.; Frederick, N. V. Rev. Sci. Insirum. 1976, 47, 1237.
2. Haynes, W. M.; Hiza, M. J. J. Chem. Thermodynamics 1977, 9, 79.
3. Haynes, W. M.; Hiza, M. J. Advances in Cryogenic Engineering Vol. 21. Timmerhaus, K. D.;
WeiUel, D. H.: editors. Plenum Press: New York. 1976. p. 516.

125
612 W. M. HAYNES

4. Hiza, M. J.; Haynes, W. M,; Parrish, W. R. J. Chem. Thermodynamics 1977, 9, 873.

5. Haynes, W. M. To be published.
6. Hiza, M. J.; Haynes, W. M. J. Chem. Thermodynamics 1980, 12, 1.
7. McCany. R. D. Nail. Bur. Stand. (U.S.). Interagency Kept NBSIR 77-867. 1977.
8. McCarty, R. D. Natl. Bur. Stand. (U.S.). Tech. Note 1030. 1980.
9. McCarty, R. D. to be published.
10. Orrit, J. E.; Laupretre, J. M. Advances in Cryogenic Engineering Vol. 23. Timmerhaus, K. D. : editor.
Plenum Press: New York. 1978, p. 573.
11. Haynes, W. M. To be published.
12. Haynes, W. M. Rev. Sci. Insirum. 1977, 48, 39.
13. Parrish, W R.; Arvidson, J. M.; LaBrecque, J. F. Natl. Bur. Stand. (U.S.). Interagency Rept NBSIR
78-887. 1978.
14. Parrish, W. R.; Hiza, M. J. Advarwes in Cryogenic Engineering Vol. 19. Timmerhaus, K. D.: editor.
Plenum Press; New York. 1974, p. 300.
15. Kidnay, A. J.; Miller, R. C; Parrish, W. R Hiza, M. J. Cryogenics 1975, 15, 531.
;

16. Hiza, M. J. Fluid Phase Equilibria 1978, 2, 27.

17. Strobndge. T. R. Natl. Bur. Stand. (U.S.). Tech. Note 129. 1%2.
18. Goodwm, R. D. Natl. Bur. Stand. (U.S.). Tech. Note 653, 1974.
19. Goodwin, R. D.; Roder, H. M.; Straty, G. C. Natl. Bur. Stand. (U.S.). Tech. Note 684. 1976.

126
6.11 Reference [34]

Four Mathematical Models

for the Prediction

of LNG Densities

Robert D. McCarty

Thermophysical Properties Division

National Engineering Laboratory
National Bureau of Standards
Boulder. Colorado 80303

U.S. DEPARTMENT OF COMMERCE, Philip M. Klutznick, Secretary

Jordan J. Baruch, Assistant Secretary for Productivity, Technology and Innovation

NATIONAL BUREAU OF STANDARDS, Ernest Ambler, Director

Issued December 1980

127
 CONTENTS

Page

Abstract 1

1. INTRODUCTION 1
.

2. EXTENDED CORRESPONDING STATES 3

3. A HARD SPHERE METHOD 6

4. A REVISED KLOSEK AND MCKINLEY METHOD 11

5. THE CELL MODEL 13

6. USE OF THE METHODS 15

7. CONCLUSIONS 19

8. REFERENCES 20

Appendix A. Experimental Data 24

Appendix B. Computer Program and Equation Parameters for the

Extended Corresponding States Model 39

Appendix C. Computer Program and Equation Parameters for the

Hard Sphere Model 46

Appendix D. Computer Program and Parameters for the Revised

Klosek and McKinley Model 50

Appendix E. Computer Program for the Cell Model 54

Appendix F. Computer Programs 55

i i i

128
 FOUR MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES*

Robert D. McCarty

Thermophysical Properties Division

National Engineering Laboratory
National Bureau of Standards
Boulder, Colorado 80303

Four mathematical models of the equation of state for LNG like

mixtures are presented. The four models include an extended
corresponding states model, a cell model, a hard sphere model and a
revised Klosek and McKinley model. Each of the models has been
optimized to the same experimental data set which included data for
pure nitrogen, methane, ethane, propane, iso and normal butane, iso and
normal pentane and mixtures thereof. For LNG like mixtures (mixtures
of the orthobaric liquid state at temperatures of 120 K or less and
containing at least 60% methane, less than 4% nitrogen, less than 4%
each of iso and normal butane and less than 2% total of iso and normal
pentane), all of the models are estimated to predict densities to within
0.1% of the true value. The revised Klosek and McKinley model is valid
only for mixtures within the range of temperature and composition
specified above while the other three models are valid for a broader
range of pressure, temperature and composition. The experimental PVTx
data set used in the optimization together with comparisons are given
and listings of computer programs for each of the models are included.

Key words: Cell model; comparisons; computer programs; corresponding states;

equation of state; hard sphere; LNG; mixtures; PVTx data; revised Klosek and

McKinley.

1. INTRODUCTION

The purpose of this report is to present in final form the four mathematical

models which were optimized to the experimentally determined orthobaric liquid

This work was carried out at the National Bureau of Standards under the
sponsorship of British Gas Corp., Chicago Bridge and Iron Co., Columbia Gas
Service Corp., Distrigas Corp., Easco Gas LNG, Inc., El Paso Natural Gas, Gaz de
France, Marathon Oil Co., Mobil Oil Corp., Natural Gas Pipeline Co., Phillips
Petroleum Co., Shell International Gas, Ltd., Sonatrach, Southern California Gas
Co., Tennessee Gas Pipeline, Texas Eastern Transmission Co., Tokyo Gas Co., Ltd.,
and Transcontinental Gas Pipe Line Corp., through a grant administered by the
American Gas Association, Inc.

IZ9
PVTx data of Miller and Hiza [25], Haynes and Hiza [12], Haynes, et al . [11],

Hiza, et al, [14], Hiza and Haynes [15] and Haynes [9]. Interim results of the

project have been reported by Haynes, et al . [13] and by McCarty [23]. The

models reported in these two interim publications differ only slightly from those

presented here. A companion archival document, McCarty [24] with the same

results but in much less detail has been submitted to the Journal of Chemical

Thermodynamics. The intent of the documentation here is more in the vein of a

user's handbook. The above experimental data are for the liquid phase of

nitrogen, methane, ethane, propane, normal and isobutane and various mixtures

thereof.

The goal of the project (the project included the above referenced experi-

mental work) was to produce one or more computer models which would predict the

density of LNG to within 0.1% of the true value from a knowledge of the

temperature, pressure and composition of the LNG. At the beginning of this study

LNG was defined as mixtures of the above components {N2, CH^, CgHg, C^Hg,

n and iC^H^q) and only the saturated liquid between 95 - 150 K was to be

considered. Near the end of the project n and iCgHj^g were added to the list of

allowable components but no experimental PVTx of pure n and iCgHj^2 binary

systems containing n and iCgH^^g were measured as part of the project. The

inclusion of these two components is based on data from Orrit, et al . [29] and

Orrit, et al . [30].

Four models were considered: the extended corresponding states model, a

hard sphere model, a cell model and a Revised Klosek and McKinley model. With

the exception of the Revised Klosek and McKinley model only pure fluid and binary

system data were used to optimize the models. In the case of the Revised Klosek

and McKinley model multi component PVTx data were used in the optimization

process.

130
 Over a normal range of LNG composition and temperature all four of the

models predict densities which agree to within 0.1% of experiment. This is true

of all of the experimental PVTx measurements on LNG like mixtures made as part of

this project.

No equation or mathematical model based on experimental data can be more

accurate than the original data and therefore the extent to which the original

goal of 0.1% accuracy has been met depends entirely upon the accuracy of the

experimental data referenced above.

There is no reason to doubt the experimental data and therefore there is

every reason to believe that the goal of the project has been achieved.

2. EXTENDED CORRESPONDING STATES

The thermodynamic equations for the extended corresponding states method are

developed in a paper by Rowlinson and Watson [35] and only a very brief

description will be given here. Leach [20] developed transformation functions

for hydrocarbons which are called shape factor functions. Mollerup [27] and

Mollerup and Rowlinson [26] combined the earlier work with the equation of state

for methane by Goodwin [8] to produce a computer program to calculate the density

of LNG mixtures, which was further modified by Mollerup [28].

The computer program in Appendix F for the calculation of LNG densities

based on the extended corresponding states method is an extensive revision of the

Mollerup program. Earlier revisions were reported by McCarty [23] and Haynes,

Hiza and McCarty [13].

The extended corresponding states method is defined by the following

equations:

VP'^^ =
yp^i.o^-i.o'^/^ii,o^

131
Si^P'T] =
fii,o VP Ni,o/^ii.o> T/fii,o]
tn(h,,_„) (2)

where Z is the compressibility factor, G is the Gibbs free energy. Pis pressure

and T is temperature. The subscript o denotes the reference fluid and the

subscript i denotes the fluid for which properties are to be obtained via the

equation of state for the reference fluid and the transformation functions

f and h.. „. The double subscript ii is introduced now to allow extension

n ,0
. .

n ,0

of the method to mixtures. The two defining eqs (1) and (2) are necessary since

there are two transformation functions. In this case the equation of state for

methane by McCarty [22] was chosen for the reference fluid. During the course of

the study it was necessary to modify the equation of state by McCarty [22] to

give a realistic vapor liquid phase boundary down to a temperature of 43 K. This

modification was necessary to acconmodate the very low reduced temperatures of

the heavier hydrocarbons and was accomplished without changing the performance of

the equation of state above the triple point of methane. The equation of state

is given in Appendix B.

The f. „ and h.. ^ are defined as

n .

,0 11,0

and

Ni.o = (V?i,o/^S' *ii.o (Tp-V («)

where

^i.o = 1 - w^)[ni - n2 m T^_ + (n3 - r^^/^U^^ - n^)] (5)

and

132
 The and are reduced temperature and volume, (i.e., =
^/"'^ii

~ ^^^^ fluid requires a unique which was estimated using pure fluid
^r. ^^^ii^

experimental data. A single set of the n^'s are used for all fluids. The n^. 's

were estimated using all of the pure fluid experimental data from this study. The

Z^/Z^ is the ratio of the compressibility factors (Z^ = PqV^/RT^) at the critical

point. The parameters n . , n., w. and Z.. are given in Appendix B. All of

these parameters were estimated using the experimental PVT data set from this

laboratory and least squares estimation techniques.

The extension of the above to mixtures is now accomplished by the following

application of the following combining rules:

f . . = (f • . f • •
)^^^ (9)

Nj,o -
^ij (2 ^i.o ^ 2 ^-0,0)

The £;^.j
and the n^^ are binary interaction parameters determined by least

squares from the PVTx data for binary mixtures. These parameters are given in

Appendix B.

133
 This method works quite well as may be seen in the comparisons in

Appendix A. It has indeed reproduced all of the present experimental data set to

within + 0.1% except for 14 out of a total of 285 experimental data points. Of

these 14 points, 11 are judged to have an uncertainty greater than 0.1%.

Figure 1 presents the deviations between the calculated and experimental

densities for these 14 points. Appendix A contains comparisons of calculated and

experimental densities for the entire data set. This is the best performance of

the four models presented here. No pressure, temperature or composition

restrictions have been placed on this model.

In the interim publications by McCarty [23] and Haynes, Hiza and

McCarty [13] some doubt about the accuracy of the calculated densities was

expressed because of the disagreement with a few binary and multi component

systems containing methane and butane. This disagreement has since been resolved

by additional measurements (Haynes [9], Haynes [10] and Miller and Hiza [25] on

some of the systems which agree with the predictions of the model but disagree

with the previous measurements. The net result of the new measurements is a very

slight change in binary interaction coefficients of the methane-butane and

nitrogen-butane system. These changes have no practical effect on LNG like

mixtures where the concentrations of N2 , iC^Hj^Q and nC^H^^Q are individually

less than 5%. In other words either the models presented here or those in the

interim publications may be used to predict the density of a LNG like mixture to

within 0.1% of the true density.

3. A HARD SPHERE ^ETH0D

The model of Rodosevich and Miller [33] is one of many modifications of the

Longuet-Higgins and Widom [21] model, and was chosen to be included in this study

134
 1 . .

as a representative example of the application of the hard sphere equation of

state concept to the correlation of PVTx data. The equation of state by

Rodosevich and Miller [33] is

pv . - 1 •«•
y + y _ _a_
(11)

where the y = b/4V and a, b, and c are adjustable parameters, P is pressure, V is

specific volume, T is temperature and R is the gas constant. The equation is

applied to mixtures by assuming the one-fluid theory and applying the following

combining rules.

a„
m
= S S
^.
a.
ij
. x.x.
1 J
(12)
J.

b„
m
= 2 2
^
b.. x.x.
ij 1 J
(13)
j

Cm
m
= 23 Sc..
ij
X.X.
1 J
(14)
^
j

The mixing rules are:

T 3
1

2
^ (1 -J-ij) (15)

b
2 ^ 1/2
r
1/2 11 (1 - (16)
(^i^jj) b..b .

n JJ

c . . + c .

(17)

The parameters j- • and k.. are in this case the binary interaction

parameters. The a's, b's, c's, j^^'s and k..'s are given in Appendix B. The

excess volume is now calculated using the equation of state and

135
V, (18)
E

where V and the V. are calculated via the eqs 11 through 17 and then

V, V.x. . V, (19)
m

where the is from eq (18) and the are from experimental data. The

values of in this case were calculated from the equations for the liquid

density of the pure fluids given in Appendix C.

The above equations are those of Rodosevich and Miller [33] and Rodosevich

[34] and only the J^^-'s and k^^-'s have been revised on the basis of the
'0 'J

present new data set, and only binary systems data were used to estimate via

least squares the L^'s and k. .'s.

As the method is used here it is an excess volume method, and consequently

when the temperature of the mixture approaches the critical temperature of one of

the component fluids, the method fails. Since the critical temperature of

nitrogen is about 126 K, this method should not be used for mixtures containing

nitrogen at temperatures above 120 K. Eliminating the data points for mixtures

which contain nitrogen at temperatures above 120 K reduces the set from 285 to

251 PVTx points. Figure 2 is a percentage deviation plot containing all of the

data points from the set of 251 for which densities calculated by the hard sphere

method differ from the experimental density by more than 0.1%. Two things are

readily seen in comparing figs. 1 and 2; first, even though total number of

points has been reduced in the comparison set, the number of points for which

deviations exceed 0.1% in the hard sphere comparison, fig. 2, is far more than

for the extended corresponding states comparison, fig. 1. Second, the hard

sphere method becomes more uncertain for all mixtures, regardless of components

as the temperature exceeds 115 K.

13b
 (U
4-1 CO
« 4-1
iH CO
3 T3
y
iH
H
m
o CO
u
00
CM

C
CO •
< ho <
iH 4J

•H
CO (U X
4-1 e •H
c c
ID CO C
(U
a
a o s
•H
U
a>
u
(U
a
< CO
a>
CO
4J <
cn
X c
ifi
in
m CO w 00
c
c •H CO
0)
c 3 (U
$
T)
c
o
4J
c
1-1
I-
•5 o <
4J Q.
CO
. o
a
c 04 ^ re
CD
o <
u
o
4J

a 111
.-1
• u o
CO

CO Q. o <u

z c
CO
x:
<U

c
4-1

UJ 4J
o (U
4J
of

O < - u
<ii
X
1—1
4-1 1—1
CO (U CO
0) x:
u 4J CO

+ DO to
00
•H

CO c 4-1

c •H <U
o CO
o •H
4.1
CO
3
3
CO CO O
•H 01 n
> •H
o o o
I T)
4-1
•H
t-l

CO
CO
4-1
CO a c
iH C B •H
cu o o
AllSNaa Nl NOIIVIABQ % < 13 CJ a

3
00
•H

137
 1 1

0):::::::::::::::
c
a.

Oi-ieMfn«*ONOmooo«3-iriv£>«vjvDPn
0)

0)

B
Q (no
c
•H
<— c
V to
to

'
o w to
•H c QD
1 1
o 4J
o
o
fH (U
(d ID to o
o c
t> < •o
I.

CO
to ) 1

(0 •H CO
Ui

- o
1— to P
to o.
CO Q
c o o
\ o C4-(

•H d) CD
Hi x: 4J •
> < Id
01 c
oc O. •H
3 a o
O
CM
O I- c
r-*

CM < 0)
(U

o
(0
4J
CO
60
C
s •H
lU T5
O 3
0. Xi Id CD iH
O U
s r*
c
* Wp •H

> + • • - O H o
c
4J
a
0)
c
CO
CO CO o
• 4J
ac X (U

AO
Id -<
1 • 0)

« to •H
CO

to
c a>
ha o &0
c >-i

3
X o n
to
3 a
a.
JJ
to
c
o
CO
<u
< u
9)
•H c a
1 . 1
1-1
•H s
to

CM CM •H to CO
> c
o o
I
Q
0) V (0
to

AiiSNaa Ni NOiiviAaa %
3
60
•H
1X4

10

138
 .

4. A REVISED KLOSEK AND MCKINLEY METHOD

The Klosek and McKinley method [18] is a totally empirical recipe for

calculating the density of a LNG-like mixture given the temperature and

composition. Pressure is not taken into account. However, this does not seem to

be a serious omission. The procedure proposed by Klosek and McKinley [18] is as

follows:

V„,, = E X,V, - kX,„^ (20)

where V„.
mix
volume of the i
th
component, X^.^ is
11
is the volume of the mixture, X. and V. are the mole fraction and

the mole fraction of methane and k is a

correction factor obtained from a table or graph. The and k are obviously

temperature dependent and in addition k is dependent upon the molecular weight of

the mixture.

Using the present data set k was calculated for all of the expe.'imental data

points "where methane was present in the mixture and excluding all data points

where N2 was present in greater than S% concentration. Figure 3 shows a typical

isotherm for k, with N2 present (labeled k2) and without Ng present (labeled k^)

All of the isotherms available show similar behavior, i.e., all of the k's for

mixtures containing nitrogen (of about 5%) fall on one line and all of those for

mixtures without nitrogen fall on another. Since all of the mixtures with

nitrogen have about the same amount of nitrogen present (about 4.5%), the method

was modified by adding a term to take into account the nitrogen when it is

present. The equation becomes

\U - 2 ''i^
- * (^2 - "!> '<n/-°*25]
X,„^ (21)

where everything is the same as in eq (20) except that k, is read from one

11

139
140
curve and k2 is read from the other. Appendix D gives tables of values for

the V^, and ^2 which are spaced such that linear interpolation is adequate

in both variables (i.e., temperature or molecular weight). The k factors in

Appendix D have been obtained graphically from the mul ti component PVTx data of

Hiza and Haynes [15] and Miller and Hiza [25] as well as densities calculated

from the extended corresponding states method of section 2.

The limits of compositions of the revised Klosek and McKinley method are the

most severe of any of the methods given here. This method should not be used for

mixtures other than LNG like mixtures and for LNG like mixtures only when they

contain at least 60% methane, less than 4% nitrogen, less than 4% each of iC^H^Q

nC^H^Q and less than 2% total of ^^^^^12 "^5^12*

There are 40 experimental PVTx points from the original set of 285 which may

be considered LNG like and fall within the composition limits outlined above.

Figure 4 shows all of the deviations between calculated and experimental

densities in this 40 point comparison set which exceeds the 0.1% criterion. The

deviation trends for the revised Klosek and McKinley method (fig. 4) are very

similar to those of the hard sphere method (fig. 2) and in fact all of the

deviations in fig. 4 occur at temperatures at or above 115 K, therefore the

method can only be considered as accurate as the others for LNG like mixtures at

temperatures below 115 K.

5. THE CELL MODEL

The cell model considered here was originally proposed by Renon,

et al . [32]. In a paper by the same three authors which appeared simultaneously

(Eckert, et al . [7]), the cell model was applied to mixtures via Scott's [36]

two-fluid theory and a three parameter corresponding states theory. Albright

[2] further modified the method by modifying the mixing rules on the basis of a

13

141
 1 t

c
o
o (0
w
fll
TO
4J
TO ce TO
a
B
O
V
o
to
(J
he /\
H
cn
CO
•
I—
HI

o rS
CD s 1

<D

UJ p c w
fC
(J c
<U
(D
•>
T3 vO
t> c
jj CO (0
HI vg
UJ OJ
0. 00
o
1^
Li es
ta

j: CO 4J
4J
>
c o
1—
Ui c ac
0) o -a-

«
o. u
0) B •H
o
u 4J CJ
60 •H
60 4-1
(0 C iH
c •H 3
o CO E
•H
4J
S
iH
o
.-4
CO CO i-H a:
01 CO
> •H u
4J CO c CO
•o •H •H

Hc
CO >«s
c
AllSNaa Nl NOIlVIAiQ % <
4.J
V
CO
-a-

V O

0)

s
60

14

142
proposal by Yuan [38] and by inserting a pressure dependence based on the

experimental liquid ethane data by Pope [31].

The optimization of this method was carried out by M. Albright [l] at

Phillips Petroleum Company in Bartlesville, Oklahoma and the details of this work

will be published elsewhere. The model is included here because it was optimized

to the same data set as the others and therefore the comparisons between

experimental and calculated densities given here in fig. 5 together with figs. 1,

2 and 4 provide a conrion basis of comparison with the other three methods. A

listing of the computer program is given in Appendix F.

The same data set as was used in the hard sphere method for comparison has

been used here, i.e., all of the data points for mixtures containing nitrogen at

temperatures 120 K and above have been taken out of the original 285 points

leaving a total of 251 data points.

As in the case of the other methods fig. 5 shows all of the points for which

the calculated and experimental densities differ by more than 0.1%.

6. USE OF THE METHODS

When the project started in 1972, the atomic weights of nitrogen, carbon and

hydrogen were taken from the 1961 carbon 12 scale, lUPAC [16]. During the course

of the investigation a revision, Atomic Weights of the Elements [3], to this

scale appeared. The revision changed slightly the atomic weights of carbon and

hydrogen, but since the changes were small (the maximum difference in any of the

densities used here is 0.003%), and because changing the atomic weights would not

change the relative results, the changes were not made. Therefore when using

the tables and programs in the appendices, the molecular weights given in the

tables and programs should be used to maintain consistency.

15

143
 O mixture 3

+ 11

12
>- • 13
j-
14

z 15
lU 16
a V 17

35

39
e 40
<
• 42
>
U 45
Q * 52

53

55

-0.2 - 56

Appendix A

100 110 120 130 140

TEMPERATURE, K

Figure 5. Deviations greater than 0.1% between experimental and calculated

densities using the Cell model. The comparison set is all of the data
in Appendix A except those data points for mixtures containing N2 at
temperatures above and including 120 K. (251 data points)

16

144
 .

The critical parameters used here are from: CH^, McCarty [22]; CgHg,

Sliwinski [37]; C3Hg, Das, et al . [4]; iC^H^Q, Das, et al . [5]; nC^H^Q, Das,

et al . [6]; Kudchadker, et al . [19]; and Jacobsen, et al . [17].

^0^^i2*
Errors in the input variables will of course, cause errors in the density

predicted by the models. In general, the error in density caused by an error in

the input varibles is a function of those input variables, and must be treated on

an individual basis. However, for LNG like mixtures certain general trends are

found. An error in the pressure must be at least 50% before it will have any

effect at all on the resulting density. An error in composition, unless it is of

the order of several percent, will cause the same relative error in density as it

will cause in the molecular weight of the mixture, i.e., if an error in

composition causes a 0.1% error in the resulting molecular weight, it will also

cause a 0.1% error in the predicted density.

The error in the calculated density due to an error in the input temperature

is a function of the composition and the temperature. Table 1 gives resulting

errors in density for a 1% error in temperature, for three hypothetical LNG like

mixtures

"In general the errors in density caused by an error in temperature are the

largest for mixtures containing a high concentration of the most volatile fluids,

CH^ and and correspondingly the errors decrease as the concentration of

the heavier hydrocarbons increases in the mixture. These errors are not a

function of which model is being used.

When using the extended corresponding states method, one should keep in mind

that twelve significant figures are required by the methane equation of state.

The hard sphere model also uses the methane equation from McCarty [22] and the

nitrogen equation of Jacobsen, et al . [17] to calculate compressibilities and

17

145
 TABLE 1. Errors in Density Caused by an Error in the

Input Temperature of 1%.

Temoerature % Error in Density

*
K Mix A* Mix B Mix C*

95 0.28 0.25 0.20

100 0.30 0.27 0.22

105 0.32 0.29 0.24

110 0.35 0.32 0.29

115 0.39 0.34 0.31

Mix A = 0.95 CH^, 0.05 N2

Mix B* = 0.9 CH4, 0.02 C^}^^, 0.02 C3Hg, 0.02 iC^H^Q, 0.02 nC^H^Q, 0.02

Mix C* = 0.6 CH^, 0.3 C2Hg, 0.02 CjHg, 0.02 iC^H^q, 0.02 nC^H^Q, 0.02

0.02 iC5H^2*

*Arbitrary LNG like compositions assumed for the purpose of illustrating the

effect of an error in the input temperature.

18

146
 .

therefore requires twelve significant figures to insure the accuracy of the

calculated density. The other two models require only eight significant figures

to be carried along in the calculations.

7. CONCLUSIONS

On the basis of the performance of the four models given here and subject to

the composition and temperature restrictions already noted, it is estimated that

given the pressure, temperature and composition of LNG, any one of the four

models may be used to predict the density to within 0.1% of the true value. As

has already been mentioned (see section 1) the above accuracy statement is

dependent entirely upon the accuracy of the experimental data in Haynes, et al

[11], Haynes, et al . [13], Hiza, et al . [14], Haynes [9], Hiza and Haynes [15],

Miller and Hiza [25] and Haynes [10]. These data have been estimated by the

authors to be accurate to within 0.1% of the true value with a precision of a few

hundredths of a percent. The work on the models given here have provided no

basis for questioning the claims of the experimenters, in fact the ability of the

models to predict the densities of the multi component mixtures to within 0.1% of

the measured values tends to support the accuracy claims of the experimenters.

Interim results of this study were reported by Haynes, et al . [13] and

McCarty [23], both of which contain earlier versions of the mathematical models

given here. These earlier versions are only slightly different than the final

ones and for the purposes of calculating LNG densities either of the versions may

be used. The reader is, however, cautioned to read the limitations of each model

as defined in the earlier sections.

Computer programs for the four models are available at the Thermophysical

Properties Division of the National Bureau of Standards in Boulder, Colorado.

19

147
 .

8. REFERENCES

[I] Albright, M. A., Private cormunicati on (1976).

[2] Albright, M. A., A Model for the Precise Calculation of Liquefied Natural

Gas Densities, NGPA Technical Publication TP-3 (1973).

[3] Atomic Weights of the Elements 1973, Pure and Applied Chemistry, Vol 37,

589-603 (1974).

[4] Das, T. R., Eubank, P. T., Advances in Cryogenic Engineering Vol 18,

Plenum Press, New York (1973), p. 208.

[5] Das, T. R., Eubank, P. T., Advances in Cryogenic Engineering Vol 18,

Plenum Press, New York (1973), p. 253.

C63 Das, T. R., Eubank, P. T., Advances in Cryogenic Engineering Vol 18,

Plenum Press, New York (1973), p. 244.

C73 Eckert, C. A., Renon, H. and Prausnitz, J. M., Molecular Thermodynamics of

Simple Liquids, lEC Fund 6_, 58 (1967).

[8] Goodwin, R. D., The Thermophysical Properties of Methane, from 90 to 500 K

at Pressures up to 700 Bars, Nat. Bur. Stand. (U.S.), Tech. Note 653

(1974)

C93 Haynes, W. M. , Orthobaric Liquid Densities of Methane + Isobutane and

Methane + Normal Butane Mixtures at Low Temperatures (to be published).

[10] Haynes, W. M., Measurements of Orthobaric Liquid Densities of

Mul ti component Mixtures of LNG Components (N2, CH^, ^2^6' ^3^8' "^^4^10'

nC^H^Q, iCgH^^' '^^5^12^ Between 110 and 130 K, to be published.

[II] Haynes, W. M., Hiza, M. J. and Frederick, N. V., A Magnetic Suspension

Densimeter for Measurements on Fluids of Cryogenic Interest, Review of

Scisntific Instruments 47, No. 10, 1237-50 (Oct 1976).

[12] Haynes, W. M. and Hiza, M. J., Measurements of the Orthobaric Liquid

Densities of Methane, Ethane, Propane, Isobutane, and Normal Butane, J.

Chem. Thermodyn., 9, 179 (1977).

148
[13] Haynes, W. M. , Hiza, M. J. and McCarty, R. D., Densities of LNG for

Custody Transfer, (Proc. Fifth International Conference of LNG,

Dusseldorf, Germany), J. W. White, W. McGrew, S. Farmer, D. Hansen, H.

Jacobshagen, editors; Institute of Gas Technology, Chicago, 111., Vol. 2,

Paper II, Section III (1977).

[14] Hiza, M. J., Haynes, W. M. and Parrish, W. R., Orthobaric Liquid Densities

and Excess Volumes for Binary Mixtures of Low Molecular Weight Alkanes and

Nitrogen Between 105 and 140 K, J. Chem. Thermodyn. 9_, 873 (1977).

[15] Hiza, M. J. and Haynes, W. M., Orthobaric Liquid Densities and Excess

Volumes for Multi component Mixtures of Low Molar Mass Alkanes and Nitrogen

Between 105 and 125 K, J. Chem. Thermodyn. 12, 1 (1980).

[16] International Union of Pure and Applied Chemistry, 1961 Carbon 12 Scale.

[17] Jacobsen, R. T., Stewart, R. B., McCarty, R. D. and Hanley, H. J. M.,

Thermophysical Properties of Nitrogen from the Fusion Line to 3500 R

(1944 K) for Pressures to 150,000 psia (10342 x 10^ N/m^), Nat. Bur.

Stand. (U.S.), Technical Note No. 648 (Dec 1973).

[18] Klosek, J. and McKinley, C, Densities of Liquefied Natural Gas and of Low

Molecular Weight Hydrocarbons, Proc. First Int. Conf. on LNG, IGT, Chicago

(1968).

[19] Kudchadker, A. P., Alani, G. H. and Zwolinski, B. J., The Critical

Constants of Organic Substances, Chem. Rev. 68, 659-735 (1968).

[20] Leach, J. W., Molecular Structure Correction for Applying the Theory of

Corresponding States to Non-Spherical Pure Fluids and Mixtures, Ph.D.

Thesis, Rice University (1967).

[21] Longuet-Higgins , H. C. and Widom, B., A Rigid Sphere Model for Melting of

Argon, Mol. Phys. 8, 549 (1964).

21

149
 .

[22] McCarty, R. D., A Modified Benedict-Webb-Rubin Equation of State for

Methane Using Recent Experimental Data, Cryogenics 14_, No. 5, 276-80 (May

1974).

[23] McCarty, R. D., A Comparison of Mathematical Models for the Prediction of

LNG Densities, Nat. Bur. Stand. (U.S.), Interagency Report NBSIR 77-867

(Oct 1977).

[24] McCarty, R. D., Optimization and Comparisons of Mathematical Models for

the Prediction of Liquefied Natural Gas Densities, J. Chem. Thermodyn. (to

be published, 1980).

[25] Miller, R. C. and Hiza, M. J., Experimental Molar Volumes for Some LNG

Related Saturated Liquid Mixtures, Fluid Phase Equilibria 2^, 49-57

(1978).

[26] Mollerup, J. and Rowlinson, J. S., The Prediction of Densities of

Liquefied Natural Gas and of Lower Molecular Weight Hydrocarbons, Chem.

Engng. Sci. 29 (1973).

[27] Mollerup, J., Correlated and Predicted. Thennodynamic Properties of LNG and

Related Mixtures in the Normal and Critical Region, Adv. Cryog. Eng. 20,

172 (K. D. Timmerhaus, editor. Plenum Press, New York, 1975).

[28] Mollerup, J., Thermodynamic Properties of Natural Gas, Petroleum Gas, and

Related' Mixtures Enthalpy Predictions, Adv. Cryog. Eng. 23, 550 (K. D.

Timmerhaus, editor. Plenum Press, New York, 1978).

[29] Orrit, J. and Olives, J. F., Density of Liquefied Natural Gas and Its

Components, Distributed at the 4th International Conference on Liquefied

Natural Gas, Algeria (1974).

[30] Orrit, J. E. and Laupretre, T. M. , Density of Liquefied Natural Gas

Components, Adv. Cryog. Eng. 23, 573 (K. D. Tinmerhaus, editor. Plenum

Press, New York, 1978)

22

150
[31] Pope, G. A., Calculation of Argon, Methane and Ethane Virial Coefficients

at Low Reduced Temperature, Ph.D. Thesis, Rice University (1971).

[32] Renon, H., Eckert, C. A. and Prausnitz, J. M., Molecular Thermodynamics of

Simple Liquids, lEC Fund. 6, 52 (1967).

[33] Rodosevich, J. B. and Miller, R. C, Calculation of LNG Excess Volumes by

a Modified Hard-Sphere Model, Adv. Cryog. Eng. 19, 339 (1974).

[34] Rodosevich, J. B., Measurements and Prediction of Liquefied Natural Gas

Densities, M.Sc. Thesis, Univ. Wyoming (1973).

[35] Rowlinson, J. S. and Watson, I. D., The Prediction of the Thermodynamic

Properties of Fluids and Fluid Mixtures - I. The Principle of

Corresponding States and Its Extensions, Chem. Engng. Sci . 24, 1565

(1969).

[36] Scott, R. L., J. Chem. Phys. 25, 193 (1956).

[37] Sliwinski, P. Z., Die Lorentz-Lorenz Funktion Von Dampff ormigem und

Fluessiyem Athan, Propan and Butan, Phys. Chem. (Frankfurt) 63, 263

(1969).

[38] Yuan, I. C, Correlation and Prediction of Excess Properties for Selected

Binaries in the C^-Cj Hydrocarbon Boiling Range of Conformal Solution

Theory, D.Sc. Dissertation, Washington University (1971).

23

151
 Appendix A. Experimental Data

The following is a list of all of the experimental PVTx mixtures data which

were measured during the course of this project. The data are from Miller and

Hiza [25]; Haynes and Hiza [12]; Haynes, et al. [11]; Hiza, et al . [14]; Hiza and

Haynes [15] and Haynes [9,10).

All of the data are for the ortnobaric liquid except for the data of Miller

and Hiza [25] which are for the single phase liquid phase, very close to the

orthobaric conditions. The units of the data are bars, moles per liter and

kelvin. The columns labeled RKM, HS, CELL and CS correspond to percentage

derivations between experimental and predicted densities by the Revised Klosek

and McKinley, hard spnere, cell and extended corresponding states models

respectively. The derivations are always calculated using

experimental-calcul ated densities.

24
 1 ,

MIXTURE NO 1 (Miller and Hiza [25])

0.85147 CH^ + 0.14853
D
r n T1 I'm RKM HS rci
l^tLL 1
uo
077 11 n nfl IS 1 9filv
10 .IcOO .13 - .02 n9
oUt .Ul m

MT YTI IDP Wn ^
NU f U 1 » a
9 ^nl za et al . ri4H
U .ooUUD 4^6
/lie oc 097 9n Kino 1 n n7
.'HO 1
iUO .uu .14 .05 .lU • U/
1 1 n nn 9n CTnQ
^U .souy .07 - .01 n9
.Ut ni
.Ul

.961 24.4612 115.00 20.5309 .05 - .04 -.03 -.03

1.380 24.1402 120.00 20.5309 .06 - .06 -.06 -.05

1 .3JU IOC nn
1^3 .UU tU csno
9n .DoUy .08 - .05 n/i
- .U4 nc
- .Ud

C. .O^U o-i cnn7

.3UU/ iOU .UU ^u c^no
9n .oouy .12 - .01 nn
.UU n9
- .Uc

MT A
nl IDT DU 0
Y TlUKt
1 H1 za et al . ri4l)
V

U .470^3 Ln^ T U.DUD/D 4^6

OO QC1Q 09 1 CI 9 n7
to . yoiy iUO .UU ^0 .1 DiO .14 .07 .11 . U/
OO CQ07 9T CI n"? n"?
.OUo iiU .UU 1
.09 .03 .Uo .uo
7AQ ^'^ AT?!? lie;
113 nn
.uu 9"? Rl10
^0 . 119 "?
.11 .03 .Ul m n'1
.uo
1
i ORn
.UoU Co • 1 993 19n .uu
icu nn 9*?
^0.13107 1 c;i
.07 - .02 — .UO - •
n'^
uo

1.500 22.8933 125.00 23.1513 .08 .00 -.05 .00

2.050 22.6290 130.00 23.1513 .14 .04 -.02 .03

9"^ IRI'^
C % 1 C\} tt .OOOi 109 .uu CO .1910 .19 .06 .Ul . U9
99 .U/
n7fic;
DO dn .uu
itu
1 nn 91
to .1910
1 m .25 .05 . nd
UH

M
MlT VA Tt DC
lUKt1
rtU 4 iniza. et al . [14])
U .0340/ l/n^ + U.D40HO
4^6
ci . iUoo 1
iuD nn
nc .uu 9c nof^c _ .04 - .07 - .ut n7/
- .u

7Q7 cc Q777
99 ,ol 1
1 1 n
iiU nn
.UU t9 no^c
9c .uyot) - .Ui n/1 - .
nc
UD nc
- .Uo

.580 22.6478 115.00 25.0965 .03 - .02 -.06 -.03

.826 22.4035 120.00 25.0965 .00 - .06 -.11 -.07

1.146 22.1872 125.00 25.0965 .10 .04 -.03 .03

1.550 21.9441 130.00 25.0965 .10 .03 -.04 .02

25

153
 MIXTURE NO 5 (Hiza, et al. [14])
0.85796 CH^ + 0.14204 ^3^8
P D T MW RKM HS CELL CS
.517 24.9622 105.00 20.0279 .01 .10 .10 .07

.817 24.6332 110.00 20.0279 - .02 .08 .07 .06

1 1 QQ lie: nn ^n n97Q .UH .Un n"?
.UJ n"?

1.785 23.9492 120.00 20.0279 - .07 .00 .00 .00

2.415 23.5942 125.00 20.0279 - .09 - .06 -.04 -.05

3.290 23.2461 130.00 20.0279 - .11 - .06 -.01 -.04

MIXTURE NO 6 (Hiza, et al. [14])

0.74920 CH^ + 0.25080 ^3^8
.478 23 .4767 105.00 23.0790 .01 .02 .03 -.02
.738 23.2064 110.00 23.0790 - .03 .00 -.01 -.03
1 OQQ CC. . 7 JDH 110 •UU cO .U/ 7U " ni
•Ul nn
.UU _ n7
- .Ut — m\jL

1.582 22.6665 120.00 23.0790 .02 .01 -.01 .01

2.216 22.3818 125.00 23.0790 - .03 - .04 -.04 -.01

3.029 22.1019 130.00 23.0790 .02 - .05 -.02 .01

MIXTURE NO 7 (Hiza, et al. [14])

0.49637 CH^ +. 0.50363 ^3^8
.384 20.4909 105.00 30.1720 - .05 .01 .04 -.03

.591 20.3046 110.00 30.1720 - .08 .01 .01 -.02

C\j .1 iOU 11 D .UU "in
OU .
1
1 77n
/ ^u .UD • Ul — n?
- •VjC - n?

1.250 19.9311 120.00 30.1720 .00 .01 -.03 .00

1.730 19.7471 125.00 30.1720 .04 .03 -.01 .04

2.320 19.5546 130.00 30.1720 - .01 .01 -.03 .05

MIXTURE NO 8 (Hiza, et al. [14])

0.29538 CH^ + 0.70462 ^3^8
.271 18.5132 105.00 35.8106 - .19 - .08 -.05 -.12
.409 18.3624 110.00 35.8106 - .24 - .09 -.10 -.12

26

154
 MIXTURE m 9 (Haynes [10])
0.92788 CH^ + 0.07212 "^4^10
p
r n J no

1.820 24.2615 120.00 19.0779 .00 .04 -.04 -.04

2.547 23 .8868 125.00 19.0779 .00 .00 -.07 -.05

3.470 23.5047 130.00 19.0779 - .01 .03 -.09 -.07

4.616 23.1225 135.00 19.0779 .04 .03 -.06 -.05

6.023 22.7284 140.00 19.0779 .15 .04 -.03 -.03

MIXTURE NO 10 (Haynes [10])

0.92780 CH^ + 0.07220
H lU
1.270 24.6285 115.00 19.0813 .01 .08 .01 -.01

1.824 24.2783 120.00 19.0813 .07 .11 .03 .04

2.549 23.8999 125.00 19.0813 .06 .07 -.01 .01

MIXTURE NO 11 (Hiza, et al . [14])

'

0.91674 CH^ + 0.08326 "^4^10

.521 25.1536 105.00 19.5467 .21 .35 .33 .22

.810 24.7960 110.00 19.5467 .12 .26 .21 .15

1 216 24.4512 115.00 19.5467 .14 .24 .18 .15

1.753 24.0889 120.00 19.5467 .10 .19 .11 .11

2.472 23.7370 125.00 19.5467 .15 .18 .10 .13

3.374 23.3789 130.00 19.5467 .17 .19 .12 .16

4.509 23.0110 135.00 19.5467 .21 .19 .15 .18

5.887 22.6391 140.00 19.5467 .35 .20 .22 .23

MIXTURE NO 12 (Haynes [10])

0.77982 CH^ + 0.22018
1.702 21.6066 120.00 25.3805 .12 .14 .11 .06

2.369 21.3549 125.00 25.3805 .11 .08 .02 .04

3.228 21.1020 130.00 25.3805 .09 .00 -.07 .03

4.291 20.8555 135.00 25.3805 .12 .03 -.10 .07

5.576 20.5978 140.00 25.3805 .20 .13 -.15 .07

27

155
 .

MIXTURE NO 13 (Haynes [10])

0.77762 CH^ + 0.22238
P D 1L T MW RKM HS CEL CS
1.179 21.8054 115.00 25.4011 .07 .15 .16 .02

I .oyy CL .3DUJ. lev .uu oc Ami

CD .Hull •U/ .U/ no

2.372 21.3164 125.00 25.4011 .10 .06 .01 .03

3.230 21 .0605 130.00 25.4011 .06 .03 -.10 .00

4.291 20.8011 135.00 25.4011 .02 .14 -.20 -.03

5.579 20.5448 140.00 25.4011 .09 .23 -.26 -.03

MIXTURE NO 14 (Hiza, et al. [14])

u oooco "^4^10
3.183 18.3058 130.00 33.3687 - .32 .11 -.15 -.08
2.342 18.4853 125.00 33.3687 - .17 .00 .00 -.03
2.281 18.4772 125.00 33.3687 - .21 .05 -.04 -.08

1.636 18.6495 120.00 33.3687 - .18 .02 .07 -.06

MIXTURE NO 15 (Haynes [10])

0.92044 CH^ + 0.07956 ^^4^10
1.254 24.3633 115.00 19.3910 - .21 .05 .06 -.01

i .OUD 120.00 19.3910 - .24 - .uo

2.521 23.6403 125.00 19.3910 - .22 .01 .04 -.03

3.434 23.2752 130.00 19.3910 - .21 .03 .08 .00

4.567 22.8920 135.00 19.3910 - .22 .00 .09 -.02

5.950 22.5037 140.00 19.3910 - .14 .02 .14 -.02

MIXTURE NO 16 (Haynes [10])

0.78329 CH^ + 0.21671 i^4"l0
.782 21.9144 110.00 25.1625 - .33 .10 .15 .01

1.164 21.6652 115.00 25.1625 - .33 .05 .07 -.01

1.671 21.4136 120.00 25.1625 - .35 .01 .00 -.02

2.329 21.1668 125.00 25.1625 - .34 .00 -.03 .00

3.170 20.9125 130.00 25.1625 - .35 .04 -.07 .00

4.208 20.6629 135.00 25.1625 - .33 .06 -.06 .04

5.474 20.4082 140.00 25.1625 - .23 .09 -.04 .07

28

15b
 .

MIXTURE NO 17 (Hiza, et al [14])

0.48687 CH. + 0.51313 ^^4^10
P D T MW RKM HS CELL CS
.629 17 .3575 110.00 37.6362 - .62 .05 .11 .02

.938 17.2076 115.00 37.6362 - .70 .00 -.01 -.02

1.361 17.0639 120.00 37.6362 - .64 - .02 -.08 -.02
1.852 16.9156 125.00 37 .6362 - .69 - .07 -.17 -.04

MIXTURE NO 18 (Hiza, et al. [14])

0.95248 CH^ + 0.04752 ^2
1.380 26.8476 105.00 16.6119 - .05 - .04 -.04 -.02

1.990 26.4052 110.00 16.6119 .06 .00 .00 .04

2.634 25.9374 115.00 16.6119 .20 .01 .02 .06

3.500 25.4522 120.00 16.6119 .26 .03 .07

4.600 24 .9496 125 .00 16 .6119 .54 .04 .07

5.830 24.4210 130.00 16.6119 .69 -5.69 .03 .05

7.300 23.8600 135.00 16.6119 .77 -3.99 -.01 -.03

9.200 23.2809 140.00 16.6119 1.09 -2.69 -.02 -.07

MIXTURE NO 19 (Hiza, et al . [14])

0.69651 CH^ + 0.30349 N2
3.450 26.8735 100.00 19.6759 - .39 .06 .09 .00

4.661 26.3393 105.00 19.6759 - .56 .02 .07 .04

6.181 25.7686 110.00 19.6759 - .92 - .05 -.02 .00

8.010 25.1790 115.00 19.6759 -1.28 .13 -.08 -.03

10.112 24.5737 120.00 19.6759 -1.56 1.88 -.07 -.04

MIXTURE NO 20 (Hiza, et al . [14])

0.50758 CH^ + 0.49242 N2
3.303 27.0801 95.00 21.9375 -2.16 .11 .14 .06

4.468 26.4588 100.00 21.9375 -1.92 - .01 .05 .04

6.383 25 .8106 105.00 21.9375 -1.56 - .15 -.05 -.01

8.461 25.1387 110.00 21.9375 -1.29 - .17 -.09 -.03

10.740 24.4431 115.00 21.9375 1.12 .28 -.04 -.01

13.983 23.7096 120.00 21.9375 -1.79 3.41 .07 -.01

17.529 22.9315 125.00 21 .9375 -1.14 * .26 .00

21 .076 22.1005 130.00 21.9375 - .72 -2.84 .05 .10

* The hard sphere solution for the density of N2 failed.

29

157
 MIXTURE NO 21 (Hiza, et al. [14])
0.67287 CgHg + 0.32713 ^3^8
P D T MW RKM HS CELL CS
18.6192 125.00 34.6588 .00 .02 -.02 .00

18.4648 130.00 34 .6588 .00 .02 .00 .01

18.3059 135.00 34.6588 .00 .00 .00 .00

18.1509 140.00 34.6588 .00 .01 .04 .02

MIXTURE NO 22 (Hiza, et al. [14])

0.50105 Hg + 0.49895 C3Hg

18.3618 105.00 37.0689 .02 .04 .02

18.2169 110.00 37 .0689 .01 .00 .00

18.0726 115.00 37 .0689 .00 -.03 -.01

17.9282 120.00 37.0689 - .01 -.05 -.01

17 .7880 125.00 37 .0689 .00 -.03 .00

17.6412 130.00 37.0689 - .02 -.04 -.02

17.4988 135.00 37 .0689 - .01 -.01 -.01

17 .3526 140.00 37 .0689 - .03 .00 -.01

MIXTURE NO 23 (Hiza, et al . [14])

0.67117 C2Hg + 0.32883 nC^H^Q
17.5047 110.00 39.2952 .00 .02 .00

17.3706 115.00 39.2952 .01 .00 .00

17.1031 125.00 39 .2952 .02 .00 .02

16.9626 130.00 , 39.2952 - .01 -.03 -.02

16.8285 135.00 39.2952 .00 -.01 -.01

16.6947 140.00 39.2952 .00 .03 .01

MIXTURE NO 24 (Hiza, et al . [14])

0.65343 C2Hg + 0.34657 nC^H^Q
17.2184 115.00 39.7929 .00 .00 .00

17.0824 120.00 39.7929 - .02 -.03 -.02

MIXTURE NO 25 (Hiza, et al . [14])

0.72436 CgHg + 0.27564 IC^H^q
17.9779 105.00 37 .8030 .01 .01 .01

17.8401 110.00 37.8030 .01 .00 .01

17.4235 125.00 37 .8030 .01 .00 .01

17.2825 130.00 37.8030 .00 .01 .01

30

158
 .

MIXTURE NO 25 (Hiza, et al . [14])

0.68939 CgHg + 0.31061 iC^H^Q
p D T MW RKM HS CELL cs
17.3716 115.00 38.7840 .00 -.03 .00
17.2344 120.00 38.7840 _ .02 -.04 -.02

MIXTURE NO 27 (Hiza, et al [14])

0.94067 C^H^ + 0.05933 N2;

3.850 21.4718 105.00 29.9481 .02 -.01 -.03

4.630 21.2912 110.00 29.9481 .03 .02 .03

5.472 21.0845 115.00 29.9481 - .01 -.04 -.04

6.383 20.8998 120.00 29.9481 .72 .01 .00

MIXTURE NO 28 (Hiza, et al [14])

0.60949 C3Hg + 0.39051
14.6487 115.00 49.5749 .01 -.02 -.01

14.5521 120.00 49.5749 .02 .00 .02

MIXTURE NO 29 (Hiza, et al [14])

0.60650 C,Hq + 0.3935:
14.1343 140.00 49.6169 .00 -.03 .00

14.0333 145.00 49.6169 .00 .00 .00

13.9346 150.00 49.6169 .02 .05 .02

MIXTURE NO 30 (Hiza, et al [14])

0.58692 C3Hg + 0.41308
14.6839 110.00 49.8915 .01 .04 .01

14.1748 135.00 49.8915 .03 -.06 -.03

14.0786 140.00 49.8915 .00 -.02 .00

MIXTURE NO 31 (Hiza, et al., [14])

0.49030 C,Ho + 0.50970 K.H,^
14.3080 105.00 51.2468 - .01 -.04 -.01

14.2136 110.00 51.2468 - .01 -.05 .00

14.1219 115.00 51.2468 - .02 -.03 .02

14.0257 120.00 51.2468 .01 -.04 .01

13.9300 125.00 51.2468 .00 -.04 .00

13.8342 130.00 51 .2468 - .01 -.04 -.01

31

159
 .

MIXTURE NO 32 (Hiza, et al. [14])

0.50326 C,Ho + 0.49674 iC.H.n
3 8 4 10
P D T MW RKM HS CELL cs
13.9718 125.00 51.0650 .01 -.03 .01

13.8737 130.00 51.0650 - .01 -.04 -.01

MIXTURE NO 33 (Hiza, et al [14])

0.97986 CjHg + 0.02014 ^2
3.567 16.2131 110.00 43.7733 .01 -.04 -.01
4.712 16.0931 115.00 43.7733 - .02 -.05 -.02

MIXTURE NO 34 (Hiza, et al. [14])

0.96206 CjHg + 0.03794
4.955 16.4638 105.00 43.4870 .02 -.03 .01

6.708 16.3410 110.00 43.4870 .00 -.04 .00

MIXTURE NO 35 (Hiza, et al. rL14J}

T \ 1 /I

0.93260 C3Hg + 0.06740

8.795 16.7084 105.00 43.0132 .20 .11 .18

6.313 16.8055 100.00 43.0132 .04 -.07 .01

MIXTURE NO 36 (Hiza, et al. [14])

0.52961 nC^H^Q + 0.47039 iC^H^o
12.6943 125.00 58.1243 .01 -.03 -.01

12.6133 130.00 58.1243 .01 -.01 .01

12.5271 135.00 58.1243 .01 -.04 -.01

12.4447 140.00 58.1243 .00 -.02 .00

MIXTURE NO 37 (Hiza and Haynes [15])

0.80284 CH. + 0.09902 C^Hg + 0.09814 C3H0
.476 24.9975 105.00 20.1852 .02 .04 .06 .02

.747 24.6696 110.00 20.1852 - .03 .00 .00 -.02

1.119 24.3515 115.00 20.1852 .01 .02 .02 .01

1.616 24.0335 120.00 20.1852 .07 .07 .07 .06

32

IbU
 MIXTURE NO 38 (Hiza and Haynes [15])
0.34242 CH4 + 0.31372 C^Hg . 0.34386 C3HJ
i

P D T MW RKM HS CELL CS
All 20.5342 110.00 30.0903 - .03 .02 .00 -.03
.636 20.3497 115.00 30.0903 - .01 .03 -.02 -.02
.638 20.3567
b W •www/ 115.00 . UJ • UD
nfi m * UX

.914 20.1786 120.00 30.0903 .12 .10 .04 .06

.914 20.1803 120.00 30.0903 .12 .11 .05 .07

1.280 19.9858 125.00 30.0903 .10 .08 .01 .04

MIXTURE NO 39 (Hiza and Haynes [15])

0.67040 CH^ + 0.16610 0.16350 N2
3.688 25.8587 105.00 20 .3301 .57 .01 .28 .00

4.813 25.4598 110.00 20.3301 .64 - .03 .27 -.01

6.181 25.0466 115.00 20.3301 .79 .02 .25 -.05
7.731 24.6359 120.00 20.3301 .70 .42 .31 -.04

MIXTURE NO 40 (Hiza and Haynes [15])

0.34140 CH. + 0.32060 ^2"6 0 33800 N

7.190 24.9084 105.00 24.5861 2.06 .01 .61 .07

9.494 24.5350 110.00 24.5861 2.55 .

- .01 .64 .04

L2.372 24.1433 115.00 24.5861 3.72 .24 .65 -.06

MIXTURE NO 41 (Miller and Hiza [25])

0 .86786 CH. + 0.08181 C-H^ + 0.05033 iC^H'10
3 8
.466 24.9239 100.00 20.4561 - .17 .04 .11 -.03
.769 24.4200 108.00 20.4561 - .17 .01 .05 -.04
1.327 23.9676 115.00 20.4561 - .18 - .01 .02 -.05

MIXTURE NO 42 (Hiza and Haynes [15])

0.79770 CH^ + 0.10280 C3H3 + 0.09950 N^
2.817 25.3950 105.00 20.1181 .17 .08 .24 -.01

3.620 25.0153 110.00 20.1181 .10 .02 .23 -.05

4.549 24.6465 115.00 20.1181 .08 .08 .33 .01

5.610 24.2667 120.00 20.1181 .02 * .44 .04

* The hard sphere solution for the density of N2 failed.

33

161
 MIXTURE NO 43 (Haynes [9])
0.89071 CH, + 0.04998 "^4^10 + 0.05931
p D T HS CELL cs

2.400 25.3450 110.00 18.8562 - .23 .00 .01 .03

3.145 24.9440 115.00 18.8562 - .25 - .04 -.02 .01

4.082 24.5383 120.00 18.8552 - .38 * -.02 .00

5.196 24.3141 125.00 18.8562 - .32 * -.04 -.04

MIXTURE NO 44 (Miller and Hiza [25])

0.85317 CH^ + 0.050// * 0.04855 C^Hg + 0.04751
4^6 S^O
1.038 24 .6056 110.04 20.1165 .02 .15 .10 .06

1.464 24 .2712 115.00 20.1165 .00 .11 .04 .02

MIXTURE NO 45 (Haynes [9])

0.85133 CH^ + 0.05759 4^6" 0.04808 C,ri„ +

3 8
0.04300
1 .180 24.3243 115.00 20.0092 - .04 .07 .00 -.01

1.700 Zd .yyob 120.00 20.0092 .00 .10 .02 .03

2.374 c6 .bboo 125.00 20.0092 .05 .11 .04 .05

3.232 23 .3108 130.00 20.0092 .04 .11 .05 .05

4.301 CL .9do4 135.00 20.0092 .08 .13 .11 .09

MIXTURE NO 46 (Haynes [9])

0.84556 CH^ + 0.07924 ^2^6 ^ 0.05060 CjHg + 0.02450 nC,H,„
4 10
1.167 24.5569 115.00 19.6051 .00 .04 .00 -.01

1.683 24.2126 120.00 19.6051 .02 .04 -.01 -.01

2.350 23.8698 125.00 19.6051 .08 .07 .03 .03

3.201 23.5204 130.00 19.6051 .12 .11 .09 .07

MIXTURE NO 47 (Haynes [9])

*

0.86040 CH^ + 0.04600 '^2"6^ 0.04790 C3Hg + 0.04570 ^^4^10

1.186 24.2654 115.00 19.9552 - .14 .01 .04 -.03

1.710 23.9371 120.00 19.9552 - .10 .04 .07 .01

2.387 23.5860 125.00 19.9552 - .10 .01 .05 -.01

3.248 23.2331 130.00 19.9552 - .13 - .01 .06 -.02

4.320 22.8637 135.00 19.9552 - .18 - .07 .05 -.07

* The hard sphere solution for the density of N2 failed.

34

162
 MIXTURE NO 48 (Haynes [9])
0.85378 CH^ + 0.05178 0.04703 CjHj + 0.04741
J ^^4"l0
P D T RKM MS CELL CS
1.190 24.2100 115.00 20.0838 - .14 .01 .04 -.03
1.706 23.8779 120.00 20.0838 - .14 .01 .04 -.02
2.379 23.5324 125.00 20.0838 - .13 - .02 .02 -.04
3.238 23.1834 130.00 20.0838 - .15 - .04 .03 -.05

MIXTURE NO 49 (Miller and Hiza [25])

0.85378 CH^ + 0.05178 0.04703 CjHj + 0.04741 ^^4^10
J

.972 24.5434 110.02 20.0838 - .14 .03 .06 -.01

1.429 24.2154 115.01 20.0838 - .12 .02 .05 -.01

MIXTURE NO 50 (Hiza and Haynes [15])

0.85260 CH^ + 0.04830 0.05070 CjHj + 0.04840 N^
J

1.581 25.8910 105.00 18.7223 .01 - .04 .03 -.07

2.138 25.5081 110.00 18.7223 - .09 - .07 .03 -.07
2.827 25.1224 115.00 18.7223 - .05 - .06 .05 -.06
-*
3.650 24.7283 120.00 18.7223 - .08 .08 -.06

MIXTURE NO 51 ( ;Haynes [9]; )

0.85892 CH^ + 0.11532 ^2^6^ 0.01341 CjHj + 0.00705 nC^H^Q + 0.00530 iC^H^g
J

1.185 25.0957 115.00 18.5565 .05 - .02 -.01 -.03

1.706 24.7131 120.00 18.5565 .03 - .06 -.05 -.07
2.372 24.3294 125.00 18.5565 .02 - .07 -.05 -.09
3.225 23.9490 130.00 18.5565 .05 - .02 .01 -.06

MIXTURE NO 52 (Hiza and Haynes [15])

0.85442 CH^ + 0.05042 C2Hg. 0.04038 CjHj + 0.02901 ^ 0.02577 iC^H^Q
J S"l0
.515 24.8775 105.00 20.1885 .09 .25 .25 .16

.818 24.5382 110.00 20.1885 .00 .17 .15 .09

1.190 24.2083 115.00 20.1885 .00 .15 .12 .08

1.695 23.8859 120.00 20.1885 .05 .19 .15 .13

MIXTURE NO 53 (
Haynes [9])
0.84558 CH^ + 0.08153 0.04778 CjHj + 0.01252 nC^H^O + 0.01259 iC^H^Q
j

1.166 24.5586 115.00 19.5838 .01 .05 .04 .01

1.680 24.2180 120.00 19.5838 .04 .07 .05 .03

2.348 23.8688 125.00 19.5838 .08 .08 .08 .04

3.188 23.5154 130.00 19.5838 .10 .10 .12 .07

* The hard sphere solution for the density of N2 failed.

35

163
 MIXTURE NO 54 (Haynes [9])
0.81249 CH^ + 0.08484 0.04931 C3Hg + 0.02708 nC^H^Q + 0.02628 N2
p
r n
u T
1 nw KNn no rn
LtLL
1

2.214 24.4562 115.00 20.0706 .01 .01 .03 -.02

2.874 24.1119 120.00 20.0706 .00 * .04 -.01
3.768 23.7507 125.00 20.0706 .02 * .02 -.05
4.793 23.3954 130.00 20.0706 - .02 -3.67 .07 -.03

MIXTURE NO 55 (Hiza and Haynes [15])

0.79090 CH. + 0.05600 CpHg + 0.05000 CjHg + 0.04770 nC^H^Q + 0.05540 N2
1.933 24.8080 105.00 20.9017 .04 .10 .12 .05

24.4664 110.00 9n om 7 fi9 n9

MIXTURE NO 56 (Miller and Hiza [25])

0.79054 CH4 + 0.05597 ^2^6 " 0.04996 C3Hg + 0.04762 nC^H^Q + 0.05591 Ng
2.158 24.8354 105.03 20.9029 .15 .21 .23 .15

MIXTURE NO 57 (Haynes [9])

CH4 + 0.04542 0.05050 CjHg + 0.04667 iC^H^Q +
3.005 24.1487 115.00 20.6355 .02 -.02 -.03 -.04
3.863 23.8075 120.00 20.6355 - .14 * .00 -.03
4.874 23.4518 125.00 20.6355 - .12 * .01 -.05

6.125 23.0893 130.00 20.6355 - .18 -5.06 .05 -.08

MIXTURE NO 58 (Hiza and Haynes [15])

U .OUDUU CH^ + 0.04680 0.04820 C3Hg + 0.05000 iC^H^q + n nAQnn n
u.utyuu
2.039 24.7803 105.00 20.7423 - .03 .08 .03 .01

2.550 24.4327 110.00 20.7423 - .09 - .02 -.06 -.06

3.135 24.1039 115.00 20.7423 .06 .00 -.02 -.02

3.790 23.7707 120.00 20.7423 - .10 * -.01 -.02

MIXTURE NO 59 (Miller and Hiza [25])

0.80545 CH4 + 0.04671 C2Hg. 0.04817 C3Hg + 0.04998 iC^H^Q + 0.04969 N2
2.273 24.7831 105.06 20.7476 .00 .10 .05 .03

* The hard sphere solution for the density of N2 failed.

36

164
 MIXTURE NO 60 (Haynes [9])
0.90613 CH^ + 0.06026 0.02154 C3Hg + 0.00306 0.00300 iC^H^g
+ 0.00601 ^2

P D T MW RKM HS CELL CS

1.478 25.3834 115.00 17.8195 .03 - .01 .01 -.01

2.043 24.9894 120.00 17.8195 .08 .00 .05 .02

2.785 24.5702 125.00 17.8195 .04 .00 .03 -.02

3.722 24.1578 130.00 17.8195 .06 -1.08 .11 .02

MIXTURE NO 61 (Haynes [9])

0.88225 CH^ + 0.07259 0.02561 C3Hg + 0.00492 ^ 0.00490 iC^H^Q
S^IO
+ 0.00973 N2

1.639 25.2023 115.00 18.3094 .09 .04 .07 .04

2.247 24.8047 120.00 18.3094 .06 .00 .04 .00

3.022 24.4022 125.00 18.3094 .04 .00 .03 .03

MIXTURE NO 62 (Haynes [9])

0.85934 CH^ + 0.08477 C^Hg. 0.02980 CjHg + 0.00707 nC^H^O + 0.00519 iC^H^Q

+ 0.01383 ^2

1.812 25 .0384 115.00 18.7496 .12 .06 .09 .06

2.441 24.6661 120.00 18.7496 .12 * .11 .07

3.223 24.2880 125.00 18.7496 .19 .00 .16 .09

4.222 23.8981 130.00 18.7496 .17 -2.12 .20 .10

MIXTURE NO 63 (Hiza and Haynes [15])

0.81300 CH^ + 0.04750 C^Hg. 0.04870 CjHj + 0.02420
J
0.02410 iC^H^g
+ 0.04250 ^2

1.834 24.8496 105.00 20.6168 .03 .12 .10 .05

2.807 24.5159 110.00 20.6168 .02 .08 .06 .03

3.384 24.1783 115.00 20.6168 .09 .06 .05 .03

6.039 23.8577 120.00 20.6168 .04 * .11 .08

MIXTURE NO 64 (Haynes [9])

0.85341 CH4 + 0.07898 C^Hg. 0.04729 CjHg + 0.00992 0.00854 iC^H^Q
+ 0.00089 "^5^12 + 0.00097 iC
5^12
.787 25.0063 110.00 19.3588 .01 .02 .02 -.01

1.172 24.6566 115.00 19.3588 .00 - .01 -.01 -.03

1.686 24.3079 120.00 19.3588 .02 .00 .00 -.02

2.351 23.9525 125.00 19.3588 .05 .01 .03 -.01

3.210 23.5883 130.00 19.3588 .06 .02 .06 .00

* The hard sphere solution for the density of N2 failed.

37

165
 «

MIXTURE NO 65 (Haynes [9])

0.75442 CH^ + 0.15401 ^2^6 * 0.06950 CjHg + 0.01057 "10 0.00978 iC.
+ 0.00083 + 0.00089 iC,5"

P D T MW RKM HS CELL cs
.723 24.2529 110.00 21 .1060 - .02 .00 -.01 .04

1.081 23.9519 115.00 21.1060 .04 .04 .03 .01

1.549 23.6535 120.00 21.1060 .07 .04 .02 .01

2.153 23.3351 125.00 21.1060 .04 .01 .00 .01

MIXTURE NO 66 (Haynes [9])

0.75713 CH^ + 0.13585 0.06742 CgHg + 0.01326 "10 0.01336 iC,
+ 0.00859 \L + 0.00216 nCcH,„ + 0.00223 iCg
^12
1.155 24.1809 110.00 21.3094 .09 .07 .08 .04

1 .byb li .o/ol 115.00 21.3094 .12 .05 .05 .02

2.155 23.5709 120.00 21 .3094 .12 .00 .08 .05

2.873 23.2644 125.00 21.3094 .15 .00 .12 .08

3.744 22.9514 130.00 21.3094 .22 -1.42 .16 .11

MIXTURE NO 67 (Haynes [9])

0.74275 CH^ + 0.16505 C^Hg-H 0.06547 CjHg + 0.00893 nC^ "10 ^ 0.00843 iC^ ^10
+ 0.00801 + 0.,00067 nC5H^2 + 0.00069 iC^
"12
1 .158 24.3141 110.00 21.0976 .02 - .02 .00 .05
1.584 24.0160 115.00 21.0976 .09 .01 .02 .01

2.093 23.6937 120.00 21.0976 .03 .00 -.03 .06

2.853 23.3804 125.00 21.0976 .05 .00 -.01 .05

MIXTURE NO 68 (Haynes [9])

0.90068 CH^ + 0.06537 C,Hg + 0.02200 C3H0 + 0.00284 nC^ "10 0.00291 iC.
+ 0.00599 Np + 0.00011 nCcH,^ + 0.00010 H.
12
1.456 25.3600 115.00 17.9026 .05 .01 .03 .01

2.024 24.9656 120.00 17.9026 .08 .00 .06 .03

2.762 24.5450 125.00 17.9026 .03 .00 .03 .03

3.698 24.1289 130.00 17.9026 .02 -1.13 .07 .02

38

166
 ,

Appendix B. Computer Program and Equation Parameters for the

Extended Corresponding States Model

The program listings in Appendix F include the extended corresponding states

method described in section 2. To use the program in its present form one must

make one of the two possible calls to SUBROUTINE PDMIX{P ,D ,T.X) (lines LNG 1

through LNG 9). The two possible calls are:

CALL PDMIX(P,D,T,X)

or

CALL PMIX(P,D,T,X)

When the call to PDMIX(P,D,T,X) is made the input variables are: P (pressure in

bars); T (temperature in kelvin); and X which is a matrix of the mole fraction of

the components of the mixture in the following order:

x(i) = mole fraction of methane

X(2) = mole fraction of ethane

X(3) = mole fraction of propane

X(4) = mole fraction of normal butane

X(5) = mole fraction of isobutane

X(6) = mole fraction of nitrogen

X(7) = mole fraction of normal pentane

X(8) = mole fraction of isopentane

No other components are allowed and if one or more of the above components are

absent, a zero should be inserted in the appropriate matrix element. The program

then calculates a density and it is returned in the argument list as D (density

in moles/liter)

When a call to PMIX(P,D,T,X) is made all of the above are the same except

that the roles of P and D are interchanged, i.e., D is an input variable and P is

calculated by the program.

39

167
 The range of the program is 90 to 150 kelvin for the saturated liquid phase

of any of the pure components of CH^ , ^2^^, ^3^8' "^4^10' ^^4^10' ^^2' "^4^12

or 'C^H^2 "lixture of those fluids. The program will extrapolate to

higher pressures (higher than saturation pressure) but the user is reminded that

such a calculation is an extrapolation and should be used with caution.

Other subprograms required:

SUBROUTINE MIX DATA, line LNG 95, Appendix F

SUBROUTINE DATA CH4, line LNG 162, Appendix F

FUNCTION FINDM, line LNG 217, Appendix F

FUNCTION SATL, line LNG 236, Appendix F

SUBROUTINE PROPS, line LNG 248, Appendix F

The equation of state from which Zq and Gq by eqs (1) and (2) may be

derived is:

40
 .

"3 + N./T
P = pRT + p^(N,T + N„T^/^ + "4' + N./T^)
r "2'

+ p^in^l + + Ng/T + Ng/T^)

+ p^(N^qT + N^^ + N^2/'^) + P^(N^3)

+ P^lN^^/T + N^g/T^) + p^N^g/T)

+ P^i^^j/T + N^g/T^) + p^(N^g/T2)

+ P^(N2o/T^ + N2t/T^) exp(- yp^) (B-1 )

+ p^(N22/T^ + N23/T^) exp(- w^)

+ P^(N24/T^ + "25/"^^) ^""P^" '^P^^

+ P^N2g/T^ + N27/t'^) exp(- yp^)

+ P^^N28/T^ + N2g/T^) exp(- yp^)

+ p^^(N3q/T^ + N3^/T^ + ^32/1^ exp(- yp^)

The computer subroutine PROPS(PP,DD,TT) , lines "LNG 248 through LNG 371 in

Appendix F" are the FORTRAN statements for eq (B-1) and the derivative

(3P/3p)t-

The parameters for eq (B-1) as applied to methane are given in table 2 and

in the FORTRAN SUBROUTINE DATA CH4, lines LNG 162 through LNG 216, Appendix F.

The parameters for eqs (5) and (6), (9) and (10) are given in table 3 and 4

and in the FORTRAN SUBROUTINE MIX DATA (IBASE), lines LNG 95 through LNG 161,

Appendix F

41

169
 Table 2. Methane Coefficients Ni for Eq B-1

R 0.08205616 -0.529609525984 x 10"

'16

-4
0.0096 '17
0.152264286004 x 10

10"
-1
= -0.187027997685 x '18
-0.109952182842 x 10

-3
0.103387108009 x 10 '19
0.191395549929 x 10

N3 = -0.155387625619 x 10' 0.386470003746 x lo'

'20

10"^ -0.157930582612 x lO'

= 0.772311478564 x '21

N5 = -0.377103300895 x lo' '22

0.195270144401 x lO'

Ng = 0.846818843475 x 10" '23

0.165996081629 x lO'

= -0.496415884529 0.603051146711
74

Ng = 0.869909352414 x 10^ •25

0.376485162808 x lO'

-2
N„ = -0.322821592493 x lo'
'26
0.125593680622 x 10

-4
'10
•0.395843026318 x 10 '27
-0.343570032513 x lO'

-1
0.266772318035 x 10 '28
-0.540945094139 x 10'
•ll

-2
'12
•0.304010057839 x 10
'29
0.185622284663 x 10

-8
'13
0,191584507536 x 10" '30
0.770786979245 x 10

-3 -5
= -0.195587933458 x 10 -0.286868318650 x 10
'31

-4
'15
0.607479967879 x 10 '32
0.372376961647 x 10

42

170
 Table 3. Coefficients for Eqs 5 and 6.

"1 "4 = -4.1419Z

= 0.919454 = 0.444850
"2 "5

= -4.01525
"3

= 0.356808 = 0.893323
"6 "8

"7 = 1.02619 = 0.761533

"9

*
d

3
w (bar) (K) (cm /mol) M No.

0.0109 45.956967 190.555 98.522 16.04303 1

0.110427 48.60314 305.5 146.2 30.07012 2

0.154837 42.445123 370. 200. 44.09721 3

0.176372 38.295398 425. 251.62 58.1243 4

0.150115 36.88998 408.1 263. 58.1243 5

0.0291791 33.542557 126.2 89 .827 28.0134 6

0.234320 33.812152 469.6 304. 72.15139 7

0.288886 31.988302 460.39 306. 72.15139 8

*Note: The large number of significant figures given for critical pressure
is necessary to reproduce the Z (Z = P V /RT ) in the least

squares fit of the data. P^'s have been converted to bar from
atmospheres.

43

171
 1 1 — 1 —
1 1

00
0000000
«£) CM t-H

00
00
cr>
• •
0 0 0 0 0 0
• • •
cy>

•
0
I—

0 CM
0 0000 r-.
00
•
CTi

•
0 0 0 0 0 0
cr>

• • •
CD
•

o
•
0 0 in
r«- <£>
0
CO 0CM
CM0 0
0 —
CM CM CO
(Ti
1

t— 0 CM
<—
0 V£) cr^
00
CO n
«a-

inCO
CM
cr>
^ in
1

0 CM
I-H «3-
0 0 CM
(

•o 00 CO
cr.
c •
0 0 0 0 0
• • • •

*o

ort

to
O"
UJ
i- • 00
o ro 00 cn o ^ CO (Ti 1—
H- ^
O CO
<x>
1^
<a- .-H
10
00
CO
CT^
0
to in CM 1— CO cri ro
4—' o o o cn I a-\
cn CTi 0 CTi
c • • • •
+-> 0 0 .—1 0
o c
14-
M- c
(U
0 >
ro CM
0
ro
ro
CM
c
0
«3-
CM
^ I-H
ro
CM
CM
in
in
4->

(D
0)

to

tft
000 r-l
.-I
O 00
•
0 0 0
1

CT>
•
00
•

J- S-
O) (U
+->
c E
>>
00
3 CM CTi o
c CM CTi
CTi in CO cr>

O O
CO IZ .-H 00 en
O cr.

CM
in
CM o
o
>-<

CM ro ^ in 10 CO •I- .-H CM m 10 00

44

172
 Table 5. Values for Checking Calculations Using Corresponding
States Equations.*

0.6975 CH^ + 0.156 C^Hg + 0.092 C3Hg + 0.029 nC^H^^ + 0.014 iC^H^Q
+ 0.0115

Temperature Density in moles/liter Pressure, bar

K

95 24.333 1

100 24.067 1.1

105 23.796 1.2

Included for check purposes only; these values are calculated from the
corresponding states model and are not experimental data.

45

173
 Appendix C. Computer Program and Equation Parameters for the

Hard Sphere Model

The program listing that follows is for the hard sphere model described in

section 3. To use the program in its present form one must make the following

reference to the computer program:

DEN = RODEN(P,T,X)

where DEN is density in moles/liter, P is pressure in bars, T is temperature in

kelvin, and X is a matrix of the mole fractions of the components of the mixture

in the following order:

X(l) = mole f racti on of methane

X(2) = mole fraction of ethane

X(3) = mole fraction of propane

X(4) = mole fraction of normal butane

X(5) = mole fraction of isobutane

X(6) = mole fraction of nitrogen

X(7) = mole fraction of normal pentane

X(8) = mole fraction of isopentane

Note: the inclusion of the pentanes is due to Rodosevich and Miller [33] and no

optimization of parameters has been included in this work for mixtures with

pentane as a component.

The range of the program is 90 to 150 kelvin for the saturated liquid phase

of mixtures of CH^, C2Hg, C2Hg, nC^H^Q, iC^H^q, N2, nC5Hj^2 ^^5^12*

program will calculate densities of any of the pure components but they will be

from a different model (i.e., some from an equation of state (CH^ and N2) and

some from the equations for saturated liquid densities. Therefore in its present

form, extrapolation to higher pressures is possible but the reliability of the

results is questionable.

46

174
 .

other subprograms required:

SUBROUTINE FM, lines LNG 435 to 529, Appendix F

SUBROUTINE ZERO, lines LNG 617 to 623, Appendix F

FUNCTION FIND VI. lines LNG 530 to 548, Appendix F

FUNCTION FIND Gl, lines LNG 549 to 560, Appendix F

FUNCTION EXCESS, lines LNG 561 to 616, Appendix F

FUNCTION FIND M, lines LNG 217 to 234, Appendix F

SUBROUTINE PROPS, lines LNG 248 to 371, Appendix F

SUBROUTINE DATA CH4, lines LNG 162 to 216, Appendix F

SUBROUTINE DATA N2, lines LNG 372 to 415, Appendix F

The parameters for eq (11), section 3 are given in table 6, and in lines

LNG 462 through LNG 466 in Appendix F.

The binary interaction parameters j.. and k.. in eqs (15) and (16),

section 3 are given in table 7 and in lines LNG 447 through LNG 461 in

Appendix F

47

175
 Table 6» Coefficients for Eq 11.

Fluid S^. Fluid

No.

CH^ 2.755x10^ 3.676x10"^ 1.00 1

CgHg 7.773 X 10^ 4.158 x 10"^ 1.50 2

C3Hg 14.165x10^ 4.644x10"^ 1.67 3

nC^H^Q 22.733 x 10^ 5.051 x 10"^ 1.83 4

iC^H^Q 21.279 X 10^ 5.056 x 10"^ 1.79 5

Ng' 1.718 X 10^ 3.546 x 10"^ 1.03 6

nC5H^2 30.550 x 10^ 5.389 x 10"^ 1.91 7

iCgH^^ 42.946x10^ 5.706x10"^ 2.11 8

b. = (2)(3.14159)(6.025 x 10^"^)S^/3

48

17b
 1

o
o

O —I
O O

—1 LO .—

o o o

r-1 CVJ
I

o
I

00
o

— 1

CTi ID
O
O <->

«3- r-

CM

Lo 1^ CO LO I—

49

177
 Appendix D. Computer Program and Parameters for the Revised

Klosek and McKinley Model

The program listing and tables that follow are for the Revised Klosek and

McKinley model described in section 4. The method may be used in two ways.

First using the equation:

The V^, and k2 may be obtained from tables 8, 9 and 10 and the volume of

the mixture calculated. For example given the mixture of 0.8130 CH^ + 0.0475

C2Hg + 0.0487 C3Hg + 0.0242 nC^H^Q + 0.0241 iC^H^q + 0.0425 N2 and a temperature

of 105 kelvin.

The X.V. and X.W. are obtained from table 8.

l^X.V. = (.8130)(. 037113) + ( .0475) ( .047267) + (.0487)( .061766)

+ (.0242) (.076100) + ( .0241 )( .077538) + ( .0425) ( .042565)

= 0.0409453

Then the E X^W. = ( .8130) (16.04303) + ( .0475) (30.07012) + (.0487) (44. 09721)

+ (.0242) (58 .1243) + ( .0241) (58.1243) + (.0425)(28.0143)

= 20.6168 the molecular weight of the mixture

from table 9 k^ = .697 x 10"^

from table 10 k2 = .849 x 10""^

plugging all this into eq (D-1) gives

-O^^BS
\u -

l/^mix - - 24.842 moles/liter

"mix

This compares to the experimental value of 24.850 (Appendix A, mixture No. 63) to

within 0.03%.

50

178
 The same result may be obtained by using the computer program in the

following way:

D = FMKM(T,X)

where T is temperature in kelvin and X is a matrix of mole fractions of the

components in the following order:

X(l) = mole fraction of methane

X(2) = mole fraction of ethane

X(3) = mole fraction of propane

X{4) = mole fraction of normal butane

X(5) = mole fraction of isobutane

X(6) = mole fraction of nitrogen

X(7) = mole fraction of normal pentane

X(8) = mol e fraction of isopentane

for the example: T = 105., X(l) = .8130, X(2) = .

X(4) = .0242, X(5) = .0241 and X(6) = .0425.

Other subprograms required:

FUNCTION VIDEAL, lines LNG 709 through 732, Appendix F

FUNCTION SAT, lines LNG 733 through 762, Appendix F

51

179
 1 1 — 11 1t1t — ( 11 < — 11 1 111 — 111 ( 1 — — 11 1 1 — — t I
1(1(1

ro
«3- in o COvo
.—
CO 00 CO
ro
«£>
CO «—
VO
I—
»—
ro ID
o O «— ro LO 00
ro ID cr>
evj
.-H a>
00 O
CJ^
1—
zn
C\J
CT>
O ro
O o o o O oo
0^
0^
ro
1— «— «— CM
vo
Lf>
CM
ro ID
r«.
ro ro ro
IT)
CSJ
CVJ
cr>
cr> t—i ro
.—
00 00 CO 00 a>
1—
O O o o o o o o O o O o o o o O o o o O O <T> cri cr> cri <Ti cyi LT)
t—

esj

ro cr> o
o VO
00 vo
o ^
00»—
roro CM CM CM ro
CM ro LO VO 00
lO
o O ro
CM ro
<— vo
VO 1^
•sS-

o VO 00 O CM vo 00 o ro in
«:J- cr> «d- cy>
.—1 ro ir> CVJ 1— ro ro
cn Ch o o o o o
CT>
CTi t— 1— I— CM CM CM CM CM ro ro .—1 f-H 1—
00 CO 00 00 0^ to
octo o O o o o o o o o O o O o O o O o o O o Ol
CTi cri cr>
o cr> (T> CTi cr>
«—

CM
I

ro o
I—
LO LO
00
r» vo 00 CM ro
•3-
00 CM CM O vo ro ro o
CM LO
00
ro
00 VO vo
t-H ir> IT)

LO o vo CM 00 LO ro r—
00 00 o O O CM vo 1—
CM CO ro LO vo
CO
o CM vo
vo
o LO CM ro
1 CTi

o
Q.
zCM o o o O o o o o o ^
f-.
ro ro ro ro ro
CTi cr>

o o o o o lO o LO o LO
o LO o VO
o lO o VO
o o
CTi •—1

00
CM

O. vo cn ro ro o vo LO vo 00 o LO CM «— o >— r>x «3- CTi

o 00 vo lO lO ^ ro ro ro ro ro LO VO 00
O) 1—1 o CM vo CO o CM vo 00 o CM <^ vo CO O CM VO 00 O ro
JZ zz vo vo vo vo vo 00 00 00 00 CO o^ O r~. r~. crv
•ej-
<T> CTi cy>
CM
o o o O o r~ O o o O O o o o o o o o O o O 00
4-> r-x r>x r~.

**-
O 1

t-H

o 00
LO C7>
-a
=3 2
cr
o CO
o cn o
.—1 LO
ro »—
a^
o — ro
LO00
LO
LO
•—1 00
ro ro CM
vo LO
vo
O
LO
0^
CO
«— 00
CM
vo
CM
vo 00 o CM lO 1— ro lO lO ro
^ CO LO ro r-. cr> CT> 1-H
1

•a LO LO LO vo vo vo vo vo vo
LO 00 00 r«» r>«. 00
p^ U
0)
o O O o O o o o O o O o O O o o o o o 1^
r»
o o r>- r-- CM
1

.—1
c
i- 00
bO

O)
CM LO LO ro CM CM
i—t VO vo <r ro ro cr> cr> (/)
LO CM o LO ro CM O
r~-.

00
VO o
ro
vo
LO ro
ro lO vo CO O CM ro LO
CO
f

cr> f—
00 vo f—
ro ^ vo 00 o CM
1 0)
3
ro
o
"S-
o o o «— I— I—
1

CM CM CM CM CM CM ro ro ro ro ro
vo vo vo vo VO vo vo vo vo VO VO vo VO VO vo vo vo vo vo vo vo
P^
1

«d-

E
O)
o o o o o o o o o o O O o O O o o O O o O O O <T>
>
o ..-

C
O
lO o ro LO 00 CM 00 LO vo 1—1
o ro a>
00 1—1 1—1 4->
CO ro vo CM 00
CTi 00 LO CM o CO LO ro CM CM 1—1 1—1 m
o 00 o 1— ro lO vo 00 o CO o CM
1

0) vo CM ro LO ro LO 1— 1—1 r». "3- c

Si
vo vo vo VO vo vo
^
r«-.
fx fx CO 00 00 00 CO CO1 o "3-
CTl <y> *o

o o o O o o O o o o o O o o o o O o o O o O «d- "d-

o
ro

00 LO CM lO lO CM VO
o O lO o LO
CTl 1—1 T— vo S-

o
vo
cr CM lO
CM lO
vo
CO
ro 00 ro
CM
VO ro
CM LO 00
o ^ CM LO CO ro
O ro vo o
CTl CTl 1-H
1-H
1-H o
LO LO lO
vo vo VO vo vo CO 00 00 cn o o o ro
cri
p~- f-«.
CTl
r>^ CTl

O o o ro
ro
o ro
o ro o ro
ro
o ro O ro o ro o CO o ro vo
ro «3-
O ro o ro
o ro
o CO o ro O ro
O o o o O «d- «d-

c=
VO o
1-H
4J
t- U
la 0)
V)
Z3 x:
o
OJ
C7> <u
0)
o CM
CTl
vo CO
CT^
o o ^
o CM o vo o o
o 00 1-H
CM
1-H 1-H
vo
1—1
00
i-H
o
CM
CM
CM
OS-
CM
vo
CM
00
CM
o
ro "o
1— E
52

180
 II — — I •

ir»if>oomoooocM
<TtoocT>CM^ootOLr>cr>i IT) o ID O CM
If)
CSJ CM
cn CM «3- o LO CM
CM CSJ CSJ CM CO

iDOomtnooooLO
CM>—i<—i«3-Lr)COLn'3-voo CO CM o CM
•a-
CM
r^cocTiOi—iCM^ixJoo-— «3- 00 O .-H CO CO CO
CTi

CM
CM CM CO CO

o o o o LO o o
LO ^ CO CM lO CO CO
in
CM
LO
,—
o
lO CO CO LO CO 00 o
CO
CSJ
v£)r- 00 o CO LO r«>
I

CT> CO
CM
o CO o o
t-H ,-( CM C^J CO CO

m LO IT) o
o LO LO LO lf> O
CM
CM
CT>
LO
VO
LO
lO
r>. 00 o o
LO
CM
00
CO
ir>
ir>
o
00 CM
CM
CTi
LO
CO
00
LO
o CM
CM
LO
00
CO
CM
CO
LO
o
CO CO t-H r-l 1-1 CM CM CO
o o

o
cn
LO lO
CO
LO
CO
LO LO
LO
O
1—
o o
cn
CM
o CM CO 00 CM o
CM
LO LO LO r-. 00 O CM
I

CO IT)
CM
LO cn o CM LO o
o o CM CO
*->
u
ID
u

c o o m LO LO O O LOo o
o
*->
o
CM
^ O
CO
LO LO
CO
LO
CM
r-^
.—
00
CMI

C3~l O O
CO
CM CO
CT> o
o
CM
o
in
<n
LO LO
LO CM
<n
CO co o o
CM LO
u o i-< CM CM
0) <u
i- i~
i-
o-
(_?
O 00
O CM
LO lO LO LO LO O O oo
cn
^
CO CO ^ ^
1^ CO
LO LO
(Ti
LO
cn CM LO
cn o cn
LO
CO
CO cn
LO r>«.
LO
cn
CO
CM
00
«3-
CM
cn
o
CM
(U

o o o o
CM LO O
LO o o
o O O
cn o
LO
CM CM 00 CM <n CM <— vo o
00 CM CM CO CO CO LO LO 00 00 CM CM
«5r
CO LO p«« cn o
.—I CM

omOLOOOOioLOO
CMCoir)LococMir><nr}-o CM vo CM LO O
^.-(.-H,—ir-tCMCMCMCO«:J- CsJ CO LO cn CO

OOOOOOr-t,— ^
LO ooooooooo
LO LO r>« r>. 00 cn co LO r»»

LO
o
o o
o o
o
IT)

o CO
CM
CM
o o LO
o o
CO
«3-
CO LO
r>»

O
cn
If) o O O
o LO t— I
LO
1—
I
O
CM
If)
CM
O
CO
LO
CO
O
cn
If)
cn
o O O
o LO t— t
If)
r—l
O
CM
lO
CJ
O
CO
LO
CO

53

181
 Appendix E. Computer Program for the Cell Model

The program listings for the cell model start at line LNG 763 and continue

on to the end of Appendix F. As is mentioned in section 4, no details of the

model are given here only the program listing. To use the program in Its present

form one must make the following reference to the computer program:

CALL ECKNON(P,D,T,X)

where P is input pressure in bars, D is the output density in moles/liter, T is

the input temperature in kelvins and X is a matrix of the mole fractions of the

components of the mixture in the following order:

x(i) = mole fraction of methane

X(2) = mole fraction of ethane

X(3) = mole fraction of propane

X(4) = mole fraction of normal butane

X{5) = mole fraction of isobutane

X(6) = mol e fraction of nitrogen

X(7) = mole fraction of normal pentane

X(8) = mole fraction of isopentane

54

182
 Appendix F. Computer Programs

Listing of computer programs for all four models. See the sections on the

individual models for a list of subprograms needed for each model.

The programs are written in FORTRAN IV and are operational on a CDC 6600

computer.

55

183
 SUBROUTINE PDMIX(P,D,T,X) LNG 1
C FOR A CALL TO PDMIX, P,T AND X ARE INPUT. P IS IN BAR.T IS IN LNG 2
C KELVIN AND D IS OUTPUT IN THE UNITS OF f^LES/LITER LNG 3
C FOR A CALL TO P MIX, D,T AND X ARE INPUT AND P IS OUTPUT, THE LNG 4
C UNITS ARE THE SAME LNG 5
C THE X MATRIX MUST CONTAIN THE MOL FRACTION OF THE ALLOWABLE FLUIDS LNG 6
C IN THE FOLLOWING ORDER, 1=C1,2=C2,3=C3,4=NC4,5=IC4,6=N2,7=NC5,8=IC5 LNG 7
C PLACE A ZERO IN THE ELEMENTS OF X WHERE THAT PARTICULAR GAS IS NOT LNG 8
C PRESENT LNG 9
DIMENSION ZATA(10,10),ATA(10,10),TC(10),VC(10),ZC(10),AC(10),W(10) LNG 10
1,PC(10) ,CF(9) LNG 11
DIMENSION THETA(10,10),TH(10,10),PHI(10,10),PH(10,10),F(10,10),FH LNG 12
1(10,10),H(10,10),HH(10,10),VR(10,10),TR{10,10),X(10) LNG 13
CONWON/DATA M/ZATA,ATA,TC,VC,W,TC0,VC0,AC0,ZC0,RR,R,0MEG0,AC,ZC,N LNG 14
1,PC,CF LNG 15
DATA(IE=0) LNG 16
D=0.0 $ PI=P/1. 01325 LNG 17
GO TO 4 LNG 18
ENTRY PMIX LNG 19
P=0.0 LNG 20
IP=1 LNG 21
GO TO 5 LNG 22
4 IP=0 LNG 23
5 CONTINUE LNG 24
IF{IE.GT.0)G0 TO 6 LNG 25
IE=1 LNG 26
CALL DATA CH4 LNG 27
IBASE=L=1 LNG 28
CALL MIX DATA(IBASE) LNG 29
6 CONTINUE LNG 30
DO 1 1=1, N LNG 31
F(I,I)=H(I,I)=1. LNG 32
THETA(I,I)=1. LNG 33
1 PHI{I,I)=1. LNG 34
DO 30 0=1,30 ^
LNG 35
HX=FXHX=0.0 LNG 36
DO 10 1=1, N LNG 37
FH{I,I)=F(I,I) LNG 38
HH{I,I)=H(I.I) LNG 39
IF(X(I).LT..0001)G0 T 0 10 LNG 40
F(I,I)=(TC(I)/TC(L))*THETA(I,I) LNG 41
H{I,n={VC{I)/VC(L))*PHI(I,I) LNG 42
10 CONTINUE LNG 43
DO 11 IA=1,N LNG 44
DO 11 IB=1,N LNG 45
IF(X(IA).LT..0001)G0 TO 11 LNG 46
IF(X(IB).LT..0001)GO TO 11 LNG 47
FAB=ZATA(IA,IB)*(F(IA,IA)* F(IB,IB))**.5 LNG 48
HAB=ATA(IA,IB)*( .5*H( I A, I A)**( 1 ./3 )+.5*H( IB, IB)**(
. 1 ./3. )
)**3 LNG 49
HX=HX+X(IA)*X(IB)*HAB LNG 50
FXHX=FXHX+X(IA)*X( IB)*HAB*FAB LNG 51
11 CONTINUE LNG 52
FX=FXHX/HX LNG 53
PR0=PI*HX/FX LNG 54
TR0=T/FX LNG 55
DEN=D*HX LNG 56
IFdP.EQ.DGO TO 8 LNG 57
DD=SATL(TR0)*1000.+1. LNG 58
9 DEN=FIND M{PR0 ,TR0 ,DD) LNG 59
8 CONTINUE LNG 60
IF(DEN.LE.O.O) GO TO 33 LNG 61
VR0=1000./DEN LNG 62
DO 12 1=1, N LNG 63

56

184
 I 4
R
3
1

IF(X( I) .LT. .OOODGO TO 12 LNG 64

VR( I , I )=VRO*PHI ( I , I )/VCO LNG 65
TR( I ,1 )=TRO*THETA( I ,1 )/TCO LNG 66
IF{VR( I ,1 ) .GT.2. )VR( I,I)=2. LNG 67
IF(VR(I,I).LT..5)yR(I,I)=.5 LNG 68
TH( I,I)=THETA(I,I) LNG 69
THETA( I I) = l + AC( I)-OMEGO)*{CF( l)-CF(21*ALnR(TR( I T U+f rF(lUrF(4
. { 70
1)/TR(I I))*{VR(I I)-CF(5))) Kir; 1
71
1 X

PH(I I)=PHI(I I) 72
PHKI I ) = 1 + AC( I)-OMEGO)*( CF(6)*( VR( I n-rF(7U-rF(81*(VR( T \\
( , ( NR 1 73
1-CF(9) )*ALOG( TR(I I) )*ZCO/ZC( ) ) \ wr;
liU 1
L. 74
12 CONTINUE LNG 75
DO
\J\J 13 1=1
A X N
11
J LNG 76
IF(X( I) .LT. .OOODGO TO 13 1NG 77
IFfABS ffFHfl I)-F(I I))/F(I lU GT OOnGD TO 30 1 Kir; 78
IFfABS ((HH(I D-Hd I))/H(I l\) GT 001 ^GO TO ^0 79
IFfABS ffTHfl D-THETAfl IlWTHETAfl l\\ GT 001 )Gn TO 30 80
IFfABS ffPHfl I)-PHIfI li)/PHIfI D) GT 001 ^GO TO ?0 1 wr; Rl
13 rONTINlIF l_ liU R?
GO TO 31 R3
30 CONTINUE 1NG 84
33 PRINT 100 P DEN T LNG 85
100 FORMATf* ITTERATION FAILED AT* 3F10 ^) LNG 86
STOP LNG 87
31 n=nFN/HX 1MR 88
THC=THETAf6 6) LNG 89
IFf IP EO 0)G0 TO 32 LNG 90
CALL PRESSf P,DEN,TRO) LNG 91
PI=1 .01325*P*FX/HX LNG 92
32 RETURN LNG 93
END LNG 94
SUBROUTINE MIX DATAflBASE) LNG 95
DIMENSION ZATAflO 10) ATAflO 10) TCflO) VCflO) ZCflO) ACflO) WflO) LNG 96
1 PCflO) CFf9) LNG 97
COMMON /DATA M/7ATA ATA TC VC W TfO VCO AfO ZCO RR R OMFGO AC 7C N LNG 98
1 ,PC ,CF LNG 99
nATAfPr=4R 3'ifi d? Qfi7R7 41 RQOOft 37 7Q4fi? 3fi 407'5ft 33 1Q3Q3) LNG 100
nATA(TC=1Q0 RtiR "^O^i ?7n 4?^ 4flfi 1 l?fi 2)
"i 01X
LNG 1X\J
nATA(W=lfi 04^0? "50 0701? 44 0Q7?1 Rft 1 ?43 "iR 1243 ?fi 0134) LNG 102
nATA(VC=Q8 R?? 1 4fi ? POO P^il fi? ?fi3 00 RQ fi?7) LNG 103
nATAfAr= mnp lin4?7 1'i4fl17 17fi'?7? l^lOll'i 0?Q17Q1) 1MR 104
!NR 1 rm
1NR 106
11 n?fi1Q ftQ??P? 7fi1^'^'^^ 1u 07/ 1

nATAf ATAf
L/Mln\nlnvljt/1«'JU31H
1 ( ATAf 1
/
? ^ = 1 01 Q?? ^ j\rMA\lj*T/"~l«
9)=] OORI d ^ j\Mlrt^ijO/~l«Ul-7c.t. ( ATAf 1
/
d 1=1 04.74^
UHt'+J \ y
/
OR
lUO
1

1 ( ATAf 1
IvninVljJj = l.
— 1 n^indft
U ( ATAf 1
jUtO ) jV"l"V /
fi 1 = I
1 jD ni noQ ) ^
1 •UlUU-?
/ /
1 09

Af ATAf? 1 =1 nn^QQ ) fATAf? dl = 1 fllfilfil 110

PfATAf? f^l = n7?fiQl fATAf? fil=1 n?1?7i fATAf^ 41 = 1 0017?!
1 1NG 111
1 1

\J\r\\r^\^y^l 01140) fATAf3 6)=1X*v<"TU\,^U/j\r^ir^V~j'^/

3fATAf3 5)=1X»UXX"U/j\riIf^\JjVJ/ 04606) fATAf4 5)= aJJ/XX"/}
997114) LNG 112
4fATAf4 61 = ?nPRf;) fATAfR fi) = l T?Rft9)
1 LNG 1 11 O 1

DATAfZATAfl ?)=1 ni1?7) f7ATAf1 "^1= QRRfiOR) fZATAfl 4)= 9R3130) 1NG 1lit
1

lf7ATA(l ^1= QRfiQ7R) fZATAfl fi)= 9'i?4'^0) (7ATAf? 3)= 999961) 1NR X1 XiJ
1

2fZATA(2 4)= 97???3) f7ATAf2 5)= 998886) fZATAf2 6)= 939622) LNG 116
3fZATAf3 4)= 98'5'i47) fZATAf3 5)=1 03099) fZATAf3 6) = . 912209) LNG 117
4fZATAf4 5)= 976416) fZATAf4 6)= 849200) fZATAfS 6)=. 857310) LNG 118
L=IBASE LNG 119
N=8 LNG 120
N1=N-1 LNG 121
PC(7)=33.37 LNG 122
PC(8)=31.57 LNG 123
TC(7)=469.6 LNG 124
TCf8)=460.39 LNG 125
VC(7)=304. LNG 126

57

ia5
f UVO / JUU • 1 NG 1?7
1 MG ?R
XlO
1

UfRi=7? TiT^Q 1 NG 1?Q

Ml/ / •tOHJtU
V '
1 NG 1 '^n
X<JU
Ar(ft^=
ML» \ O ?flRRRfi
vCOOOOU /
1 MG 1 "^1
x*?x
ATAfI 7^=ATAM RWl nfi 1 MG
ATAf? 7^=ATA(? fl^=l 0? 1 MG xoo
ataH 7^=ATAn f^)=^ m 1 NG XOH
ATAfd 7i=ATA(fl Ri=l 1 NG X03
ATAK 7^=ATAK fl^ =1 1 KIG XuU
ATA(fi7i=ATA(fi R^=1 MG xo /
Lnu
1

ATAf7 R^=1 Lnu X JO

7ATAf1 7^=7ATAf1 R)= PR KIG HQ
Lnu
1
X07
7ATA(? 7i=7ATAf? Ri= QQ LnU
1 1 40
KIG xtu

7ATAn 7\=7ATA('^ R^= QQ KIG

LPIU
1 141
XHX
7ATAM 7\s7ATAffl R>= QQ Lnu
1 14?
KIG XHC

7ATA(t; 7^=7ATA('i fli= QQ KIG 14'^

Lnu
1 XHO
ZATA(6 7i=ZATA(6 8i= 9Q 1KIG 144

7ATA(7 RWl
DO 3 J=l Ml LNG 146
J1=J+1 LNG 147
DO 3 K=J1 N LNG 148
ZATAfK J)=ZATA(J K) LNG 149
ATA(K J)=ATA{J K) LNG 150
DO 4 1=1 N LNG 151
ATAd I)=ZATA(I I)=l. LNG 152
PCO=PC(L) LNG 153
VCO=VC(
WW vwvi»/
L) LNG 154
TCO=TC(L)
l^w IwV^/ LNG 155
OMEGOsArfL^ LNG 156
7ro=prn*vcn/Tcn/R
£.ww ruw iww/ WW/ rv 1 I NG 157
X«J /

DO 5 1=1 N LNG 158

zc( n=pc( I)*VC( I ) /TC( I ) /R LNG 159
X«^ 7
RFTIIRN IfiO
X wv
END LNG 161
^IIRRdllTTNF DATA rH4 Lnu
1 KIG Xuc

INTTIAI T7F<; THF FnilATION OF ^TATF mU^TANT^ TO MFTHAMF Lnu

1 KIG xo J
DIMFN<;T0N RH?) VPfQ^ RTMI) Lnu
1 KIG X vH
1 fi4

rnNWON/nATA/R R RAMMA VP DTP Lnu

1 KIG X03
niMFW^^TfUl
u X luno X
I
AMO^
n\xui LnU
1 KIG xoo

rnMMnN/<;ATr
V/Ui iwn / on w//a
ii n 1 Lnu
1 KIG 167

LnU
1 KIR IDO

RAMMA=- nOQfi LnU

1 KIR

A(11=1Q0
r\\x/ X 555
^ » tj %j Lnu
1 KIG X f VJ

LnU
1 KIG 171

Af3^=1R 40415fi47?
X O 'TW*t X */W*T L • / Lnu
1 KIG 17?
X c1

A(4)=7 34Q8921512 t ivu 173

L> KIG X *J
1

A(5)=-l 4313160833 1 NG 174

A(6)=A(7)=0 0 LNG 175 Xi ^
U\ 1)=-
G( X/ .187027997685F-01
•XU/wkfJ^/wU^LwX LNG 176 X wf

2)= •Xw^^w/XWwwJ^'wX
G( fe/
u\ .103387108009E+01 LNG 177
G{ 3)=--155387625619E+02 LNG 178
G( 4)= 772311478564E+03 LNG 179
G( 5W- 377103300RQ5F+05 Lnu
1 RO
KIG 1xou

G( 6^= ft4fiR1Rfl4'^475F-n"? Lnu

1 NG 181
XOX
Gf 7)=_ 4Q641 5RR45?QF+nn Lnu
1 KIG XO^

G( 8)= 869909352414E+02 LNG 183

6( 9)=-.322821592493E+05 LNG 184
G( 10)=- .395843026318E-04 LNG 185
G{11)= .266772318035E-01 LNG 186
G( 12 =- .304010057839E+01
)
LNG 187
G{13)= .191584507536E-03 LNG 188
G{14)=-.195587933458E-03 LNG 189

58

186
G(15)= .607479967879E+01 1 NG 190
G ( 16 ) =- . 529609525984E-03 t 1 Q1
X 7 1.
G(17)= .152264286004E-04 1MR i.y
1Q? c
6(18)=-. 109952182842E-01 LNG 193
G(19)= .191395549929E-03 194
G{20)= .386470003746E+05 t X 7 <J
G{21)=- .157930582612E+07 LNG 196
G(22)= .195270144401E+03 LNG 197
G(23)= .165996081629E+07 LNG 198
G(24)= .603051146711E+00 LNG 199
6(25)= .376485162808E+02 LNG 200
G{26)= .125593680622E-02 LNG 201
G( 27 )=- .343570032513E+02 LNG 202
G 28 ) =- 540945094139E-05
( . LNG 203
G(29)= .185622284663E-02 LNG 204
G(30)= .770786979245E-08 LNG 205
G( 31 ) =- .286868318650E-05 LNG 206
G(32)= .372376961647E-04 LNG 207
VP(1)=4. 77748580 LNG 208
VP(2)=1. 76065363 LNG 209
VP(3)=-. 56788894 LNG 210
VP(4)=1. 32786231 LNG 211
VP(5)=1.5 LNG 212
VP(6)=.1159 LNG 213
VP(7)=90.68 LNG 214
VP(8i=190.555 LNG 215
END LNG 216
FUNCTION FIND M(P,T,D) LNG 217
SOLVES THE EQUATION OF STATE OF (ETHANE FOR DENSITY GIVEN P AND T LNG 218
DD=D LNG 219
TT=T LNG 220
DO 10 1=1 50 LNG 221
CALL PRESS (PP DD TT) LNG 222
P2=PP LNG 223
IF(ABS (P-P2)-l E-7*P)20 20 1 LNG 224
CALL DPDD(PP DD TT) LNG 225
DP=PP LNG 226
C0RR=(P2-P)/DP LNG 227
IFfABS (CORR)-l E-7*DD)20 20 10 LNG 228
DD=DD-CORR LNG 229
FIND M=0 LNG 230
RETURN LNG 231
FIND M=DD LNG 232
RETURN LNG 233
END LNG 234
FUNCTION SATL(T) LNG 235
CALCULATES THE SATURATED LIQUID DENSITY OF METHANE LNG 236
DIMENSION A(10) LNG 237
COMMON/SATC/A LNG 238
IF{T.GT.A(1))G0 TO 1 LNG 239
X=(l.-T/A(l)) LNG 240
SATL=A(2)+A(3)*X**(.35)+A(4)*X+A{5)*X**(4./3.)+A(6)*X**(5./3. LNG 241
)+A(7)*X**2 LNG 242
SATL=SATL/1000. LNG 243
RETURN LNG 244
SATL=1.E20 LNG 245
RETURN LNG 246
END LNG 247
SUBROUTINE PROPS(PP ,DD .TT) LNG 248
EQUATION OF STATE FOR METHANE AND NITROGEN LNG 249
DIMENSION X(33) LNG 250
DIMENSION B(33),G(32) LNG 251
EQUIVALENCE (B.X) LNG 252

59

18?
 COftlON/DATA/G.R. GAMMA
DATA(ID=1) LNG 253
DATA(IZ=1) LNG 254
1 CONTINUE LNG 255
D=DD LNG 256
P=PP LNG 257
T=TT LNG 258
GM=GAMMA LNG 259
D2=D*D LNG 260
D3=D?*D LNG 261
D4=D3*D LNG 262
D5=D4*D LNG 263
D6=D5*D LNG 264
D7=D6*D LNG 265
D8=D7*D LNG 266
D9=D8*D LNG 267
D10=D9*D LNG 268
D11=D10*D LNG 269
D12=D11*D LNG 270
D13=D12*D LNG 271
TS=SORT (T) LNG 272
T2=T*T LNG 273
T3=T2*T LNG 274
T4=T3*T LNG 275
T5=T4*T LNG 276
F=EXP {GM*D2) LNG 277
GO TO (100,200),K LNG 278
ENTRY PRESS LNG 279
K=l LNG 280
GO TO 1 LNG 281
100 CONTINUE LNG 282
B( 1)=D2*T LNG 283
B( 2)=D2*TS LNG 284
B( 3^=D2 LNG 285
B{ 4)=D2/T LNG 286
B( 5)=D2/T2 LNG 287
B( 6)=D3*T LNG 288
B( 7)=D3 LNG 289
B( 8)=D3/T LNG 290
B( 9)=D3/T2 LNG 291
B(10)=D4*T LNG 292
B(11)=D4 LNG 293
B(12)=D4/T LNG 294
B(13)=D5 LNG 295
B(14)=D6/T LNG 296
B(15)=D6/T2 LNG 297
B(16)=D7/T LNG 298
B(17)=D8/T LNG 299
B(18)=D8/T2 LNG 300
B(19)=D9/T2 LNG 301
B(20)=D3*F/T2 LNG 302
B(21)=D3*F/T3 LNG 303
B(22)=D5*F/T2 LNG 304
B(23)=D5*F/T4 LNG 305
B(24)=D7*F/T2 LNG 306
B(25)=D7*F/T3 LNG 307
B(26)=D9*F/T2 LNG 308
B(27)=D9*F/T4 LNG 309
B(28)=D11*F/T2 LNG 310
B{29)=D11*F/T3 LNG 311
B(30)=D13*F/T2 LNG 312
B(31)=D13*F/T3 LNG 313
B(32)=D13*F/T4 LNG 314
LNG 315
60

188
 0
1

102 P=0 LNG 316

DO 101 1=1,32 LNG 317
101 P=P+B(I)*G(I) LNG 318
P=P+R*D*T LNG 319
PP=P LNG 320
RETURN LNG 321
ENTRY DPDD LNG 322
K=2 LNG 323
GO TO 1 LNG 324
200 CONTINUE LNG 325
F1=2.00*F*GM*D LNG 326
F21=3.000*F*D2 +F1*D3 LNG 327
F22=5.000*F*D4 +F1*D5 LNG 328
F23=7.000*F*D6 +F1*D7 LNG 329
F24=9.000*F*D8 +F1*D9 LNG 330
F25=] 1 nn*c*ni n+n *m 1 LNG 331
F26=] 5 nn*r*rM 9+F1 *ni 7 LNG 332
D
D -9 nn*n*T
1 -L UU U . 1 LNG 333
D 9 LNG 334
0 -C • UU u 1

D
D -9^
J — nn*nu
UU• LNG 335
"L. nn*n/T
A
b 4 =9 UU U /
* 1 LNG 336
D c
D 0 — c. .UU U/ 1 LNG 337
n c -7 nn*n9*T
D 0 -J.UU Uc 1 LNG 338
D 1
-0 UU LNG 339
D / .

n Q
D 0 - J UU Ut:/
. 1 LNG 340
n -0
B( y -0 nn*n9
.UU uc.//t9<11 LNG 341
B 10 — H UU uo
. 1 LNG 342
B 11 -4 UU L) J
. LNG 343
B 12 =4.00*D3/T LNG 344
B 13 =5.00*04 LNG 345
B 14 =6.00*D5/T LNG 346
B 151 =6.00*D5/T2 LNG 347
B 16 =7.00*D6/T LNG 348
B 17 =8.00*D7/T LNG 349
B 18 =8.00*D7/T2 LNG 350
B 19 =9.00*D8/T2 LNG 351
B 20 =F21/T2 LNG 352
B 21 =F21/T3 LNG 353
B 221 =F22/T2 LNG 354
B 23 =F22/T4 LNG 355
B 24 =F23/T2 LNG 356
B 25 =F23/T3 LNG 357
B 26 =F24/T2 LNG 358
B 27' =F24/T4 LNG 359
B 28 =F25/T2 LNG 360
B 29 =F25/T3 LNG 361
B( 301 =F26/T2 LNG 362
B 3r =F26/T3 LNG 363
B( 321 =F26/T4 LNG 364
P==0 LNG 365
DO 201 1=1,32 LNG 366
201 P=P+B(I)*G(I) LNG 367
P=P+R*T LNG 368
PP=P LNG 369
RETURN LNG 370
END LNG 371
SUBROUTINE DATA N2 LNG 372
C INITIALIZES THE EQUATION OF STATE CONSTANTS TO NITROGEN LNG 373
DIMENSION G(32) ,VP(9) ,GI(11) LNG 374
COMMON/DATA/G R .GAMMA VP , DTP
, , LNG 375
R=8.20539E-2 LNG 376
GAMMA=-.0056 LNG 377
G( 1)= 0.136224769272827E-02 LNG 378

61

189
 G( 2) =
0.107032469908591E 00 LNG 379
G( 3) =
-0.243900721871413E 01 LNG 380
G( 4) =
0.341007449376470E 02 LNG 381
G( 5) = -0.422374309466167E 04 LNG 382
G( 6) = 0.105098600246494E-03 LNG 383
6( 7) = -0.112594826522081E-01 LNG 384
G( 8) = 0.142600789270907E-03 LNG 385
G( 9)= 0.184698501609007E 05 LNG 386
G(10)= 0.811140082588776E-07 LNG 387
G(ll)= 0.233011645038006E-02 LNG 388
G(12)= -0.507752586350986E 00 LNG 389
G{13)= 0.485027881931214E-04 LNG 390
G(14)= -0.113656764115364E-02 LNG 391
G(15)= -0.707430273540575E 00 LNG 392
G{16)= 0.751706548852680E-04 LNG 393
G(17)= -0.111614119537424E-05 LNG 394
G(18)= 0.368796562233495E-03 LNG 395
G(19)= -0.201317691347729E-05 LNG 396
G(20)= -0.169717444755949E 05 LNG 397
G(21)= -0.119719240044192E 06 LNG 398
G(22)= -0.975218272038281E 02 LNG 399
G(23)= 0.554639713151823E 05 LNG 400
G(24)= -0.179920450443470E 00 LNG 401
G{25)= -0.256582926077184E 01 LNG 402
G(26)= -0.413707715090789E-03 LNG 403
G(27)= -0.256245415300293E 00 LNG 404
G(28)= -0.124222373740063E-06 LNG 405
G(29)= 0.103556535840165E-04 LNG 406
G(30)= -0.538699166558303E-09 LNG 407
G(31)= -0.757415412839596E-08 LNG 408
G(32)= 0.585367172069521E-07 LNG 409
VP(1)=5. 1113192094 $ VP(2)=6.482667539E-1 LNG 410
VP(3)=-1. 5108730916E-1 $ VP(4)=7 .4028493342E-1 LNG 411
VP{5)=1.5 $ VP(6)=.123 $ VP(7)=63.15 $ VP(8)=126.26 LNG 412
VP(9)=0.0 LNG 413
DTP=31.0 LNG 414
-RETURN $ END LNG 415
FUNCTION VPN{TT) LNG 416
CALCULATES THE VAPOR PRESSURE OF BOTH METHANE AND NITROGEN LNG 417
DIMENSION G(32) ,VP(9) LNG 418
COMMON/DATA/G.R, GAMMA, VP .DTP LNG 419
T=TT LNG 420
X=(1.-VP(7)/T)/(1.-VP(7)/VP{8)) LNG 421
VPN=VP(6)*EXP(VP(1)*X+VP{2)*X*X+VP(3)*X**3+VP(9)*X**4+VP(4)*X* LNG 422
1(1.-X)**VP(5)) LNG 423
RETURN LNG 424
END LNG 425
FUNCTION RODEN(P,T,X) LNG 426
THE HARD SPHERE MODEL. SEE SUBROUTINE FM FOR THE ARGUMENT LIST LNG 427
DIMENSION X(10) LNG 428
CALL FM(P,T,X,V,G) LNG 429
D=28. LNG 430
V=V+V IDEL(P,D,T,X) LNG 431
R0DEN=1000./V LNG 432
RETURN LNG 433
END LNG 434
SUBROUTINE FM(0,T,X ,V9 ,G9) LNG 435
PREDICTION OF EXCESS PROPERTIES WITH LHW POTENTIAL, COMP 1-METH LNG 436
ANE,COMP 2-ETHANE,C0MP 3-PROPANE .COMP 4 N-BUTANE ,COMP 5-1 -BUT LNG 437
ANE, COMP 6-NITR06EN. COMP 7-NORMAL PENTANE.COMP 8-ISOPENTANE LNG 438
ADV CRYO ENGR. VOL. 19 (1973)-REPR0GRAMED BY R. MCCARTY .2/22/74 LNG 439
ARGUMENTS ARE X-MOLE FRACTIONS, T-TEMPERATURE.0-PRESSURE,V9-EXCESS LNG 440
VOLUME, G9-EXCESS GIBBS ENERGY, FIRST THREE ARE INPUT, LAST TWO ARE LNG 441

62

190
 ( , ,

OUTPUT, INPUT IS IN KELVINS AND BAR, LNG 442

DIMENSION A{8),B(8) ,S(8) ,V(8) ,G(8) ,C(8,8) ,D(8,8) ,K(8,8) LNG 443
1J(8,8),E(8),Y{8),0(8),X(10) LNG 444
TYPE REAL J,K,N1 LNG 445
DATA(KEY=1) LNG 446
DATA(J(1, 2)=-. 00388616), (J(l, 3)=-. 0120932) (0(1,4)=-. 0231577) , LNG 447
A( J{ 1, 5)=- .0238349), (J (1,6)=-. 00997547), (J (1 ,7)=-. 0326), (J( 1,8) = LNG 448
B-. 0458), (J(2, 4)=-. 00400910) ,( J(2, 5)=-. 00812712) ,(J( 2, 6)=-. 0143976) LNG 449
C, (J (2, 7)=- .003) ,(J(2,8)=-.004) ,(J(3,4)=+.0007615710),(J(3,5)=-.003 LNG 450
D83743),(J(3,6)=-.024014),(J(3,7)=0.0),(J(3,8)=0.0) ,(J(4,5)=.002221 LNG 451
E50),(J(4,7) = .0) ,(J(4,8) = .0),(J(5,6)=-.0576043),(J(4,6)=-.0576043), LNG 452
F(J(5,7)=.0),(J(5,8)=0.0),(J(6,7)=-.04),(J(6,8)=-.05),(J(7,8)=.0) LNG 453
G ,(J(2, 3)=-. 002162) LNG 454
DATA(K( 1,2)=. 00298830), (K( 1,3)=. 0597378), (K( 1,4)=. 110893), LNG 455
AK(1,5) = .100298) ,(K(1, 6) = . 0197290), (K( 1,7) = . 14), (K( 1,8) = 1745), . LNG 456
B(K(2,4)=.0677703) ,(K(2,5) = .0346632) .(K(2,6) = .0529034),(K(2,7) = LNG 457
C.02) ,(K(2,8) = .03),(K(3,4)=.0249291),(K(3,5)=-.00838212) ,(K(4,5) = .0 LNG 458
D199213),(K(4, 6)=. 154365 ) ,(K(4,7) = .0) ,(K(4,8) = .0) ,(K(5,6) = .154365) LNG 459
E, (K(5,7)=.0),(K(5,8)=.0),(K(6,7)=.15),(K(6,8)=.18),(K(7,8)=0.0) LNG 460
F, (K(2.3) = .014527),(K(3,6) = .14719),(K(3,7)=0.0) ,(K(3,8)=0.0) LNG 461
DATA(S=3.676E-8,4.158E-8,4.644E-8,5.051E-8,5.056E-8,3.546E-8, LNG 462
15.389E-8,5.706E-8) LNG 463
DATA(A=2.755E+5,7.773E+5,14.165E+5,22.733E+5,21.279E+5,1.718E+5, LNG 464
130.550E+5,42.946E+5) LNG 465
DATA( 0=1. ,1.5,1.67,1.83,1.79,1.03,1.91,2.11) LNG 466
THESE ARE THE ACENTRICITY FACTORS (FOR MOLECULAR SHAPES, ETC.) *** LNG 467
DATA(Y=35.,45.,60.,75.,75. ,40. ,90. ,105.) LNG 468
DATA(P1=3. 14159) ( Nl=6 .025E+23) ,(R=8.3143 )
, LNG 469
IF(KEY.EQ.0)G0 TO 1 LNG 470
KEY=0 LNG 471
DO 2 1=1,8 LNG 472
J(I,I)=0.0 LNG 473
K(I,I)=0.0 LNG 474
DO 2 M=I,8 LNG 475
J(M,I)=J(I,M) LNG 476
2 K(M,I)=K(I,M) LNG 477
1 CONTINUE LNG 478
P=0*.l LNG 479
IW=8 LNG 480
DO 10 1=1, IW LNG 481
10 E(I)=0(I) LNG 482
DO 15 1=1, IW LNG 483
15 B(I)=(2./3.)*P1*N1*S(I)**3 LNG 484
DO 20 1=1, IW LNG 485
DO 20 M=1,IW LNG 486
D(I,M) =(((B(I)**( 1./3. ) + B(M) ** ( 1./3.))/ 2. )*(1.-J{I,M) )) LNG 487
1 **3 LNG 488
20 C( I , M ) = ( 1. - K( I, M ) ) ( Ad)* A(M) ( 1. / 2. ) LNG 489
1( D( I, M )
**2 / ( B(I) * B(M) )
** ( 1./2. LNG 490
A2=0 $ B2=0 LNG 491
E2=0 $ V2=0 LNG 492
DO 25 1=1, IW LNG 493
V2=V2+X( I)*Y(I) LNG 494
DO 25 M=1,IW LNG 495
E2=E2+X(I)*X(M)*(E(I)+E(M))/2. LNG 496
+ X(I) * X(M) * C(
A2 = A2 I, M ) LNG 497
25 B2=B2+X(I)*X(M)*D(I,M) LNG 498
V6=V2 LNG 499
DO 30 I = 1, IW LNG 500
A3 = A(I) LNG 501
B3 = B(I) LNG 502
E3 = Ed) LNG 503
V2 = Yd) LNG 504

63

191
 VI = FIND Vl( A3, B3, E3, R, V2, P, T) LNG 505
Gl = FIND Gl( A3, B3, E3, R, V2, P, T) LNG 506
Vd) = VI LNG 507
30 Gd) = Gl LNG 508
A3=A2 LNG 509
B3=B2 LNG 510
E3=E2 LNG 511
V2=V6 LNG 512
V1=FIND V1(A3,B3,E3,R,V2,P,T) LNG 513
G1=FIND G1(A3,B3,E3,R,V2,P,T) LNG 514
V7=V1 LNG 515
G7=G1 LNG 516
EXCEAS VOLUME AND GIBS ENERGY LNG 517
V9=0 LNG 518
G9=0 LNG 519
DO 35 1=1, IW LNG 520
V9=V9-X(I)*V{I) LNG 521
35 G9=G9-X(I)*G(I) LNG 522
W9 = -V9 LNG 523
H9 = -G9 LNG 524
V9=V9+V7 LNG 525
G9=G9+G7 LNG 526
VIWW=V(IWW) LNG 527
RETURN LNG 528
END LNG 529
FUNCTION FIND Vl( A3, B3, E3,R, V2, P, T ) LNG 530
SOLVES THE HARD SPHERE EQUATION OF STATE FOR VOLUME GIVEN P AND T LNG 531
A2 IS THE CONSTANT A, B2 IS THE CONSTANT B, E3 IS THE ACENTRICITY LNG 532
INDEX = 0 LNG 533
1 VI = V2 LNG 534
XI = B3 / ( 4. * VI ) LNG 535
F2 = (( 1. + XI + Xl**2) / { 1. - Xl)**3) * E3 -A3 / ( VI * R * T) LNG 536
1- (P * VI) / { R*T) LNG 537
F3 = A3 / (R*T*V1**2) - P/ (R*T) LNG 538
F3 = F3 - ((( XI + 2.* Xl**2) 1 XI) + 3* ( 1. + XI + Xl**2) LNG 539
1* XI) / (( 1. - XI )*M * VD) E3 LNG 540
V2 = VI - F2/ F3 LNG 541
IF( ABS( (V2 - VI) / V2) .LT. .00001 ) GO TO 2 LNG 542
INDEX = INDEX + 1 LNG 543
IF( INDEX .LT, 250 ) GO TO 1 LNG 544
? VI = V2 LNG 545
FIND VI = V2 LNG 546
RETURN LNG 547
END LNG 548
FUNCTION FIND Gl( A3, B3, E3,R, V2, P, T ) LNG 549
CALCULATES THE GIBBS FREE ENERGY FOR THE HARD SPHERE EOS LNG 550
VI = V2 LNG 551
XI = B3 / ( 4. * VI ) LNG 552
Gl =ALOG( 1. / ( 1. - XI )) +{3. *X1) / { 1. - XI) + ( 3. * Xl**2) LNG 553
1 / (2. * ( 1. - XI )**2) LNG 554
Gl = Gl - A3 / ( E3 * R * T*V1 ) + ( P * VI ) / ( E3 * R * T ) -1 .0 LNG 555
1 - ALOG( VI ) LNG 556
Gl = R * T * E3 * Gl LNG 557
FIND Gl = Gl LNG 558
RETURN LNG 559
END LNG 560
FUNCTION EXCESS{P,DD,T,X) LNG 561
CALCULATES THE EXCESS OR IDEAL' VOLUM DEPENDING ON THE ENTRY LNG 562
DIMENSION X(10),F(10) LNG 563
KR=0 LNG 564
GO TO 1 LNG 565
ENTRY V IDEL LNG 566
KR=1 LNG 567

64
 1 CONTINUE LNG 568
CALL ZERO(F) LNG 569
IF(X(1) .LE..000001)G0 TO 2 LNG 570
CALL DATA CH4 LNG 571
IF(T.GT.190.555)G0 TO 12 LNG 572
PM=VPN(T)+. 00001 LNG 573
DELP=P-PM LNG 574
D=SAT(T,1) LNG 575
CALL DPDDfOP.D.T) LNG 576
DELD=DELP/DP LNG 577
D=DELD+D LNG 578
F(1)=X(1)*1000./D LNG 579
GO TO 2 LNG 580
12 D=FIND M{P,T.DD) LNG 581
IF{D.LE. 0.0)0=1000. LNG 582
F(1)=X(1)*1000./D LNG 583
2 IF(X(2).LE. .OOOOODGO TO 3 LNG 584
F(2)=X(2)*1000./SAT{T,2) LNG 585
3 IF(X(3) .LE.. OOOOODGO TO 4 LNG 586
F(3)=X(3)*1000./SAT(T,3) LNG 587
4 IF(X(4) .LE. .OOOODGO TO 5 LNG 588
F(4)=X(4)*1000./SAT(T,4) LNG 589
5 IF(X(5) .LE. .OOOODGO TO 6 LNG 590
F{5)=X(5)*1000./SAT(T,5) LNG 591
6 IF(X{7) .LE.. OOOODGO TO 61 LNG 592
F(7)=X(7)*1000./SAT(T,7) LNG 593
61 IF(X(8) .LE.. OOOODGO TO 62 LNG 594
F(8)=X(8)*1000./SAT(T,8) LNG 595
62 IF(X{6) .LE.. OOOODGO TO 8 LNG 596
CALL DATA N2 LNG 597
IF(T.GT.126.6)G0 TO 7 LNG 598
PN=VPN(T) + 000001
. LNG 599
DELP=P-PN LNG 600
D=SAT(T,6) LNG 601
CALL DPDD(DP,D,T) LNG 602
F{6)=X{6)*1000./(D+DELP/DP) LNG 603
GO TO 8 LNG 60'!
7 D=FIND M{P,T,DD) LNG 605
IF{D.LE. 0.0)0=1000. LNG 606
F(6)=X(6)*1000./D LNG 607
8 V=1000./DD LNG 608
VS=0 LNG 609
DO 21 1=1,8 LNG 610
21 VS=VS+F(I) LNG 611
EXCESS=V-VS LNG 612
IF(KR.GT.O)EXCESS=VS LNG 613
CALL DATA CH4 LNG 614
RETURN LNG 615
END LNG 616
SUBROUTINE ZERO(X) LNG 617
C INITIALIZES THE COMPONENT MATRIX TO 0 LNG 618
DIMENSION X(10) LNG 619
DO 1 1=1,10 LNG 620
1 X(I)=0.0 LNG 621
RETURN LNG 622
END LNG 623
FUNCTION FMKM(T,X) LNG 624
C THE REVISED KLOSEK AND MCKINLEY METHOD, THE INPUT IS TEMPERATURE LNG 625
C AND THE COMPONENT MATRIX. TEMPERATURE IS IN KELVIN .OUTPUT IS LNG 626
C DENSITY IN MOLES PER LITER. THE ALLOWABLE COMPONENTS ARE C1,C2,C3 LNG 627
C NC4,IC4,N2,NC5,IC5 IN THAT ORDER. THIS METHOD SHOULD NOT BE USED LNG 628
C FOR MIXTURES WITH LESS THAN 60% METHANE, OR FOR MIXTURES CONTAININ LNG 629
C MORE THAN 4% NITROGEN OR MORE THAN ^% EACH OF NC4 OR IC4 OR MORE LNG 630

65

193
 )

THAN 1% TOTAL OF NC5 AND IC5. LNG 631

DIMENSION TM(IOO) ,TN(100) ,X{10) ,0(8) LNG 632
DATA( 0=16 .04303 ,30 .07012 ,44 .09721 ,58 .1243 ,58 .1243 ,28 .0134 ,72 .1513LNG 633
19,72.15139) LNG 634
DATA{(TM( I), 1=1. 10)=- .005,. 12,. 22,. 34,. 43,. 515,. 595,. 66,. 725,. 795)LNG 635
DATA((TM( I), 1=11, 20)=-. 006,. 135,. 26,. 38,. 5,. 59,. 665,. 74,. 81,. 885) LNG 636
DATA((TM( I), 1=21, 30)=- .007,. 15,. 3,. 425,. 575,. 675,. 755,. 83,. 91,. 99)LNG 637
DATA{{TW( I) ,1=31, 40)=-. 007,. 165,. 34,. 475,. 635,. 735,. 84,. 92, 1.045, LNG 638
A1.12) LNG 639
DATA{(TM(I) 1=41 ,50 )=- .008 .19 .375 , .535 .725 , .835 , .95 ,1 .055 ,1 155
, , , . LNG 640
A,U245) LNG 641
DATA{(TM{I) 1=51, 60)=-. 009,. 22,. 44,. 61,. 81,. 945, 1.065, 1.18, 1.28, LNG 642
A1.38) LNG 643
DATA((TM(I) 1=61, 70)=- .01,. 25,. 5,. 695,. 92, 1.055, 1.205, 1.33, 1.45, LNG 644
A1.55) LNG 645
DATA((TM(I) 1=71, 80)=-. 013,. 295,. 59,. 795, 1.035, 1.21, 1.385, 1.525, LNG 646
11.64,1.75) LNG 647
DATA{(TM(I) 1=81 ,90)=- .015,. 345,. 7,. 92, 1.2, 1.37, 1.555, 1.715, LNG 648
Al. 86, 1.99) LNG 649
DATA{(TM{I) 1=91, 100)=-. 017,. 4,. 825, 1.06, 1.39, 1.59, 1.8, 1.95, LNG 650
A2. 105, 2. 272) LNG 651
DATA((TN{I) 1=1, 10)=-. 004,. 1,. 22,. 35,. 5,. 6,. 69,. 78,. 86,. 95) LNG 652
DATA((TN(I) 1 = 11, 20)=- .005,. 12,. 28,. 43,. 59,. 71,. 83,. 94, 1.05, 1.14) LNG 653
DATA((TN(I) 1=21 ,30)=-. 007,. 16,. 34,. 49,. 64,. 79,. 94, 1.08, 1.17, 1.27) LNG 654
DATA((TN{I) 1=31 ,40)=- .01 ,. 24,. 42,. 61,. 75,. 91, 1.05, 1.19, 1.33, 1.45) LNG 655
DATA{(TN(I) 1=41, 50)=- .015,. 32,. 59,. 77,. 92, 1.07, 1.22, 1.37, 1.52, LNG 656
11.71) LNG 657
DATA({TN(I) 1=51, 60)=-. 024,. 41,. 72,. 95, 1.15, 1.22, 1.3, 1.45, 1.65, 2.) LNG 658
DATA((TN(I) 1=61, 70)=- .032,. 6,. 91, 1.23, 1.43, 1.63, 1.85, 2. 08, 2. 3, LNG 659
12.45) LNG 660
DATA((TN{I) 1=71, 80)=-. 043,. 71, 1.13, 1.48, 1.73, 1.98, 2. 23, 2. 48, 2. 75 LNG 661
1,2.9) LNG 662
DATA((TN(I) 1=81, 90)=- .058,. 95, 1.46, 1.92, 2. 2, 2. 42, 2. 68, 3., LNG 663
A3. 32, 3. 52) LNG 664
DATA((TN(I) 1=91, 100)=-. 075, 1.3, 2. ,2. 4, 2. 6, 3., 3. 4, 3. 77, LNG 665
A3. 99, 4. 23) LNG 666
IF(X(1).LT. 00001) GO TO 20 LNG 667
AW=0.0 LNG 668
DO 1 1=1,8 LNG 669
1 AW=AW+X(I)*Q{I) LNG 670
VI=VIDEAL{T,X) LNG 671
0=1 LNG 672
IF(.T.GE.95.)J=11 LNG 673
IF{T.GE.100.)J=21 LNG 674
IF(T.GE.105.)J=31 LNG 675
IF(T.GE.110.)J=41 LNG 676
IF{T.GE.115.)J=51 LNG 677
IF(T.GE.120.)J=61 LNG 678
IF(T.GE.125.)J=71 LNG 679
IF{T.GE.130.)J=81 LNG 680
JJ=J+9 LNG 681
W=15. LNG 682
DO 5 1=0 ,JJ LNG 683
W=W+1. LNG 684
IF(AW.GT.W)GO TO 5 LNG 685
GO TO 6 LNG 686
5 CONTINUE LNG 687
1=00 LNG 688
6 DIF1=AW-W LNG 689
0=1-1 LNG 690
FK=(TM( I )-TM{0 ) )*DIF1+TM( I LNG 691
FK1=(TM( I+10)-TM(0+10) )*DIF1+TM( I+IO) LNG 692
IT=(T+. 00001 )/5. LNG 693

194
 DIF2=T-IT*5 LNG 694
IF(T.GE.135.)DIF2=T-130. LNG 695
IF(T.LT.90.)DIF2=T-90. LNG 696
FK=FK+{FKl-FK)*DIF2/5. LNG 697
IF(X(6) .LT..0001)G0 TO 17 LNG 698
FKN={TN(I)-TN(J))*DIF1+TN(I) LNG 699
FK1=(TN(I+10)-TN(J+10))*DIF1+TN(I+10) LNG 700
FKN=FKN+(FKl-FKN)*DIF2/5. LNG 701
FK=FK+{FKN-FK)*X(6)/.0425 LNG 702
17 FK=FK/1000. LNG 703
FMKM=1./(VI-FK*X(1)) LNG 704
RETURN LNG 705
20 FMKM=0.0 LNG 706
RETURN LNG 707
END LNG 708
FUNCTION V IDEAL(T,X) LNG 709
CALCULATES THE IDEAL VOLUME OF A MIXTURE FOR THE K AND M METHOD LNG 710
DIMENSION X(10) LNG 711
V=0 LNG 712
J=0 LNG 713
IF(X(6).GT. .0001. AND. T.GT. 115. )J=1 LNG 714
DO 10 1=1,8 LNG 715
IF(X( I) .LE. .OOOOODGO TO 10 LNG 716
IF(J.GT.O.AND.I.E0.6)GO TO 10 LNG 717
V=V+X( I)/SAT(T,I) LNG 718
10 CONTINUE LNG 719
IF(J.E0.1)V=V+X(6)/SATN2(T) LNG 720
VIDEAL=V LNG 721
RETURN LNG 722
END LNG 723
FUNCTION SATN2{T) LNG 724
CALCULATES A PSEUDO SATURATED LIQUID DENSITY FOR N2 ABOVE 115 K LNG 725
IF{T.LT.115.)G0 TO 1 LNG 726
DELT={T-115.) LNG 727
SATN2=SAT(115.,6)+DELT*(SAT(115.05,6)-SAT(114.95,6))/.l LNG 728
RETURN LNG 729
1 SATN2=SAT(T,6) LNG 730
RETURN LNG 731
END LNG 732
FUNCTION SAT(T,n LNG 733
CALCUALTES THE PURE FLUID DENSITIES FOR THE K AND M METHOD LNG 734
UNITS ARE DEG K AND MOLES/LITER LNG 735
DIMENSION A(7,8) LNG 736
DATA((A{I) ,1=1, 7)=190. 555, 10. 16, 18. 65812322, 6. 712030737, LNG 737
1-. 9472019702, 0.0,0.0) LNG 738
DATA((A(I) ,I=8,14)=305.33, LNG 739
1 6.86,12.55205121,13.43284373,-19.00461066,11.07715985,0.0) LNG 740
DATA((A( I) ,1=15, 21) =369. 82, 5., 8. 684458671, 18. 04085714, -29. 46261356 LNG 741
1,16.43559114,0.0) LNG 742
DATA({A(I) ,I=22,28)=425.16,3.92, LNG 743
1 7.286062567,11.96307859,-19.87591962, LNG 744
211.60211932,0.0) LNG 745
DATA((A(I),I=29,35)=408.13,3.8, LNG 746
1 7.657535400,8.145251283,-13.10582462, LNG 747
28.145894091,0.0) LNG 748
DATA{(A( I) ,1=36, 42) =126. 2, 11. 21, 19 .39216835, 26. 01408462, -39. 497587 LNG 749
191,23.32977312,0.0) LNG 750
DATA( (A( I) ,1=43, 49)=469. 6, 3. 285, -.0362004993, 59. 00202990, LNG 751
1-93.44193819,43.66780833,0.0) LNG 752
DATA({A{ I), 1=50, 56)=460. 39, 3. 271, 2. 946310456, 35. 50770979, LNG 753
1-57.41242993,28.15898339,0.0) LNG 754
IF{T.GT.A(1,I))G0 TO 1 LNG 755
X=(1.-T/A(1,I)) LNG 756

67

195
 SAT=(A(2,I)+A(3,I)*X**(.35)+A(4,I)*X+A(5,I)*X**{4./3.) LNG 757
l+A(6,I)*X**(5./3.)) LNG 758
RETURN LNG 759
1 SAT=l.E+20 LNG 760
RETURN LNG 761
END LNG 762
SUBROUTINE ECKNON(PIN,DOUT,TIN,Q) LNG 763
MAIN: ECKERT-RENON DENSITY MODEL LNG 764
LNG 765
LNG 766
THIS PROGRAM IS DESIGNED FOR THE CALCULATION OF LNG 767
LNG LIQUID DENSITIES ONLY AND WILL NOT LNG 768
COMPUTE DENSITIES -ABOVE 160 K. LNG 769
LNG 770
LNG 771
THIS PROGRAM IS CALIBRATED SOLELY AGAINST DATA LNG 772
tCASURED BY LNG 773
UNITED STATES DEPARTMENT OF COtWERCE LNG 774
NATIONAL BUREAU OF STANDARDS LNG 775
CRYOGENICS LABORATORY LNG 776
BOULDER, COLORADO LNG 777
LNG 778
LNG 779
LNG 780
LNG 781
>>>>>> >>>>>>>>NOTICE<<<< <<<<<<< <<<<<<<<<<<<<<<<< LNG 782
NO WARRANTY IS MADE OR IMPLIED AS TO THE ACCURACY LNG 783
OF DENSITIES CALCULATED BY USE OF THIS MODEL. LNG 784
LNG 785
USERS ARE AT THEIR OWN RISK IN THE USE OF THIS PROGRAM. LNG 786
LNG 787
LNG 788
LNG 789
COMPONENT ID NUMBERS ARE LNG 790
1 METHANE LNG 791
2 ETHANE LNG 792
3 PROPANE LNG 793
4 N-BUTANE LNG 794
5 I-BUTANE LNG 795
6 NITROGEN LNG 796
7 N-PENTANE LNG 797
8 I-PENTANE LNG 798
LNG 799
LNG 800
LNG 801
LNG 802
INPUT VARIABLES ARE LNG 803
PIN=PRESSURE IN BAR LNG 804
TIN=TEMPERATURE IN K LNG 805
0=MATRIX OF MOLE FPRACTIONS OF COMPONENTS LNG 806
OUTPUT VARIABLE IS LNG 807
DOUT=DENSITY IN MOLES/LITER LNG 808
DIMENSION VH0LD(12) ,VPH0LD(12) ,0(8) LNG 809
TYPE INTEGER CNTl ,CNT2 .SWITCH LNG 810
TYPE REAL MWM,INCR,KIJ ,MW,MOL,LAM,LAMB LNG 811
COmon /RUN/ID(12),X(12) ,NAM(2,12),C12(12,12),R(12) ,Z(12),S(12) ,VS LNG 812
1(12) .TS(12) ,US(12) ,SG(12),EP(12),LAM(12),NC,MW(12),AIJ(12,12),PST( LNG 813
212),TCT(12) LNG 814
COMMON /DAT/TBI (12) ,VBI(12) ,TBIM(12) ,VBIM{12) ,C(12) ,SGIM{12) ,SIM(1 LNG 815
12),EPIM(12) ,PIM(12) ,SIN(12) ,VSIM( 12) ,USIM( 12) ,TSIM( 12) ,DVBIM( 12 ) ,P LNG 816
2D(12) ,PV(12) ,RH0{12) ,VBIMP{12) ,TTM(12) ,MWM,CNT1 ,CNT2,T0LD(12) ,INCR LNG 817
3(12) , SWITCH ,JPC,JPCS,JMIX, DENS, VEX, HEX, GEX,TMP,VMP, TP LNG 818
C0W0N/PAR/KIJ( 8, 8),AjI( 8, 8) LNG 819
68

196
 T
T ) ) ) )

COMMON/UNITS/ITC,IPC,T,TMAX,DT,P,PMAX,DP ,PS LNG 820

SWITCH = 0 LNG 821
CALL ZEROl 1NG
THE NEXT VALUES ARE SET TO PREVENT UNDER/OVER FLOWS LNG 823
TMAX = 100.0 LNG 824
DT = 100 0 1 WG OCyJ
PMAX = 100 0 1 NG
1 NG ft?7
DP = 100 0 oco
PS = 100 0
TALL INPlITfO)
rnnF=n
K^KJU I- \J

ITC=0
IPC=0 NG
L_)VU
1 fi13
P=PIN/1 01325
T=TIN LNG 835
IF(T GT (160 GO TO 19 ) ) LNG 836
CALI PZERO(CTMX) 1NG 837
MWM = 0 1NG 838
DO 9 I = 1 NC LNG 839
MWM = MWM + (X(I)*MW(I)) LNG 840
CONTINUE LNG 841
DO 11 I = 1,NC LNG 842
TBK I) = T / TS( n LNG 843
CALL VOLUME(TBI(I) VBI(I) JPUR NC) LNG 844
TBIM( I) = T / TSIM( I) LNG 845
CALL VOLUME( TBIM( I) VBIM(I) JMIX NC) LNG 846
rONTINUE LNG 847
VM = 0.0 LNG 848
no 1? T = 1 NC LNG 849
VM = VM + Xf I ^ *VBIM( n*VSIM( T ) 1 ( LNG 850
CONTINUE LNG 851
TMP = T * 1 8^ - (459 671
( 1MG ft5?
no 13 I = 1 wr LNG 853
vHOi n( T 1 = vRT n 1 NG
VPHOI n( 1) = VR TM( ) 1 MG O J J
1 NG R5fi
VRINiPf \)
vuiiir\i/ VRTf
1 u =
T 1
I i. \ 1
1 NG 857
TAI 1 PRF^( PP VR TMPf T ^ TRT f T 1 MPl O 30
VR I n = VRTMP
( ) ( 1 NG 859
pp = (p * T^TMfTU / ll"sTM(Tl 1 MR
VRTMP( I ) = VRTM( I LNG 861
CALL PRES(PP VRIMPfll TBIM(Il tiC) LNG 862
VBIM( I ) = VBIMP( I LNG 863
CONTINUE LNG 864
VMP = 0.0 LNG 865
DO 14 I = 1 NC LNG 866
VMP
fill = VMP
fill + (X( XI )*VBIMP( I)*VSIM(
vuiiir\i/
' *«jxii\x//
\I) /\ \ 1 LNG 867
PD( I = VBIMP( I
)
* VSIM( I ) LNG 868
CONTINUE LNG 869
DENS=MWM/VMP LNG 870
D0UT=DENS*1000 /MWM . LNG 871
RETURN LNG 872
DOUT=0 0 LNG 873
RETURN LNG 874
END LNG 875
SUBROUTINE ZEROl LNG 876
COMMON /RUN/A(361) LNG 877
COMMON /DAT/B(224) LNG 878
DO 1 I = 1,361 LNG 879
A(I) = 0.0 LNG 880
DO 2 I = 1,224 LNG 881
B(I) = 0.0 LNG 882
69

197
 RETURN LNG 883
END LNG 884
SUBROUTINE INPUT(Q) LNG 885
DIMENSION NNO( 8) ,SIGM( 8),EPSI( 8),LAMB( 8),NAME(2, 8),SN0{ 8) ,M0 LNG 886
1L( 8),CT{ 8),TCH( 8),VCH( 8),ECH( 8),AR( 8),AZ( 8), 0(8) LNG 887
TYPE INTEGER CNTl ,CNT2 .SWITCH LNG 888
TYPE REAL MWM,INCR,KIJ,MW,MOL, LAM, LAMB LNG 889
COMMON /RUN/ID(12),X(12),NAM{2,12),C12(12,12),R(12),Z(12),S(12),VS LNG 890
1(12) ,TS(12) ,US(12),SG(12),EP(12),LAM(12),NC,MW(12).AIJ(12,12),PST( LNG 891
212),TCT(12) LNG 892
COMMON /DAT/TBI(12) ,VBI(12),TBIM(12),VBIM(12),C(12),SGIM{12),SIM{1 LNG 893
12),EPIM(12) ,PIM(12) ,SIN{12) ,VSIM(12) ,USIM(12) ,TSIM(12) ,DVBIM(12) ,P LNG 894
2D(12) ,PV(12) ,RH0(12) ,VBIMP( 12 ,TTM( 12 .MWM.CNTl ,CNT2 ,T0LD( 12 , INCR
) ) )
LNG 895
3(12),SWITCH,JPC,JPCS,JMIX,DENS,VEX,HEX,GEX,TMP,VMP,TP LNG 896
COMMON /PAR/KIJ( 8, 8),AJI( 8, 8) LNG 897
DATA SIGM/ LNG 898
*0. 991000, 1.029000, 1.155000, 1.278000. 1.388752, 1.392995, LNG 899
*1. 47162, 1.47217/ LNG 900
DATA EPSI/ LNG 901
*0. 640000. 0.909000. 1.69800, 2. 237000, 2. 705262, 2. 545907, LNG 902
*3. 76195 .3.68336/ LNG 903
DATA LAMB/ LNG 904
*1. 053604. 0.986325. 1.227117, 1.408519. 1.473346. 1.461676, LNG 905
*1. 698790, 1.695012/ LNG 906
DATA MOL/ LNG 907
*28. 01600, 16. 04200, 30.06800. 44. 09400. 58. 12000. 58. 12000. LNG 908
*72. 146000, 72. 146000/ LNG 909
DATA SNO/ LNG 910
no. 00000. 10. 00000, 10. 00000, 10.00000, 10. 00000. 10.00000. LNG 911
*10. 00000. 10. 00000/ LNG 912
DATA CT/ - LNG 913
126.0600, 190. 5600, 305. 4300,369.8200, 425. 1600, 408. 0300, LNG 914
469.65,460.39/ LNG 915
DATA TCH/ LNG 916
112.7699, 170. 9645, 256. 8311, 294. 7255, 340. 7867. 322. 6109. LNG 917
411.0100,403.3200/ LNG 918
DATA VCH/ LNG 919
26.01846,29.04010,40.88188,54.60301,68.72075.69.44272. LNG 920
84.90197.84.99830/ LNG 921
DATA ECH/ LNG 922
1393.000,1977.000,3695.000.4867.000,5886.652,5539.895, LNG 923
8186.000,8015.000/ LNG 924
DATA AR/ LNG 925
1 . 000000 , 1 .000000 , 1 .000000 , 1 .000000 , 1 .000000 , 1 .000000 , LNG 926
1.000000,1.000000/ LNG 927
DATA AZ/ LNG 928
10.00000,10.00000,10.00000,10.00000,10.00000,10.00000, LNG 929
10.00000,10.00000/ LNG 930
DO 4 1=2,6 LNG 931
4 X(I)=0(I-1) LNG 932
X(l)=0(6) LNG 933
X{7)=Q(7) LNG 934
X(8)=0(8) LNG 935
NC=0 LNG 936
DO 1 1=1,8 LNG 937
IF(X(I).LE.O.O)GO TO 1 LNG 938
NC=NC+1 LNG 939
X(NC)=X(I) ' LNG 940
ID(NC)=I LNG 941
1 CONTINUE LNG 942
DO 2 I = l.NC LNG 943
J = ID(I) LNG 944
R(I) = AR(J) LNG. 945

70

198
 Z(I) = AZ(J) LNG 946
S(I) = SNO(J) LNG 947
TS(I)= TCH(J) LNG 948
US(I)= ECH{J) LNG 949
SG(I) = SIGM(J) LNG 950
TCT(I) = CT(J) LNG 951
EP(I) = EPSI(J) LNG 952
LAM(I) = LAMB(J) LNG 953
VS(I) = VCH(J) LNG 954
MW(I) = MOL(J) LNG 955
2 CONTINUE LNG 956
DO 3 I = l.NC LNG 957
DO 3 J = 1,NC LNG 958
M = ID(I) LNG 959
L = ID{J) LNG 960
C12(I,J) = KIJ(M.L) LNG 961
C12(J,I) = KIJ(L,M) LNG 962
AlJd.J) = WKM.L) LNG 963
AIJ(J,I) = AJKL.M) LNG 964
3 CONTINUE LNG 965
RETURN LNG 966
END LNG 967
END LNG 968
SUBROUTINE PZERO(TRT) LNG 969
TYPE INTEGER CNTl , CNT2 , SWITCH LNG 970
TYPE REAL MWM, INCR ,KI J ,MW ,MOL ,LAM,LAMB LNG 971
COf^ON /RUN/ID(12),X(12),NAM(2,12),C12(12,12),R(12),Z(12),S(12),VS LNG 972
1(12),TS(12),US{12),SG{12) ,EP(12) ,LAM(12) ,NC ,MW( 12 ) ,AIJ 12 ,12 ,PST( ( ) LNG 973
212) ,TCT{12) LNG 974
COKNON /DAT/TBI ( 12 ; ,VBI( 12) ,TBIM( 12) ,VBIM( 12 ) ,C{ 12) ,SGIM( 12) ,SIM(1 LNG 975
12) ,EPIM(12) ,PIM(12) ,SIN(12) ,VSIM(12) ,USIM( 12 ) ,TSIM( 12 ,DVBIM( 12 ) ,P
)
LNG 976
2D(12) ,PV(12),RH0(12),VBIMP(12),TTM(12),MWM,CNT1,CNT2,T0LD(12),INCR LNG 977
3(12) , SWITCH ,JPC,JPCS,JMIX, DENS, VEX, HEX, GEX,TMP,VMP, TP LNG 978
COMMON/PAR/KIJ( 8, 8),AJI( 8, 8) LNG 979
COMMON /UNITS/ITC,IPC,T,TMAX,DT,P,PMAX,DP,PS LNG 980
IF(NC.EO.l) TRT = T/TCT(1) LNG 981
IF(NC.GT.l) CALL TCM(TRT) LNG 982
IF(NC.GT.l) TRT = T/TRT LNG 983
IF(NC.EQ.l) CALL PURE (TRT, TRT, R( 1 ) ,RXX) LNG 984
DO 1 I = 1,NC LNG 985
TCXX = T/TCT{I) LNG 986
IF(NC.GT.l) CALL PURE( TCXX ,TRT,R( I ) ,RTR) LNG 987
JPC = 0 LNG 988
IF{TRT.GT.( .87)) JPC = 2 LNG 989
IF(TRT.GT.(1.0)) JPC = 1 LNG 990
IF(NC.EO.l) RTR = RXX LNG 991
THERE ARE SEVERAL OTHER FORMS OF THESE EQUATIONS LNG 992
REFER TO ORIGINAL ARTICLES FOR VALUES TO USE LNG 993
S(I) = RTR * Z(I) - 2. * RTR + 2. LNG 994
SIM(I) = S(I) LNG 995
LNG 996
SG(I) = (VS(I)/RTR ) **(l./3.) LNG 997
EP(I) = US(I) / S(I) LNG 998
C(I) = LAMd) * (2176. /((185.6)*(1. 98726))) LNG 999
TTM(I) = TS(I) LNGIOOO
TOLD(I) = TS(I) LNGlOOl
INCR(I) = (20.0) LNG1002
1 CONTINUE LNG1003
JPUR = 0 LNG1004
CNTl = 1 LNG1005
CNT2 = 1 LNG1006
JMIX = 0 LNG1007
2 DEN = 0 LNG1008
71

199
 )

DO 3 I = l.NC LNG1009
DEN = DEN + (X(I) SIM(I)) LNGlOlO
CONTINUE LNGlOll
DO 4 I l.NC LNG1012
PIM(I) = (X(I) * SIM(I)) /DEN LNG1013
CONTINUE LNG1014
DO 6 I = l.NC LNG1015
SGIM(I) = 0.0 LNG1016
DO 6 J = l.NC LNG1017
M = I LNG1018
N = J LNG1019
IF(I.GT.J) M = LNG1020
IF(I.GT.J) N = LNG1021
IF(I.EQ.J) GO TO 5 LNG1022
TRUDX = (AIJ(N,M)/(T/TTM(I))) LNG1023
IF(TRUDX.LT. (-180.0)) GO TO 5 LNG1024
SGIM(I) = SGIMd) +{(PIM(J) *(((SG(I)**(l./3.) + SG(J)**(l./3.)) LNG1025
1/(2.0)) **3))* (AIJ(M,N) * EXP(AIJ(N,M)/(T/TTM(I))))) LNG1026
GO TO 6 LNG1027
5 SGIM(I) = SGIM(I) +{PIM(J) *( ( (SG{ I )**(1 ./3. ) + SG(J)**(l./3.)) LNG1028
l/(2.0)))**3 ) LNG1029
6 CONTINUE LNG1030
DEN = 0.0 LNG1031
DO 7 I = l.NC LNG1032
DEN = DEN + (PIMd)* SGIM(I)* SGIMd)) LNG1033
7 CONTINUE LNG1034
DO 8 I = l.NC LNG1035
SINd) =(S(I) * (SG1M(I)**(2.))) / DEN LNG1036
8 CONTINUE LNG1037
LNG1038
DO 9 I = l.NC LNG1039
TEST = (1.0) - (SIMd) / SINd)) LNG1040
TEST = ABS (TEST) LNG1041
IF(TEST.GT.(0.00001)) GO TO 10 LNG1042
9 CONTINUE LNG1043
GO TO 12 LNG1044
10 DO 11 I = l.NC LNG1045
SIMd) =(SIM(I) + SINd)) /(2.0) LNG1046
11 CONTINUE LNG1047
CNTl = CNTl + 1 LNG1048
IF(CNT1.GT.250) GO TO 12 LNG1049
GO TO 2 LNG1050
12 DO 14 I = l.NC LNG1051
SIMd) = SINd) LNG1052
EPIM(I) = 0.0 LNG1053
DO 14 J = l.NC LNG1054
K = I LNG1055
L = J LNG1056
IF(J.LT.I) K = J LNG1057
IF(J.LT.I) L = I LNG1058
IFd.EO.J) GO TO 13 LNG1059
TRUDX = C12(L,K) / (T/TTM(I)) LNG1060
IF(TRUDX.LT. (-180.0)) GO TO 13 LNG1061
EPIMd) = EPIMd) +((PIM(J) * SORT{EP{I)*EP(J)))* LNG1062
1(C12(K.L) * EXP{(C12(L.K) /(T/TTMd ) ) ) ) ) LNG1063
GO TO 14 LNG1064
13 EPIMd) = EPIMd) + (PIM(J) * SQRT(EP{I) * EP(J))) LNG1065
14 CONTINUE LN61066
ASSUME NUMBER OF MOLS OF MIXTURE = 1.0 LNG1067
DO 15 I = l.NC LNG1068
VSIMd) = Rd) * (SGIM(I)**(3.0)) LNG1069
USIMd) = SIMd) * EPIMd) LNG1070
TSIM(I) - USIMd) / ((1.98726) * C(I)) LNG1071
72

200
 - ) ,

15 CONTINUE LNG1072
DO 16 I = 1,NC LNG1073
TEST = (1.0) - (TSIM(I)/nM(I)) LNG1074
TEST = ABS(TEST) LNG1075
IF{TEST.GT. (0.00001)) GO TO 17 LNG1076
16 CONTINUE LNG1077
RETURN LNG1078
17 DO 20 I = 1,NC LNG1079
LNG1080
IF(TSIM(I) - TTM(I))18,20,19 LNG1081
18 TNEW = TTMd)
- INCR(I) LNG1082
IF(TNEW.EQ.TOLD(I)) INCR(I) = INCR(I)/ (2.0) LNG1083
IF(TNEW.EQ.TOLD( I)) GO TO 18 LNG1084
TOLD(I) = TTMd) LNG1085
TTM(I) = TNEW LNG1086
GO TO 20 LNG1087
19 TNEW = TTM(I) + INCR(I) LNG1088
IF(TNEW.E0.TOLD(I)) INCR(I) = INCR(I) / (2.0) LNG1089
IF(TNEW.E0.TOLD( I)) GO TO 19 LNG1090
TOLD(I) = HMd) LNG1091
TTMd) = TNEW LNG1092
20 CONTINUE . LNG1093
CNT2 = CNT2 + 1 LNG1094
IF(CNT2.GT.250) RETURN LNG1095
GO TO 2 LNG1096
END LNG1097
SUBROUTINE PURE(TZ,T,Z,R) LNG1098
DIMENSION A(15) LNG1099
IF(TZ.LT.( .8653525)) R = Z LNGllOO
IF(TZ.LT.( .8653525) GO TO 2 ) LNGllOl
IF(T.LT.( .8653525)) R = Z LNG1102
IF(T.LT.( .8653525)) GO TO 2 LNG1103
IF{T.GT.(1.0)) TA = 1.0 LNG1104
IF(T.LE.(1.0)) TA = T LNG1105
DATA A/. 9184780, -.1530647, 1090050 , .8073883 , 1 .441803 LNG1106
1-10.85944 ,-6 .041687 ,51 .26758 .9062500,-108.9805 ,28 .88672,
, LNG1107
2106.1406,-45.78125,-38.43750,20.56250/ LNG1108
R = 0.0 LNG1109
n= (TA - (.9267292)) / .7267517E-01( LNGlllO
DO 1 K = 1,14 LNGllll
R = R + A(16-K) LNG1112
R = R * TT LNG1113
1 CONTINUE LNG1114
R = R + Ad) LNG1115
2 RETURN LNG1116
END LNG1117
SUBROUTINE VOLUME(T,V ,J ,NC) LNG1118
TR = T LNG1119
IF(TR.GT.(1.00)) GO TO 4 LNG1120
V = 0.5 LNG1121
LNG1122
1 VT = d.O) + (0.1*TR*V**(4.0)) LNG1123
ERR = (1.0) - (V/VT) LNG1124
TEST = ABS{ERR) LNG1125
IF{TEST.LE.(0.00001)) GO TO 2 LNG1126
V = VT LNG1127
GO TO 1 LNG1128
2 V = V**(3.0) LNG1129
3 RETURN LNG1130
4 V = (((({(10.06600 *TR)-24.79837)*TR)+23.260722)*TR)-6. 686880) LNG1131
GO TO 3 LNG1132
END LNG1133
SUBROUTINE BETA(P,DV,T,V,KK) LN61134

73

ZOl
 , ,

TYPE REAL AF,BF,CF,DF,EF,AG,BG,CG,DG,P,DV,T,V,F»F1,G,G1,DEN,EG LNG1135

TYPE REAL C,B1,B2,B3 LNG1136
DIMENSION Cdl) LNG1137
DATA AF/0.11566564E+02/.BF/-0.53510144E+01/,CF/-0.74598207E-01/, LNG1138
lDF/0 .67068653E+00/ ,EF/-0 11939B87E+00/ , AG/0 .62721813E+01/,
. LNG1139
2BG/0 .47698365E+00/ ,CG/-0 .16023080E+01/ ,DG/0 .49837746E+00/ LNG1140
3EG/-0.42639183E-01/ LNG1141
F « AF +(BF*V)+(CF*V*V)+ {DF*V*V*V)+(EF*V*V*V*V) LNG1142
IF(F.LT.(0.0)) GO TO 1 LNG1143
Fl ' (BF+{CF*V*2.E+00)+{DF*V*V*3.E+00)+(EF*V*V*V*4.E+00))* LNG1144
IT * EXP{F) LNG1145
G ' AG+{BG*V)+{CG*V*V)+(DG*V*V*V)+(EG*V*V*V*V) LNG1146
IF(G.LT.{0.00)) GO TO 1 LNG1147
Gl = (BG+(CG*V*2.E+00)+{DG*V*V*3.E+00+(EG*V*V*V*4.E+00))) * EXP(G LNG1148
1) LNG1149
Gl = -Gl LNG1150
DEN = Fl + Gl LNG1151
DV = (l.E+OO) / DEN LNG1152
IF(DV.GT.(0.0).0R.DV.LT.(-1.0E+00)) GO TO 1 LNG1153
GO TO 2 LNG1154
DATA C/- .118659822E+02 - .5509099E+00 .4172102E+01 .- .8996686E+00
,
, LNG1155
l-.1500376E+01,-.1958324E+00,-.2788988E+01,.5195283E+00, LNG1156
2.3734878E+01,.1361904E+00,.4671948E-01/ LNG1157
1 Bl = C(l) + C(2)*T + C(3)*V LNG1158
B2 = C{4)*((V+C(5))**(2 ))*EXP((C(6))*({T+C(7))**(2 ))) LNG1159
B3 = C(ll) * (V+C(8))**(C(9))/(T**(C{10))) LNG1160
DV = Bl + B2 + B3 LNG1161
DV = EXP(DV) LNG1162
DV = - DV LNG1163
2 RETURN LNG1164
END LNG1165
SUBROUTINE PRES(A,B,C,NC) LNG1166
TYPE REAL H,P,V,T,K1,K2,K3.K4,VT,PT LNG1167
P = A LNG1168
V = B LNG1169
T = C LNG1170
H = l.E-02 LNG1171
IF(H.GT.P) H = P LNG1172
ASSIGN 3 TO KK LNG1173
1 CALL BETA{P,K1,T,V,NC) LNG1174
Kl = H * Kl LNG1175
VT = V + ((0.5E+00) *K1) LNG1176
PT = P + ((0.5E+00)*H) LN61177
CALL BETA(PT,K2,T,VT,NC) LNG1178
K2 »= K2 * H LNG1179
VT = V + ((0.5E+00) *K2) LNG1180
CALL BETA(PT,K3,T,VT,NC) LNG1181
K3 = K3 * H LNG1182
PT = P + H LNG1183
VT » V + ( K3) LNG1184
CALL BETA(PT,K4,T,VT,NC) LNG1185
K4 = K4 * H LNG1186
V = V + (((Kl + (2.E+00*K2) + {2.E+00*K3) + K4) / {6.E+00)) ) LNG1187
P «= P - H LNG1188
IF(P.EQ.{0.0)) GO TO 4 LNG1189
IF(P.LT.(0.0)) GO TO 2 LNG1190
GO TO 1 LNG1191
2 GO TO KK , (3,4) LNG1192
3 P « P + H LNG1193
H « P LNG1194
ASSIGN 4 TO KK LNG1195
GO TO 1 LNG1196
4 A = P LNG1197
74

zoz
 B = V LNG1198
C = T LNG1199
RETURN LNG1200
END LNG1201
SUBROUTINE TCM{TMC) LNG1202
TYPE REAL NUM LNG1203
TYPE INTEGER CNTl ,CNT2, SWITCH LNG1204
TYPE REAL MWM,INCR,KIJ,MW,MOL, LAM, LAMB LNG1205
DIMENSION VCI(6) ,TH(12) LNG1206
COWION /RUN/ID(12),X(12),NAM(2,12),C12(12,12),R(12),Z(12),S(12),VS LNG1207
1(12),TS{12) .US(12) ,SG{12),EP(12),LAM(12),NC,MW(12),AIJ(12,12),PST( LNG1208
212),TCT(12) LNG1209
COMMON /DAT/TBI(12) .VBI(12) ,TBIM(12) .VBIM( 12 ) ,C(12) ,SGIM( 12) .SIM(1 LNG1210
12),EPIM(12),PIM(12),SIN(12) ,VSIM(12) ,USIM(12) ,TSIM(12) ,DVBIM( 12) ,P LNG1211
2D(12) ,PV(12) ,RH0(12),VBIMP(12),nM{12),MWM,CNTl,CNT2,T0LD(12),INCR LNG1212
3(12), SWITCH .JPC.JPCS.JMIX, DENS, VEX, HEX.GEX.TMP.VMP, TP LNG1213
C0MM0N/PAR/KIJ( 8, 8),AJI( 8, 8) LNG1214
COMMON/UNITS/ITC,IPC,T,TMAX,DT,P,PMAX,DP,PS LNG1215
DATA VCI/ 1.44,1.59,2.27,3.18,4.03,4.21/ LNG1216
DO 1 I = l.NC LNG1217
J = ID(I) LNG1218
V = VCI(J)**(2./3.) LNG1219
TH( I) = X( I) * V LNG1220
L CONTINUE LNG1221
nZ = 0.0 LNG1222
DO 2 I = l.NC LNG1223
nZ TTZ + THd)
= LNG1224
I CONTINUE LNG1225
DO 3 I = l.NC LNG1226
TH(I) = TH(I) / nZ LNG1227
5 CONTINUE LNG1228
SUMl = 0.0 LNG1229
SUM2 =0.0 LNG1230
K = NC - 1 LNG1231
DO 5 I = 1,K LNG1232
L = I + 1 LNG1233
DO 4 J = L,NC LNG1234
TTZ = (TCT(I)-TCT(J))/{TCT(I)+TCT(J)) LNG1235
TTZ= ABS(TTZ) LNG1236
T12 = (((((({TTZ*(-3.038))+(5.443))*TTZ)+(-1.343))*TTZ) LNG1237
1 + (0.287))*TTZ) - (.0076) LNG1238
T12 = T12*{TCT(I) + TCT(J)) * (0.9) LNG1239
SUM2 = SUM2 + ((2.)*TH(I)*TH(J)*T12) LNG1240
\ CONTINUE LNG1241
SUMl = SUMl + (TH(I)*TCT(I)*(1.8)) LNG1242
5 CONTINUE LNG1243
TMC = SUMl + SUM2 + (TH(NC)*TCT(NC)*{1.8)) LNG1244
V = 0.0 LNG1245
DO 6 I = 1,NC LNG1246
LNG1247
J = IDd) LNG1248
TH(I) = X(I) * VCI(J) LNG1249
V = V + TH(I) LNG1250
5 CONTINUE LNG1251
DO 7 I = 1,NC LNG1252
TH(I) = TH(I) / V LNG1253
7 CONTINUE LNG1254
TMM = 0.0 LNG1255
DO 9 I = l.NC LNG1256
DO 8 J = l.NC LNG1257
M = ID(I) LNG1258
N = ID(J) LNG1259
NUM = (VCI{M)**(l./3.)) * (VCI(N)**(l./3.)) LNG1260
75

203
 NUM = SORT(NUM) LNG1261
DEN = {0.5)*((VCI(M)**(l./3.))+(VCI(N)**{l./3.))) LNG1262
NUM = NUM / DEN LNG1263
NUM = NUM**(3.) LNG1264
AKIJ = (1.0) - NUM LNG1265
TCIJ = (1.0 - AKIJ) * S0RT(TCT(I)*TCT{J)*1.8*1.8) LNG1266
TMM = Ttfl + TH(I)*TH(J)*TCIJ LNG1267
8 CONTINUE LNG1268
9 CONTINUE LNG1269
TCMP = TMM + (10.0) LNG1270
ICT = 0 LNG1271
TPO TMM + (10.0)
= LNG1272
10 TR = (T*1.8)/TCMP LNG1273
NUM = (2901.01) -{(5738.92)*TR) +( (2849.85)*TR*TR) LNG1274
1 + ((1.74127)/ (1.01 - TR)) LNG1275
NUM = NUM * (TR - (1.0)) LNG1276
IF(NUM.LT.(-180.)) DD = 0 LNG1277
IF(NUM.LT.(-180.)) GO TO 11 LNG1278
DD = EXP{NUM) LNG1279
THE STATEMENT ABOVE MAY RESULT IN AN UNDERFLOW SENSE LIGHT LNG1280
ON SOME OPERATING SYSTEMS WHEN THE NUMBER NUM IS A LARGE LNG1281
NEGATIVE NUMBER. THE LARGE NEGATIVE VALUE IS PROPER AND LNG1282
THE CORRECT ANSWER FOR DD IS ZERO. LNG1283
11 TCMP = TMM + ((TMC-TMM)*DD) LNG1284
TEST = (1.0) - (TCMP/TPO) LNG1285
TEST = ABS(TEST) LNG1286
IF(TEST.LT.( .0001)) GO TO 12 LNG1287
ICT = ICT + 1 LNG1288
LNG1289
IF(ICT.GT.250) GO TO 12 LNG1290
TPO = TCMP LNG1291
GO TO 10 LNG1292
12 TMC = TCMP/(1.8) LNG1293
RETURN LNG1294
END LNG1295
SUBROUTINE BLOCK LNG1296
TYPE REAL KIJ.AJI LNG1297
DIMENSION KIJ(8,8 ) ,AJI(8,81 LNG1298
COMMON /PAR/ KIJ, AJI LNG1299
DATA KIJ/ LNG1300
*0. 000000,. 293E-07 ,.911E-06.,.0030498 .0.001000 ,0.001000 ,0.001000, LNG1301
*0. 001000, 1.088608 ,0.000000,,.188E-10 ,.463E-06 ,.462E-08 ..120E-03, LNG1302
*0. 001000, 0.001000 ,1.098880 ,1.078910 .0.000000 ,.517E-06 ,.535E-06, LNG1303
*.188E-10, 0.001000 ,0.001000 ,.8578232 ,1.146026 ,1.009559 ,0.000000, LNG1304
*.546E-06,.136E-05 ,0.001000 ,0.001000 ,0.995000 ,1.287762 ..9868486, LNG1305
*. 9690485, 0.000000 ,.541E-06 ,0.001000 ,0.001000 ,1.178993 ,1.266327, LNG1306
*1. 016231,. 9976706 ,.9435714 ,0.000000 ,0.001000 ,0.001000 ,0.950000. LNG1307
1.000000,1.000000 ,1.000000 ,1.000000 ,1.000000 ,0.000000 ,0.001000, LNG1308
*0. 995000. 1.000000 ,1.000000 ,1.000000 .1.000000 ,1.000000 ,1.000000, LNG1309
0.000000/ LNG1310
DATA AJI/ LNG1311
*0. 000000,. 161E-04 , .257E-06 ,.0003462 .0.001000 ,0.001000 ,0.001000, LNG1312
0.001000, 1.014967 ,0.000000 ,.188E-10 ,.197E-06 ,.184E-05 ,.335E-06, LNG1313
0.001000,0.001000 ,1.008061 ,1.004105 ,0.000000 ,.142E-06 ,.103E-04, LNG1314
.188E-10, 0.001000 ,0.001000 ,.9863463 ,1.011399 ,1.001862 ,0.000000, LNG1315
.105E-04,.119E-05 ,0.001000 ,0.001000 ,0.995000 ,1.024894 ,1.003289, LNG1316
.9987221,0.000000 ,.119E-05 ,0.001000 ,0.001000 ..9566020 ,1.025269, LNG1317
1.005949,1.001225 ,.9968984 ,0.000000 ,0.001000 ,0.001000 ,0.995000, LNG1318
1.000000,1.000000 .1.000000 .1.000000 ,1.000000 ,0.000000 ,0.001000, LNG1319
0.995000,1.000000 ,1.000000 .1.000000 ,1.000000 ,1.000000 ,1.000000, LNG1320
0.000000/ LNG1321
END LNG1322

76

204
6.12 Reference [35]

A-246
J. Chem. Thermodynamics 1982, 14, 837-854

Mathematical models for the

prediction of liquefied-natural-gas
densities^

R. D. Mccarty
Thermophysical Properties Division, National Engineering Laboratory,
National Bureau of Standards, Boulder. Colorado 80303. U.S.A.

(Received 27 October 1980; in revised form 25 January 1982)

Three mathematical models of the equation of state for liquid mixtures simulating liquefied
natural gas (LNG) are discussed and compared. The adjustable parameters for each model have
been optimized using the same set of experimental data, consisting of over 280 new (p, V. T. x)
points taken at the National Bureau of Standards in Boulder. Colorado. It is estimated that each
of the models will predict LNG densities over its range of validity to within O.I per cent of the
true values, given the pressure, temperature, and compyosition of the mixture. Deviation plots
and a detailed performance evaluation are given for each model. The range of validity varies
slightly among the models but in general the range of the study included the saturated liquid
from 90 to 135 K.

1. Introduction

A project was begun at the National Bureau of Standards in Boulder, Colorado in the

summer of 1972, with an ultimate goal of obtaining one or more models of the
equation of state of liquefied-natural-gas-like (LNG-like) mixtures which could be
used to predict the density of any LNG mixture with a total uncertainty of less than
0.1 per cent given the pressure, temperature, and composition of the liquid. LNG-like
mixtures for the purposes of this paper are defined as mixtures of N2, CH4, CjHft,
CjHg, «-C4Hio, and 1-C4H10 with mole fractions of N2, A1-C4H10, and i-C4H,o
restricted to 0.05 or less. The project was sponsored by a consortium of 18
international energy companies through a grant administered by the American Gas
Association, with the hope of providing a means to an equitable custody transfer of
large quantities of LNG. A series of papers reporting the results of experimental
measurements performed during this project have been previously published in this
Journal and the present paper is a part of that series.*'"*' The purpose of this paper is
to report an analysis of three models which are believed to have achieved the 0.1 per
cent total-uncertainty goal.
No equation of state can be more accurate than experimental data. Therefore,
experimental (p, V, T, x) values of comparable or greater accuracy and of adequate
scope were a necessary condition to the success of the project. At the time when this

° This work is a contribution of the National Bureau of Standards, not subject to copyright.

205
838 R. D. Mccarty

project was initiated a survey of the existing values revealed that such a set did not
exist,and a major experimental program was initiated. Over the next six years
comprehensive (p, K T, x) results for the saturated liquid state of the major
components (Nj, CH4, CjH^, CjHg, j-C4Hio, M-C4H10) of LNG as well as their
binary and multicomponent mixtures were obtained. These results are reported
elsewhere,""^' and will not be repeated here. However, a synopsis of the experimental
values is given in a later section.

2. Selection of models

A preliminary investigation of the existing equations of state of a mixture, revealed a

multitude of possibilities from which to choose. Rather than relying on a single model
it was decided to select several which would represent the various categories of
models found in the literature, thus enhancing the possibility of achieving the 0.1 per
cent accuracy goal of the project.
The choices were made on the basis of past performance for other fluids, and
likelihood of success. An efibrt was also made to choose one model which would be
representative of a general category of approaches which could be identified as being
basically different from each other. On these bases three models were selected : the
extended corresponding-states model*^' which, while it is complex, allows a great deal
of flexibility; a hard-sphere model*^* which has demonstrated capabilities; and the
Klosek and McKinley*'°' model which is very simple in principle and totally
empirical. From time to time as the project continued other methods and models
were suggested by co-workers and the sponsoring companies but only two of these
were investigated in any detail: (a) Watson's*' method, investigated to the point
where became obvious that it was unsuitable without some major modifications;
it

new empirical method, found to work well for restricted ranges of

(b) Hiza's*'^'
composition and temperature.

3. Data used
The experimental densities for saturated liquid of the pure components of LNG given
in references 1 and 7 have been used here both as data for optimization procedures
and as pure-fluid data for calculating the molar volumes of the components (see
equations 19 and 21). Table presents these data in the form of coefficients to an
1

equation for the saturated liquid density and table 2 summarizes the data used from
references 1, 2, 3, and 7.
References 5 and 6 contain 110 (p, V, T, x) points for a total of 27 different
multicomponent mixtures which, with one exception, were used for comparison
purposes only. The data for the ternary system: (0.05931N2 O.8907ICH4 + -I-

0.04998«-C4Hjo) from reference 6 were used to obtain the N2-n-C4H,o binary

interaction parameters, which are also used for the N2-/-C4H10 binary interaction
parameters given in sections 4 and 5. No other (p, K T, x) values have been used in
the development of the mathematical models in the following sections.
The use of values from a single source was the intent from the beginning of the

206
 MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES 839

TABLE 1. Saturated-liquid densities of the pure fluids

3

(p-p,)/(moldm-') = a(l-r/r,)°" -(-

^ b,{l-T/Ty

where and are the critical temfjerature and amount-of-substance density

CH4 C^H, C3H8 i-C4H,o

a 18.65812 12.55205 8.684459 7.657535

^ 6.712030 13.43284 18.04086 8.145251
-G.9472020 - 19.00461 -29.46261 - 13.10582
0 11.07716 16.43559 8.145894
TJK 190.555 305.33 369.82 408.13
pJ{mo\dm'^) 10.16 6.86 5.00 3.80
M/lgmoi" ')° 16.04303 30.07012 44.09721 58.1243

"-QHio '-C5H,/ n-CiHii'

a 7.286063 19.39217 2.946310456 -0.0362004993

by 1 1 ,96308 26.01408 35.50770979 59.0020299
b2 - 19.87592 -39.49759 -57.41242943 -93.44193819
b, 11.60211 23.32977 28.15898339 43.66780833
TJK 425.16 126.20 460.39 469.6
^^/(mol dm"-') 3.92 11.21 3.271 3.285
M/(gmol"')'' 58.1243 28.0134 72.15139 72.15139

" TTie project under which the work reported here was done started in The molar masses M were
1972.
calculated using the values for C, H. and N accepted at that time. The values for H and C were changed
slightly by the 1976 Commission on Atomic Weights. For the sake of consistency we have continued to use
the previous values. The maximum error in density of any of the fluids resulting from using the newer
values is 0.003 per cent.
^ Towards the end of the project, f-pentane and ^-pentane were added to the list of possible components
in the LNG-like mixtures. No experimental (p. V. T) values for these two pure fluids were taken as part
of this project; the constants in table 1 for the saturated amount-of-substance densities of these fluids were
derived from the results of McClune.'^*'

project, the purpose being to eliminate systematic differences which almost always
occur when values from more than one source are used. The development of the
mathematical models was done simultaneously with the acquisition of the
experimental values; this procedure allowed almost immediate consistency checks
between the experimental results for various different compositions of a binary
'^~*^'
mixture as well as for comparisons with data from other sources.*'°'^^ These
comparisons verified, in many cases, the anticipated systematic differences in

multisource data.

4. Extended corresponding states

A development of the thermodynamic equations for the extended

theoretical
corresponding states method may be found in a paper by Rowlinson and Watson''^'
and only those equations necessary for this particular application will be given here.
Other important work on this method has been done by Leach'^' in developing
models for the shape factors and by MoUerup"^' and Mollerup and Rowlinson'^''

207
 0

840 R. D. McCARTY
TABLE 2. Experimental measurements used, their temperature ranges T, to Tj, and the ranges of the mole
fraction x of the most volatile component in binary mixtures

Fluid or bmary mixture Reference /,/K X No. of points

/ 120 pure fluid 19

L LI
H4 1
1 IUj 160 pure fluid 11
1 lUU 270 pure fluid 22
1Art
CjHs 1 100 288 pure fluid 16
1 300 pure fluid 12
LI
I-L4M10
;
2 113 300 pure fluid 12
28 148 249 pure fluid 21
'-C3H,, 28 125 254 pure fluid 20
.XN2 + (1 -X)CH4 2 105 140 0.05 to 0.5 21
2 105 120 0.06 4
xN, + (l-x)C3Hg 2 105 115 0.02 to 0.07 6
xCH4 + (l-x)C,H, 2 105 140 0.35 to 0.68 20
xCH4 + (l -xjCjHg 2 105 130 0.3 to 0.85 20
xCH4 + (l-x)i-C4H,o 2, 3 110 140 0.48 to 0.92 17
xCH4 + (l -x)n-C4H,o 2. 3" 115 140 0.58 to 0.93 23
xC,H, + (l-x)C,H8 2 105 140 0.50 to 0.67 10
xC2H, + (l-x)/-C4H,o 2 105 130 0.69 to 0.72 6
xC,H^ + (l-x)^-C4H,o 2 110 140 0.65 to 0.67 8
xCjH« + (l-x)i-C4H,o 2 105 130 0.49 to 0.50 8
xCjH, + (l-x)n-C4H,o 2 110 150 0.58 to 0.61 8
xi-C4H,o + (l -x)n-C4H,o 2 125 140 0.47 4

" The results for (0.91674CH4 + 0.08326n-C4H,o) from reference 3 were not used because they appear to
be inconsistent with later results'*' for nearly the same composition.

who combined the work of Leach'**' with an equation of state for methane by
Goodwin*^*" to produce a model for LNG-hke mixtures. The same general approach
as used by Mollerup'^^' is used here except that the shape-factor functions of Leach'^*
have been refitted to the pure obtained at this laboratory
fluid values experimentally
on this project**'^' and an equation of
by McCarty'^'' has been used for
state
methane. Interim results of this study can be found in references 21 and 22. The
differences between the interim results mentioned above and the final results, given
here, are minor and the previously published equations may be used. More details on
the performance of the extended-corresponding-states method are given later in this
section.
The extended-corresponding-states method is defined by the equations:

z,(p,r) = Zo(p/i.,o/A.o.7-/y;,oK d)
G,(p, T) =/i,,oGo(p^,.o/yi-,.o^ T/yi,,o)-«^ln(Vo). (2)

where Z is the compression factor, G is the Gibbs free energy, p is pressure, and T is
thermodynamic temperature. The subscript 0 denotes a reference fluid (in this case
methane) and the subscript denotes the fluid for which properties are to be obtained
^

via the equation of state for the reference fluid and the transformation variables

and /i„ 0- The double subscript is introduced now to facilitate the extension later to
mixtures. Two equations, (1) and (2), are necessary to define uniquely the

2Ub
 MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES 841

transformation variables o and ''/..o- The equation of state for methane by

McCarty*^^' was constructed using a hypothetical saturation line for both liquid and
vapor down to 43 K. This construction is necessary to represent realistically the very
low reduced temperatures of the heavier hydrocarbons and was accomplished
without changing the performance of the equation of state above the triple-point
temperature of methane. The equation of state for the reference fluid is

p = pRT + pHNiT + N2T"^ + N,+NJT + N^/T^) + p\N(,T + N,+Ns/T + Ng/T^)

+ ANioT + N,,+N,,/T) + pHN,,) + p^(N^JT + N,JT') + p'(N,JT)
+ p'(N^,/T + N,s/T') + p^(N,^/T') + pHN,o/T' + N,JT')cxp{-yp')
+ pHN22/T'+N,,/T^)exp(-yp') + p'{N,JT' + N,s/T')exp(-yp')
+ p^{N2JT' + N2,/r')exp(-yp') + p''{N2s/T'+N2<,/T')exp(-yp')
+ p''{N,o/T'+N,JT'+N,2/T*)exp{-yp'l (3)
where p is pressure, T is thermodynamic temperature, and p is amount-of-substance
density. The coefficients N, are given in table 3.

TABLE 3. Coefficients N-, of equation (4) for CH4''

R 0.831434X 10"' MPa dm' mol"' K"', 0.0096 dm^ mol"

N, -0.189506118654 x 10"' dm^ mor^ K"', 0.104756987190 X 10 dm" mor' K-'",
-0.157446511658 X 10' dm*mor^ N4 0.782544605655 x 10' dm" mor' K,
-0.382099919632 X 10* dm^ mor' K\ 0.858039193151 x 10"* dm' mor^ K-',
N7 -0.502993394999 X 10"' dm' mol"'. 0.881435651333 X 10' dm" mol-^ K.
-0.327098978594 X 10* dm'mor^K^
N.o -0.401087946417 X 10 'dm" mol-* K-
0.270307051249 x 10"' dm'^ mor*. N.2 -0.308038191105 X 10 dm" mol * K,
N.3 0.194123002261 x 10"* dm'5 mor'. A/l4 -0.198179473576 X 10"* dm"* moI-" K,
Nl5 0.615529077453 x 10 dm'« mor* K\ N.6 -0.536626852203 X 10"* dm" mor'-K,
N.7 0.154281787794 X 10"' dm'* mor« K, N.8 -0.111409049265 X 10"' dm'* mol-« K'.
0.193931540966 X 10"* dm'" mor' K\ N20 0.391590731296 X 10* dm« moj-' K',
N21 -0.160023162832 X 10" dm' mor^ K\ N22 0.197857473814 X 10' dm" mol-' K',
^23 0.168195529711 x 10" dm'^ mor' K*. ^24 0.611041574405 x 10"' dm" mor' K',
0.381473591215 x 10' dm" mo!"' K^ N26 0.127257796890 X 10"^ dm'" moI-' K',
N27 -0.348122335444 X 10' drn'^ mol"' K*, N28 -0.548112616636 X 10"" dm" mol"" K'
0.188081779935 x 10"' dm" mor" K\ ^30 0.780999906720 x 10-' dm" mol-'^ K'.
N3, -0.290669323872 X 10"" dm" mor" K\ N32 0.377310956389 X 10"* dm^ mor'^ K*.

° An existing wide-range equation of state for CH4 was used for the base fluid to facilitate the extension
of this work to a wider range of pressures and temperatures in the future. The large number of digits in the
coefficients given here is necessary to insure the validity of the calculated pressure in some areas of the
(p, T) surface.

The fa 0 and /j,i,o are then defined as

^..o = (^^/.o/l^o')0...o(77,K^K (5)

where the superscript " denotes the critical property and 7^' = 7/7^- and Vi = F/l^--

The 0„(7;\ V,') and 0,..o(T7, V^) are called shape factors, where

209
842 R D. McCARTY
and

(/),,o
= (Zyzn[l+K-H'o);6,(l/'-b,)-M^1'-/'4)ln77l], (7)

where the Tl and V- are as before and w, is a fluid-dependent parameter.

The d/s and b-s are fluid-independent parameters and Zq/Z- is the ratio of the
compression factors at the critical point. The a/s. i»/s, and Wi's were estimated by
least squares using the experimental data from (1) and (2) and simultaneous fitting
techniques. The parameters a^. bj, h^. and the critical properties are given in table 4.

TABLE 4. Coefficients for corresponding-states equations (6) and (7)

t;, = -0.109495. = 0.919454. = -4.01525. = -4.14192, = 0.444850

6, = 0..^56808. b. = 1.02619. = 0.893323. b^ = 0.761533

7; M
w ' Fluid no.
MPa K cm^ mol"' g mol
'

0.029179 3.3542557 126.2 89.827 28.0134 6

CH4 0.0109 4.59570 190.555 98.522 16.04303 1

0.110427 4.860314 .^05.5 146.2 .30.07012 2

C,H« 0.154837 4.2445123 370 200 44.09721 3
0.176372 3.8295398 425 251.62 58.1243 4
'-QHio 0.150115 3.688998 408.1 263 58.1243 5
n-C.H,, 0.234320 3.3812152 469.6 304 72.15139 7
'-C5H,, 0.288886 3.1988302 460.39 306 72.15139 8

" The large number of significant figures given for critical pressure is necessary to reproduce the
(Zj = p^V^lRT^) in the least-squares tit.

Equations (6) and (7) were originally proposed by Leach'*'*' with h', as the acentric
factor,which allows the extension of the equations to any fluid with a minimum of
information about that fluid. Here, u', was treated as an adjustable parameter to
provide for the greatest possible accuracy in the limited range of pressure and
temperature of interest.
Yhe extension of the above to mixtures is accomplished through the equations:

/'U.0
= '/u(i''/i'o + i/'];:o)'. (9)

''x.o = Z,I,^.-^j''u.O' (10)

L.o^h.o = Z, Z, •^i-^j/v.o''.j.O' (11)

where .x, and are the mole fractions of the pure components.
The e,j and the r}^^ are binary interaction parameters determined by least squares
from (p, V, T, x) values for binary mixtures as outlined in section 3. The 0 '^^'^ ''x.o

are now the transformation variables to transform the equation of state of the
reference fluid to an equation of state for the mixture specified by p, 7, and x. The ^,/s
and £,,'s for the corresponding-states method are given in table 5.

Of the three models presented here, the extended-corresponding-states model is

superior to the other two in most respects. It covers the widest range of composition.

210
 MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES 843

TABLE 5. Binary interaction coefficients for corresponding states equations (8) and (9). Fluid numbers /

are given in table 4

i.j \ 2 3 4 5 6 7 8

1 1 1.00514 1.01922 1.04243 1.05048 1.01009 1.06 1.06

2 1 1.00599 1.01616 1.02369 1.02127 1.02 1.02
3 1 1.00172 1.01140 1.04606 1.01 1,01
4 1 0.997114 1.20955 1,0 1.0
'
5 1 1,13889 1,0 1.0
6 1 1.0 1.0
7 1 1.0
8 1

i.j \ 1 3 4 5 6 7 8

1 1 1.01127 0.988608 0.983130 0.986978 0.953430 0.98 0,98

2 1 0.999961 0.972223 0.998886 0.939622 0.99 0.99
3 1 0.985547 1.03099 0.912209 0.99 0.99
4 1 0.976416 0.849200 0,99 0,99
5 1 0,857310 0.99 0.99
6 1 0,99 0.99
7 1 0.99
8 I

the widest range of pressure and temperature, and is probably the most accurate,
although the last is hard to substantiate. Of the approximately 285 experimental
[p. V. T. x) points reported in references 2 to 6, only 14 differ by more than 0.1 per
cent in density from the value predicted by this extended-corresponding-states model.
Of these 14 points 11 (see below) are judged to have an experimental uncertainty
greater than the 0.1 per cent. Figure 1 presents the deviations between calculated and
exf>erimental densities for these 14 points. Deviations between predicted and
experimental densities of the other 271 (p, K 7, .x) points are published in

reference 23. No restrictions as to pressure, temperature, or composition (within the

data from references 2 to 6) need be placed on the model. The other two models have
pressure-, temperature-, and composition-range restrictions. In preliminary accounts
(references 21 and 22), some reservations were expressed about the accuracy of the
calculation methods because of the disagreement between experimental and
calculated densities for a few binary and multicomponent system which contained
methane and butanes as components. This disagreement has since been resolved by
more experimental measurements which have been made'*-^' on mixtures of similar
composition and in some cases the same composition (more measurements on the
same sample). As a result of the additional measurements'*-^' some of the
unpublished multicomponent mixture results for systems containing methane and
butanes were discarded,'^' and some of the already published binary systems
containing (methane -I- n-butane) were judged to be uncertain by more than the 0.1
per cent in density. The published results'^ *' which are believed to be uncertain by
more than 0.1 per cent are detailed in table 2. As a result of the additional values

211
844 R. D. Mccarty

+ +
+
8 +
9 +
V

8 A A
-0.2
100 110 120 130 140
r/K
FIGURE 1, Relative deviations exceeding 0.1 per cent between densities calculated by the extended-
all of the (p, K 7, x) mixture values taken in this project.'^'*' +,
corresponding-states method and
0.91674CH4 0.8326w-C4H,o:
-I- ,
0.06740N2 + 0.93260C,He; A. 0.0O859N2 0.75713CH4 + -)-

O.lBSSSC^He 0.06742CjH8 + 0.01326n-C4H,o + 0.01336i-C4H,o + 0.00216f7-C5H, ^

-I- 0.00223/- -I-

C'sH.j; O. 0.85442CH4 + 0.05042C2H6 O.O4038C3H8 -I- 0.02901n-C4H,o -I- 0.02577i-C4H,o V. ;

O.7492OCH4 O.25O8OC3H8.
-I-

obtained in program, subsequent to the models published in

the experimental
references 21 and 22, very slight changes were made in the binary interaction
parameters: CH4-/-C4H,o, CH^-n-C^HiQ, N2-1-C4H10. and N^-n-C^Hio- These
changes have a negligible effect on density calculations for LNG-like mixtures, i.e.
where the mole fractions of N2, i-C4H,o. and n-C4H,o are individually less than 0.05.
Therefore, either the models presented here or those in references 21 and 22 may be
used to predict the density of an LNG mixture to within 0.1 per cent of its true value.
The accuracy claim of 0.1 per cent in density must be qualified by specifying ranges
of pressure and temperature as well as components. For each of the three models the
qualifications will be different; in the case of the extended-corresponding-states
model the temperature range is 95 to 135 K for saturated liquid pressures. The
component restrictions are minimal, i.e. any combination of mole fractions of Nj,
CH4, CjH^, CjHg, «-C4H,o, and i-C4Hio, as long as liquid-liquid phase separation
does not occur. The total mole fraction of «-C5H,2 and /-C5H12 must not be more
than 0.02. The above restrictions are based on the performance of the model when
compared to the set of results from this laboratory Comparisons with other
sources have been made.'^^*'^^^^' With a few exceptions the comparisons have
revealed our values to be internally consistent to better than the 0.1 per cent
accuracy claim given here. Rodosevich's values**^' are agreement with ours to in
within 0.1 per cent in density but do not extend the range of pressure and temperature
upon which the extended-corresponding-states model is based. A comparison of
calculated and experimental densities from Rodosevich*'^' and Rodosevich and
Miller'^' is given in figure 2. Orrit and Laupretre'^"*' have published (p. V. T. x) values
for LNG-like mixtures which agree reasonably well with this model as shown in
figure 3, but again containno points which would expand the scope of the set used
here.
Nunes et al.^^^^ have measured (p, V, T, x) for (methane -I- nitrogen) at low
temperatures in the compressed liquid to several tens of megapascals. These results

212
 MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES 845

0.08

^ 0.04 -

-0.04 -

T/K
FIGURE 2. Typical relative deviations between densities calculated by the extended-correspwnding-
states method and experimental densities from Rodosevich"*' and Rodosevich and Miller."'
O. O.92O6CH4 + 0.0794«-C4H,o; A. 0.9262CH4 + 0.0738CjHg; 0.9737CH4 + 0.0263C,Hs; V. .
O.92O6CH4 + 0.0794n-C4H,o; .
0.9462CH4 + 0.0538i-C4H,o; 0.8409CH4 + O.IO86C2H6 +
0.0505N,; A. 0.8303CH4 + O.lOOlCjHe + O.O493C3H8 0.0203n-C4H,o. -(-

0.2 1 1 1 1 1

On ^ CQ On

0 0 OA A ^ A
A ^ 0

8 0

-0.2 1 1 1 1 I

100 110 120

T/K.

FIGURE 3. Typical relative deviations between densities calculated by the extended-corresponding-

statesmethod and experimental densities from Orrit and Laupretre.*^*' V. O.OOQNj O.95CH4 -1- -I-

O.026C2H6 + O.OlSCjHg; ,
O.OO9N2 + O.9OCH4 + O.053C2H6 + O.O28C3H8 0.012n-C4H,o; A,
-t-

O.OO9N2 + O.93CH4 + O.O26C2H6 O.OI5C3H8 -(- -I- 0.012n-C4H,o + 0.008i-C4H,o; O. 0.006Nj +

O.89CH4 O.OSeC^H^, + O.O24C3H8 + 0.012n-C4H,o
-I- 0.012i-C4H,o
-I- O.OOSi-CsH.j; O. O.OIN^
-I-
-1-

O.7OCH4 + O.I6C2H6 + O.O9C3H8 + 0.029«-C4H,o + 0.014i-C4H,o -H O.OOOSn-CsH,, + O.OOlIi-

CjHij

agree very well with the extended-corresponding-states model with the exception of
one isotherm (see figure 4) which again suggests some internal systematic differences.
Finally Straty and Diller'"' have measured several hundred (p, F, 7, x) points for
(methane -i- nitrogen) over a wide range of p, V, and T including compressed liquid
and gas. Figure 5 is a summary of the agreement of the corresponding-states model
discussed here with these results. Figure 5 demonstrates that the model is well
behaved and reasonable when extrapolated beyond the range of the present data set.
but that the extrapolation is limited by the range of p and T of the equation of state

213
846 R. D. Mccarty

0.4

S 0.2 T= 110.09 K

r= 116.3 K

:^ 0

-0.2
200 300
p/MPa
FIGURE 4. Typical relative deviations between densities calculated by the extended-corresponding-
states method and experimental densities from Nunes et a/.'^*' for (0.497CH4 + 0.503N2).

-0.06 '

.(b) (c) .

-0.08 . i>~25.9 moi'dm

/7~12.6 mol-dm
-0.10

-0.12

t
-0.14
170 180 190 200 210 220 240 260 280 300 110 120 130
T/K T/K. T/K
FIGURE 5. Typical relative deviations between densities calculated by the extended-corresponding-
states method and experimental densities from Straty and Diller'^"" for (0.501 15Nj -i- 0.49884CH4).

for the base fluid, 90 to 350 K with pressures to 35 MPa. No comparisons with or
i.e.

thermodynamic properties of mixtures have been made with any of

calculations of the
the models presented here and calculations of this sort are not recommended because
of the extreme emphasis placed on (p, K T, x) and the complete absence of
consideration of other thermodynamic properties during the development of the
models. Table 6 contains amount-of-substance densities for purposes of checking the
method.

TABLE 6. Values for checking calculations using corresponding-states equations. These are included for
check purposes only; they are calculated from the corresponding-states model and are not experimental
values

0.6975CH^ -l-0.156CjHe O.O92C3H8 0.029n-C4H,o -h 0.014(-C4H,o -(-O.OllSNj

T/K /)/(mol dm ') p/MPa

95 24.333 0,1013250
100 24.067 0.1114575
105 23.796 0.1215900

214
 :

MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES 847

5. Hard-sphere model

The hard-sphere model presented here is a refit of the model proposed by Rodosevich
and Miller'^' one of many modifications of the Longuet-Higgins and Widom
which is

model,'^^* and was chosen to be included in this study as a representative example of

the application of the hard-sphere equation-of-state concept to the correlation of
(p, V,T,x) data.
The equation of state from reference 9 is

pV/RT=c(l+ y + y^)/{l-y)^-a/RTV, (12)

where a, b, and c are adjustable parameters and y = fe/4K The equation is applied to
mixtures by assuming the one-fluid theory and applying the mixing rules:

flm = (13)

<^n, = I,I/oXiX^- (15)

The combining rules are

b, j=m!' + b]i')(\-j,^,\ (16)

a„=(fl„fl,,)''^(65/^.^>)"'(l-^.,K (17)

c. ,= (c, + c^,)/2. (18)

The parameters j.^ and A:,j are now the binary interaction parameters. The as, b\. c's.

y,/s, and kijS are given in tables 7 and 8. The excess volume is calculated using the
equation of state and

y^ = K-lM (19)

where and the V, are calculated via the equations (12) through (18) and then

K„ = I,^^x, + F^ (20)

where the is from equation (19) and the are experimental. The in this case

were calculated from the equations for the liquid density of the pure fluids given in
table 1.

The above equations are the same as they originally appeared in Rodosevich and
Miller'^* and only the ji-s and /c./s have been revised on the basis of the (p, K T, x)

TABLE 7. Coefficients for the hard-sphere model of equation (12): h = 2nLSf/3 where L denotes
Avogadro's constant and the Si's are the pure-fluid coordination numbers

Fluid fly(dm*mor^•MPa) Sy(dmmor c, Fluid no. 1

N, 1.718X 10' 3.546 X 10-' 1.03 6

CH, 2.755 X 10' 3.676 X 10"' 1 1

CjH, 7.773 X 10' 4.158 X 10"' 1.5 2

CjHb 14.165 X 10' 4.644 X 10"* 1.67 3
n-C^H.o 22.733 X 10' 5.051 X 10"' 1.83 4
<-C4H,o 21.279 X 10' 5.056 X 10"* 1.79 5

^15
 .

848 R D. McCARTY

TABLE 8. Binary interaction parameters for the hard-sphere model of equations (16) and (17). See table 4
for identification of and ) i

hi
ij 1 2 3 4 5 6 7 8

I 0 -0.00388616 -0.0120932 -0.0231577 -0.0238349 - 0.00997547 -0.0326 0.0458

2 0 -0.002162 -0.00400910 -0.00812712 -0.0143976 -0.003 -0.004
3 0 0.000761571 -0.00383743 -0.024014 0 0
4 0 0.00222 ISO -0.0576043 0 0
5 0 -0.0576043 0 0
6 0 -0.04 -0.05
7 0 0
8 0

i-j 1 2 3 4 5 6 7 8

I 0 0.00298830 0.0597378 0.110893 0.100298 0.0197290 0.14 0.1745

2 0 0.014527 0.067703 0.0346632 0.0529034 0.02 0.03
3 0 0.024291 -0.00838212 0.14719 0 0
4 0 0.0199213 0.154365 0 0
5 0 0.154365 0 0
6 0 0.15 0.18
7 0
8

binary-mixture set as outlined in section 3. An attempt was made to optimize the a, b,

and cparameters for each pure fluid on the basis of the new pure-fluid values;
however, the results were disappointing, and no perceptible improvement in the
prediction of mixture prop»erties was achieved. The results did suggest, however, that
the introduction of a temperature dependence to the a, b. and c parameters would
improve the performance of the equation of state for pure fluids.
This method has one disadvantage not present in the extended-corresponding-
states method which causes the method to be restricted to temperatures below 120 K
for mixtures which contain Nj. As the method is applied here, it is an excess-volume
method, and therefore when the temperature of the mixture approaches the critical
temperature of one of the component fluids the method fails. Since the critical
temperature of is about 126 K, calculations for mixtures containing N2 should not

be made with this method for temperatures above 120 K. This added restriction on
the method reduces the original set of 285 experimental (p, K 7, .x) points'^"''' to 251
Figure 6 shows all of the points within the new set of 251 for which calculated and
experimental densities differ by more than 0.1 per cent. Conclusions may be drawn
from the deviations shown in figure 6. First, even with a reduced total number of
points for comparison, the total number of deviations between calculated and
experimental densities for the hard-sphere model which exceed the 0.1 per cent
is far more than for the extended-corresponding-states model. Second, the
criterion,
method degrades for all mixtures regardless of components as the temperature
becomes greater than 115 K.

216
 :

MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES 849

0.4 1 1 1 1

0
r, V „ -

» 8

•
•

1 1 1
1

100 110 120 130 140

T/K
FIGURE 6. and densities
Relative deviations exceeding 0.1 per cent between experimental densities
The comparison set is about 250 (p, V, T. x) points from references 2
calculated by the hard-sphere method.
to 6. a. 0.92780CH4 + 0.07220n-C«H,o V. 0.91674CH4 + 0.08326n-C*H,o : C- 0.77982CH4 + ;

0.2201 8n-C4H,o; 0.77762CH4 + 0.22238n-C4H,o; O.58828CH4 + 0.41 172fl-C4H,o O, :

0.30349Nj + 0.69651CH4; •. 0.49242N2 + O.50758CH4; V, 0.06740N2 + 0.93260C3Hg; .

O.34242CH4 O.31372C2H6
-I- 0.34386C3H„; 9. 0.33800N2 + 0.34140CH4 + 0.32060CjH6;
-I- ,
0.85317CH4 + 0.05077C2He 0,04855C3H8 + 0.0475 ln-C4H,o; A. O.84566CH4 + 0.07924CjH6 +
-I-

O.O5O6OC3H8 + 0.02450n-C4H,o; A,0.85133CH4 + 0.05759C2H<, + O.O48O8C3H8 + 0.04300n-C4H,o

-l-,0.85442CH4 + O.O5042C2H6 + 0.04038C3He + 0.02901n-C4H,o + 0.02577i-C4H,o <>.0.05540N2 ;

-I-O.7909OCH4 + O.O56OOC2H6 + O.O5OOOC3H8 + 0.04770n-C4H,o; »,0.04250N2 + O.8I3OOCH4 +

0.04750CjH6 + O.O487OC3H8 + 0.02420n-C4H,o + 0.02410i-C4H,o.

Figure 7 shows deviations greater than 0.1 per cent between experimental densities
and those calculated by the hard-sphere method for all mixtures in the 251 -point
comparison set which are LNG-like, i.e. all multicomponent mixtures with 60 moles
per cent or more of CH4 but with 5 moles per cent or less of each of Nj, j-C4Hio, and
n-C^HiQ. The points above 120 K for those mixtures containing have been

FIGURE 7. Relative deviations exceeding 0.1 per cent between experimental densities and densities
calculated by the hard-sphere method. The comparison set is about 60 (p. V. T, x) points for LNG-like
mixtures from references 4, A. O.84566CH4
5, and
0.07924C2Ht7. O.O5O6OC3H8 + 0.02450n--1- -1-

C4H,o; A. 0.85133CH4 0.05759C2Hfc + 0.04808CjH6 + 0.04300n-C4H,o -t-, 0.85442CH4

-t- ; -I-

0.05042C2H<, O.O4038C3H8 + 0.02901n-C4H,o + 0.02577i-C4H,o .0.04250Nj + O.SBOOCH* +

-(- ;

0.04750CjH6 0.04870CjH8 + 0.02420n-C4H,o + 0.02410i-C4H,o-

-I-

217
 )

850 R. D. McCARTY

excluded. This newly-defined comparison set now contains 60 (p, K T, x) points.

Excluding the 5 points (see above) which have been judged to be experimentally
uncertain by more than the 0.1 per cent criterion as explained in the section on the
extended-corresponding-states model, we are still left with 4 points outside the 0.1 per
cent tolerance which did not appear on the extended-corresp)onding-states deviation
plot. All of these four deviations occur at temperatures greater than 120 K. Therefore
the restriction of 120 K as an upper limit in temperature must be placed on the
method regardless of the comp>osition.

6. A revised Klosek and McKinley method

The Klosek and McKinley method" is a totally empirical method and was included
in this study because of its widespread use in the LNG industry.
simplicity and its

Very eady in the study it became apparent that the original form of the method would
not work for mixtures in which was present, even at mole fractions of 0.04 or less.
The original method was proposed as

Kn,x = I,^.^-MCHJ (21)

where F^j, is the volume of the mixture, x, and V, are the mole fraction and molar
volume of component and the summation is taken over all components including
/

CH4, x(CH4) is the mole fraction of CH4, and A: is a correction factor obtained from a

TABLE 9. Molar volumes V, of saturated liquid of the pure components at temperatures T at intervals
close enough to allow linear interpolation

CH4 C3H, n-C«Hio <-C,H„

T/K mor '

90 0.035441 0.046081 0.060461 0.074708 0.076084 0.037543 0.089173 0.089243

92 0.035649 0.046235 0.060632 0.074891 0.076274 0.038081 0.089379 0.089454
94 0.035861 0.046390 0.060804 0.075075 0.076466 0.038650 0.089586 0.089666
9^ 0.036077 0.046547 0.060977 0.075259 0.076659 0.039254 0.089793 0.089878
98 0.036298 0.046704 0.061151 0.075445 0.076853 0.039897 0.090000 0.090091
100 0.036524 0.046863 0.061325 0.075631 0.077047 0.040586 0.090208 0.090304
102 0.036755 0.047023 0.061501 0.075818 0.077243 0.041327 0.090416 0.090518
104 0.036992 0.047185 0.061677 0.076006 0.077440 0.042128 0.090624 0.090733
106 0.037234 0.047348 0.061855 0.076194 0.077637 0.043002 0.090833 0.090948
108 0.037481 0.047512 0.062033 0.076384 0.077836 0.043963 0.091042 0.091163
110 0.037735 0.047678 0.062212 0.076574 0.078035 0.045031 0.091252 0.091379
112 0.037995 0.047845 0.062392 0.076765 0.078236 0.046231 0.091462 0.091596
114 0.038262 0.048014 0.062574 0.076957 0.078438 0.047602 0.091673 0.091814
116 0.038536 0.048184 0.062756 0.077150 0.078640 0.049179 0.091884 0.092032
118 0.038817 0.048356 0.062939 0.077344 0.078844 0.050885 0.092095 0.092251
120 0.039106 0.048529 0.063124 0.077539 0.079049 0.052714 0.092307 0.092470
122 0.0394O4 0.048704 0.063309 0.077734 0.079255 0.054679 0.092520 0.092690
124 0.039710 0.048881 0.063496 0.077931 0.079462 0.056797 0.092733 0.092911
126 0.040025 0.049059 0.063684 0.078128 0.079671 0.059085 0.092947 0.093133
128 0.040350 0.049239 0.063873 0.078327 0.079880 0.061565 0.093161 0.093355
130 0.040685 0.049421 0.064063 0.078526 0.080091 0.064263 0.093376 0.093578

218
 " '

MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES 851

table or graph. The and k are temperature dependent and in addition k is

dependent upon the mean molar mass of the mixture.
The revised equation is

K„,x = I.^iK-{'^i+(^2-'^iWN2)/0.04251x(CHJ, (22)

where quantities are the same as in equation (21) except that a second correction
factor has been added and x(N2) is the mole fraction of N2. Tables 9, 10, and 1 1 give
values for the V^^k^, and 2, which are spaced so that linear interpolation is adeqate in
A.

both variables {i.e. temperature and molar mass).

The pure-fluid volumes in table 9 are taken from the equations in section 3 with the
exception of that of Nj above 115 K which is a linear extrapolation of the 115 K
point using the slope of the curve at that point. The k factors in tables 10 and 1 1 have
been obtained graphically from the multicomponent (p, F, 7, x) data of references 4
and 6 and from multicomponent densities calculated using the extended-
corresponding-states method.

TABLE 10. Correction factor for the Klosck and McKiniey method as a function of temperature T and
molar mass M
M/(gmol ') 16 17 18 19 20 21 22 23 24 25
7/K 10^fc,/(cm' mor' )

90 -0.005 0.120 0.220 0.340 0.430 0.515 0.595 0.660 0.725 0,795
95 -0.006 0.135 0.260 0.380 0.500 0.590 0.665 0.740 0.810 0,885
100 -0.007 0.150 0.300 0.425 0,575 0.675 0.755 0.830 0.910 0,990
105 -0.007 0,165 0.340 0.475 0.635 0.735 0.840 0.920 1.045 1,120
110 -0.008 0.190 0.375 0.535 0.725 0.835 0.950 1.055 1,155 1.245
115 -0.009 0.220 0.440 0.610 0.810 0.945 1.065 1.180 1,280 1.380
120 -0.010 0.250 0.500 0.695 0.920 1.055 1.205 1.330 1,450 1,550
125 -0.013 0.295 0.590 0.795 1.035 1.210 1.385 1.525 1.640 1.750
130 -0.015 0.345 0.700 0.920 1.200 1.370 1.555 1.715 1.860 1.990
135 -0.017 0.400 0.825 1.060 1.390 1.590 1.800 1.950 2.105 2.272

TABLE 11. Correction factor k. for the Klosek and McKiniey method as a function of temperature Tand
molar mass M
M/(g mol
•
') 16 17 18 19 20 21 22 23 24 25
~
T/K lO^kjAcm^ mol •
)

90 -0.004 0.10 0.22 0.35 0.50 0.60 0.69 0.78 0.86 0.95
95 -0.005 0.12 0.28 0.43 0.59 0.71 0.83 0.94 1.05 1.14
100 -0.007 0.16 0.34 0.49 0.64 0.79 0.94 1.08 1.17 1.27
105 -0.01 0.24 0.42 0.61 0.75 0.91 1.05 1.19 1.33 1.45
110 -0.015 0.32 0.59 0.77 0.92 1.07 1.22 1.37 1.52 1.71
115 -0.024 0.41 0.72 0.95 1.15 1.22 1.30 1.45 1.65 2.00
120 -0.032 0.60 0.91 1.23 1.43 1.63 1.85 2.08 2.30 2.45
125 -0.043 0.71 1.13 1.48 1.73 1.98 2.23 2.48 2.75 2.90
130 -0.058 0.95 1.46 1.92 2.20 2.42 2.68 3.00 3.32 3.52
135 -0.075 1.30 2.00 2.40 2.60 3.00 3.40 3.77 3.99 4.23

219
852 R. D. Mccarty

-1 1 1

% 0.2 - —

-J 1 I

110 120 130 140

r/K
FIGURE 8. Relative deviations exceeding 0.1 per cent between experimenul densities
and densities
and McKinley method. The comparison set is about 60 (p, V, T. x) points
calculated by the revised Klosek
for LNG-like mixtures from references 4, 5, and 7. O. O.84566CH4 + 0.07924C2H6 + O.OSOeOCjHg +
0.02450n-QH,o; +, 0.85934CH4 + 0.08477CjH6 + 0.02980CjH8 + 0.00707«-C4H,o + 0.00519i-
C4H,o; A,0.00859Nj + 0.75713CH4 + 0.13585CjHe, + O.O6742C3H8 + 0.01326n-C4H,o + 0.01336i-
C4H,o + 0.00216n-C5H,j + 0.00223i-C5H,j.

The composition limits of the revised Klosek and McKinley method are the most
severe of any of the three methods given here. This method should not be used for
LNG-like mixtures, unless they contain at least 60 moles per cent of CH4, less than 4
moles per cent of Nj, less than 4 moles per cent of each of /-C4H10 and n-C4H,o, and
less than 2 moles per cent in all of /-C5H,2 and n-CsHij.

There are 61 experimental (p, V, T, x) points from references 4 and 6 which fall
within the composition limits outlined above. Figure 8 shows all of the deviations
between calculated and experimental densities in this 61-point comparison set which
exceed the 0.1 per cent criterion. Figure 8 exhibits the same sort of deviation trend as
is found in figure 7. In figure 8 all of the deviations are at 1 15 K and above; therefore,

the claim of 0.1 per cent accuracy with this method must be restricted to temperatures
below 115 K.

7. Conclusions

On the basis of the performance of the three models and subject to composition and
temp)erature range restrictions as outlined earlier in this paper; it is estimated that,
given the pressure, temperature, and composition of LNG, any one of the three
models presented here may be used to predict the density to within 0.1 per cent of the
true value. As was mentioned before, no model or method can be any better than the
experimental values used to develop it; therefore the above accuracy claim depends
strongly on the accuracy of the experimental values. The estimated uncertainty of the
experimental values''"^' used here is 0.1 per cent or less, with a precision of a few

hundredths of a per cent. With the exceptions already detailed in sections 3 and 4, the
correlations presented here provide no basis for questioning the claims of the
experimenters.
This entire study including the experimental work has provided valuable
experience in the problem of mathematically modelling the equation of state of a
mixture, and the lessons learned therefrom are worth passing along to future workers
in the field. Having accurate and internally consistent experimental results has always

220
 MATHEMATICAL MODELS FOR THE PREDICTION OF LNG DENSITIES 853

been an important factor in equation-of-state development for pure substances but in

the case of the equation of state of mixtures it is absolutely essential. Without
internally consistent results the job is hopeless, especially if the accuracy of the
equation of state is important.
The value of conducting the experimental and modelling program simultaneously
has also been clearly demonstrated, for without the parallel efforts in this project the
experimental problems with some of the mixtures containing butane would never
have been detected.
Of the three methods presented here, the extended-corresponding-states model has
proved to be more accurate and more versatile than either of the other two. However,
over the very restricted range of composition and temperature of liquefied natural
gas, no one method seems to be more accurate than the other.
Interim results of this study were reported by Haynes et al!^^^ and McCarty''"
both of which contain earlier versions of the mathematical models presented here.
The earlier versions in references 21 and 22 are only slightly different from the present
ones, the main difference being in the CH4-C4H,o binary interaction parameters for
the extended-corresponding-states method and the hard-sphere method however, for
;

calculations of the density of LNG, the difference is inconsequential. The user of any
of these models is cautioned to read carefully the limitations associated with each
model given in the relevant individual section.

Computer programs for the three models presented here are available from the
Thermophysical Properties Division of the National Bureau of Standards in Boulder,
Colorado. A complete listing of these programs and more detailed instructions on
how to use them is published elsewhere.'^^'

REFERENCES
1. Haynes.W. M.. Hiza. M, J. J. Chem. Thermodynamics 1977, 9, 179.
2. Hiza.M. J.; Haynes, W. M.; Parrish. W. R. J. Chem. Thermodynamics 1977, 9. 873.
3. Haynes. W. M. Orthobaric liquid densities of methane + isobulane and methane + normal butane
mixtures at low temperatures (to be published).
4. Hiza.M. J.; Haynes. W. M. J. Chem. Thermodynamics 1980. 12. 1.

5. Haynes. W. M. J. Chem. Thermodynamics 1982, 14. 603.

6. Miller. R. C; Hiza. M. J. Fluid Phase Equilibria 1978, 2. 49.

7. Haynes, W. M.; Hiza. M. J.; Frederick, N. V. Rew Sci. Instrum. 1976, 47. 1237.
8. Leach, J. W. Molecular structure correction for applying the theory of corresponding states to non-
spherical pure fluids and mixtures. Ph.D. Thesis, Rice University. 1967.
9. Rodosevich, J. B.; Miller, R. C. Adv. Cryog. Eng. 1974, 19, 339.
10. Klosek, J.; McKinley, C. Proc. First Int. Conf. on LNG. Chicago. 1968, paper 22.
11. Watson, K. M. Ind. Eng. Chem. 1943, 35. 4. 399.
12. Hiza, M. J. Fluid Phase Equilibria 1978, 2, 27.
13. Shana'a. M. Y.; Canfield, F. B. Trans. Faraday Sac. 1968, 64. 2281.
14. Bloomer, O. T.; Parent, J. D. Chem Eng. Prog. Symp. Ser. 1952, 49, 11.
15. Orrit, J.; Olives, J. F. Distributed at 4th International Conference on Liquefied Natural Gas. Algeria.
1974.
16. Rodosevich, J. B. Measurements and prediction of liquefied natural gas densities. M,Sc. Thesis,
University of Wyoming. 1973.
17. Rowlinson. J. S.; Watson. I. D. Chem. Eng Sci. 1969, 24, 1565.
18. Mollerup. J. Adv. Cryog. Eng. 1975, 20, 172.
19. Mollerup. J.; Rowlinson. J. S. Chem. Eng. Sci. 1974. 29. 1373.
20. Goodwin. R. D.; Prydz. R. J. Res. Sail Bur. Stand. (L .S.) Sect. A 1972. 76, 1-81.

221
854 R. D. Mccarty
21. McCany. R. D. Natl Bur. Stand. (U.S.). Interagency Report NBSIR 77-867. 1977.
22. Haynes. W. M.; Hiza. M. J.; McCarty. R. D. Proc. Fifth International Conference on LNG,
Diisseldorf Vol. 2. Institute of Gas Technology, Chicago, 111. 1977, Section III, Paper 11.
23. McCarty, R. D. Natl Bur. Stand. (.U.S.). Tech. Note 1030. 1980.
24. Ornt, J. E.; Laupretre, J. M.
Adv. Cryog. Eng. 1978, 23, 573.
25. Nunes da Ponte. M.; Streett, W. B.; Staveley, L. A. K. J. Chem. Thermodvnamics 1978, 10, 151.
26. Longuet-Higgins. H. C; Widom. B. Mol. Phys. 1964, 8. 549.
27. Slraty, G. C: Diller, D. E. J. Chem. Thermodynamics 1980, 12, 937.
28. McClune. C. R. Cryogenics 1976, 16, 289.

222
6.13 Reference [25]

Fluid Phase Equilibria, 2 (1978) 27-38 27

© Elsevier Scientific Publishing Company, Amsterdam — Printed in The Netherlands

AN EMPIRICAL EXCESS VOLUME MODEL FOR ESTIMATING

LIQUEFIED NATURAL GAS DENSITIES

M.J. HIZA
Cryogenics Division, Institute for Basic Standards, National Bureau of Standards,
Boulder, Colorado 80302 (U.S.A.)

(Received August 22nd, 1977)

ABSTRACT

Hiza, M.J., 1978. An empirical excess volume model for estimating liquefied natural gas
densities. Fluid Phase Equilibria, 2: 27—38.

The mathematical model presented herein was developed to represent excess volumes at
saturation formulticomponent liquid mixtures of nitrogen and the low molecular weight
alkanes between 105 and 120 K. Parameters of the model were determined from experimental
excess <volumes for binary liquid mixtures of nitrogen, methane, ethane, propane, isobutane,
and normal butane. Comparisons made with selected experimental excess volumes reported
in the literature for multicomponent liquid mixtures of the above components demonstrate
the predictive capability of the model in two simple forms. An extension of the model to
include mixtures containing isopentane and normal pentane is also proposed. Pure component
molar volumes are given at 0.5 K intervals from 105 to 116 K to facilitate the use of the
present model in estimating liquefied natural gas (LNG) densities.

BACKGROUND

Analysis of consistency and the prediction of orthobaric (saturated) liquid

molar volumes of multicomponent mixtures requires a reliable mechanism to
account for the effect of each component as a function of composition and
temperature. Of specific interest here are multicomponent liquid mixtures of
natural gas components; viz. nitrogen, methane, ethane, propane, isobutane,
and normal butane. A simple and reliable empirical model in this specific area
is most attractive because it offers an alternative to the more complex theoret-

ical models for predicting orthobaric LNG (Liquefied Natural Gas) densities
for custody transfer.
At temperatures where all components present in the liquid mixture are sub-
critical, the logical subject of an empirical model is the liquid mixture excess
volume. The development of such a model into a useable (predictive) form
requires a consistent set of density data for the pure components and for the

223
28

possible binary systems, each at several compositions, in the temperature range

of interest.
The empirical models of Klosek and McKinley (1968) and of Miller (1974)
are of this type. The first, using mixture molecular weight as the correlating
parameter, does not properly account for the presence of nitrogen in the mix-
ture nor does it include adequate information on the temperature dependence
of excess volume. The second, basically a more sound approach, treats the
presence of nitrogen separately, uses an effective ethane mole fraction as the
correlating parameter for the hydrocarbon contribution, and includes signifi-
cantly more information on temperature dependence, as well as pressure
effects. However, both models were based on relatively incomplete data sets
and some of the features of each were, of necessity, based on educated guesses.
For example, accurate orthobaric liquid density data for nitrogen + ethane,
nitrogen + propane, and binary mixtures of ethane, propane, isobutane, and
normal butane were not available at the time these earlier models were devel-
oped. In addition, both models were intended only for use with methane-rich
mixtures. Neither will correctly predict the excess volume at saturation of a
simple equimolar liquid mixture of methane, ethane, and propane.
The empirical excess volume model proposed here was developed with due
consideration of the deficiencies in the models of Klosek and McKinley
(1968) and of Miller (1974). It is based on the self-consistent orthobaric liquid
densities for the pure components (Haynes et al., 1976; Haynes and Hiza,
1977) and for the binary mixtures (Hiza et al., 1977) obtained in the NBS LNG
density project. The excess volumes for the methane + butanes and nitrogen +
methane reported by Miller (1974) and Rodosevich and Miller (1973) were
also used. The excess volumes reported by Miller (1974) and Rodosevich and
Miller (1973) are generally consistent with those from the NBS data within
an imprecision equivalent to about 0.1 percent in the mixture molar volumes
(Hiza et al., 1977). All excess volumes reported in these references were calcu-
lated from:

= v„ -Ex,[y, + /3,v.(P,-Pn.)] (1)

where V is the molar volume, P is the saturation pressure, x is the mole frac-
tion, and j3 is the isothermal compressibility. Subscripts m and i refer to the
mixture and to the pure components, respectively. In all cases the isothermal
compressibilities were taken from Rowlinson (1969) for nitrogen and methane,
and from Miller (1974) for ethane. Isothermal compressibility adjustments to
the molar volumes of propane, isobutane, and normal butane were not made
in deriving excess volumes from the NBS data. Neglecting isothermal compres-
sibility adjustments to the molar volumes of propane and the high-molecular
weight alkanes has no significant effect on the results.

FORMULATION OF THE EXCESS VOLUME MODEL

The present model is based on excess volume ratios relative to the well-
defined methane + ethane and nitrogen + methane binary systems, treating

Z2A
 29

the hydrocarbon + hydrocarbon and nitrogen + hydrocarbon contributions

separately. Eqns. 2, 3 and 4 describe this model in general terms.

Vf = Vic + Vgc (2)

where

X5
(3)
12

and

In these equations excess molar volume and x is the component mole

is

fraction. The subscripts CC, and NC refer to total, hydrocarbon + hydro-

t,

carbon, and nitrogen + hydrocarbon, respectively. The subscript N refers to

nitrogen and 1, 2, 3, 4, and 5 refer to methane, ethane, propane, isobutane
and normal butane, respectively. In eqn. (3), the excess volumes for methane +
ethane ( V12), etc. are taken at a pseudo mole fraction of methane, x\, equal
to the sum of the actual methane and nitrogen mole fractions, x^ + jc^. In
eqn. (4), the excess volumes for nitrogen + methane (V^i), etc. are taken at
the actual nitrogen mole fraction in the mixture, x^. Obviously, if no nitrogen
is present in the mixture Vf = Vcc eqn. (4) is not needed.
In this approach, it is assumed that contributions from heavy hydrocarbon +
heavy hydrocarbon interactions are negligible. In the temperature range of
interest, this assumption is valid since experimental excess volumes for binary
mixtures of ethane, propane, isobutane, and normal butane are nearly zero
(Hiza etal., 1977).
The values of 1^12 and V13 in cm^/mol in eqn. (3) are evaluated from a
temperature dependent Redlich— Kister expansion of the following form:

Vl = XiXi[{ao + a^T + a^T^) + (bo + b^T + h^T^){2Xi - 1)

+ (Co + Cir+C2r2)(2x,-l)2] (5)

atx, = x\ = x-y + jcn and Xj = 1 —

x\. T is the temperature in Kelvin. The co-
efficients of eqn. (5) obtained from the NBS data for the methane + ethane
and methane + propane systems are given in Table 1 The coefficients for the
.

methane + propane system are the same as those presented earlier (Hiza et al.,
1977); those for the methane + ethane system are slightly different from the
earlier values (Hiza et al., 1977) since a few points, obtained by vapor recircu-

225
30

TABLE 1

Coefficients of Eq. (5) for the methane + ethane and methane + propane systems

methane + ethane methane + propane

flo —14.89469 —11.96725

ai 0.2770596 0.236748
02 —0.001473116 —0.001429237
60 —37.98245 16.59465
61 0.6427122 —0.3212125
62 —0.002783724 0.001398221
Co 0 —52.93613
Ci 0 0.9887147
C2 0 —0.004667488

lation and composition analyses of liquid samples (Hiza and Haynes, 1978),
were added to determine the coefficients given here. The applicable tempera-
ture ranges for these coefficients are 105 to 135 K for methane + ethane and
105 to 130 K for methane + propane. For convenience values of A = Qq +
aiT + a2T^,B = 60 + &iT+ &2^^,and C= Co + c^T^ CzT^ are given at a few
specific temperatures in Tables 2 and 3 for methane + ethane and methane +
propane, respectively.
There are few data points available for the methane + isobutane and meth-
ane + normal butane systems. Thus, it is not possible to represent the excess
volumes for these systems analytically. The ratios of excess volumes for the
methane + butane systems to the excess volume for the methane + ethane sys-
tem were determined at several compositions in the methane rich region at
approximately 108 K. For purposes of this correlation these ratios have been
assumed to be constant. These are

vfjvh = 2.0 (6)

and

F?5/VL = 2.5 (7)

where subscripts 4 and 5 refer to isobutane and normal butane, respectively.

Thus eqn. (3) reduces to

X4 ^ 1.5x5
yEc(@.;) = vf,[i.(J-i),-j^ ) il-xl) (l-xl)j
(8)

With some loss in precision, the ratios of the excess volumes for the methane
+ propane system to the methane + ethane system above 30 mole percent
methane can be represented as a fimction of or J only. With this modification,

226
 31

TABLE 2

Redlich— Kister coefficients at specific temperatures for the methane + ethane system

T A B
(K) (ao + aiT + a2T^) (60 + 6iT + 62^2)

105 —2.044536 —1.188226

108 —2.154678 —1.038889
108.15 —2.160881 —1.032738
110 —2.242838 —0.967168
115 —2.514795 —0.885297
120 —2.860408 —0.942612

TABLE 3

Redlich— Kister coefficients at specific temperatures for the methane + propane system

T A B C
(K) (ao + fliT + aaT^^) {bo ^ biT + baT^) (Co + CiT + C2r2)

105 -2.866048 —1.717276 —0.580142

108 -3.069086 -1.787450 —0.596522
108.15 -3.079914 —1.790298 —0.599547
110 -3.218738 -1.820251 —0.654118
115 -3.642889 -1.853315 —0.961468
120 —4.138503 -1.816468 -1.502193

eqn. (8) becomes

^3
1 + (O.eOlbxl + 0.1664)

1.5Xe
(9)

In examining the nitrogen + hydrocarbon data, it was noted that the excess
volumes for the nitrogen + ethane and nitrogen + propane systems are essen-
tially the same; and, at low nitrogen content, these and the excess volumes for
nitrogen + methane are almost linear functions of the nitrogen mole fraction
(up to about 10 mole percent for nitrogen in methane). Without the benefit
of experimental data, it was assumed that excess volumes for the nitrogen +
butane systems are approximately the same as those for the nitrogen +
propane system. Therefore, eqn. (4) was reduced to
0.81(1 -xl)-
1 + (10)

where
•

134.196
2.8842 (11)
126.2- T

227
32

The applicable temperature range of eqn. (11) is 91 to 120 K. For temperatures

above 115 K, this equation should not be used for nitrogen concentrations in
excess of 5 mole percent.

MODEL LIMITATIONS

In the applicable temperature range, i.e. 105 to 120 K, limitations of the

present empirical model are due primarily to insufficient or lack of data for
certain binary systems. The available data are inadequate to describe the tem-
perature and composition dependence of excess volumes for either of the
methane + butane systems. Above 115 K, more data are needed for the nitro-
gen + methane system to adequately describe the composition dependence of
excess volumes for mixtures containing more than 5 mole percent nitrogen.
And finally there is a lack of any experimental information on properties of
the nitrogen + butane systems.
Therefore, betv^^een 105 and 115 K, this model should not be used for multi-
component LNG mixtures containing more than 10 mole percent of nitrogen,
isobutane, or normal butane. Between 115 and 120 K, the uncertainties of
predicted excess volumes are expected to become unacceptably large (>0.15%
of the maxture molar volume) for multicomponent LNG mixtures containing
more than 5 mole percent nitrogen. For mixtures containing only methane,
ethane, and propane, the model should be applicable from 30 to 100 mole
percent of methane from 105 to 120 K, and probably to somewhat higher
temperatures. It is believed that, within these limits of composition and tem-
perature, the uncertainty in predicted excess volumes for multicomponent
LNG mixtures equivalent to about ±0.15% of the mixture molar volume.
is

Best results are expected in the middle of the applicable temperature range,
i.e., 110—115 K, particularly if eqn. (9) is used instead of eqn. (8).

Some comparisons between experimental excess volumes for selected multi-

component mixtures of LNG components from the literature and predicted
results usingboth eqns. (8) and (9) are given in Table 4. In almost all cases, the
predicted excess volumes are less than the experimental values, and only for the
mixture containing 72 mole percent of methane with ethane and propane
(Shana'a and Canfield, 1968) does the difference exceed 0.15% of the mixture
molar volume. The average difference between predicted and experimental ex-
cess volumes is about —0.09% of the mixture molar volume. Comparisons are
also being made with excess volumes from experimental multicomponent mix-
ture measurements in the NBS LNG density project, and these will be published
with the data in the near future.

ESTIMATING DENSITIES OF LNG, INCLUDING PENTANES

Use of the present model for estimating the orthobaric molar volume, and
thus the density, of a liquefied natural gas mixture at a desired temperature
and saturation pressure follows directly from eqn. (1). A consistent set of pure

228
 ( 1 1

o
Ho
1-1 00 mo Oi 05
o
o
1-H o o o I-H o o
o
T? ? ? T ?? ?T ?

I> 00
m OS
o
in
;d r-
o in
o o
in o
CD CD
i-H
in
CO
rf in to CD in CD in 00

? ? T ? ?T ?

(N T— o in in 05 CD Ol
o o l-H rH CM
qo qq qq
o dd d dd d
I I I

00 CD o> O in CD CD
o
in in
CO
CD
00 CD
in CD in CD
o CO
in 00
in

o
I T T

00 o
CO 05
in CD
CD
in CD
CD in
in 00
o
in
t>
d d d d

(N CD 00 O 7-1 o CT>
e<i Oi 00 CD in CM in CO o to
CO q CM T-H 05 r~ q CM in
cx) d D- 1— d d
CO
d 00
CO CO
d oo"
CO CO

in in

00 in 00 00 00 in 00 in 00 in 00
o »—
tH
o
T-H
o o T-H
o
I-H
o I-H
o

<C 00 CD M CO 00 «3:
3- c^ E cmI CM CO
o u u u u u U U U u u y U U U c z u u Z U U
01 w
S 0 coinoi -^coo oooo'* O200CO coi-Hcoco ino5cD r-inc~
a CDC^O OO'-hO C0CD03 C^i-hO oooio OOOO -^inoi
rroin oocMCT) CMCDO cDooin coo-^c^ m^fO TfO'^
E OO'-hO COCOPJ [>.-h>-h OOOO OO-HOO OOO-H OCTJO
0
o dd d dd d ddd d d d OOOO OOO OOO
00 00 CO
CD CO r-
T5 CO
1 " o>
I-H
05
1-H
05
tH c
cs
C
B3
CO

05 rr o> Ol
-o xT t~ x:
CJ CJ CJ O
'> <T>
sevii
01 nfie sevi Her, Her,
M — 2 «
tA
L-
Ca do 01 do do ii
0 0 0 0
BS CO OS b;

229
34

component molar volumes adjusted to the mixture vapor pressure at the same
temperature are needed to make this calculation.
The saturated liquid molar volumes of nitrogen and the alkanes through the
butanes are readily obtained from the expression

(12)

with the parameters given by Haynes et al. (1976) and Haynes and Hiza (1977).
In this expression p is the density in mol/1, T is the temperature in Kelvin, and
the subscript c denotes the critical point. For isopentane and normal pentane,
saturated liquid molar volumes between 105 and 120 K can be obtained by
extrapolation of the data of Orrit and Laupretre (1978) or of McClune (1976).
Densities from these investigators are generally within 0.1% of the NBS values
for the lower molecular weight alkanes. In his paper, McClune gives the constants
for a quadratic expression in temperature which he used to extrapolate the den-
sities for isopentane and normal pentane down to 93 K. In the present paper,
the densities given by Orrit and Laupretre for isopentane (from 124.925 to
253.794 K) and for normal pentane (from 148.879 to 249.011 K) were fitted
to eqn. (12) with values of Pc and taken from Kudchadker et al. (1968).
These parameters of eqn. (12) are given in Table 5. Compared with molar
volumes between 120 and 105 K from McClune's extrapolation, molar volumes
obtained from the fit of the data of Orrit and Laupretre are larger for isopentane
by 0.14—0.15% and smaller for normal pentane by 0.09—0.29%. With the small
concentrations of pentanes normally encountered in LNG mixtures, these differ-
ences are not significant.
Of common interest is the molar volume or density at a temperature where
the total vapor pressure of the liquid mixture is about atmospheric pressure.
For mixtures containing mole fractions of 0.85—0.97 methane, 0.03—0.09
ethane, and up to 0.016 nitrogen, 0.03 propane, 0.01 butanes, and 0.0023 pen-
tanes, Giarratano and Collier (1977) report a calculated boiling point tempera-
ture range from about 106.5 K with a nitrogen mole fraction of 0.016 to
about 113.2 K with no nitrogen. The pure component molar volumes given in
Table 6 at 0.5 K increments from 105 to 116 K and at 120 K can be used

TABLE 5

Parameters of eqn. (12) for isopentane and normal pentane

isopentane normal pentane

a 2.946310 —0.03620055
35.50771 59.00203
-57.41243 —93.44194
28.15898 43.66781
Tc (K) 460.39 469.60
Pc (mol/1) 3.271 3.285

230
00<-<eOiftiOOONlf>t^CT>N'^t^ONif5t~OC»5irtCO>-i'^t-
e<iaoeooocooiTt05'^05inoift'-it£>i-i;De>^it~c^t~rtooo
t~_t-^oqooa)05 0 0'-H,-ie^rte»3'><i<T)'ir5ino«Dt~t~aoco«
OO O" O OO 1-J i-I r-I ,-( ^ ^ tH ,-i rH r-i i-J T-i ^' rH r-i (N

Orj«aOi-Hifto>p:r-f-iifto>'«tooNtf)i-Hino5T}<eowt-cqo
^a)Tfoirto;Di-it~Nt^rtcoTto5ifiOini-i«£)c^c~c<5t-
00 00 05 oo e<i N CO CO -"I" ift in to r-. t>- 00 00 05 a> o
O" OO T-i 1-i 1-i ^ i-( ,-i ^" i-i r-i r-i ,h' r-I r-( »H r-i »-I rH i-i N

o
o
(N 05 to
n •<t tH o> to in o
05 t~ in <9>
•><< 05 m on
<t
rH ,—1 n
00 CO r- (N
C«5
c~
t
in to to
to rH to
ifi00 00
in
05
in in
rH in
o o
o o o
«D <£> to to to to to to to to to to to to to to to t~ c- t~
t~ c~ t~ c~ t- I> r~ t- t~ r- t- t~ l> r-

00 00 t~
n 00 n
00
to to to to in in in in to to to to t~ t~
n
00 CO 00 00 00 CO 00 n
00 00 n n w
00 00 05
CO 00 CO 05
oo
05
in in to to c- c~ 00 00 a> 05 rH rH esi e<i oo n m <t in in to o
t- t~ t~ t~ c- t~ t~ 00 00 00 00 00 00 00 00 00 00 00 00 OS
t~ c~ c~ t- c- t- t~ t> t- t~ t~ t~- t- t~ r- c~ t~

to ©in OS 00 CO 00 00 CO 00 o> in to «t o
to rH in Oi 00 CO to rH in o> CO rH CO rH in INO X
t~ 00 00 00 05 <y> rH rH (N o o
CO CO CO Tf in in CO CO C~ t> I-H t
rH rH rH rH rH CO (N IN IN (N IN IN (M IN CO
to to to CO to to to to to to to to to to to CO CO to to CO CD to CD to

CD 00 Ol rH o CO in CO 00 05 1-1 CO in OS rH •>1' to 05 rH 05
CO o 00 CO t> rH CO 05 CO r» rH to o Tr rH in OS 00
CO CO CO in in in to to r- 00 00 00 OS 05 o o o
rH I-H in
t~ c~ t> t~ t~ t~ t> 00 00 00 00 00 00
•t rl' t t •t t

(N CO in OS rH t- rH in o o
in rH 00 in CO 00 CO to o
rH c~ CO OS in rH 00
N o t~ CO o
CD CO OS CD CO OS CO CO OS CO CO o
,-H rH CO <* in CO CD 00 00 OS 05 o
,H rH eo CO CO rf in rH
t~ t> t- t~ c- t- r~ c~ c~ r~ t- 00 00 00 00 00 00 00 05
CO CO CO CO CO co CO co CO eo CO CO CO CO co CO CO CO CO CO CO CO CO CO

00 00
CO o00
t> CD 00 rH t~
I> t~ 00 rH
00
OS
rr 00 00 OS
CD * to rH
O OS OS rH
to 00
o
OS CO
CO in CO in
CO
CO 00
r-I to Ol CO in OS CO in 05 CO rH CD o in o
to CO 00 in CO o 00
CO CO CO CO Tj- in in in CD CO c~ 00 00 OS OS o o
rH CO CO
t t •<*<
<r <t t t
in in in in in to

o in o in o in o in o in o in o in o in o in o in o in oo
in in to to c~ 00 00 OS OS o o rH rH CO CO CO in in to o
o o o oo o o o oo CO

231
 :

36

directly to calculate the ideal contribution to the mixture molar volume when
the mixture vapor pressure is about one bar. The molar volumes given in Table
5 for nitrogen were adjusted to one bar using isothermal compressibilities given
by the following expression

/3N2(bar-^) = -0.001032 + 0.051368/(126.2 - T) . (13)

The molar volumes given for all of the other fluids are the orthobaric (satura-
tion) values. Values for isopentane and normal pentane were calculated from
eqn. (12) and the parameters given in Table 5. Though the isothermal com-
pressibility correction has little effect on the methane molar volume at the
conditions considered here, one can determine the magnitude of the pressure
effect using the following expression to obtain the isothermal compressibility
for methane

/3cH4(bar-i) = -0.0001595 + 0.029488/(190.555 - T) (14)

Between 105 and 120 K, eqns. (13) and (14) represent isothermal compres-
sibilities listed by Rowlinson (1969) within 4% for nitrogen and within 2%
for methane. At the same temperatures eqn. (14) gives isothermal compres-
sibilities for methane that are smaller but within 3% of those taken from the
more recent data of Goodwin (1974).
There are no experimental data in the literature from which excess volume
ratios of methane + pentanes to methane + ethane can be derived. However,
for the present purpose estimates of these ratios were made by inspection of
the excess volume ratios of methane + propane, methane + isobutane, and
methane + normal butane to methane + ethane. Based on these estimates,
eqn. (9) can be revnritten to include the pentanes as follows:

+ (0.6015x1 + 0.1664) -
V^ci@xl) = Vf2 1

1.5x5 2.2x6 3-1^7

(15)
n-xi) (1-xi) (i-xi)j
In eqn. (15), subscripts 6 and 7 refer to isopentane and normal pentane,
respectively. For the nitrogen + hydrocarbon contribution, Vnc» to the total
excess volume, Vf, of mixtures containing small amounts of pentanes, eqns.
(10) and (11) should be used without modification.

CONCLUSION

For liquid mixtures containing nitrogen and the low molecular weight
alkanes all at subcritical temperatures, the mathematical excess volume model

proposed here combined with the consistent set of experimental pure compo-
nent densities now available, allow reasonable estimates of the orthobaric
liquid densities of mixtures containing these components within defined com-
position limits. For the special case involving some typical LNG compositions

232
 37

over a limited temperature range (105 to 115 K), liquid mixture densities can
be predicted with the present method with an inaccuracy of ±0.15% or less.
Additional LNG type mixtures data, consistent with the pure component and
binary mixtures data used here, would more clearly identify the composition
and uncertainty limits of the present model.

ACKNOWLEDGMENTS

The contribution of M.J. Brown to the reduction of the data is gratefully

acknowledged. Partial support of this work was provided by British Gas Corp.,
Chicago Bridge and Iron Co., Columbia Gas Service Corp., Distrigas Corp.,
Easco Gas LNG, Inc., El Paso Natural Gas, Gaz de France, Marathon Oil Co.,
Mobil Oil Corp., Natural Gas Pipeline Co., Phillips Petroleum Co., Shell Inter-
national Gas, Ltd., Sonatrach, Southern California Gas Co., Tennessee Gas
Pipeline Co., Texas Eastern Transmission Co., Tokyo Gas Co., Ltd., and Trans-
continental Gas Pipe Line Corp., through a grant administered by the American
Gas Association, Inc.

REFERENCES

Giarratano, P.J. and Collier, R.S., 1977. Evaluation of capacitance densitometry for LNG
mixtures with low nitrogen composition. Ind. Eng. Chem., Proc. Design and Develop-
ment, 16: 330—336.
Goodwin, R.D., 1974. The thermophysical properties of methane, from 90 to 500 K at
pressures to 700 bar. Nat. Bur. Stand. (U.S.), Tech. Note 653, 274 pp.
Haynes, W.M. and Hiza, M.J., 1977. Measurements of the orthobaric liquid densities of
methane, ethane, propane, isobutane, and normal butane. J. Chem. Thermodynam.,
9: 179-187.
Haynes, W.M., Hiza, M.J. and Frederick, N.V., 1976. Magnetic suspension densimeter for
measurements on fluids of cryogenic interest. Rev. Sci. Instrum., 47: 1237—1250.
Hiza, M.J. and Haynes, W.M., 1978. Liquid mixture excess volumes and total vapor pres-
sures using a magnetic suspension densimeter with compositions determined by
chromatographic analysis: methane • ethane. In: K.D. Timmerhaus (Ed.), Advances in
Cryogenic Engineering, Vol. 23, Plenum, New York, paper M-6 (in press).
Hiza, M.J., Haynes, W.M. and Parrish, W.R., 1977. Orthobaric liquid densities and excess
volumes for binary mixtures of low molar-mass alkanes and nitrogen between 105 and
140 K. J. Chem. Thermodynam., 9: 873-896.
Klosek, J. and McKinley, C, 1968. Densities of liquefied natural gas and low molecular
weight hydrocarbons. In: J.W. White and A.E.S. Neumann (Editors), Proceedings of
First International Conference on LNG, Institute of Gas Technology, Chicago, 111.,
Paper 22, Session 5.
Kudchadker, A. P., Alani, G.H. and Zwolinski, B.J., 1968. The critical constants of organic
substances. Chem. 659—735.
Rev., 68:
McClune, C.R., 1976. Measurement of the densities of liquefied hydrocarbons from —100
to -180°C (173 to 93 K). Cryogenics, 16: 289-295.
Miller, R.C., 1974. Estimating the densities of natural gas mixtures. Chem. Eng., 81
(No. 23): 134-135.
and Laupretre, J.M. 1978. Density of liquefied natural gas constituents. In: K.D.
Orrit, J.
Timmerhaus (Ed.), Advances in Cryogenic Engineering, Vol. 23, Plenum, New York,
paper M-4 (in press), (see also: Orrit, J. and Olives. J.F., 1974. Density of liquefied

233
38

its components. Unpublished report distributed at 4th International

natural gas and
Conference on Liquefied Natural Gas, Algeria, 16 pp).
Rodosevich, J.B., 1973. Measurement and prediction of liquefied natural gas densities.
M.S. Thesis, Univ. of Wyoming, Laramie, 83 pp.
Rodosevich, J.B. and Miller, R.C., 1973. Experimental liquid mixture densities for testing
and improving correlations for liquefied natural gas. A.LCh.E.J., 19: 729—735.
Rowlinson, J.S., 1969. Liquids and liquid mixtures. 2nd edn., Butterworths, London, 371 pp.
Shana'a, M.Y. and Canfield, F.B., 1968. Liquid density and excess volume of light hydro-
carbon mixtures at — 165°C. Trans. Faraday Soc, 64 (No. 549, Part 9): 2281—2286.

234
6.14 Reference [22]

A concentric cylinder capacitor hu been uaed to metture the orthobaric liquid

dielectric constants of multicomponent mixtures of the major components of liquefied
natural gas ILNG} to an accuracy of approximately ± 0.05% at temperatures from 1 10 to
130 K. These mixtures ranged from a ternary mixture containing nitrogen, methane, and
normal butane to four to eight component methane rich (74 to 90 mol %l mixtures con-
taining up to 5 mol % of nitrogen, 16 mol % of ethane, 7 mol % of propane, 5 mol % of
the butanes, and 0.44 mol % of the pentanes. Some of these mixtures were prepared to
simulate commercial LNG compositions. Experimental densities previously reported for
these mixtures have been combined vitith the mixture dielectric constant data to calculate
values of the Oausius-Mossotti (CM) function and the excess CM function. Pure compo-
nent experimental CM functions for LNG components except for propane and isobutane
have been combined with the mixture data in the development of a simple calculational
technique for the prediction of LNG densities to an uncertainty of approximately ± 0. 15%
based on a knowledge of the composition and dielectric constant of the liquid mixtures.
In fitting the data, pseudo values of the CM function are derived for the slightly polar
components, propane and isobutane, while constraining the mixture excess CM function
to be zero.

Prediction of liquefied natural gas (LNG) densities

from new experimental dielectric constant data
W.M. Haynes and R.D. McCarty

Key words: liquafiad natural gas, dansity, dialaetrie oonttant

A large-scale project has been carried out at this for LNG components at low temperatures to analyse
laboratory to provide erne or more mathematical models to property the dielectric constant results. Thus, one of the
predict the density of a liquefied natural gas (LNG) mixture major purposes of this paper is to present the dielectric
to within an uncertainty of 0.1%, given the temperature, constant data for multicomponent mixtiues of LNG com-
composition and pressure of the liquid mixture. Such models ponents, for which the pressure density temperature
will serve as a basis for custody transfer. The mathematical properties have been reported previously.
models developed for this purpose have ranged from simple The dielectric constant of a fluid is closely related to
empirical models'"* to those that are theoretically based its density through the Qausius-Mossotti (CM) function.
and, consequently, much more complex. In fact, dielectric constant measurements can serve as a
In order to develop and test such models, a compre- simple and reliable substitute for density measurements.
hensive, accurate and consistent set of density data has been In the custody transfer of LNG, densities are usually
obtained for the major components*-' of LNG and for determined either by direct density measurements with a
mixtures'"*" of these components. Most of these data were commercially available instrument or indirectly using a
obtained with a magnetic suspension densimeter.'*''" Prior calculation based on measurements of composition and
to the completion of the LNG density project, the apparatus temperature. One technique employed in commercial
was completely rebuilt because of the need to extend the densimeters is based on a density determination through
pressure capability of the apparatus.'^ During the recon- the appUcation of capacitance (or dielectric constant)
struction of the apparatus, it was decided to include a measurements.
capacitor inside the sample space so that simultaneous The multicomponent
dielectric constant data for
measurements of dielectric constant and density could be mixtures of LNG components have been combined with
performed on the same fluid samples. The new apparatus the previously reported density data to develop a simple
was subsequently used to complete the measurements for calculational scheme for the prediction of LNG densities
the LNG density project. The measurements with the new based on measurements of the dielectric constant and com
apparatus included density and dielectric constant data position of the liquid mixture. This procedure provides an
for binary mixtures of methane with either isobutane or alternative approach to the complex models^"* that have
normal butane'" and for multicomponent mixtxires' with been previously used for prediction of LNG densities.
up to eight components, some simulating commercial LNG Additionally, the procedure presented here can be used as a
compositions. The experimental densities for the multicom- relatively simple check against any of the other LNG
ponent mixtures were reported as a function of temperature density determination methods, whether by direct or
and pressure in an earlier paper.' The dielectric constant indirect mearis. It should be emphasized that, in most cases,
measurements that were performed at the same time were it is much easier to perform accurate and precise

not published along with the density data since, at the time measurements of dielectric constant than of density.
of the measurements and analysis of the density data, The prediction of LNG and LPG (liquefied petroleum
there were insufficient pure fluid dielectric constant data gas) densities and heating values from dielectric constant

0011-2275/83/080421-06 $03.00® 1983 Burterworth & Co (Publishers) Ltd.

CRYOGENICS .AUGUST 1983 421

235
measurements has been investigated by Miller and bons, the vapour composition was assumed to be pure
colleagues.""" However, their work for LNG" was based methane. For mixtures with nitrogen as a constituent, the
primarily on binary mixture data, along with two ternary vapour composition was estimated from partial-pressure
mixtures, and did not include pentanes. The present study calculations using available phase-equilibrium data for
probably includes the only experimental dielectric constant nitrogen-methane mixtures."'"
data that have been obtained for multicomponent mixtures The experimental dielectric constants, e, have been
simulating commercial LNG compositions. combined with the experuiiental densities, p, to calculate
values for the Qausius-Mcssotti function CM, from the
Experimental relation,

The experimental apparatus, procedures, and uncer-

tainties have been described in detail elsewhere.''*"'" The
dielectric constant measurements were performed using a
™ • (rfl)
? ")

concentric cylinder capacitor with a vacuum capacitance

The excess Clausius-Mossotti function (CM^ ), analogous to
of approximately 20 pF. The dielectric constant was
the definition of the excess volume of a liquid mixture, is
obtained from measurements of the ratio of the capacitance
given by the expression,
with liquid between the electrodes to the vacuum
capacitance. Capacitances were measured to a resolution of
ICP pF, wtoch is equivalent to the stability of the vacuum CM^ = CM -^xfiM,, (2)
capacitance. Based on reproducibility of binary mixture i

dielectric constant measurements'" with the same capacitor,

where CM, as defined above, refers to the Clausius-Mossotti
the uncertainty in the dielectric constant data for the
work fiinction of the liquid mixtw'e at a given temperature at
multicomponent mixtures investigated in the present
saturation pressure, x^ is the mole fraction of component i,
should be no larger than 0.05%.
Density data' were obtained simultaneously with the
and CMj is the Qausius-Mossotti function of component i
at the same temperature and pressure as the mixture. Along
dielectric constant results for the mixtures investigated
here. The devices for measuring density and dielectric
an isotherm, the CM function is a very slowly varying
function; therefore, adjustments of the pure component
constant were contained in a single sample cell; therefore,
CMf$ to the saturation pressure of the mixture are sufficiently
measurements of these properties were made on the same
small to neglect.
liquid samples. During the course of the mixture measure-
In the calculation of CM['s, the pure Quid orthobarlc
ments, data were obtained tor liquid methane, prior to
filling the ceil with each new mixture sample, as a control
Uquid densities were obtained from expressions given in
references 1, 5, and 6. The dielectric- constants for nitrogen
on the measurement process for determining density. The
'dielectric constant data for liquid methane (usually at
were taken from Ely and Straty" for methane from Straty
;

1 20 K), obtained along with the density data, agreed to and Goodwin'* for ethane from Weber^ and for propane,
; ;

better than 0.02% with the results of Straty and Goodwin.'* isobutane, and normal butane from Haynes and Younglove.^'

Vacuum capacitances were measured for each new filling

For the pentanes, no dielectric constant data could be found
in the literature for temperatures less than 223 K. There-
of the cell.
From an analysis of the density measurement tech- fore, dielectric constant values for the pentanes at tempera-

nique, it has been shown that the total systematic error in tures between 110 and 130 K have been estimated by a
'*'"'"
the density deter.nination is approximately 0.05%.' linear extrapolation of higher temperature results.

Combined with the 0.05% uncertainty in the dielectric The values used from
for saturated liquid isopentane ranged

constant, it is estimated that the maximum uncertainty in 2.138 at 1 10 K 130 K, while for normal pentane
to 2.105 at

the derived values of the Qausius-Mossotti function is 0.15%. the values varied from 2.123 at 1 10 K to 2.092 at 130 K.
These extrapolated values are estimated to be accurate to
better than ± 1%. An uncertainty of 1% in the dielectric
Results constant for either of the pentanes corresponds to a change
of less than 0.0008 cm^ moF' in the mole fraction average
The experimental orthobarlc Uquid dielectric constants
of the CMi'% (2 XfCMi) for mixture o, which contains the
of mixtures of LNG components are presented in Table 1
as a function of temperature (IPTS-68), pressure, and largest amount of the pentanes (approximately 0.2 mol% for
density. As noted in a preceding section, the vapour pressure each pentane) of the mixtures investigated here. A change
and density data that were obtamed simultaneously with of 8 X 10"* cm^ mof' represents less than 0.01% of the
the dielectric constant results have been discussed in detail.' mixture CM function.
The compositions of the mixtures were determined gravi- From the results presented in Table I several observa- .

metrically except those for mixtures m, n, and p. For these tions concemmg trends in the behaviour of the excess CM
mixtures the compositions were determined by analyses of function can be noted. For the measurements of the present
the prepared mixtures using a gas chromatograph with a study, the GVf^'s are generally small and positive and
thermal conductivity detector. decrease with increasing temperature. (For the same mix-
Small adjustments were made to the observed dielec- tures, the excess volumes were negative, increasing in magni-
tric constants to account for differences in compositions tude (more negative) with increasing temperature.') The
between the prepared gas mixtures and the liquid mixtures maximum observed excess CM
function for the mixtures of
condensed mto the equilibrium cell, which resulted from this study was approximately 0.25% of the mixture CM
vapour occupying a small part of the sample space. The fimction. Those mixtures that contam the largest fractions
maximum correction was less than 0.03%, which was for of the polar components, propane and isobutane, are charac-
mixture a at 1 25 K. Ail other corrections were of the order terized by the largest excess CM functions. This behaviour
of 0.01% or less. For mixtures containing only hydrocar- was expected based on dielecmc constant and density

422 CRYOGENICS AUGUST . 1983

236
Table1. Onhobaric liquid dielectric oonstanti e and Clautiui-Motsotti functioni CM of multioomponent mixture of LNG
componenti

r, K MPa p„p„ mol dm"^

3
e
3
CM, cm^ mol"^
1^31
CAr.cm^mol"'
10'(P«<Pt-Pc*c> I0^{p„ot-P»

a - 0.05931 Na + 0.89071 CH* + 0.04998 nC4H,o

(MW- 18.8562)

110.00 0.2400 25.3450 1.65978 7.1130 0.0068 - 0.095 - 0.058

115.00 0.3145 24.9440 1.64742 7.1160 0.0069 - 0.097 - 0.100
120.00 0.4082 24.5383 1.63497 7.1188 0.0071 -0.100 - 0.140
125.00 0.5196 24.1141 1.62203 7.1218 0.0069 -0.097 -0.181'

b - 0.86040 CH4 + 0.04600 CjHg + 0.04790 C3H8 + 0.04570 /C4H10

(MW' 19.9552)

115.00 0.1186 24.2654 1.70983 7.8852 0.0092 - 0.117 0.036

120.00 0.1710 23.9371 1.69759 7.8815 0.0043 - 0.055 0.082
125.00 0.2387 23.5868 1.68523 7.8832 0.0042 - 0.053 0.061^
130.00 0.3248 23.2331 1.67269 7.8836 0.0026 - 0.033 0.056^
135.00 0.4320 22.8637 1.65999 7.8870 0.0041 -0.052 0.013^

c - 0.85378 CH4 + 0.05178 CaHe + 0.04703 CaHg + 0.04741 /C4H10

(Af»V- 20.0838)

115.00 0.1191 24.2100 1.71285 7.9304 0.0115 - 0.145 0.005

120.00 0.1706 23.8779 1.70061 7.9288 0.0088 - 0.111 0.026
125.00 0.2379 23.5324 1.68823 7.9296 0.0078 -0.098 0.071^
- 0.069 9^
130.00 0.3238 23.1834 1.67570 7.9293 0.0055 0.01

d - 0J5133 CH4 + 0.05759 CjHg + 0.04808 CgHe + 0.04300 nC4Hio

(AfkV- 20.0092)

115.00 0.1180 24.3243 1.70951 7.8632 0.0090 - 0.114 - 0.009

120.00 0.1700 23.9965 1.69758 7.8619 0.0052 - 0.066 0.008
125.00 0.2374 23.6586 1.68553 7.8621 0.0024 - 0.031 0.006^
130.00 0.3232 23.3108 1.67330 7.8631 0.0003 - 0.004 - 0.007^
135.00 0.4301 22.9634 1.66120 7.8645 - 0.0012 0.015 - 0.025'
e - 0.84566 CH4 + 0.07924 CjHe + 0.05060 CaHg + 0.02450 nC4H 10
(MW' 19.6051)

115.00 0.1167 24.5569 1.70224 7.7241 0.0028 -0.036 0.077

120.00 "
0.1683 24.2126 1.69016 7.7244 0.0008 - 0.010 0.073
125.00 0^350 23.8698 1^7791 7.7219 - 0.0024 0.031 0.106'
130.00 0.3201 23.5204 1.66557 7.7198 - 0.0098 0.127 0.132'

f - 0.04801 Nj + 0.80940 CH4 + 0.04542 CjHe + 0.05050 C3HB + 0.04667 /C4H10

{ArtV- 20.6355)

115.00 0.3005 24.1487 1.69935 7.8284 0.0195 - 0.249 - 0.090

120.00 0.3863 23.8075 1.68708 7.8273 0.0172 - 0.220 - 0.075
125.00 0.4874 23.4518 1.67464 7.8285 0.0165 - 0.211 -0.092'
130.00 0.6125 23.0893 1.66152 7.8248 - 0.043'

g - 0.02628 + 0.81249 CH4 + 0.08484 CjHg + 0.04931 C3H8 + 0.02708 /7C4H10

(AfW- 20.0706)

115.00 0.2214 24.4562 1.69916 7.7283 0.0150 - 0.194 - 0.083

120.00 0.2874 24.1119 1.68701 7.7278 0.0120 -0.155 -0.077
125.00 0.3768 23.7507 1.67487 7.7322 0.0134 - 0.173 - 0.133'
130.00 0.4793 23.3954 1.66264 7.7331 -0.145'

h - 035892 CH4 + 0.11532 CaH, + 0.01341 C3H8 + 0.00530 /C4H10 + 0.00705 /7C4H10
(MW- 18.5565)

115.00 0.1185 25.0957 1.68091 7.3711 0.0006 - 0.008 0.028

120.00 0.1706 24.7131 1.66863 7.3749 0.0021 -0.028 - 0.022
125.00 0.2372 24.3294 1.65622 7.3771 0.0012 - 0.016 - 0.052'
130.00 0.3225 23.9490 1.64364 7.3760 - 0.0030 0.041 -0.037'

CRYOGENICS AUGUST . 1983 423

237
Table 1. continued

7, K P, MPa p„pt,moldm^ e CAf, cm-* mol ^

CW.cm^mol ^

Paxpt Pexpt

i - 0.84558 CH4 + 0.081 53 C2H6 + 0.04778 CaHg + 0.01259 /C4H10 + 0.01252 /JC4H10
(MW= 19.5838)

115.00 0.1166 24.5586 1.70334 7.7334 0.0094 -0.122 -0.000

120.00 0.1680 24.2180 1.69122 7.7323 0.0062 -0.080 0.014
125.00 0.2348 23.8688 1.67899 7.7322 0.0035 -0.045 0.014^
130.00 0.3188 23.5154 1.66656 7.7309 -0.0006 0.008 0.032^

j -0.00601 N2 + 0.90613 CH4 + 0.06026 CjHg +0.02154 CsHg + 0.00300 /•C4H10 + 0.00306 nC4Hio
(/WIV= 17.8195)

115.00 0.1478 25.3834 1.65987 7.1030 0.0057 -0.080 -0.025

120.00 0.2043 24.9894 1.64698 7.0991 -0.0005 0.007 0.031
125.00 0.2785 24.5702 1.63408 7.1013 -0.0011 0.016 -0.001^
130.00 0.3722 24.1578 1.62098 7.0989 0.033*

k - 0.00973 N2 + 0.88225 CH4 + 0.07259 C2H6 + 0.02561 CaHg + 0.00490 /C4Hio + 0.00492 nC4Hio
(MW= 18.3094)

115.00 0.1639 25.2023 1.67011 7.2448 0.0081 -0.112 0.155

120.00 0.2247 24.8047 1.65731 7.2456 0.0067 -0.093 0.144
125.00 0.3022 24.4022 1.64434 7.2455 0.0037 - 0.051 -0.055^

I - 0.01383 Nj + 0.85934 CH4 + 0.08477 CjHg + 0.02980 C3H8 + 0.00519 /C4H10 + 0.00707 /7C4H10
(MW= 18.7496)

115.00 0.1812 25.0384 1.67817 7.3638 0.0076 -0.103 -0.027

120.00 0.2441 24.6661 1.66553 7.3609 0.0024 -0.033 0.012
125.00 0.3223 24.2880 1.65268 7.3569 -0.0045 0.061 0.066^
130.00 0.4222 23.8981 1.63968 7.3542 0.103^

m- 0.85341 CH4 + 0.07898 C2H6 + 0.04729 C3H8 + 0.00854 /C4H10 + 0.00992 /7C4H 10
+ 0.00097 /CgHis + 0.00089 /7C5H 12
(MW= 19.3587)

110.00 0.0787 25.0063 1.71025 7.6552 0.0100 -0.131 0.013

115.00 0.1172 24.6566 1.69823 7.6572 0.0098 -0.128 -0.013
120.00 0.1686 24.3079 1.68611 7.6573 0.0078 -0.10? -0.014
125.00 0.2351 23.9525 1.67385 7.6576 0.0054 -0.071 -0.017^
130.00 0.3210 23.5883 1.66137 7.6578 0.0028 -0.037 -0.020^

n - 0.75442 CH4 + 0.15401 CjHe + 0.06950 C3H8 + 0.00978 /C4H10 + 0.01057 nC^H^Q
+ 0.00089 /C6H 12 +0.00083 ^CgH 12
(AfkV= 21.1060)

110.00 0.0723 24.2529 1.74897 8.2374 0.0193 -0.234 -0.053

115.00 0.1081 23.9619 1.73738 8.2338 0.0137 -0.166 -0.011
120.00 0.1549 23.6535 1.72575 8.2353 0.0131 -0.159 -0.028
125.00 0.215:^ 23.3351 1.71406 8.2390 0.0139 -0.169 -0.074^

o- 0.00859 Nj + 0.75713 CH4 + 0.13585 C2H6 +0.06742 C3H8 + 0.01336 /C4H 10

+ 0.01 326 nC4Hio + 0.00223 /C5H ,3 + 0.00216nC5Hi2
(MW= 21.3094)
110.00 0.1155 24.1809 1.74764 8.2502 0.0136 -0.165 0.015
115.00 0.1595 23.8731 1.73600 8.2521 0.0135 -0.164 -0.008
120.00 0.2155 23.5709 1.72406 8.2486 0.0080 -0.097 0.033
125.00 0.2873 23.2644 1.71227 8.2473 0.0038 -0.046 0.049^
130.00 0.3744 22.9514 1.70017 8.2447 0.082^

p - 0.00801 Nj + 0.74275 CH4 + 0.16505 CjHg + 0.06547 C3H8 + 0.00843 /C4H10

+ 0.00893 nC4H 10 + 0.00069 /C5H 12 + 0.00067 /JC5H12
(yVfkV= 21.0976)
110.00 0.1158 24.3141 1.74469 8.1790 , 0.0154 -0.188 -0.015
115.00 0.1584 24.0160 1.73307 8.1767 0.0109 -0.133 0.014
120.00 0.2093 23.6937 1.72140 8.1816 0.0137 -0.167 -0.046
125.00 0.2853 23.3804 1.70971 8.1826 0.0117 -0.143 -0.058'

424 CRYOGENICS AUGUST 1983

.

238
Tabia 1. oontinuad

K MPa dm -3 CW.cm^mol-^ ^° (P^ofPc^) 10 (P«.ot-PFiT)

T, P, p„ot, mol C/lf^,cm='mo|-^
^•xpt P«j<pt

q - 0.00599 Nj + 0.90068 CH4 + 0.06537 C2H6 + 0.02200 C3H8 + 0.00291 /C4H10

+ 0.00284 ^CaHio + O.OOOIO/C6H12 +0.00011 nCsHu
(MW= 17.9026)

115.00 0.1456 25.3600 1.66189 7.1274 0.0031 0.043 0.013

120.00 0.2024 24.9656 1.64901 7.1242 - 0.0024 0.034 0.058
125.00 0.2762 24.5450 1.63624 7.1286 -0.0008 0.012 .
-0.004^
130.00 0.3698 24.1289 1.62319 7.1284 -o.oor

AflV,molecular weight; T. temparatura (IPTS-68); P. preuure;p,^, experimsntal deniitv; CM^, exceu aausiuj-MoMorti function calculated
from eauation 2.p^^^^, density calculated from equation 3 using experimental CM{% from literature; Pfit. density calculated from aquation 3
using CM\ values for propane and isobutane determined from least squaras by fitting multicomponent mixture data.
^Data not used in fit to equation 3

measurements of binary mixtures of LNG components con- should be defined. The normal boiling point of LNG
taining either propane or isobutane.'"-'^ For those mixtures mixtures is approximately the same as that of pure methane
containing nitrogen, excess CM functions are not given for (112 K); thus, the temperature range for the caiculationai
temperatures above 125 K since the critical temperature of scheme was selected to be 1 10 to 120 K. The compositions
nitrogen approximately 126.2 K,
is of the mixtures from the present study coyer ranges that
Combining (1) and (2) and assuming that the CM encompass almost any that might be encountered in LNG
function of the mixture is a mole fraction average of the custody transfer applications. The extent to which the
pure component CM/'s, ie, CM^ = 0, then a density (Pcaic) caiculationai procedure can be used for compositions
of the mixture can be calculated from the relation, outside the range of the present data is not known.
Since the CM function varies slowly as a function of
temperature (or density) for a liquid along the saturation
Pcaic xfiMi (?)
e + 2 curve, the CM value for each of the pure components of
LNG is assumed to be constant frcm 1 10 to 1 20 K. The
Differences between experimental and calculated densities
CM values for the saturated liquid at 1 1 5 K are used for
each of the nonpolar components. Then, new pseudo values
from (3) are presented in the next to last column of Table 1.
of the CM function are determined for propane and isobu-
These differences are equivalent to the ratios of the excess
CM functions to the mixture CM functions (CM^/CM). In tane (given in Table 2 along with the CMfs for the nonpolar
components) by a least squares fit of (3) using the mixture
other words, assuming CM^ = 0, it is possible to predict the
data in Table 1 for temperatures between 110 and 120 K.
densities of the LNG-like mixtures investigated here within
These pseudo values for the CM function for propane and
0.25% from measurements of the dielectric constant, com-
isobutane are determined by constraining the CM function
position, and temperature of the liquid mixture. This would
for LNG mixtures to be equal to the mole fraction average
not meet the 0.1% uncertainty level desired in LNG custody
of the pure component CM values, ie, for CM^ = 0. The new
transfer applications. However, by making minor modifica-
pseudo values for the CM function determined from fitting
tions to the CMj's in (3), it is possible to reduce the density
the data are approximately I.l and 0.4% larger than
uncertainty to 'near the desired 0.1% goal, as discussed in
experimental CM values^' for propane and isobutane, res-
the next section.
pectively. The pseudo values are substantially smaller (1 .5
to 2.0%) than the infinite dilution values deduced by Pan
et al." from binary mixture data for methane with either
Caiculationai technique
propane or isobutane.
In the preceding section, it has been observed that the In the last column of Table 1 comparisons between
,

densities calculated from (3), using available experimental experimental densities (Pexpt) and values (pfit) calculated
CM data for the pure components, can differ by as much as from (3) with CW,'s from Table 2 are presented. At tem-
0.25% from the experimental densities for the LNG-like peratures from 110 to 120 K for the mixtures studied here,
mixtures investigated here. In general, those mixtures that all deviations between the experimental and calculated den-

exhibit the largest differences contain significant amounts sities, using the pseudo values for the CM function for

of the slightly polar components, propane and isobutane. propane and isobutane, are less than 0.1% except for mixture
Subsequently, an effort was initiated to develop a simple a at 120 K (0.14%) and mixture k at 115 K (0.15%) and at
mathematical procedure that could be used to predict LNG 120 K (0.14%). The deviation of greater than 0.1% for
densities to within an uncertainty of approximately 0.1% mixture a is likely the result of using a constant CM value
from an input of dielectric constant and composition. This for nitrogen over a 10 K range for a mixture that contams a
caiculationai technique would incorporate new pseudo relatively significant amount (6%) of nitrogen. For tempera-
values of the CM functions of propane and isobutane to tures between 1 10 and 120 K, the average deviation or bias
account for the relatively large CM^ values observed for between experimental and calculated densities is - 0.001%,
mixtures containing these constituents. while the average absolute deviation is 0.044%. Above 1 20 K,
The ranges of temperature and composition over at temperatures for which the data were not used in the fit,
which the caiculationai technique would be applicable the average absolute deviation was 0.058%. Maximum

CRYOGENICS AUGUST . 1983 425

^39
Table 2. ValuM of tha ClauHU»4Motsottj function for tion of the mixtures is gratefully acknowledged. This work
LNG componants to ba utad in (3) to caicuiata LNG was Bureau of Standards under
carried out at the National
dansitias the sponsorship of British Gas Corp., Chicago Bridge and Iron
Co., Columbia Gas Service Corp., Distrigas Corp., Easco Gas
w'**!^ em^
CM- Will mol"^
IIIUI LNG, Inc., El Paso Natural Gas, Gaz de France, Marathon Oil
Co., Mobil Oil Corp., Natural Gas Pipeline Co., Riillips Petro-
4.3984 leum Co., Shell International Gas Ltd., Sonatrach, Southern
Msthanfi 6.5490 California Gas Co., Tennessee Gas Pipeline Co., Texan
Ethane 11.063 Eastern Transmission Co., Tokyo Gas Co., Ltd., and Trans-
Propane (16.1612) continental Gas Pipe Line Corp., through a grand adminis-
Isobutane (21.2305) tered by the American Gas Association, Inc.
Normal butane 20.303
Isopentane 25.151
Normal pentane 24.868 References

Pwudo values for propane and Itobutana (In p«rwitt>atai vwereI 1 Kzz^HJ. Fluid Phase Equilibria 2 (\^l%)n
datarminad from lean iquarvs analysii by fitting tfta multicompo- 2 McCarty, RJ). Nat Bur Stand (US), Interagency Report
nant mixture dru of present rtudy. The CM functions for the other NBSIR 77-867 (1977)
componanu are experimental nturatad liquid v«lua* at 11 5 K 3 McCany, RJ). Nat Bui Stand (US), Technical Note 1030
(1980)
4 McCarty, RJ). / Oiem Thermodynamics 14 (1982) 837
deviation (0.18%) was for mixttue a at 125 K; this mixture
5 Haynes, W.M., Hiza, MJ., Fredenck, N.V. Rev Sci Insrrum
contains a relatively large amount of nitrogen, which has a 47 (1976) 1237
critical temperature of approximately 126.2 K. 6 Haynes, WJ*L, Hiza, MJ. / Chem Thermodynamics 9 (1977)
179
Conclusion 7 Hiza, MJ., Haynes, W.M., Paxriah, W.R. J Oiem Thermo-
dynamics 9 (1977) 873
8 Hiza, MJ., Hayne*, W.M. / Oiem Thermodynamics 12
Dielectric constant data are reported for multicom-
(1980) 1
ponent mixttues of LNG components at temperatures from
9 Haynes, W.M. /Oiem Thermodynamics 14 (1982) 603
1 10 to 130 K. Density results obtained simultaneously with 10 Haynes, W.M. Otthobaric liquid densities and dielectric
the dielectric constant data for these mixtures had been constants of (methane isobutane) and (methane * normal
published previously. These density and dielectric constant butane) at low temperatures, submitted to J Chem Thermo-

data have been used to develop a simple, alternative approach dynamics

11 Haynes, W.M. /?«v Sci /njrrum 48 (1977) 39
for determining densities of LNG mixtures to within an
12 Haynes, W.M. Apparatus for density and dielectric measure-
uncertainty of less than 0. 1 5% based on measurements of ments to 35 MPa on fluids of ayogenic interest, submitted
the dielectric constant and composition of the Liquid 10 J Res Nat Bur Stand (US)
mixture. Composition measurements are required in deter- 13 Pan, WJ., Mady, MJL, MiUet, tLCAIOiE J 21 (1975) 283
14 Thompson, Ji., R.T., Miller, R.C Advances in Cryogenic
mining LNG densities with any correlation or prediction
Engineering 25 (1980) 698
method. It is a relatively simple and straightforward task 15 Uo, CC Miller, R.C Cryogenics 21(1981 ) 85
to perform accurate and precise dielectric constant measure- 16 Straty, G.C, Goodwin, R.D. Cryogenics 13 (1973) 712
ments compared to direct density measurements. The 17 Patmh, W.R., Hiza, MJ. Advances in (Ilryogenic Engineering

extent to which the calculation can be exuapolated beyond 19 (1974) 300

18 fCidnay, AJ., Miller, R.C., Paniih, W.R., Hiza, MJ.
the composition ranges of the mixtures investigated here is
Cryogenics 15 (1975) 531
unknown. However, the compositions of the present study 19 Ely, J.F.. Stiaty, G.C / Chem Phys 61 (1973) 712
were selected to cover the ranges that would be 20 Weber, 1_A./ Oiem /^ys 65 (1976)446
encountered in most LNG custody transfer applications. 21 Haynes, W.M., Vounglove, B«A. Advances in Cryogenic
Engineering 27 (1982) 883
Authors 22 Btaziet, D.W., Freeman. G.R. Can J Chem 47 (1969) 893
23 MopsilcFJ./Oiem/^yj 50 (1969 ) 2559
The authors are at the Chemical Engineering Science 24 Moliton, A., Gerbier, J. Revue de Physique Appliquee
6 (1971) 273
Division, National Engineering Laboratory, National Bureau
25 Betinskii, B.A., Iknmov, Sh. Kh. Sovier Physics-Acousrtcs
of Standards, Boulder, Colorado 80303. Paper received 18 (1973) 300
2 March 1983. The contributions of MJ. Hiza in the prepara- 26 Dagg,I.R., Reesot,GX.Qjn//*>'j 52 (1974) 29

426 CRYOGENICS .AUGUST 1983

240
 ,

NBS-IMA (REV. 2-.ec)

U.S. DEPT. OF COMM. 1. PUBLICATION OR 2. Performing Organ. Report No. 3. Publication Date
REPORT NO.
BIBLIOGRAPHIC DATA
SHEET (See instructions) NBS Monogr. 172 October 1983
4. TITLE AND SUBTITLE
LIQUEFIED NATURAL GAS DENSITIES: SUMMARY OF RESEARCH PROGRAM AT THE NATIONAL BUREAU
OF STANDARDS

5. AUTHOR(S)
W. M. Haynes, R. D. McCarty and M. J. Hlza
6. PERFORMING ORGANIZATION (If joint or other tiian NBS. see /n struct/on sj 7. Contract/Grant No.

NATIONAL BUREAU OF STANDARDS

DEPARTMENT OF COMMERCE 8. Type of Report & Period Covered
WASHINGTON, D.C. 20234 Final

9. SPONSORING ORGANIZATION NAME AND COMPLETE ADDRESS (Street. City. Stote, ZIP)
British Gas Corp.; Chicago Bridge and Iron Co.; Columbia Gas Service Corp.; Dlstrigas
Corp.; Easco Gas LNG, Inc.; El Paso Natural Gas; Gaz de France; Marathon Oil Co.; Mobil
Oil Corp.; Natural Gas Pipeline Co.; Phillips Petroleum Co.; Shell International Gas,
Ltd.; Sonatrach; Southern California Gas Co.; Tennessee Gas Pipeline Co.; Texas
Eastern Transmission Co.; Tokyo Gas Co., Ltd.; Transcontinental Gas Pipe Line Corp.
10. SUPPLEMENTARY NOTES

Library of Congress Catalog Card Number: 83-600608

Document describes a computer program; SF-185, FIPS Software Summary, is attached.

11. ABSTRACT (A 200-word or less factual summary of most significant information.
If document includes a si gnifi cant
This report summarizes tiie results of a project concerning the densities of liquefied natural gas (LNG)
and its components. This project, initiated in the Properties of Fluids Section of the Cryogenics Division
of the National Bureau of Standards in July 1972, was carried out under the sponsorship of a consortium of
eighteen energy companies.
The experimental part of this project has included the following accomplishments: (a) development of a
magnetic suspension densimeter for absolute density measurements on liquids, including liquid mixtures in
equilibrium with their vapor, at temperatures from 90 to 300 K; b) orthobaric liquid density measurements
(

on the major components of LNG, which include nitrogen (95-120 K) , methane (105-160 K) , ethane (100-270 K)
propane (100-288.7 K) , isobutane (115-300 K) , and normal butane (135-300 K) ; (c) orthobaric liquid density
measurements on approximately thirty-five binary mixtures of the above components for all combinations
except nitrogen + butane systems, primarily in the temperature range of 105 to 130 K; and (d) orthobaric
liquid density measurements on twenty-seven mul t icomponent mixtures (105-120 K), including several LNG-like
mixtures with up to eight components. The total uncertainty of a single density measurement is approxi-
mately 0,1 percent at low temperatures and decreases to approximately 0,06 percent at rocm temperature. The
estimated standard deviation of a single density measurement is less than 0,02 percent.
The density data have been used to optimize, test, and compare several mathematical models as to their
suitability for the calculation of LNG densities for custody transfer. Models selected for optimization and
testing included an extended corresponding states method, a hard sphere model, a cell model, and an
empirical model due to Klosek and McKinley, The ultimate goal of this project was to produce one or more
mathematical models that could be used to predict the density of any LNG mixture to within an uncertainty of
0,1 percent from an input of pressure, temperature, and composition. After revisions based on the new
experimental data from this project, each of the models investigated here satisfy this goal for typical LNG
compositions. The limitations and ranges of validity of the various models are discussed. Also presented
are techniques for predicting LNG densities from dielectric constant measurements and from excess volume
ca leu at ions. I

The last section of this report consists of publications that provide a complete and detailed account of
the results of this project,
12. KEY WORDS (Six to twelve entries; alphabetical order; capitalize only proper names; and separate key words by semicolons)

Binary mixtures; density; experimental data; liquefied natural gas; magnetic

suspension densimeter; mult icomponent mixtures; prediction methods; pure fluids.

13. AVAILABILITY 14. NO. OF

PRINTED PAGES
[X] Unlimited
For Official Distribution. Do Not Release NTIS
to 241
I I

[X] Order From Superintendent of Documents, U.S. Government Printing Office, Washington. D.C.
20402. 15. Price

Order From National Technical Information Service (NTIS), Springfield, VA. 22161

USCOMM-DC 6043-P80
 .

NBS TECHNICAL PUBLICATIONS

PERIODICALS Building Science Series —
Disseminates technical information
developed at the Bureau on building materials, components,
JOURNAL OF RESEARCH— The Journal of Research of the systems, and whole structures. The series presents research results,
National Bureau of Standards reports NBS research and develop- lest methods, and performance criteria related to the structural and
ment in those disciplines of the physical and engineering sciences in environmental functions and the durability and safety charac-
which the Bureau is active. These include physics, chemistry, teristics of building elements and systems.
engineering, mathematics, and computer sciences. Papers cover a
broad range of subjects, with major emphasis on measurement Technical Notes — Studies or reports which are complete in them-
methodology and the basic technology underlying standardization. selves but restrictive in their treatment of a subject. Analogous to
Also included from time to time are survey articles on topics monographs but not so comprehensive in scope or definitive in

closely related to the Bureau's technical and scientific programs. treatment of the subject area. Often serve as a vehicle for final
As a special service to subscribers each issue contains complete reports of w ork performed at N BS under the sponsorship of other
citations to all recent Bureau publications in both NBS and non- government agencies.
NBS media. Issued six times a year. Annual subscription; domestic
$18; foreign $22.50. Single copy. $4.25 domestic; $5.35 foreign.
Voluntary Product Standards —
Developed under procedures
published by the Department of Commerce in Part 10, Title 15, of
the Code of Federal Regulations. The standards establish
NONPERIODICALS nationally recognized requirements for products, and provide all
concerned interests w ith a basis for common understanding of the
Monographs — Major contributions to the technical literature on characteristics of the products. NBS administers this program as a
various subjects related to the Bureau's scientific and technical ac- supplement to the activities of the private sector standardizing
tivities.
organizations.
Handbooks — Recommended codes of engineering and industrial
—
Consumer Information Series Practical information, based on
practice (including safety codes) developed in cooperation with in-
terested industries, professional organizations, and regulatory
N BS research and experience, covering areas of interest to the con-
sumer. Easily understandable language and illustrations provide
bodies.
useful background knowledge for shopping in today's tech-
Special Publications —
Include proceedings of conferences spon-
nological marketplace.
sored by NBS, NBS annual reports, and other special publications
appropriate to this grouping such as wall charts, pocket cards, and
Order the above NBS publications from: Superintendent of Docu-

bibliographies.
ments. Government Printing Office. H'ashingion. DC 20402.

Applied Mathematics Series — Mathematical tables, manuals, and

Order the following SBS publications — FIPS and NBSlR's—from
studies of special interest
the National Technical Information Services. Springfield, VA 22161
to physicists, engineers, chemists,
biologists, mathematicians, computer programmers, and others Federal Information Processing Standards Publications (FIPS
engaged in scientific and technical work.
PUB) — Publications in this series collectively constitute the
National Standard Reference Data Series —
Provides quantitative Federal Information Processing Standards Register. The Register
data on the physical and chemical properties of materials, com- Govern-
serves as the official source of information in the Federal
piled from the world's literature and critically evaluated. ment regarding standards issued by NBS pursuant to the Federal
Developed under a worldwide program coordinated by NBS under Property and Administrative Services Act of 1949 as amended.
the authority of the National Standard Data Act (Public Law Public Law 89-306 (79 Stat. 1127), and as implemented by Ex-
90-396). ecutive Order 1717 (38 FR 12315, dated May II, 1973) and Parl6
1

of Title 15 CFR (Code of Federal Regulations).

NOTE: The principal publication outlet for the foregoing data is NBS Interagency Reports (NBSIR) —A special series of interim or
the Journal of Physical and Chemical Reference Data (JPCRD) final work performed b\ NBS for outside sponsors
reports on
published quarterly for NBS by the American Chemical Society (both government and non-government). In general, initial dis-
(ACS) and the American Institute of Physics (AlP). Subscriptions, tributionis handled by the sponsor; public distribution is by the

reprints, and supplements available from ACS, 155 Sixteenth St.,

1 National Technical Information Services, Springfield, VA 22161,
NW, Washington, DC 20056. in paper copy or microfiche form.
U.S. Department of Commerce
National Bureau of Standards

Washington, D.C. 20234

Official Business
Penalty for Private Use $300

POSTAGE AND FEES PAID

U S DEPARTMENT OF COMMERCE
COM-215

SPECIAL FOURTH-CLASS RATE

BOOK