DEVELOPMENT

LOW-COST,LOW-WEIGHT CNG CYLINDER

Final Report
t

Prepared by

Mark E. Richards

Institute of Gas Technology

1700 South Mount Prospect Road
Des Piaines, Illinois 60018-1804

c (IGT Project Nos. 30691 and 61112)

for

U.S. Department of Energy

Office of Advanced Automotive Technologies
1000 Independence Avenue, SW
Washington,"DC 20585

Contract No. DE-FC02-97EE50460

DOE Project Manager

Mr. Lucito B. Cataquiz
General Engineer

and,

Gas Research Institute

, 8600 W. Bryn Mawr Avenue
Chicago, IL 60631.

Contract No. 5094-290-31 00

GRI Project Manager

William Liss
Team Leader, Energy Conversion

September 1999
 ,
'l%i~ rrport y?s prrpartd S .El 8CCOWt of W0r)C r p O n s O d by Ul8gC4lCy Of the
Uaitcd States Govnnmtnt Neither the United States Govtrprncat. aor aay agency
thereof, nor any of @ar cmployecs, makes my mmnty, express or implied, or
gssuxses h y kgal liability or rrtpoasibi&ty for 16e accuracy, cornpleterren, or use-
fulness of my infomation, apparatus product, or p m disdssed, or rtprcsents
that i!s use would not infringe privately owned rights. Refereact he& to my rpe-
CiTj commercial product, ptoccss, or Cexvict by trade aamt, trademark, aaanufac-
turcr, or otherwise does not necessarily constitute or imply its cadorsuncat, rrcom-
ma&tion. or favoring by the Uatted States oovwnmtat or my agency thereof.
The views and opinions of authors expressed. hatin do not a m s d y a t e or
refka those of the United States Govanmat or any agency thereof.
 f

DISCLAIMER

Portions of this document may be illegible

in electronic image products. Images are
produced from the best available original
documentt
 LEGAL NOTICE

This report was prepared by the Institute of Gas Technology as an account of work sponsored by Gas
Research Institute (GRI).Neither GRI, members of GRI, nor any person acting on behalf of either:

a. MAKES ANY WARRANTY OR REPRESENTATION, EXPRESS OR IMPLIED WITH

RESPECT TO THE ACCURACY, COMPLETENESS, OR USEFULNESS OF THE
INFORMATION CONTAINED IN THIS REPORT, OR THAT THE USE OF ANY
INFORMATION, APPARATUS,METHOD, OR PROCESS DISCLOSED IN THIS
REPORT MAY NOT INFRINGE PRIVATELY OWNED RIGHTS, OR

b. ASSUMES ANY LIABILITY WITH RESPECT TO THE USE OF, OR FOR ANY AND
ALL DAMAGES RESULTING FROM THE USE OF, ANY INFORMATION,
APPARATUS,METHOD, OR PROCESS DISCLOSED IN THIS REPORT.
Reference to trade names or specific commercial products, commodities, or services in this report does
not represent or constitute an endorsement, recommendation, or favoring by GRI or its contractors of the
specific commercial product, commodity, or service.
 Form Approved
REPORT DOCUMENTATION PAGE OMB NO.0704-0188

I September1999
I
I Final Report, 12/96-11/98
I
4. TITLE AND SUBTITLE 5. FUNDING NUMBERS

Low-Cost, Low-Weight CNG Cylinder Development DE-FC02-97EE50460(US. DOE)

5094-290-3 100 (GRI)

M. Richards, K.Melford, J. Wong, L. Gambone

7. PERFORMING ORGANfZATlON NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION

REPORT NUMBER
Institute of Gas Technology
1700 South Mount Prospect Road
Des Plaines, Illinois 60018-1804

0. SPONSORING/MONITORI" AGENCY NAME@) AND ADDRESSES(ES) 10. SPONSORINGlb!ONlTORNG

AGENCY REPORT NUMBER
U.S.Department of Energy Gas Research Institute
Office of Advanced Automotive Technologies 8600 West Bryn Mawr Avenue GRI-99/0211
lo00 IndependenceAvenue, SW Chicago, Illinois 60631
Washington, DC 20585
11. SUPPLEMENTARY NOTES

12a. DlSTRIBUTlONlAVAltABIIJWSTATEMENT 12b. DISTRIBUTIONCODE

13.ABSTRACl(Max/~~~)
This program was established to develop and commercialize new high strength steel-lined, composite hoopwrapped
compressed natural gas (CNG) cylinders for vehicular applications. As much as 70% of the cost of natural gas vehicles can be
related to on-board natural gas storage costs. The cost and weight targets for this program represent significant savings in each
characteristic when compared to comparable containers available at the initiation of the program.

The program objectives were to optimize specific weight and cost goals, yielding CNG cylinders with dimensions that should,
allowing for minor modifications, satisfy several vehicle market segments. The optimization process encompassed material,
design, and process improvements. In optimizing the CNG cylinder design, due consideration was given to safety aspects
relative to national, international, and vehicle manufacturer cylinder standards and requirements.

The report details the design and development effort, encompassing plant modifications, material selection, design issues,
1 .
tooling development, prototype development, and prototype testing. Extenuatingcircumstancesprevented the immediate
commercialization of the cylinder designs, though significant progress was made towards improving the cost and performance
of CNG cylinders. A new low-cost fiber was successfully employed while the weight target was met and the cost target was
missed by less than seven percent.

14. SUBJECT TERMS 15. NUMBER OF PAGES

Natural gas vehicle, compressed natural gas, container, cylinder 96
..
I
.P
18. PRICE CODE

17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. UMtTAnON OF ABSTRACT
OF REPORT OF THIS PAGE OF ABSTRACT
I I I
ISN 7540-01-280-5500 Standard F m 288 (Rev 2-89)
Prescribedby ANSI Sld 239-18
298-102
 SUMMARY
RESEARCH
Title Low-Cost, LowaWeight CNG Cylinder Development
Contractor Institute of G a s Technology
1700 South Mount Prospect Road
Des Plaines, Illinois 60018-1804
Principal M.E. Richards, K.Melford, J. Wong, L.Gambone
Investigators
Report Period December 1996 - November 1998
Objective This program was established to develop and commercializenew high strength
steel-lined, composite hoop-wrapped compressed natural gas (CNG) cylinders for
vehicular applications. The program objectives were to optimize specific weight
and cost goals, yielding two CNG cylinders with dimensions that should, allowing
for minor modifications, satisfy several vehicle market segments.
Technical A major obstacle confronting the widespread acceptance of natural gas vehicles
Perspective (NGV) is their substantial cost premium over conventionally fueled vehicles. As
much as 70 percent of the cost premium can be related to on-board fuel storage
costs. Market growth is dependent on making NGVs more affordable and storage
costs are the primary element. In 1996, the Gas Research Institute published a
report co-authored by the Institute of Gas Technology and Powertech Labs
concerning the market for and economics of fuel storage containers for natural gas
vehicles. The report identified and assessed the market potential of compressed
natural gas storage technologies and presented a number of cylinder optimization
options with the greatest potential to reduce cylinder cost and weight while
maintaining a high level of safety. This program was initiated as an outgrowth of
the recommendations of the 1996 GRI report.
i . Results Several production cylinders were fabricated and tested though extenuating
circumstances prevented the immediate commercializationof the designs.
Significantprogress was made towards improving the cost and performance of
CNG cylinders. A new lowcost fiber was successfully employed while the weight
target was met and the cost target was missed by less than seven percent.
Technical The design optimization process encompassed material (higher strength steel and
Approach chemically resistant composite reinforcing fibers), design (wall thickness
optimization and fabrication process selection), and process improvements (heat
treatment and filament winding operations). In optimizing the CNG cylinder
design, due consideration was given to safety aspects relative to national,
international, and vehicle manufacturer cylinder standards and requirements.
Project Although the project did not immediately lead to a commercial product, important
Implications progress was made towards improving the cost and performance of NGV cylinders.
The testing of chemically-resistantglass fibers will be of benefit to the industry in
general and the frequent discussions with OEMs will likely lead to their acceptance
of cylinders made with these fibers once they become widely available.
William Liss
Team Leader, Energy Conversion
Gas Research Institute
, ,

,
I . '
iv

i .
 TABLE
OF CONTENTS

Introduction ................................. ........ n ............................................................... 1

Objective........... ................ .......................................... ...........1

Plant ..........Modifications
......"..o ..........o
.........o
..o
..........................................................................n
... ..2
Material Selection and Issues ............................. n ...........
o
.n
...*.........o
n
..o. n 3
STEELLINER .............................................................................................................................................................. 3
COMPoSlTEREINFORCEMENT .................................................................................................................................... 3
C o s t Studies ........................................................................................................................................................... 4
Environmental Performance.................................................................................................................................. 6
Design Issues ........................................................................................................................................ 15
......................................................................................................................................................
BENCHMARKING 16
Tooling Development. ............................................................................................................................... 16

ProtoUype Development......................................................................................................................
" 17 ...
Prototype Testing ..................................................................................................................................... 17
h 4 B l E N T CYCLING ................................................................................................................................................... 18
BURST..................................................................................................................................................
HYDRAULIC 18
Other Considerations ........................ ..... n. ........................................................................................... 22

Conclusions .................................................... " ........................... 22

Appendix A: Lucas Aerospace CNG Cylinder 3600 psi 163 Inch Diameter Design Summary

Appendix B: Lucas Aerospace CNG Cylinder Fiber Reinforcement Study

.
i .

t
i.

b. .
V

O Q
 LISTOF FIGURES
. W E I Gvs
FIGURE 1 CNG CYLINDER ~ COST . ............................................................................................................... 2
. .
FIGURE2 COSTVS PRODUCT’IONVOJAME .... ............................................................................................................ 3
8..

FIGURE3.GLASSFIBER ACID TE~TRESULTS ................................................................................................................. 6

FIGURE 4.REF’RESENTATIONOF CYLINDER MARKINGS AND PRECONDmONING LocAnONS.......................................... 8
FIGURE 5 .SMPLEASTM CHP RATINGS...................................................................................................................... 9
FIGURE 7.AMBENT CYCLING -SURE PROFILE......................................................................................................... 9
FIGURE 8.HIGH-TEMPERATURE CYCLING PRESSURE PROW ..................................................................................... 10
FIGURE 9.HIGH-TEMPERATURE CYCL3NGTEMPERATURE PROFlLES............................................................................ 10
FIGURE10.LOW-TEMPERATURE CYCLING PRESSURE .ORP ....................................................................................11
FIGURE 12.LOW-TEMPERATURE CYCLING TEMPERATURE PROF~LES ........................................................................... 12
FIGURE 13.ACID APPLICATION AREA. CYLINDER A ..................................................................................................... 12
FIGURE 14.ACIDAPPLICATION AREA. CYLINDER B ..................................................................................................... 12
FIGURE 15.BURSTPRESSURE PROW. CYLINDER A ................................................................................................... 13
FIGURE 16.BURSTPRESSURE PROFILE.CYLINDER B ................................................................................................... 13
FIGURE 17.CYLINDER B AFTER BURST TEST............................................................................................................... 14
FIGURE 16.CYLINDER B.OPPOSITE SIDE ..................................................................................................................... 14
FIGURE 19.CYLINDER B.D ~ MENDE ........................................................................................................................... 15
FIGURE 18.PRonrrYPE CYLINDERS (F*G. H) ............................................................................................................. 18
FIGURE 19.AMBENTCYCLING PROFILE, EARLY TESTING ........................................................................................... 19
FIGURE 20.BURST PRESSURE PROFILE. CYLINDER F ................................................................................................... 19
FIGURE 21.BURSTPRESSURE PROFILE.CYLINDER G ................................................................................................... 20
FIGURE 22.BURSTPRESSURE PROFILE. CYLINDER H................................................................................................... 20
FIGURE 23.CYLINDER F AFTER BURSTTEST ............................................................................................................... 21
FIGURE 24. CYLINDER G AFTER BURSTTEST ..............................................................................................................21
FIGURE 25. CYLINDER H h ” E R BURSTTEST .............................................................................................................. 22

LISTOF TABLES
TABLE1.COST& WEIGHTGOALS ................................................................................................................................. 2
2.FIBERS
TABLE C O N S I D .......................................................................................................................................
~D 4
TABLE .
3 PRICE~QUANT~TY FOR THIOKOIJFORTAF~LPREP ~ w.......................................................................................
i 4
4.CARBON
TABLE PREPREG C~STCOMPARISONS .......................................................................................................... 5
TABLE .
5 NGVUGM DRAFT 6 HYBRID ENVIRONMENTAL TEST.................................................................................... 7
6. GRAVELOMETER
TABLE IMPACTRE~ULTS ................................................................................................................. 8
.
TABLE 8 COMPOSm THICKNESS AND ESTIMATED CYLJNDER SPECIFIC WEIGHTS ...................................................... 16
TABLE .
9 FINAL COST AND WEIGHT ESTIMATES ........................................................................................................... 22

i . .

z .

vi

.
 INTRODUCTION
A major obstacle confronting the widespread acceptance of natural gas vehicles (NGV) is their
substantial cost premium over conventionally fueled vehicles. As much as 70 percent of the cost premium
can be related to on-board fuel storage costs. Market growth is dependent on making NGVs more affordable
and storage costs hre the primary element. In 1996, the Gas Research Institute (GRI) published a report co-
authored by the Institute of Gas Technology (IGT) and Powertech Labs concerning the market for and
economics of fuel storage containers for natural gas vehicles.’ The report identified and assessed the market
potential of compressed natural gas storage technologies that would significantly reduce the cost of on-board
storage while also reducing weight. Based on weight, cost, and safety evaluations, the report presented a
number of cylinder optimization options with the greatest potential to reduce cylinder cost and weight while
maintaining a high level of safety. This report details a program involving Lucas Aerospace, IGT, and
Powertech Labs initiated as an outgrowth of the recommendationsof the 1996 GRI report.
OBJECTIVE
This project’s objective was to develop two new Type 2 cylinders made from high-strength steel liners,
hoop-wrapped with a composite fiber and resin matrix. Two sizes were to be developed: a 16-inchdiameter,
6 l-inch-long cylinder and a 16-inchdiameter,34-inch-long cylinder. The longer cylinder was suitable for
undercarriage mounting in a truck or van and in-bed mounting in pickup trucks, while the shorter cylinder
was better suited for mounting in the trunk area of a passenger car. Both cylinders were intended for 3,600
psig service.
Figure 1 shows how typical cylinders in the market at the initiation of the project performed relative to
weight and cost. Also shown are the request for proposal (W) target weight and cost for Type 1 and Type 2
cylinder designs under which this project was operating. Cylinder weight and cost (when expressed per
standard cubic foot (scf) of gas storage volume) are very dependent on the dimensions of the cylinder.
Smaller cylinders are naturally penalized because there is always some “fi~edcost” in any cylinder design,
and this fixed cost is more noticeable with small storage volumes. As the near-term expected market for NGV
trucks and vans is much larger than that for passenger cars, the longer cylinder was the primary focus of this
project. It is also generally much easier to make a shorter cylinder from an existing long cylinder design than
to do the reverse. Qualification (to NGV2 standards2)of the shorter cylinder is also simpler once the longer
cylinder is qualified.
Lucas recently manufactured steel-lined E-glass hoop-wrapped cylinders for the Ford Motor Company
(Ford). The weight and cost performance of these cylinders appear in Figure 1 as Current Designs. Because
some of these cylinders are somewhat small, they appear outside the area representing typical Type 2
designs. Two boxes also appear in Figure 1, each representing the proposed possible ranges of cost and
weight that each of the cylinder designs undertaken in this project could achieve. The boxes should be
interpreted as follows: it was predicted that the upper right comer of each box could be achieved with some
certainty. As one moves towards the lower left comer (lower weight and cost), there was more uncertainty
involved. The lower-left comer of each box represented approximately a 50 percent probability of success.
Table 1 lists the bounding values of the two boxes described above.

I Richards, M.E.,et al., Compressed Natural Gas Storage Optimizationfor Natural Gas Vehicles,Final Report to the
Gas Research Institute (GRI-96/0364),December 1996.
Basic Requirements for Compressed Natural Gas Vehicle (NGV)Fuel Containers, American National Standards
Institute (ANSYAGA NGV2-98). 1998.
e
 Figure 1. CNG Cylinder Weight vs. Cost

Table 1. Cost & Weight Goals

Design (lb./;cf)
Cost I
100%Probability 50%Probability
Weight Weight
($/scf) (lb./scf) ($/scf)
Cost I
16 x 61 0.19 0.40 0.17 0.32
16 x 34 0.23 0.54 I 0.21 0.48 I
The new cylinder designs continue to use (American Iron and Steel Institute) AIS1 4130x steel as it is
widely available in commercial quantities from a number of suppliers. Lucas has a great deal of experience
using 413Ox steels in deep draw processes for the manufacture of pressure vessels. Other alloys, such as
chrome-vanadium and others, would require an exorbitant amount of testing and in-vehicle service
experience to derive the same level of confidence in the final product as is exhibited by 4130x steels.
The cost goals detailed in Table 1 were based on an initial estimated production volume of 3,000 units
per year. Figure 2 shows the expected effects of changes in production volume on product cost, expressed
relative to base costs at 3,000 units per year. Full capacity for Lucas's facility, depending on the sizes of
cylinders produced, was approxiwtely 30,000units per year. At an annual production volume of 30,000
units per year, costs could decrease by about 15 percent, yielding a cost between $0.27 to $0.34 per scf for
the larger cylinder. At half capacity, cost for the larger cylinder falls between $0.28 to $0.36 per scf.
The program plan called for Lucas to manufacture 17 new cylinders for testing and evaluation during the
course of the project. The plan also included over 60 pieces of material released into production, with the
process development components being used for various trials and laboratory tests. Up to 10 additional
cylinders would be made available to original equipment manufacturers (OEM)for evaluation. The following
discussions are grouped by topic rather than chronologically.
P L A N T MODIFICATIONS
During the project, Lucas made extensive modifications to their heat treatment facility to accommodate
larger diameter cylinders of up to 16 inches in diameter and 72 inches in length. These modifications
included widening of the heat-treating furnace doors, relining of the chambers with thinner insulation, and
alterations to the process control logic.

2
 Figure 2. Cost vs. Production Volume

‘1105%
8
5 100%
sg 95%
u

0
t4-
% 90%
85%
crrE
80%
0 3 6 9 12 15 18 21 24 27 30
Annual Units (000s)

Lucas worked with hltrex (their winding machine supplier) to modify the fiber feed system to allow the
use of internal pull spools. This change was made to accommodate the use of Advantexo glass fibers
manufactured by Owens Coming. Several material handling issues were addressed, including the design of
the disc transportation pallet and the establishment of a handling requirement checklist.
MATERIAL SELECTION AND ISSUES
Steel Liner
United States NGV container standards have, until recently, placed limits on the strength of steel one
could use in a container. The intention of these limitations was to avoid possible sulfide stress cracking
(SSC) problems in service. Changes to the standards replaced the specific strength limits with SSC test
requirements. The strength level of the steel for the Lucas cylinders was determined by the steel’s ability to
pass the SSC test requirements of the NGV2 standard.
Lucas fabricated six 4130X steel test specimens in two compositions (melts) and three heat treatment
temperatures. One sample was maintained within NGV2-92 requirements (strength less than 140 hi). The
three heat treatment levels corresponded to nominal strengths of 140 ksi, 150 h i , and 170 h i . These
specimens were evaluated in regards to their fracture toughness and resistance to sulfide stress corrosion.
Initial results indicated that strength in the neighborhood of 150 ksi, allowing for process variation (at most f
10 hi), would pass the sulfide stress corrosion test. Further testing refined the target strength to a value of
148 ksi. Several steel samples heat-treated to this strength successfully completed SSC testing as well as
tensile and Charpy impact testing. Also, fracture performance tests were perforrhed on steel samples to meet
the requirements of the NGV2-98, clauses 18.13 & 18.4.
Lucas examined various changes to their requirements for steel composition and determined that in the
areas where improvements could be achieved, these improvements would not significantly affect either the
steel’s quality or its SSC resistance. A new Lucas specification (402-3084) was issued for the single-source
supply of plate and disc from British Steel.
Composite Reinforcement
The original intent of this project was to use carbon fiber as the basis of the new cylinder design’s
composite reinforcement. carbon fiber has been shown to be more resistant than fiberglass to chemical
P attack, although some concerns have been expressed about the impact resistance of carbon fiber-based

.%t
composites. Additionally, some OEMs favored carbon fiber-based cylinders. At the time the project was
proposed, it was believed future prices for recently introduced types of carbon fiber would allow for a cost
effectivedesign. Later discussions between the project team and OEMs revealed that the OEMs-General
Motors (GM) in particular-might be willing to use lowercost acid-resistant glass fibers instead of carbon
fibers. Fiber selection and liner development were highly interdependent as carbon and glass fibers yield
appreciably different composite thickness for the same loadcarrying ability. As the maximum outer diameter
of cylinder was limited, fiber selection would determine the outer diameter of the liner and tooling
parameters.
The project team decided to examine several types of fibers because potential customers were willing to
consider fiber systems other than carbon and there was some uncertainty concerning future carbon fiber
pricing. The availabilityand pricing for the six fiber types in Table 2 were investigated. Pricing information
was obtained for production levels of 3, 10, and 30 thousand cylinders per year. Vetrotex was used in Lucas’s
existing designs. Tow refers to the number of individual fiber strands that make up the fiber roving (e.g., 12k
tow consists of 12,000 individual fibers).
Table 2. Fibers Considered
Pe Supplier
Advantex glass (ECR) Owens Coming
Zentron glass (ECR) Owens-Coming
Vetrotex glass Vetrotex
12k-tow carbon Toray
Lowcost 48k-tow carbon Zoltek
Lowcost 5Ok-tow carbon Akzo Nobel
Two fiber application methods were also considered: wet winding, in which the fiber is passed through
an epoxy bath prior to application onto the metal liner; and pre-impregnated (prepreg) winding, in which the
epoxy and fibers are applied as a single entity as received from the manufacturer. Although prepreg material
is more costly than the combined material costs of fiber and resin in a wet winding system, processing
simplifications and improved yields using a prepreg system could yield net savings.
Cost Studies
The team evaluated the “as wrapped” costs to determine which fibers could meet the cost and
performance goals of the project. As mentioned above, the original thrust of the project was towards carbon
fiber. Team members met with representatives from Thiokol (now Cordant Technologies) to discuss the
possibility of using Thiokol’s prepreg carbon fiber technology. Thiokol quoted a range of prices (Table 3)
dependent on purchased quantity. The quotes were developed using Fortafil SOk-tow carbon fiber from Akzo
Nobel.
Table 3. PricdQuantity for ThiakovFortafil Prepreg
(Quod March 1997)

Thiokol also provided estimated costs and composite weights for a variety of prepreg systems based on a
12-inch cylinder gable 4). The costs were per cylinder and did not account for increases or decreases in
other processing costs or required plant modifications. There was little benefit to using large-tow fiber

4
compared to some traditional grades of carbon fiber. This was due to differences in physical properties of the
fibers and in lower realizable strengths of large-tow grades once applied to a liner (i.e., translation strength).
Table 4. Carbon Prepreg Cost Comparisons
Composite Cost
Composite Weight Relative to
Composite Type Designation per Cylinder (Ib.) Large-Tow Prepreg
Large Tow Carbon 3C-50k 16.5 -
Traditional Carbon T700-24k 10.7 $7.76
Traditional Carbon T700-12k 10.3 $14.09
Traditional Carbon G30-700 8.3 ($45.99)
Traditional Carbon M30S-18k 7.6 $56.37
Aramid Twaron 2200 13.2 ($51.67)
Traditional glass fibers were not considered from the outset due to their vulnerability to environmental
attack. ECR glass, a more chemically resistant form of commonly used E-glass, had been available on a
limited basis for some time, but its cost per pound was typically twice that of E-glass. During the project,
Owens Coming announced that a new form of ECR glass (dubbed Advantex) would soon be available.
Owens Coming’s plans included initial production at three facilities in Canada, Belgium, and France. The
new boron-free formulation came about in part because it minimizes air pollutants during the manufacturing
process, helping Owens Coming meet environmental regulations. It was claimed Advantex would have .
strength and rnodulus properties on par with E-glass and would be priced similarly.
Advantex had certain advantages in that it cost much less per pound than carbon fiber and it could be
used in Lucas’s existing production processes. Wet winding carbon fiber posed potential production
problems related to wet-out speed and carbon dust contamination of electrical equipment. Advantex’s cost
advantage was only slightly offset by its lower modulus (which requires a thicker wrap and reduces cylinder
capacity for a given envelope) and greater density (which increases weight). The cost advantage in terms of
dollars per scf was much more significant than the weight disadvantage in pounds per scf. Some
modifications to Lucas’s winding equipment were required as Advantex was packaged as an internal pull
spool compared to Lucas’s external pull setup.
Preliminary comparative cost figures showed the use of prepreg carbon fiber resulted in an increase in
cost of about $200 per cylinder compared to wet-wound carbon fiber. The latter was about $200 more per
cylinder than wet-wound Advantex glass fiber cylinder. These figures were for the 61-inch long cylinder and
did not include some plant and process modification costs. The use of prepreg Advantex was not
economically viable due to the relatively large amount glass needed and low cost of the Advantex fiber.
The cost analyses yielded two directions in which the project could go:
1. An Advantex-based composite using Lucas’s existing or slightly modified manufacturing processes
(wet winding).
2. A prepreg carbon fiber-based composite using a modified manufacturing process.
OEMs were willing to consider Advantex as an alternative to carbon fiber (assuming all qualification
tests could be passed) and that the cost analyses favored Advantex. Thus, it was decided that further
discussions with Thiokol regarding the prepreg carbon fiber approach would be suspended until the
environmental performance of Advantex could be gauged.
Some consideration was also given to Twaron, an aramid fiber produced by Akzo Nobel. Twaron is
similar to Kevlar, a DuPont product. Twaron was priced more competitively in Europe than in the United
States due to US.import tariffs. Initial estimates of system cost showed that a Twaron-based system would

5
be markedly less expensive than a carbon fiber-based system, though still more expensive than an
Advantex-based system.
Environmental PerJomtance
Initial environmental evaluations of Advantex fibers placed fabricated tensile samples under constant
load exposed to a room temperature acidic environment. These tests showed that Advantex glass fibers were
appreciably more resistant to acid attack than conventional glass fibers. Figure 3 shows the time to failure for
Advantex and Lucas's existing glass fiber (Vetrotex) under different loading conditions. One Advantex
specimen failed in the grip area, which is not considered a valid test. The letters D, W,M, and Y in Figure 3
indicate day, week, month, and year lengths of time.
Figure 3. Glass Fiber Acid Test Results

120

loo
e
n
& 80

6
8 6 0

56: 40
I
I
I
I I 1 I Ill
I I I I Ill
I Illlll
I I 1 1 1 1 1 1
I
I
I
I
I I I I Ill
I I I 1111
I llllll
I I11111
I
I
I
I I 1 1 1 1 1
I 1 1 1 1 1 1
1 1 1 1 1 1
I 1 1 1 1 1
1
1
I
I
I
I
I I11111
I I I 1 1 1 1
1 1 1 1 1 1 1
1 1 1 1 1 1 1
I
I
I
I I 1 1 1 1 1
I I I l l U
1 1 1 1 1 1 1
I I I11111
I 1 1 1 1 1 1 1 I I 1 1 1 1 1 1 I 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
I 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 I IIIIIII
20 I 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1
I I 1 1 1 1 1 1 I I I11111 1 1 I1111 1 I I I IIIII I I 1 1 1 1 1 1
I I Illlll I I I11111 I I I1111 1 I I I I1111 I I I 1 1 1 1 1
I I 1 1 1 1 1 1 I I I11111 I I 1 1 1 1 1 1 I I I I1111 I I Illlll
0
0.1 1.o 10.0
- 100.0 -- l-;iM0.0 10,boo.0
Time to Failure (hr)
Vetrotex A Advantex

Based on the improved acid resistance at room temperature of Advantex in comparison to conventional
E-glass, the team decided to forgo high-temperature acid resistance testing in favor of sub-scale prototype
environmental testing. Lucas fabricated three cylinders (12-inch diameter by 60-inch length) using existing
liners, epoxy, and processes-the only differences were the substitution of Advantex for Vetrotex fibers and
the omission of coatings. All the cylinders went through environmental testing: two through a hybrid of the
NGV2-98 and G M s Draft 6 environmental tests and one through the Canadian Standards Association (CSA)
B5 1-97 environmental test.'
The CSA B51-97 environmental test was performed at Powertech. The cylinder was pressurized to 1.25
times service pressure and held for 100 hours while subjected to battery acid. The residual burst pressure was
10,600 psig, far in excess of the 7,650 psig required by the test and typical of the burst pressures observed on
virgin cylinders. There was no evidence of damage to the Advantex fibers after the 100-hour acid exposure.
IGT discussed the future direction of Draft 6 with GM and received material from GM that indicated the
likely revisions to Draft 6. As portions of Draft 6 are likely to be changed to be more similar to NGV2, those
elements expected to change were performed to NGV2-98 rather than Draft 6. The test performed was a
hybrid of NGV2-98 and GM Draft 6 and the specifics appear in Table 5.

'High Pressure Cylindersfor the On-BoardStorage of Natural Gas as a Fuel for Automotive Vehicles,Canadian
Standards Association (CSA B5 1-97 Part 2). 1997.
6
 Table 5. NGVWGM Draft 6 Hybrid Environmental Test
ents I
Gravel impacts in the five “other fluid’, (OF)areas and three on
NGV2 and bottom. Pendulum impacts in the five OF areas, three on bottom in
GM Draft 6 cylindrical section. (Dome impacts were omitted as the cylinders are
Type 2). All preconditioning was performed at room temperature
rather than -40°F prescribed in Draft 6.
Two flaws were cut in the bottom area of the composite, one 0.030
inches deep by 8 inches long and the other 0.050 inches deep by one
I inch long.
GM Draft 6 will likely use NGV2 fluid specifications but its future
three times application method is not clear.
The cycle sequence of NGV2 will be used (7,500 cycles at ambient,
3,750 each at high and low temperatures) though the high portion will
be p e r f o d before the cold portion as it is believed that the high
temperature cycling is the most injurious. GM Draft 6 pressure limits
will be used (100 psi lower level rather than 10%of service pressure).
GM Draft 6 will likely change its cycle sequence to match that of
NGV2.
Temperature will be measured on the cylinder skin (composite) and
controlled to GM Draft 6 specifications: 190°F. 3,750 cycles will be
performed in one step as per the discussion above.
Low Temperature GM Draft 6 Temperature will be measured on the cylinder skin (composite)and
Cycling controlled to GM Draft 6 specifications: -60°F. 3,750 cycles will be
performed in one step as per the discussion above. The upper pressure
limit will be 80% of service pressure as GM Draft 6 will likely adopt

=--r
this specification (rather than 125%of service pressure).
GM Draft 6 No leak test will be performed (a GM Draft 6 requirement) as this is
directed at Type 4 cylinders, Type 2 cylinders that survive cycling
should not leak.
Both cylinders were labeled according to Section 18.4.1 of NGV2-98. Figure 4 shows how the cylinders
were marked. The thick line indicates the mid-line of the cylinder while the dashed line indicates the fill line
when the cylinder is immersed in the saltwater solution. The dashed circles do not need to be drawn on the
cylinders.
Preconditioning
Gravel impacts were performed on the five “other fluid” (OF)areas and three on the bottom. Pendulum
impacts were performed on the five OF areas and on five areas on the bottom of the section. All
preconditioning was performed at room temperature rather than -40°F prescribed in Draft 6.Two flaws were
cut in the bottom area of the composite, one 0.030inches deep by 8 inches long and the other 0.050 inches
deep by one inch long. The drops specified by Draft 6 were also omitted as they are primarily directed at
Type 4 containers.

7
 Upon completion of the gravelometer impact tests, each impact area was evaluated and assigned an
ASTM chip rating.' Table 6 lists the impact areas for each cylinder and their correspondingchip ratings.
Figure 5 shows sample ASTM chip ratings.
Figure 4. Representation of Cylinder Markings and Preconditioning Locations
(P = Pendulum, G = Gravelometer)
H2S04 P M e O H I G a n NaOH
- n p '

.--, .--, / / /
e--,

P1,Gl) P2,GZ) P3,G3\

#- - 0

Table 6. Gravelometer Impact Results

Area Cylinder A Cylinder B
PlGl 4B 4c
P2G2 4B 4c
P3G3 4B 3c
P6G4 5B 3c
P7G5 4B 3c
P8G6 4B 5c
P9G7 4c 3c
PlOG8 4B 3c
Fluid Application
It was expected that Draft 6 would use NGV2 fluid specifications, though the exact application method
was still unknown. Therefore the application method and fluid compositions specified in NGV2 were used.
Cycle Testing
The cycle sequence of NGV2 was used: 7,500 cycles at ambient, 3,750 each at low and high
temperatures. Draft 6 was expected to change to match the cycle sequence of NGV2. Because it was believed
that the high temperature cycling is the most injurious the high portion was performedbefore the low portion.
Draft 6 pressure limits were used (100 psig at the low lower level rather than 10% of service pressure). Draft
6 also specified an exact cycle profile of 15 seconds to pressurize, 60 seconds held at pressure, and 15
seconds to depressurize. This pressure profile was not strictly adhered to because it was believed that the
NGV2 requirements were sufficient and the Draft 6 requirements were neither more nor less severe. To
expedite testing, both cylinders were cycle tested simultaneously.

i
Standard Test Method for Chipping Resistance of Coatings, American Society for Testing and Materials
: (D317&87(1996)el), 1997.
I
8
t
::
e,*
 Figure 5. Sample ASTM Chip Ratings
5B

0 e
0 0 a

4c
I . a
e
8
0 0 0 tr, 4:,

Figure 6 shows the typical pressure profile versus time during the ambient cycle test segment.
Figure 6. Ambient Cycling Pressure Profile

lime (minutes)

9
 Inspection of the cylinders after the ambient cycling portion of the test did not reveal any signs of stress
corrosion cracking, composite discoloration, delamination, loose fibers, or cylinder bulging.
High Temperature Qcling
Temperature was measured on the cylinder skin (composite) and controlled to GM Draft 6 specifications
of 190°F. A buffer volume of hydraulic fluid was used to help maintain cylinder surface temperature. Figure
7 shows the typical pressure profile versus time graph during the high-temperature cycle test segment.
Figure 7. High-Temperature Cycling Pressure Profile

5,000
4,500
4,000
-
8 3,500
I 3,000
2,500
t 2,000
n 1,500
1.000
500
0
0 1 2 3 4 5 6 7 8
Time (minutes)

Figure 8 shows a sample of the temperature traces during the high-temperature cycling segment. The time
axis is time of day, not test duration. The trace of hydraulic fluid manifold temperature was measured before
the fluid buffer, which exhibited a much lower temperature than the cylinder surfaces. The variation in this
trace was due to flowing “cold” fluid from the pump on pressurization and “hot” fluid from the downstream
system on depressurization. The temperature of the surrounding environment was only slightly higher than
the cylinder surfaces.

e
200
190
180
170
Figure 8. High-Temperature Cycling Temperature Profiles

---------
________________________________________-----_-------__-__
-

________________________________________-----------------__--
9
--
A

---
9
*

-
1393 1396 13:09 13:12 13:14 13:17 1320 1323 1326 1329
Time (hh:mm)
 Upon completion of the high-temperature cycles, inspection of the cylinders did not reveal any signs of
+ - stress corrosion cracking, delamination, loose fibers, or cylinder bulging.
Low-Temperature Cycling
Temperature was measured on the cylinder skin (composite) and controlled to GM Draft 6 specifications
of -60°F. A buffer volume of hydraulic fluid was used to help maintain cylinder surface temperature. The
upper pressure limit was 80%of service pressure as it was expected that Draft 6 would likely adopt this
specification (rather than 125% of service pressure). No leak-related interruptions were experienced.
Cycling at -60°F presented a particular challenge as that temperature was very close to the pour point of
the hydraulic fluid used. Even with the buffer volume and the surroundingenvironment in the neighborhood
of -lOO°F, it was not possible to cycle indefinitely. Hydraulic fluid heating due to flow friction and heat of
compression would gradually raise the temperature ofthe cylinder surfaces. Cycling was halted until such
time that the cylinder surfaces were sufficiently below -60°F to continue.
Figure 9 shows the typical pressure profile versus time graph during the low-temperature cycle test
segment.
Figure 9. Low-Temperature Cycling Pressure Profile
3,500 1 I I I I
I I I I I . I I
I I I I I I I
3,000

-
'ii; 2,500

=e!
n
2.000

$ 1,500
g!
Q 1,000

500

0
0 1 2 3 4 5 6 7 8
Time (minutes)

Figure 10 shows a sample of the temperature traces during the low-temperaturecycling segment. The
three Cham Tx traces are of thermocouples within the chamber. These temperatures typically varied between
-70°F and -100 OF. The uppermost trace (Man TZ)is the hydraulic fluid temperature before the fluid buffer.
The oscillation in temperature was due to a combination of warm fluid coming in from the pump, fluid
heating due to friction and compression, and subsequent cooling during the one-minute holds and
depressurization.Man 72 is the hydraulic fluid temperature after, or on the cylinder side of,the fluid buffer.
Note the gradual rise in Man 72 over time. The non-steady-state characteristics shown here are typical of
cold-temperaturecycling.
At the completion of the low temperature cycling phase, inspection of the cylinders did not reveal any
signs of stress corrosion cracking, delamination, loose fibers, or cylinder bulging. Figure 11 and Figure 12
show the acid application areas of each cylinder at the completion of the cycle segments. Some hairline
cracking and minor discoloration were evident, but these conditions are typically benign.

11
 Figure 10. Low-Temperature Cycling Temperature Profiles

Time (hh:mm)

--cCylA-E-CylB+ManTl ++ManT2+ChamTl +ChamT2+ChamT3

Figure 11. Acid Application Area, Cylinder A Figure 12. Acid Application Area, Cylinder B

No leak test was performed (Draft 6 requirement) as this is aimed at Type 4 cylinders. Type 2 cylinders
that survive cycling should not leak. Draft 6 required that cylinders hold two times service pressure rather
than the 1.8 times service pressure required by NGV2. Cylinder A was pressurized to slightly over two times
service pressure and held there for approximately 20 seconds. The cylinder did not fail at two times service
pressure and remains available €or inspection. Figure 13 shows the pressure profile for Cylinder A. Cylinder
B was pressurized to failure. Figure 14 shows the burst pressure profile for Cylinder B. The maximum
recorded pressure in the test of Cylinder B was 9,865 psig, well in excess of the 7,200 psig requirement.

12
 Figure 13. Burst Pressure Profile, Cylinder A
8

7
/I I I
I

E 5

E 4

f 3

0
0 100 200 300 400 500 600 700
Tlme (seconds)

Figure 14. Burst Pressure Profile, Cylinder B

1 1 I

I I I I I
c------*------+------+------+--------
I I I I I
1 I I I I
I I I I I
I I I I I
1 I I I I

0 200 400 600 300 1000 1200 1 0

Figure 15 through Figure 17 show Cylin letion of the burst test. It appeared as though
the compositefailed in the area of the 8-inch by 0.030-inch induced flaw (Figure 16).There was also some
composite separationat the location of the 1-inch by 0.050-inch induced flaw (Figure 17) but it was not as
dramatic.

13
 Figure 15. Cylinder B After Burst Test

i 1

Figure 16. Cylinder B,Opposite Side

14
 Figure 17. Cylinder B, Dome End

Environmental Testing Conclusions

According to Lucas, the burst strength of Cylinder B was nearly that of a virgin cylinder-no more than
10% lower. It is highly probable that the acid, as applied in this test, did not affect the composite in an
appreciable manner. The degradation in burst strength could be attributable to the number of pressure cycles
endured during the test. It could also be attributable to the cut induced in the cylinder at the start of the test.
An informal inspection indicated that the failure might have initiated at this location. It is important to keep
in mind that the cylinders tested were uncoated. Traditional E-glass cylinders, if uncoated, could be expected
to fail either NGV2-98 or Draft 6 environmental tests.
Lucas and IGT made separate presentations to GM/Impco and to Ford at their facilities regarding the
environmentaltesting of the Advantex-wrapped prototype cylinders. The test results were well received,
though GM would still require production cykinders to pass Draft 6. The interim test report written by IGT
was distributed to both organizations.
During the above meeting GM explained that they were reviewing Draft 6 to bring it more in line with
NGVZ and that they may require a1 cylinders for testing to both the existing and future versions of
Draft 6.
DESIGNISSUES
Starting stock plate thickness was established early in the project as it affected tooling design, the longest
lead-time item. A review of end geometry using an elliptical design indicated no major improvement to the
target project requirements. Design calculations for cylinder geometry were reviewed to establish the needed
plate thickness. The press used for the first cupping operation fixed the maximum initial disc diameter. A
larger diameter disc allowed some reduction in initial plate thickness and reduced the volume of the metal in
the dome end of the cylinder. However, this did not markedly affect the target project requirements. Also, the

15
existing Lucas cylinder design report for their existing line of Ford cylinders was reviewed to determine the
necessary changes to optimize the design.
Lucas Visited with potential customers to discuss targets for cylinder size, weight, and cost. As a result of
these discussions, it was decided that the cylinders would have an outer diameter of 16.3 inches and lengths
of 61 and 34 inches. In discussions regarding the impact of using Advantex versus carbon fiber composites,
potential customers indicated that a slightly undersized cylinder could be accommodated whereas a slightly
oversized cylinder would be undesirable. This topic was addressed with potential customers because a liner
designed for a fiberglass composite would be smaller than one designed for a carbon fiber composite. Once a
final decision was made regarding liner diameter and tooling development commenced, significant changes
in liner diameter would result in costly and time consuming tooling redesign.
A finite element model was prepared and stress analyses were completed based on a process minimum
steel strength of 140 ksi (150 i 10 ksi). The design liner wall thickness was 0.250 inches for all fiber types.
Three composite systems were examined-Table 7 lists the thickness of each system's composite wrap and
an estimate of the specific weight of the resulting 16.3-inch diameter, 61-inch long cylinder. The design
report appears in Appendix A.
Table 7. Composite Thickness and Estimated Cylinder Specific Weights
Composite Thickness Specific Weight
Composite System (inches) (pounds per scf)
Advantex/Epoxy 0.250 0.173
~oo/EpoXY 0.150 0.1m
Twaron/Epoxy 0.225 0.163
Process tolerances for the liner were determined and led to reduction in wall thickness variability by
approximately 50 percent compared to present Lucas production. The design wall variation at the cylinder's
mid-length was fixed at 0.014 inches. Results from the units manufactured indicated a process variation of
0.012 inches was possible. Process drawings for the liner, composite overwrap, and label stages were issued.
Benchmarking
Lucas benchmarked competitors' manufacturing processes to support design and manufacturing
decisions. The benchmarking effort included a comparison between Lucas's existing deep drawing and
ironing processes and flow forming cylinders direct from hot drawn tube. It was found that the most cost
effective.and least technologically risky manufacturing method should remain deep drawing and ironing from
steel plate stock. The benchmark effort also led to the development of a cost model that was used to derive
the product cost and selling price for the project targets in terms of dollars per scf.
Lucas examined the feasibility of adding a flow forming process to better control wall thickness. As
Lucas did not have flow forming equipment of the proper size in-house, the advantages gained from the flow
forming step would be outweighed by the increased processing cost, time, and transportation cost. Also, the
potential uncertainty of flow former availability from a third party was problematic.
TOOLING
DEVELOPMENT
The tooling concept design was c leted, which led to tool design and tool stress analysis. The draw
reduction ratios were computed, enabling starting stock and draw concepts for the liner to be calculated.
Lucas's first cup subcontractor (Royal Ordnance) assisted in the development of the early cupping stage (tool
design, handling systems modification, and production development). Initial tooling design placed the
maximum material on both the punch and die. During tooling and prototype development, material was
removed from the die to achieve the desired container wall thickness.

16
 A tothl of twelve discs were plasma cut, spheroid&, and phosphated in preparation for Cup 1
production at Royal Ordnance. This was the Tool Try Out (TTO)batch from which the three cylinders for
long term testing were produced. The twelve initial discs were successfully advanced to the Cup 2 stage at
Royal Ordnance. One unit was retained at Royal Ordnance for mechan a1 handling trials and the remaining
eleven units were returned to Lucas for further processing.
The increased diameter hot-spin heating coil was installed. Lucas moved its ultrasonic testing to after the
external shot blast process step. This change yielded more consistent results. The test piece used to calibrate
the ultrasonic testing equipment had a known defect 0.012 inches deep. This was equivalent to only 4.5
percent of the liner wall thickness and an improvement on the BS504S requirement of 5 percent. Machining
was brought partially in-house and it was expected that the majority of machining would be performed
in-house. The outside contractor was retained to provide backup machining capacity.
PROTOTYPE
DEVELOPMENT
A dummy unit, simulating the new liner design, was manufactured and the unit successfully passed
through the proposed production operations.
Following the completion of tooling development, eleven initial units successfully completed the Draw 3
operation. The Draw 4 punch was installed and these eleven units completed the Draw 4 operation. Two units
were used to obtain the hot spin parameters and the remaining nine units passed through the hot spin, internal
clean, and hardedtemper processes. Hardness tests were performed at three places along the liners. The
results from the nine units showed a hardness range of HE3 293 to 300.This is equivalent to an ultimate
tensile strength (UTS)range of 141 to 158h i . The target tolerance range was 140to 160ksi. Test samples
taken from a h e r after heat treatment showed a 151 ksi UTS.Three impact test pieces were also taken from
the sample and yielded Charpy impact values of 105,110, and 115 J/cm2.
After neck machining, the nine units were internally cleaned, grit blasted externally, ultrasonically tested,
weighed, dimensionally inspected, and primer coated prior to filament winding. The liner dimensional
inspection indicated a wall thickness towards the top end of the required range. The average wall thickness
mid-way along the liners was 0.275 inches. The filament winding program was confirmed by dry winding
and the nine units were successfully filament wound and cured. Because the steel liner was towards the top
end of the required range, the thickness of the overwrap was targeted to the minimum requirement (0.250
inches). The nine cylinders were autofrettaged, weighed, dimensionally inspected, and given a 0.001-inch top
coat of polyurethane varnish. These nine cylinders were prepared for shipping to IGT.
One liner used for obtaining the hot spin parameters was heat treated and sent for neck machining prior
to its use as a liner burst sample.
Twenty-nine additional discs were pressed into Cup 1at Royal Ordnance. This, together with the Cup 2
unit previously retained at Royal Ordnance, provided sufficient units to complete the contract. Thirty units
were progressed to Cup 2 should further cylinders
TESTING
PROTOTYPE
Due to Lucas’s decision to discontinue NGV cylinder production (see Other Considerations), the
planned qualification testing program was scaled back. It was felt that the prospects were slight for a sale of
the entire plant. If, for instance, another company acquired only Lucas’s design to produce with their own
equipment and tooling, it would be necessary to qualify cylinders manufactured on that equipment. Thus it
was establishedjointly between Lucas, IGT, and GRI that six cylinders would be used for evaluation and
testing at IGT. Three cylinders were subjected to ambient cycling (section 18.3 of NGV2-98) and three others
to hydrostatic burst testing (section 18.5 of NGV2-98).

Tmnsportable Gas Containers, British Standards Institution (BS5045, Part 1). 1982.

17 0
 Ambient Cycling
The NGV2-98 ambient cycling test requires that finished containers be pressure cycled at ambient
temperature to failure or 45,000 cycles. The pressure range for cycling is from 12.5 to 125 percent of service
pressure. The cycling rate is limited to no more than ten cycles per Pninute. The containers shall not fail
before reaching a number of cycles equal to 750 times the design life of the containers in years (e.g., 15,000
cycles for a container with a 20-year design life). After this point, the containers may fail by leakage but the
composite fibers are not allowed to fail. To expedite testing, all three cylinders were cycle tested
simultaneously (Figure 18).
Figure 18. Prototype Cylinders (F, G, H)

The cycling rate was rather slow due to the large size of the containers and that three were being cycled
simultaneously. Figure 19 shows the typical pressure profile versus time approximately 2,000.cycles into the
test.
All three cylinders successfully completed 15,OOO cycles. At approximately 24,520 cycles, one of the
cylinders developed a leak under the composite approximately one foot from the edge of the composite wrap.
The composite structure was intact. A second cylinder failed in a similar fashion at approximately 30,690
cycles. The remaining cylinder successfully completed 45,000 cycles. All three cylinders met the
requirements of the ambient cycling test of NGV2-98.
Hydraulic Burst
The hydraulic burst test in NGV2-98 requires that containers be pressurized to failure at a rate not
exceeding 200 psi per second at pressures below SO percent of minimum required burst pressure, and not
exceeding 50 psi per second at pressures above. For Type 2 cylinders, a liner is also burst and must achieve
125 percent of service pressure (4,500 psig for a 3,600 psig service pressure cylinder).
The burst pressure for the liner was 5,829 psig, If the minimum UTS were to be encountered, a burst
pressure of 5,440 psig would be expected (5,829 x 140 / 150). If, in addition, the minimum wall thickness
were encountered, a burst pressure of 4,945 psig would be expected (5,440 x 0.25 / 0.275). Therefore, the
margin of safety at minimum UTS and minimum liner wall thickness is 1.1 (4,945 / 4,500).

18

c s
 Figure 19. Ambient Cycling Profile, Early Testing
t i B
5,000
4,500
4,000
-
6 3,500
m
a 3,000
U

g 2,500
2,000
n. 1,500
1,000
500
0
0 1 2 3 4 5 6 7 8
Time (minutes)

Three cylinders were pressurized to failure-failure here meaning failure of the composite. Typically, the
composite of a Type 2 cylinder will rupture, causing a slight increase in the volume of the liner that results in
a sudden and large pressure drop (assuming that the pumping equipment does not have so much flow capacity
to very quickly catch up). Once this pressure drop is observed, the test is halted. The liner is usually intact at
this point. If pressurization were to continue, the maximum pressure realized after the initial composite
rupture will not be greater than the pressure realized at the time of composite failure.
Figure 20 shows the pressure profile for Cylinder F, which achieved a maximum recorded pressure of
9,435 psig. The slight dip just past 2,000 seconds was attributable to the test equipment, not any change in
the cylinder. Figure 21 shows the same profile for Cylinder G,which achieved a maximum recorded pressure
of 9,206 psig. Finally, the profile for Cylinder H appears in Figure 22. Cylinder H reached a maximum
recorded pressure of 9,430 psig. Minimum burst pressure for this design was 9,000 psig, thus all cylinders
met the test requirements.
Figure 20. Burst Pressure Profile, Cylinder F

0 lo00 1500 2000 2500 3000

Tlme (seconds)

19
 Figure 21. Burst Pressure Profile, Cylinder G
10
I I I I I I I
Q
8

0
0 500 1000 1500 2000 2500 3000
Time (seconds)

0 so0 1000 1500 2000 2500 3000 3500

T h e (seconds)

Figure 23 through Figure 25 show Cylinders F, G, and H r completion of the burst tests.
 Figure 23. Cylinder F After'Burst Test

I
Figure 24. Cylinder G After Burst Test
 Figure 25. Cylinder H After Burst Test

OTHER
CONSIDERATIONS
F On May 22,1998, Lucas informed their workforce that the Group had decided to withdraw from both the
industrial gas and NGV cylinder markets and close that area of the business situated in Burnley. IGT (and
others) were informed that Lucas would complete a number of cylinders for this contract but would not take
the project forward into commercial production.
A conference call took place on June 4,1998. Representativesfrom Lucas, IGT, GRI, and DOE
participated. The purpose of the conference call was to discuss Lucas’s decision to exit the NGV and
industrial gas cylinder markets and to discuss possible avenues of commercialization for the cylinders
developed during this project. At the completion of the project, Lucas held the rights to the design. The
disposition of the plant and tooling is not known at this time, though some of the equipment may not be
completely unrestricted with respect to its sale or transfer. Lucas has been discBssing the transfer of the
design and some equipmentholing with interested parties though the talks have been preliminary in nature.
GRI and IGT developed a list of potential container manufacturers who may be interested in acquiring
Lucas*sdesign and equipment. The list was forwarded to Lucas.
CONCLUSIONS
Although the project did not immediately lead to a commercial product, important progress was made
towards improving the cost and performance of NGV cylinders. The testing of Advantex glass fibers will be
of benefit to the industry in general and the frequent discussions with OEMs will likely lead to their
acceptance of cylinders made with Advantex once they become available.
Based on the nine finished cylinders produced (16.3-inch diameter, 61-inch length), the final weight was
0.18 pounds per scf which met the target of the RFP and exceeded the project goal. The expected cost was
somewhat above the RFP target. The expected “factory gate” cost (excluding packing, transportation, and
U.S.import duty costs) was $0.45 per scf. /
Table 8. Final Cost and Weight Estimates
RFP Target Project Goal Final (Expected) Value
Weight (Ib/scf) 0.18 0.19 0.18
cost ($/scf) 0.46 0.40 0.49

22
 Appendix A
Lucas Aerospace CNG Cylinder 3600 psi 163 Inch Diameter Design Summary
 Powertech G -
Powertech Labs Inc. 12388 88th Avenue, Surrey, B.C. Canada V3W 7R7
I ,

POWERTECH LABS INC.

FINAL REPORT

LUCAS AEROSPACE CNG CYLINDER

3600 PSI 16.3 INCH DIAMETER
DESIGN SUMMARY

Subtask of
Gas Research Institute High Strength Steel
Type 2 Cylinder Project

Powertech Project 9081-32

k .. Principle Investigators:

Joe Wong
Livio Gambone

h
Project: 908 1-32 14 November 1997 Page 1 of 11
a ml087jyw
e,
Lacas Aerospace CNG Cylinder Powertech Labs Inc.

TABLE OF CONTENTS

Section

1.o INTRODUCTION 3

2.0 DEVELOPMENT OF MODEL 3

2.1 Computer Model 3
2.2 Steelhperties 3 . .
2.3 Laminate Properties 3
2.4 Pressures 4

3.0 STEELLINERTHICKNESS 5

4.0 FIBERWRAPTHICKNESS
4.1 Carbon Fiber Design
4.2 Aramid Fiber Design
4.3 Glass Fiber

5.0 OPTIMIZED GLASS FIBER DESIGN 9

5.1 Maximum Liner Stresses 10
5.2 Maximum Fiber Stresses 10
5.3 Maximum Liner Strains 10

6.0 STRESS CONTOURS 11

7.0 RECOMMENDATIONS 11

8.0 REFERENCES 11

Appendix 1 Drawings

Appendix 2 Stress Contour Plots

Project: 9081-32 14 November 1997 Page 2 of 11

m1087jyw
 Lucas Aerospace CNG Cylinder Powertech Labs Inc.

1.0 INTRODUCTION i * '

The purpose of this report is to summarize the design and optimization of the Lucas Aerospace
16.3-inch diameter NGV cylinder for use at a service pressure of 3600 psi. The cylinder design
will utilize a steel liner hoopwrapped with a composite wrap. Designs were developed using
fibers made of glass, carbon, and aramid.

2.0 DEVELOPMENT OF MODEL

2.1 Computer Model

Nonlinear finite element analysis techniques using the COSMOSM analysis software package,
developed by Structural Research and Analysis Corporation, were used to estimate the stresses
and strains in the cylinder at various pressures.

, The finite element model was developed using the minimum cylinder dimensions taken from the
drawing in Appendix 1. The cylinder was discretized using 4 node axisymetric isoparametric
two-dimensional planar elements.

2.2 Steel Properties

The material properties for the steel are based on the results of the tensile tests performed on
cylinders produced to date (see Appendix 1):

Yield strength - 125 - 135 ksi

Tensile strength - 140- 160 ksi
Elastic Modulus - 30 Msi
Plastic Modulus - 133,333 psi (bi-linear stress strain curve)
Shear Modulus - 12.0 Msi
Poisson's Ratio - 0.32
I

2.3 ,Laminate Properties

Three types of forcing fibers were used for this study:

1. glass fibers 1
2. carbon fibers
3. arcvnid fibers

The mechanical properties of the fibers used for this analysis are shown in Table 1

Project: 9081-32 14 November 1997 Page 3 of 11

m 1087jyw
h a s Aerospace CNG Cylinder Powertech Labs Inc.

Table 1
Mechanical Properties of Reinforcing fibers

The tensile strengths of fibers vary depending on the type of fiber, the fiber manufacturer, the
grade of fiber, and the 4ranslation efficiency when the fiber is filament wound. The selection of
fibers is addressed in a separate report (1).

2.4 Pressures

Calculations were performed for the following design pressures:

Autofrettage pressure - 7000 psi

zero pressure - 0 psi
Min operating pressure - 360 psi
Service pressure - 3,600 psi
Test pressure - 5,400 psi
Minimum burst pressure - 9,000 psi

The minimum burst pressure of the cylinder design is determined by the stress ratio requirements
for the particular fiber. The stress ratio in CSA B51-95 Part 2 for a type NGV-2 E-glassfiber
design is set at a value of 2.75. For a working pressure of 240 bar (3,480 psi) at 15"C, the 2.75
value is equivalent to the ANSUAGA NGV2-92 stress ratio requirement of 2.65 at a service
pressure of 248 bar (3,600 psi) at 21OC. Similarly, the stress ratio in CSA B51-95 Part 2 for a
type NGV-2 carbon fiber design is set at a value of 2.35. For a working pressure of 240 bar
(3,480 psi) tit 15OC, the 2.35 value is equivalent to the ANSVAGA NGV2-92 stress ratio
requirement of 2.25 at a service pressure of 248 bar (3,600 psi) at 21OC. The calculations for this
analysis were performed at 248 bar (3,600 psi) at 21OC.

Project: 9081-32 14 November 1997 Page 4 of 11

mlOS7jyw
 Lucas Aerospace CNG Cylinder Powertech Labs fnc.

3.0 STEEL LINER THIC

Burst pressures were estimated for various liner thicknesses using the following formula:

P = S (D2- d2)/ (1.3 D2+ 0.4d2)

Where:
P = pressure
S = tensile strength
D = outside diameter of liner
d = inside diameter of liner

Figure 1 shows the range of burst pressures the 140 ksi and 160 ksi UTS steel liners at
different wall thickness. The minimum burst for the liner is 4,500 psi. At 140 ksi UTS,a
liner with 0.215 wall thickness will burst at With a liner wall thickness of 0.25 inch,
the burst pressure is estimated to be 5,209 UTS and 5,950 psi at 160 ksi UTS.

A wall thickness of 0.25 inch was margin of safety over the minimum
liner burst. Note that the burst diameter 3,600 psi design is
approximately 6,300 psi (134 ksi UTS).

i .

7000
e 6500
6000
5500
3 5000
' & 4500
E 4000
3500
3000
0.15 0.17 0.19 0.21 0.25 0.27 0.29
i

f
Figure 1. Burst pressure calculations of the steel 'ner

F Project: 9081-32 14 November ,997 Page 5 of 11

i m1087jyw
i+
Lucas Aerospace CNG Cyrinder Powertech Labs Inc.

4.0 FIBER WRAP THICKNESS

A simplified finite element model of the cylinder sidewall was used to establish the wall
thickness of the composite wrap using the three types of fibers. Keeping the outer diameter at
16.3 inches, the FB model was adjusted for different wrap thicknesses while maintaining a liner
wall thickness of 0.25 inches. An autofrettage pressure of 7,000 psi was used.

4.1 Carbon Fiber Desi m

Figure 2 shows the results of the stress analysis for the carbon fiber cylinder design. The
selection of the wall thickness will depend on the tensile strength of fiber chosen. It can be seen
that the fiber stress at a minimum burst pressure of 9,OOO psi is approximately 450 ksi at a wrap
thickness of 0.150 inch. This wrap thickness may be adequate for a higher-grade fiber such as the
Toray T700. The lower grade “big tow” carbon fibers will require a thicker wrap. Prototype
cylinders will have to be burst tested in order to obtain the tensile strength of the fibers in the
laminate.

carbon fibers

1000 c I I i- 3
900 . 2.9
. 2.8
800 . 2.7
700 - 2.6
- 2.5 -0
600 - 2.4
500 - 2.3
- 2.2 %
400 - 2.1
300 -2
- 1.9
200 - 1.8
100 - 1.7
- 1.6
- 1.5
0.05 0.1 0.15 0.2
wrap thickness

Figure 2. Stress analysis of carbon fiber cylinder

Project: 9081-32 14 November 1997 Page 6 of 11

m1087jyw
 Lucas Aerospace CNG Cylinder Powertech Labs Inc.

4.2 Aramid Fiber Design

Figure 3 shows the results of the stress analysis for the aramid fiber cylinder design. As in the
carbon fiber design, the selection of the wrap thickness will depend on the tensile strength of the
fibers. Assuming that the tensile strength of the fiber in the laminate will achieve 400 ksi, the
wrap thickness required in the design is 0.225 inch.

-0
01 Q15 a2 a3

Figure 3. Stress analysis of aramid wrap

r Project: 9081-32 14 November 1997 Page 7 of 11

li m1087jyw
Lucas Aerospace CNG Cylinder Powertech Labs Inc.

4.3 Glass Fiber

Figure 4 shows the results of the stress analysis for the glass fiber cylinder design. It can be seen
that the fiber stresses at the burst pressure decreased to the ultimate tensile strength of the fiber
(330ksi) at a wrap thickness of 0.25 inch. However, the stress rad0 did not improve substantially
when the wrap thickness was increased from0.2 to 0.25 inches. The stress ratio may improve if
the autofrettage pressure is lowered.

16.3 inch Glass Fibre

500 3
e 2.8
,$ 400 2.6 .p
L
2.4 3 -t- p3600
8 300 2.2 fY
E 2 % p9000
z 200 1.8
stress ratio
1.6 5
x 100
I

YI
1.4
1.2
- +a

0 I
0.1 0.15 0.2 0.25 0.3
wrap thickness

Figure 4.Stress analysis of glass fiber wrap

Project: 9081-32 14 November 1997 Page 8 of 11

m1087jyw
 ~ ~~~

&as Aerospace CNG Cylinder Powertech Labs Inc.

5.0 OPTIMIZED GLASS FIBER DESIGN

A full finite element model including the end domes was generated for the 16.3 inch OD glass
fiber design with a wrap thickness of 0.25 inch. The autofrettage pressure was lowered to
6750 psi.

Table 2 summarizes the maximum stresses and strains at various cylinder pressures.

Table 2
Summary of Maximum Stresses and Strains

Steel Liner Liner

Pressure O
'n Inner Fiber Fiber Stress
Yield Hoop Long. Mises. Long. Strain
Strain @si)
@si) @si) (ksi) (hi)
Autofrettage
125 141 71 128 0.01189
6750 psi
145 165 138 146 0.007569 187.0 0.003227
Zero I
125 -48 -26 49 0.007375 '84.8 0.003621
0 psi
145 -23 -17 24 0.003059 35.1 0.001754
Min.Oper.
125 -39 -21 43 0.007616 87.5 0.003699
360 psi
145 -13 -12 14 0.0033 38.0 0.001832
Fill
3600 psi 125 84 28 76 0.00978 112.5 0.004407
145 92 80 84 0.005465 62.8 0.002539
Cycling/Max
125 100 40 90 0.01038 119.3 0.004603
4500 psi
145 112 97 101 0.006066 75.9 0.002736
HydroTest
125 117. 52 104 0.01098 126.3 0.004799
5404) psi
145 1 113 119 0.006667 .76.6 0.002932
Min. Burst 125
113 132 133 0.02682
9000 mi
I 145 1 165 I 159 I 152 10.02131 213.1 10.009638

Project: 9081-32 14 November 1997 Page 9 of 11

m1087jyw
 &as Aerospace CNG Cylinder Powertech Labs Znc.

5.1 Maximum Liner Stresses

The stresses in the liner at 3,600 psi will be used for the fracture mechanics assessment to
determine the minimum fatigue life and leak-before-burst assessment.

5.2 Maximum Fiber Stresses

At the minimum burst pressure, the maximum fiber stress was calculated to be 308.4 h i .

5.3 Maximum Liner Strains

The cylinder will exhibit an increase in diameter after undergoing autofrettage. The amount of
change will depend on the yield strength of the steel liner. Based on the strains calculated by the
FE model at zero pressure, figure 5 shows the diameter of the cylinder prior to autofrettage such
that the final liner diameter will be 15.8 inches.

Diameter before autofrettage

:g 150
5 145
z
140
E 135
3 130
125
-E120

i -

Project:9081-32 14 November 1997 Page 10of ll

m1087jyw
f
. .

LucasAerospace CNG Cylinder Powertech Labs Inc.

6.0 STRESS CONTOURS

The stress contours in the cylinder with liner yield strength of 140 ksi and 160 ksi at various
pressures are included in Appendix 2. It can be seen that the maximum stresses in the liner are
located near the end dome transition zone and the maximum stresses in the composite are located
near the liner composite interface.

7.0 RECOMMENDATIONS
Based on the preliminary design analysis of the 16.3 inch cylinder the following
recommendations can be made:

1. The liner thickness of the cylinder should be 0.250 inch.

2. The minimum thickness for the glass wrap should be 0.250 inch.

3. The autofrettage pressure for the glass wrap design should be 6750 psi,

4. The minimum thickness for the carbon wrap should be 0.15 inch.

5. The minimum thickness for the aramid wrap should be 0.225 inch.

6. The wrap thickness for the cylinders cannot be finalized until prototype cylinders have been
burst tested to obtain the strength of the laminate.

8.0 REFERENCES
1. Powertech report PR165-lrg ‘‘Lucas Aerospace CNG Cylinder Fiber Reinforcement Study”

f
. .

Project: 908 1-32 14 November 1997 Page 11 of 11

mlO87jyw
Lucas Aerospace CNG Cylinder Powertech Labs Inc.

Appendix 1

Drawings

Project: 9081-32 14 November 1997

m1087jyw
LucasAerospace CNG Cylinder Powertech Labs lnc.
x
n
Project: 9081-32 14 November 1997
m1087jyw
 Lucas Aerospace CNG Cylinder Powertech Labs Inc.

Appendix 2

Stress Contour Plots

2
z
P
t
i
I .

i
L
.n

a
I
&

Project: 9081-32 14 November 1997

*
Y
m1087jyw
i
t
e *
steel Liner Properties f ;

Elastic Modulus 30 Msi

Plastic Modulus 133,333 Psi
Yield Stress 125 Ksi
Poisson’s Ratio 0.32

ComDosite Properties

Laminate Hoop Modulus 6.8 Msi

Transverse Modulus 0.01 Msi
Axial Modulus 0.01 Msi
Shear Modulus 0.01 Msi
Poisson’s Ratio 0.25

Pressures at Time Stm

Calculations were performed for the following pressures with designated time steps.

Pressure Value Time Step

Initial 0 psi 0
Autofrettage pressure 7,000 psi 40
Zero pressure 0 psi 60
Minimum operating pressure 360 psi 70
Service pressure 3,600 psi 80
Service pressure 4,500 psi 90
Test pressure 5,400 psi 100
Minimum burst pressure 9,000 psi 130
 6
U
r
8 .
L
. .. . , . ,.... ... ..
I 1 . .

GeoStar 1.75A : 1163-a4

Win 1 Main (Active)

l L l n STRESS S t e p : 4 8 =4Q

Von Hlses Von Hlses

QQ5

QQ5

004

QQ4

91E+QQ4

45E+QQ4

8QE+QQ4

55E+Q84
 GeoStar 1.75A : 1163-a4

Win 1 Main (Active1

l L i n STRESS Step:60 =60

Von Mlses Von Hlses

am4

004

004

3.66E 014 ;.66E+OJQ4

2.93€+184 93E+0Q4

2.20€+104 2 1E+00 4

1.47€+004 47E+Q04

7.33E+003 I 33E+003

1.92000Qb
 GeoStar 1.75A : 1163-a4

Win 1
i L l n STRESS step:7m =7m

Von Mlses Von Mlses

0m4

0m4

0m4

Y
A
 ._” “.. -,. .
, ,.i ... . . I

* - .. .

GeoStar 1.75A: 1163-a4

Win 1 Main (Active)

V o n Mlses Von Mlses

7.47E+Q84 7.47E+hl84

6.6 1Et884 6.6 1E+QQ4

5.75E+6184

4.88E+Q84 4.88E+QQ4

4.m2~+~m4 4 .82€+884

3.15~+0m4 3.15E+O)Q4

2.29E+884 2.29E+Q84

1.43E+QQ4 1.43E+884

5.6 I E+883 5.61E+Q83

 GeoStar 1 . 7 5 A : 1163-a4

Win 1
L l n STRESS Step:S8 =98

Von Mlses Von Mlses

.
 GeoStar 1.75A : 1163-a4

Win 1 Main (Active)

l L l n STRESS S t e p . - l Q Q = 1 Q Q

Von Mlses Von Mlses

1 .QlE+QQ5 1 .QlE+QQ5

8.97E+QQ4 8.97E+Q04

7.81E+Q04 7.81E+QQ4

6.65E+QQ4

5.49E+BQ4

4.32E+884

3.16E+QQ4

2 .QQE+QQ4

8.42E+QQ3
" % 1 V I

GeoStar 1.75A : 1163-a4

Win 1
I L l n STRESS Step:l3m =130

Von M l s e s Von Mlses

1 .ie~+mm5

9.75E+rnQ4

7.66E+Q84

5. s ~ ~ + m m 4

3.49~+mm4

I .4m~+mm4
Steel Liner ProPerties

Elastic Modulus . 30 Msi

Plastic Modulus 133,333 Psi
Yield Stress 135 Ksi
Poisson’s Ratio 0.32

Comuosite Properties

Laminate Hoop Modulus 6.8 Msi

Transverse Modulus 0.01 Msi
Axial Modulus 0.01 Msi
Shear Modulus 0.01 Msi
Poisson’s Ratio 0.25

Pressures at Time Stm

Calculations were performed for the following pressures with designated time steps.

Pressure Value Time Step

Initial 0 psi 0
Autofiettage pressure 7,000 psi 40
Zero pressure 0 psi 60
Minimum operating pressure 360 psi 70
Service pressure 3,600 psi 80
Service pressure 4,500 psi 90
Test pressure 5,400 psi 100
Minimum burst pressure 9,000 psi 130
 I
(*
u
 GeoStar 1.75A: 1163-a4

Win 1 Main (Active)

L l n STRESS S t e p : 4 @ =4Q

Von Mlses Von Mlses

1.37E+QQ5

i.21~+~rn5

1 .Q5E+QQ5
*,.. pi. .,. . " .. I . . .I .. . . ,

GeoStar 1.75A: 1163-a4

Win 1 Main (Active)

Lln STRESS Step:6m =6Q

Von fllses Von fllses

4.3m~+m4

3,76~+mm4

3.n~+mrn4

2.69E+QQ4

2.15€+884
m u "

1.6LE+Q14

.
1 ~EE+(DQ~

S.~EIE+~Q~

Q .424QQQb
 GeoStar 1.75A: 1163-a4

Win 1
J L l n STRESS S t e p : 7 8 =7Q

Von M l s e s Von M l s e s
4.34E+884

3 . 8 1 E+Q)Q4

JX_*

2.74€+084 2.74E+884

2.2OE+884 2.~~IE+BQ~

i.66~+aa14 1,66E+084

1 .13E+Q84 1.13E+QQ4

5.92E+QO3 5.92€+8Q3

5 6 1 .OQOOQ 56 . Q Q Q O I
 GeoStar 1.75A: 1163-a4

Win 1 (Active) Main

lLln STRESS Step:Em =Em

Von Mlses Von Mlses

ma4

ma4 mm4

ma4 ma4
’
mm4 ma4

24~+mm4

32 E +m a 4

4m~+ma4

48~+mm4
 GeoStar 1.75A: 1163-a4

Win 1 Main (Active)

L l n STRESS Step:lQB =1QO I L l n STRESS Steo:l@Q =lo38 HI
Won Mlses Won Mlses
 GeoStar 1.75A: 1163-a4

Win 1 (Active) Main

L l n STRESS Step:130 =130

Von Mlses Von Mlses

.59€+005 1.59E+005

.4 1EtQ05 1.41E+005

.23E+005 1.23E+005
" '

.05E+005 1 .05E+b)Q5

8.65E+OQ4

6.84E+004

5.03E+084

3.21E+004

1.4QE Q04
 Appendix B
Lucas Aerospace CNG Cylinder Fiber Reinforcement Study
 Powertech G Powertech Labs Inc. 590-7500
12388-8ghAve.,surrey, B.C. Canada (~504)

POWERTECH LABS INC.

FINAL REPORT

LUCAS AEROSPACE CNG CYLINDER

FIBER REINFORCEMENT STUDY

Sub-TSk
of
Gas Research Institute
High Strength Steel
Type 2 Cylinder Project

Powertech Project 9081-32

Principle Investigators:
Joe Wong
Livio Gambone
Craig Webster

Project: 9081-32 14 November 1997 Page 1 of 20

m1088jyw
k a s Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

'CABLE OF CONTENTS

Section Pag;e

1.o INTRODUCTION 3

2.0 mBERPERFORMANCE 4
2.1 GlassFibers 4
2.2 CarbonFibers 8
2.3 AramidFibers 11

3.0 POWERTECH ACID TEST RESULTS 14

4.0 RECOMMENDATIONS 17

5.0 REFERENCES 18

Appendix A Owens Coming Advantexm Acid Test Results

Project:9081-32 14 November 1997 Page 2 of 20

m1088jyw
Lucas Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

1.0 INTRODUCTION

The objective of the current program is to develop an optimized high strength steel hoop-
wrapped (Type 2) compressed natural gas (CNG) cylinder for Lucas Aerospace. The new design
will incorporate a high strength steel liner as permitted in the proposed revision (July 1997) to
the ANSUAGA NGV2 standard. Fiber reinforcement will be selected based on the results of the
current fiber study.

Technical information pertaining to the performance of fibers under CNG service conditions was
gathered from several fiber suppliers in an effort to determine which fiber system would be most
suitable for the optimized Type 2 cylinder design. The fiber types considered included glass,
carbon and aramid and encompassed the following specific manufacturers and products:

Glass Fibers Carbon Fibers Aramid Fibers

Vertrotex RO99 P122 E-glass Toray T700 Akzo Nobel TwaronB 2200
Owens Corning ZenTronTM Akzo Nobel Fortafd B 3(C)
Owens Coming AdvantexTM Zoltek PANEX 8 33

Consideration was given only to the above products since they appeared to offer the greatest
potential for the intended application [l]. Where possible, the performance characteristics of
Vetrotex RO99 P122 &glass fibers were used as a basis for comparison since these fibers are
utilized in the present Lucas Aerospace CNG cylinder design.

Project:9081-32 14 November 1997 Page 3 of 20

m 1088jyw
 k a s Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

2.0 FIBER PERFORMANCE

2.1 Glass Fibers

\
2.1.1 General

Glass fibers are the most widely used reinforcement for the manufacture of CNG cylinders due to
their mechanical properties and attractive selling price. The inherent low cost of glass fibers is
achieved by virtue of the simplicity of the manufacturing process, which in essence can best be
described as the drawing of filaments from molten sand.
? '
. .
1.
3
The most common and least expensive glass fiber type used is E-glass, which is a calcium
2,
alumino-borosilicate glass. S-glass is a magnesium alumino-silicate glass originally developed
for aerospacdaircraft applications due to its higher strength and excellent thermal stability. A
lower cost version (SZglass@)manufactured by Owens Coming using less stringent (non-
military) specifications is also available and offers similar mechanical performance. ZenTronm
is a new high strength glass fiber product line developed by Owens Coming, which has a similar
formulation to S2-glass" and offers comparable performance.
i
The susceptibility of E-glass fibers to stress corrosion cracking (SCC) in the presence of acidic
F
environments led to the development of ECR-glass, a boron-free version of E-glass. Owens
Corning currently markets a glass fiber (Advantexm) that combines the mechanical properties
and lower cost of traditional E-glass with the acid corrosion resistance of ECR-glass. The
6 Advantexm fiber system replaces Owens Coming's existing E-glass and ECR-glass product line.

2.1.2 Mechanical Pror>erties

f .
s
k
ical properties of the glass fiber products under consideration are included in

Project: 9081-32 14 November 1997 Page 4 of 20

m1088jyw
LucasAerospace CNG Cylinder Fiber Reinforcement Study Powertech krbs Inc.

Table 1
Mechanical Properties of Glass Fibers [2 41 -
Vetrotex Owens
Property RO99 P122 Corning

ksi (MPa)
Strand Tensile Modulus 10.6 (73) 11.7 (81) 13.5 (94)

Strand Elongation at 4.6 I 4.2

Composite Strength
Translation Efficiency
(%) I
70t I 52'

* Impregnated single strand (measured)

t Unidirectional laminate in epoxy (measured)

2.1.3 Environmental Resistance

Glass fiber reinforced composites are susceptible to a number of moisture-induced degradation

mechanisms [5 - 141. For example, exposure to moisture causes a decrease in glass transition
temperature and plasticization of the resin, which leads to reductions in strength, stiffness, and
impact properties. Resin swelling attributed to moisture introduces stresses into the composite
laminate, particularly at the fiberhesin interface. In addition, microcracking, void formation,
crazing, and fiberhesin debonding have been reported. Researchers have confirmed that silane
coupling agents used to promote interfacial bonding between the glass fibers and epoxy can
dissolve in water.

Water can also cause the chemical degradation of glass fibers resulting in lower fracture energies.
This degradation mechanism reportedly involves an ion exchange mechanism whereby acidic
hydrogen ions present in water or dilute acid solutions are exchanged with the larger alkali metal
cations in the glass fiber surfaces. This causes surface shrinkage, which in turn creates tensile
stresses leading to new surface flaws. In the presence of service loading this can lead to SCC. A
great deal of study has been devoted to examining the brittle fracture of glass fiber reinforced
-
materials due to SCC [15 211. In general, the evidence indicates that acids such as sulfuric,
nitric, hydrochloric, and to a lesser extent, hydrogen bromide, hydrogen iodide, and some organic
acids such as oxalic, promote the corrosion of glass fibers. In addition, alkali chemicals such as
sodium hydroxide have been shown to be aggressive corrosion agents at high pH levels.

Project: 9081-32 14 November 1997 Page 5 of 20

m1088jyw
 k a s Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

A number of failures of metal-lined composite CNG cylinders have been attributed to SCC of the
glass reinforcing fibers [l]. In all cases reported, the type of fibers used in the cylinder designs
was E-glass. Accordingly, the NGV industry has shown significant interest in the potential
substitution of E-glass with glass fibers more resistant to the environmental service conditions to
which CNG cylinders are exposed.

Owens Coming has subjected both their AdvantexTMand ZenTronTMglass fibers to validation
tests in water and acidic environments. Table 2 lists glass fiber weight loss as a function of
exposure time to sulfuric acid. Both AdvantexTMand ZenTronm fibers exhibit significantly less
weight loss than E-glass fibers under these conditions.

Table 2
Glass Fiber Weight Loss Due to Acid Exposure [3,22,23]

Fiber Weight Loss (%)

Exposure Time (10% HtSOd, 96'C)
mays) Owens Corning Owens Coming Owens Coming
AdvantexTM ZenTronTM
1 39 6 --
7 42 10 4

Results from tensile strength retention tests performed by Owens Coming also confirm that
AdvantexTMfibers outperform E-glass fibers in water and mild acid exposure (see Table 3).
Similar performance has been confirmed with respect to flexural strength retention. More
information regarding the Owens Coming test specimens, exposure conditions and test results is
available in Appendix A.

It is important to note that the data in Table 3 was ge ed by first exposing glass fiber
reinforced pkels to the given environment for the specified time, then performing tensile tests on
samples prepared from these panels. These test conditions are not representative of the SCC
failure mechanism observed in NGV service. As a result, Powertech Labs has performed a
number of true SCC tests on both the Vetrotex E-glass and Owens Coming AdvantexTMfiber
systems. Details regarding the test procedure and test results are summarized in Section 3.0.

f Project: 9081-32 14 November 1997 Page 6 of 20

m1088jyw
s!%3
 Lucas Aerospace CNG Cylin&r Fiber Reinforcement Study Powertech Labs Inc.

Table 3
d
Glass Fiber Reinforced Compos!ke Tensile Strength Retention in Water and Acid [24,25]

2.1,4 Availability and Cost

The glass fiber industry has three major global producers: Owens Coming, PPG Industries and
St. Gobain (Vetrotex in the U.S.). Together, these companies contribute 60% of the glass fibers
to the global market. The remainder of the market is served by a number of regional suppliers.

Glass fiber production is capital intensive, making sufficient plant loading critical to profitability.
The industry has recently been operating near full capacity due to the growth in composites use.
In support of this growth, the glass fiber industry is expected to add more than one-half million
tons of capacity in the next four years. Owens Coming will account for about 50% of this total.

Current prices for E-glass fibers are about $0.85 to $1.00/lb. Owens Coming AdvantexTMfibers
are priced at the same level as E-glass fibers. In contrast, Owens Corning high strength
ZenTronTM fibers range from $5.50 to $6.75/lb depending on order volumes.

Owens Coming manufactures its Advantexm glass fiber product in North America (Guelph,
Ontario) and in Europe (Battice, Belgium and L'Ardoise, France). ZenTronm glass fibers are
available from the Nor& American Specialty Fibers Division in Huntington, Pennsylvania.
I

2.1.5 Filament Winding Considerations

Owens Coming's AdvantexTMand ZenTronm fibers are equally compatible with all of the most
L - widely used resin systems for filament winding. In addition, a glass fiber sizing system has been
developed to provide good cyclic and burst strength perfomance for filament wound epoxy
pressure vessels. Both fiber systems are available in center-pull and outside-pull spools.

Project:9081-32 14 November 1997 Page 7 of 20

r'
ml08Sjyw

~
LucasAerospace CNG Cylindkr Fiber Reinforcement Study Powertech Labs Inc.

2.2 Carbon Fibers

2.2.1 General

Carbon fibers are produced by the thermal decomposition of various organic fiber precursors
such as cellulose, polymerized acrylonitrile (PAN) and mesophase pitch. More than 90% of all
carbon fibers on the market are manufactwed from PAN precursors. These precursors provide a
carbon fiber yield of 45 to 50% (typically 2.2 lb PAN per lb carbon fiber produced). The current
price of commercial grade carbon fibers is approximately $12.00 to $20.00/lb. The capital-
intensive nature of the manufacturing process coupled with the high cost of raw material (PAN
precursor) and relatively low conversion yield is primarily responsible for its high cost.

In general, carbon fibers offer the highest modulus and strength of all reinforcing fibers. The
fibers have high fatigue strength and do not suffer from SCC or stress rupture failures. However,
carbon fiber'exhibits low failure strain, a property which imparts low impact resistance to carbon
fiber reinforced composites.

2.2.2 Mechanical Prowrties

The mechanical properties of the carbon fiber products under consideration are included in
Table 4.

Table 4
Mechanical Properties of Carbon Fibers [26-281

* Unidirectional laminate in epoxy (measured)

Project:9081-32 14 November 1997 Page 8 of 20

m1088jyw
hcas Aerospace CNG C y l i h r Fiber Reinforcement Study Powertech Labs Inc.

2.2.3 Environmental Resistance

In general, carbon fibers are unaffected by moisture, solvents, bases and weak acids at room
temperature [29]. In addition, they are essentially immune to SCC and stress rupture at room
temperature. They also offer outstanding strength and modulus retention over long periods at
elevated temperatures (>15OoC).

2.2.4 Availability and Cost

Worldwide carbon fiber shipments totaled over 17 million pounds in 1995. Military and
commercial aerospace applications consumed nearly half of that amount, while the U.S.
consumed over 40% over all applications [l]. Worldwide PAN-based nameplate capacity is
approximately 22.7 million pounds, which misleadingly suggests an excess market capacity. In
actuality, the carbon fiber industry is currently operating in the oversold condition. This
condition has been exacerbated by the exit of Courtaulds and BASF from the carbon fiber
industry due to the reduction and/or elimination of military programs in the early 1990s.

As a result, several carbon fiber manufacturers have plans to expand their production capacity
[1,30]. Akzo Nobel has already expanded their plant by about 3.3 million pounds a year. Ammo
is planning to start the old BASF plant, which will contribute 2 million pounds per year. R.K.
Carbon is considering an expansion of between 1 and 2 million pounds, while Hercules is
initiating several de-bottlenecking efforts to expand capacity. Toray plans to boost its production
capability at its Ehime plant to 10.4 million pounds by the spring of 1998. In addition, Toray is
considering new carbon fiber production capacity of about 4 million pounds for the U.S.by 1999.
The total increase in carbon fiber capacity represented by these potential expansions is
approximately 15 million pounds (prior to the year 2000).

The high cost of carbon fiber has limited its entry into the commercial marketplace and this has
limited any potential price reduction from increased t exposure. The current price of
commercial grade carbon fibers ranges from $12 to $20/lb, whereas the price of aerospace grade
fibers can be as high as $55/lb. The cost of PAN raw material is partly responsible for this high
cost since it is produced from propylene, which is a hydrocarbon derived from the catalytic
cracking of crude oil. Another major factor in the cost of carbon fibers is the very high level of
investment required of producers due to manufacturing capital and operating requirements.

A number of carbon fiber suppliers such as Akzo Nobel, Zoltek, and R.K.Carbon offer an
industrial grade large tow (48K or 50K) low cost carbon fiber product manufactured using a
textile acrylic fiber-based PAN precurser. Additional research is required to investigate whether

Project: 9081-32 14 November 1997 Page 9 of 20

m1088jyw
 k a s Aerospace CNG C y l i h r Fiber Reinforcement Study Powertech Labs Inc.

this product is technically equivalent to the high strength 12K tow products used in current CNG
cylinder designs. Technical issues which remain to be resolved include fiber windability, resin
wetting, fiber tension and flatness. Carbon fiber costs as low as $10 to $1Mb are achievable for
these lower cost textile-based fibers.

2.2.5 Filament Winding Considerations

The availability of inexpensive large tow carbon fibers such as Akzo Nobel's Fortafil@'3(C) is
attractive to the current CNG cylinder optimization program due to the potential for finished
cylinder cost reduction and increased filament winding through-put. As discussed previousl). in
Section 2.2.4, the wet winding of CNG cylinders using large tow fibers requires considerable
future experimentation. In addition, the manufacturing process reportedly incurs the risk of
carbon particulate contamination of sensitive electrical equipment. In contrast, pre-impregnated
(prepreg) unidirectional composite tapes made from large tow fibers offer the possibility of cost
competitive and solvent free processing, ambient temperature storage, and improved composite
strength translation efficiency. This latter feature is particularly helpful for the Fortafil@3(C)
carbon fiber system as it suffers from a low translation efficiency (see Table 4).

Thiokol's TCRm prepreg product, which is available for carbon, aramid, and glass fiber systems,
may be suitable for the optimized Lucas Aerospace hoop-wrapped CNG cylinder [33]. A
detailed prepreg fiber feasibility study comparing the finished weight and cost parameters of the
fiber and the conventional 12K or 24K Toray "ROO fiber would be essential to
large tow Fortafil@'
the manufacturing selection process. Thiokol's preliminary estimates for carbon fiber weight per
f"
cylinder (12inch diameter x 60 inch long) and associated cost using TCRm prepreg are shown in
Table 5.
I
I

The preliminary estimates in Table 5 suggest that TCRm prepregs manufactured with Fortafil@
3(C) carbon fibers offer both a minimal cost savings and incur a significant weight penalty
compared to those manufactured with Toray T700 fibers.

For completeness, a similar weight and cost study should be prepared the case of wet winding
of conventional 12K or 24K Toray "ROO fibers.

b
k

j Project: 9081-32 14 November 1997 Page 10 of 20

m1088jyw
 hcas Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

Akzo Nobel Toray

Fortafil@3(C) woo
5OK 24K
Carbon Fiber Weight 16.5 11.8
(W
Carbon Fiber Cost 13.97 20.06
($fib)
Total Carbon Fiber 230.5 1 236.71
Cost ($1 I I
i '

23 AramidFibers

2.3.1 General

Aramid is a generic term for a class of aromatic polyamide fibers introduced commercially
during the 1970s. Organic fibers such as Kevlar@aramids were developed by E.I. DuPont de
Nemours. Akzo Nobel also manufactures an aramid fiber product under the trade name
Twaron@. As a result of their highly aligned polymer chain, these fibers offer high tensile
strength and modulus with moderate elongation. Aramid fibers exhibit excellent fatigue
resistance and they are resistant to flame, high temperatures and chemicals such as organic
solvents, fuels and lubricants. Their limitations include a tendency to absorb moisture, a
I' susceptibility to ultraviolet degradation and poor adhesion to matrix resins.
ih

\
2.3.2 Mechanical Properties

The mechanical properties of the aramid fiber product under consideration are included in
Table 6.

Project: 9081-32 14 November 1997 Page 11 of 20

m1088jyw
 Lucas Aerospace CNG Cylinder Fiber Reinforcement S t d y Powertech Labs Inc.

Table 6
Mechanical Properties of Aramid Fiber [31]

Property Akzo Nobel

Twaron@
2200
Strand Tensile Strength 457 (3150)
ksi (MPa)
Strand Tensile Modulus 16.7 (115)
Msi (GPa)
Strand Elongation at 2.6
Break (96)
Composite Strength
Translation Efficiency 46*

* Unidirectional laminate in epoxy

(estimated)

2.3.3 Environmental Resistance

Twaron" aramid fibers generally offer good resistance to most organic chemicals, and acids and
bases in the pH range 3 to 10 [31]. As with most fiber-reinforced composites, the limiting factor
for chemical resistance is usually the less resistant polymer matrix. Aramid fibers alone can
absorb significant amounts of moisture when exposed to high humidity. However, the total
moisture absorbed by an aramidepoxy composite may not be substantially greater than that
absorbed by other epoxy composites [32].

The ultraviolet resistance of bare unprotected Twaron" aramid fibers is relatively poor [31].
However, this behavior has limited impact on fiber reinforced systems since the fibers are
embedded in an epoxy or polyester matrix. Resin-rich topcoats, which offer high ultraviolet
absorption, can protect aramid fibers more than adequately [32].

Twaron" aramid fibers exhibit minimal creep behavior at 85OC [31].

2.3.4 ptvailability and Cost

The global market for aramid s approximately 20,000 tons annually and U.S. demand is
expected to grow 6 to 8% per for several years. The largest use for aramid fibers is for the

Project: 9081-32 14 November 1997 Page 12 of 20

m1088jyw
P
 Lucas Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

reinforcement of firefighter's breathing apparatus bottles. Other uses include aircraft oxygen
bottles and a variety of aerospace appiications.

There are only a handful of suppliers of aramid fibers: DuPont, Akzo Nobel, Teijin, and Hoechst
Celanese. DuPont (Kevlar") and Akzo Nobel (Twaron") dominate the market. DuPont's share
of the world market is above 608, while Akzo Nobel holds about a 30% share. Prices for aramid
fibers have recently settled to the $7 to $9/lb range.

2.3.5 Filament Winding Considerations

Akzo Nobel's Twaron@aramid fiber can be filament wound using wet processing methods at
rapid speeds (up to 325 fdmin.). This is facilitated by the use of a conical pultrusion-type die
through which the aramid fibers are forcibly wetted. Although the resulting winding speeds
would likely be too high for the manufacture of CNG cylinders, some increased throughput can
be expected. Prior to winding, the aramid fibers are pre-heated in a ceramic infrared heater to
250 to 300OC to remove moisture from the fibers, thereby improving fiber wetting. This latter
processing feature shortens curing time since the initial cure temperature can be higher due to
reduced moisture in the composite.

Twaron@aramid fibers would also be suitable prepreg candidates for the Thiokol TCRm prepreg
tape system.

Project: 9081-32 14 November 1997 Page 13 of 20

m1088jyw
h
 Lucas Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

3.0 POWERTECH ACID TEST RESULTS

In order to complement the acid test work performed by Owens Corning (see Section 2),
Powertech initiated a series of SCC tests for the Vetrotex E-glass and Owens Corning
Advantexm fiber systems, The test involved subjecting epoxy resin impregnated tows of each
fiber system to a range of constant loads in a sulfuric acid environment and monitoring the time
to failure of individual samples. The test was designed to duplicate the exposure of a Lucas
Aerospace hoop-wrapped CNG cylinder to battery acid.

The fiber tows were impregnated with a Shell Epon 8132 epoxy resinlAncamide 506 hardener
and cured at room temperature for 16 hours followed by a post-cure of 2 hours at 100°C. Cross-
sections of several cured glass fiber tows were examined in the scanning electron microscope
(SEM)to confirm adequate resin distribution/fiber wetting. In addition, the image analyzer
attachment to the SEM was used to estimate fiber cross-sectional area.

The acid solution consisted of 19% by volume sulfuric acid. The test was conducted at room
temperature. The effects of elevated temperature acid exposure and varying acid concentration
were not evaluated in the current test program.

The Powertech acid test results are presented graphically as fiber stress versus log time-to-failure
in Figure 1. Based on the consistent but limited data, the Advantexm fibers clearly outperformed
the Vetrotex E-glass fibers under these exposure conditions. At fiber stresses associated with
typical Lucas Aerospace hoop-wrapped CNG cylinder operating pressures (60 to 100 ksi), the
SCC data suggests that Lucas cylinders hoop-wrapped with Vetrotex fibers are affected by acid
exposure, whereas those hoop-wrapped with Advantexm fibers may be unaffected.

Figures 2 and 3 are SEM micrographs of fracture surfaces of broken Vetrotex and Advantexm
glass fibers showing the characteristic features of brittle fracture due to SCC. The fracture
surfaces consist of a small fracture mirror (indicative of high applied stress), and small ridges
oriented in a direction parallel to crack propagation, which eventually merge into similar but
larger ridges.

bject: 9081-32 14 November 1997 Page 14 of 20

* r
m1088jyw
 hcas Aerospace CNG Cyrinder Fiber Reinforcement Study Powertech Labs Inc.

120

I00

80

60

40

20

0
I 10 100 1000 10000 100000 1000000 10000000

Time (min)

Figure 1: Plot of fiber stress versus log time-to-failure in 19%by volume sulfuric
acid solution.

Project: 9081-32 14 November 1997 Page 15 of 20

m1088jyw
 Lucas Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

Figure 2: SEM micrograph of broken Vetrotex E-glass fiber after 646 minutes exposure to 19%
by volume sulfuric acid at 51.4 ksi. Characteristic brittle fracture surface features
caused by SCC are visible (fracture mirror identified with an arrow).
Manification: 3500X

Figure3: SEM micrograph of broken Owens Corning Advantexm glass fiber after
24,480minutes exposure to 19% by volume sulfuric acid at 84.8 ksi. Characteristic
brittle fracture surface features caused by SCC are visible (fracture mirror identified
with an arrow). Magnification: 5000X

Project: 9081-32 14 November 1997 Page 16 of 20

ml088jyw
h
Lucas Aerospace CNG Cyrinder Fiber Reinforcement Study Powertech Labs Inc.

4.0 RECOMMENDATIONS

A review of the technical information pertaining to the performance of selected glass, carbon,
and aramid fiber systems under CNG service conditions has been performed. With respect to
fiber reinforcement selection for an optimized Lucas Aerospace Type 2 CNG cylinder design, the
following course of action is recommended:

1. Given the favorable cost, superior acid performance, and European availability of Owens
Corning AdvantexTMfibers, it is recommended that prototype cylinders be fabricated and
tested using this fiber system. The prototype cylinders should be subjected to the
environmental, accelerated stress rupture and flaw tolerance tests as described in the latest
draft (July 1997) of the ANSUAGA NGV2 standard.

In order to expedite the test program, the prototypes should be wet wound using Lucas,
existing process parameters (e.g. resin system, fiber tension, wind speed, etc.). Should this
fiber system prove successful in cylinder prototype tests, these process parameters can be fine
tuned in consultation with Owens Coming at a later date. In addition, to improve fiber
wetting and increase throughput, the use of a conical pultrusion-type fiber feed die should be
considered.

2. Concurrent with the above recommendation, tests should be conducted on AdvantexTMfibers

to confirm their elevated temperature acid performance and the effect of varying acid
concentration. In addition, it is recommended that the Owens Coming Application
Development Center in Granville, Ohio be contacted for more information regarding their
existing acid exposure test program.

3. Should the Owens Coming AdvantexTMfiber system prove unsuitable for the optimized
Lucas Aerospace CNG cylinder, then additional prototype cylinders should be fabricated
using Toray T700 carbon fibers in either prepreg tape format or through a wet winding
process, pending favorable results from a detailed weight and cost study as suggested in
Section 2.2.5. Further investigation will be required to determine if sensitive electrical
equipment can be protected from carbon fiber particulate contamination for the case of wet
winding of Toray T700 fibers, should this approach prove more weightkost effective than the
prepreg processing of these fibers.

Project: 9081-32 14 November 1997 Page 17 of 20

rn 1088jyw
Lucas Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

5.0 REFERENCES

1. Richards, M.E., Blazek, C., Webster, C.; Wong, J., and Gambone, L., “Compressed Natural
Gas Storage Optimization for Natural Gas Vehicles”, Gas Research Institute Report 96/0364,
December 1996.

2. Hunter, T.,Vetrotex (UK) Limited, February 13,1997.

3. AdvantexTM
Glass Fiber, Product Information, Owens Corning, 1996.

4. ZenTronTMHigh Strength Fiber, Product Information, Owens Corning, 1996.

5. Apicella, A., Nicolais, L. and de Cataldis, C., “Characterization of the Morphological Fine
Structure of Commercial Thermosetting Resins Through Hygrothermal Experiments”,
Advances in Polymer Science, Vol. 66,1985, pp 189-207.

6. Apicella, A. and Nicolais, L., “Effect of Water on the Properties of Epoxy: Matrix and
Composite”,Advances in Polymer Science, Vol. 72,1986, pp 69-77.

7. Antoon, M.K. and Koenig, J.L., Journal of Polymer Science: Polymer Physics Edition, Vol.
19,1981, pp 197-212.

8. Antoon, M.K. and Koenig, J.L., Journal of Polymer Science: Polymer Physics Edition, Vol.
19,1981, pp 1567-1575.

9. Strait, L.H., Karasek, M.L. and Amateau, M.F., “Effects of Seawater Immersion on the
Impact Resistance of Glass Fiber Reinforced Epoxy Composites”, Journal of Composite
Materials, Vol. 26, No. 14,1992, pp 2118-2133,

10. Adams, D.F.and Miller, A.K., “Hygrothermal Microstresses in a Unidirectional Composite

Exhibiting Inelastic Behavior”, Journal of Composite Materials, Vol. 11, 1977, pp 285-299.

11. Illinger, J.L. and Schneider, N.S., “Water in Polymers”, ACS Symposium Series 127,
Washington, D.C., 1980.

12.Shirel1, C.D., Leisler, W.H.ans Sandow, F.A., “Nondestructive Evaluation of and Haw
Criticality for Composite Materials”, ASTM STP 696, Baltimore, 1979.

Project: 9081-32 14 November 1997 Page 18 of 20

m1088jyw
k a s Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

13. Ishida, H. and Koenig, J.L., J o u m l of Polymer Science: Polymer Physics Edition, Vol. 18,
1980, pp 1931-1943.

14. Lee, S.M. and Schile, R.D., “An Investigation of Material Variables of Epoxy Resins
Controlling Transverse Cracking in Composites”, Journal of Materials Science, Vol. 17,
1982, pp 2095-2106.

15. Streat, N., Chan, A., Wong., J.Y., Nadeau, J.S. Romilly, D.P. and Powsartip, A., “Loadings
on Aerial Lift Devices and Their Effects on FRP Booms”, Canadian Electrical Association
Distribution R&D Report 124 D447, February, 1987.

16. Akhtar, A., Nadeau, J.S., Wong, J.Y., Romilly, D.P. and Taggart, C., “Brittle Fracture of
Nonceramic Insulators, Canadian Electrical Association Transmission R&D Report 186
T350, September, 1985.

17. Akhtar, A. and Wong, J.Y., “Failure Analysis of Brittle Fracture in Nonceramic Insulators**,
Journal of Composites Technology and Research, Vol. 9, No. 3, Fall 1987, pp 95-100.

18. “Suspension and Tension Composite Insulators for Overhead Lines: Brittle Fracture at Low
Mechanical Stress”, CIGRE Study Committee 22, WG 22-10,22.8 (IWD 10) 23, June 1980.

19. Chandler, H. and Jones, R., “Stress Corrosion Failure of Composite Long Rod Insulators”, 4*
International Symposium on High Voltage Engineering, Athens, Greece, 5-9 September,
1983.

20.French, M.A. and Pritchard, G., “Environmental Stress Corrosion of Hybrid Fiber
Composites”, Composites Science and Technology, Vol. 45, No.3,1992, pp 257-263.

21. White, R.J. and Phillips, M.G., “Environmental Stress-Rupture Mechanisms in Glass
FiberRolyester Lamin **, 5fi International Conference on Composite Materials, ICCM-V,
San Diego, CA, 29-30 July, 1 August, 1985, pp 1089-1099.

22. Walling, J., Owens Corning, Private Communication, February 19,1997.

23. Shaw, W., Owens Corning, Private Communication, October 9,1996. k

24. Matzeg, R., “Guelph Tso Corrosion Study”, Owens Corning, Novemebr 26,1996.

Project: 9081-32 14 November 1997 Page 19 of 20

m1088jyw
Lucas Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

25. Hartmann, D.R., Greenwood, M.E. and Miller, D.M., “HighStrength Glass Fibers”, Owens
Corning Technical Paper, 1994.

26. Kayaba, K., Toray Industries Inc., Private Communication, March 6,1995.

27. Akzo Nobel Technical Data Sheet 931A, 1993.

28. Dry,A., Zoltek Corporation, Private Communication, September 23, 1996.

29. Judd, N.C.W., “The Chemical Resistance of Carbon Fibers and a Carbon Fiber/Polyester
Composite”, Proceedings of the Is’ International Conference on Carbon Fibers, Plastics
Institute, 1971, p. 258.

30. “ED0 Partner Expands”, Fleets and Fuels, Vo1.3, N0.18, August 26, 1996, p.4.

31. Dalmolen, B., Akzo Nobel, Private Communication, September 15, 1997.

32. Rosen, B.W., “Analysis of Material Properties”, Engineered Materials Handbook, Volume 1,
Composites, 1987, p. 190.

33. Tidwell, J., “Low Cost, Low-Weight Type 2 CNG Development Program”,Thiokol Proposal
to Lucas Aerospace, August 28,1997.

Project: 9081-32 14 November 1997 Page 20 of 20

mlO88jyw
Lucas Aerospace CNG Cylinder Fiber Reinforcement Study Powertech Labs Inc.

Appendix A

Owens Corning Advantex TM

Acid Test Results

Project: 9081-32 14 November 1997

m1088jyw
 Guelph Tso Corrosion Studv

paw Materials:
-
1. Chopped Strand Mat 3 layers each [approximate thickness 0.127
a) -
E glass M723 from Mexico 450 g d s q m
b) -
3709 glass M723 from Guelph Feb 1995 production
c) -
5075 glass M723 from Guelph March 1995 production

2. Continuous roving [approximate thickness 0.307

-
a) E glass Amarillo 1) 3578 AC 21 1 (2400 TU()
2)495 DF 208 (2362 TU()

b) 3709 glass - Guelph 1) 3578 AC 2400

2)495 AA 2400

-
c) 5075 glass Guelph 1) 3578 AC 2400
2)495 AA 2400

3. Resin - Alpha-Owens Coming E704-BI lsophthalic Polyester

Exposure: 1 Normal Sulfuric Acid (about 5%) at 20 deg.C

Note: All laminates were post cured at 90 deg.C for 3 hrs.

- -
Tests: 1. Tensile ASTM 0638 Test Type I
-
2. Flexural ASTM D790
-
3. Glass Fibre Content ASTM D2584
-
4. Density ASTM 0792
5. Weight Loss
6. B a r d Hardness

The exposure began May 24,1995. Testing dates in days are 0, 1,30,120,500, and 1440.
 - 3709
--c 5075

100 200 300 400 500 600

Exposure Time (days)

Normalized to 30 Weight % Glass Prcpntcd hy Rick Metzcg 27/11/96

 357 Continous Rovin? Flexural Strength Retention

I I

I : I

--
-+E

-
3709
5075-

200 300 400

Exposure Time (days)

Normalized to 30 Weight % Glass Prcpnrd hy Rick Ma~zcg 2711 IN6

 1

. -

357 Continous Roving Tensile Strength

0 100 200 300 400 500

Exposure Time (days)

Normalized to 30 Weight 9b Glass Prcparcd by Rick Maixcg 27/11/Y6

 a

357 Continous Rovinp Tensile Strength Retention

I IO% .-

100%
. . .

.....
:
. ,,. ' ._...,. *;
.' i , ,- ,

.
,!/ . .
.....
. .
- IC
I
;, . .I
... .:
-',I...,._ ..
:I ...
._ ii i...i
..
. (..._ . . . . . .
... . . ..
.. .. .!.. .:. ... ... ... .. . . .
> ~

--
-.---_-

3709
. .
. . . .
.r ..
. . 5075
.......
I
I .
. . .... ..,
. . . .
...... . . .I

.. ....
/I

..: ., , .
. .., .. (. _. .. .. . .
. . . . .

I I
60%

SO%
0 100 200 300 400

Exposure Time (days)

Normrlizcd to 30 Weight % Glass

T
8
0
 495 Continous Roviw Flexural Strenpth Retention

100%

- 3709

D 200 300 410 SO0 600

Exposure Time (days)

Normalized to 3 0 Weight % Glass Prcparcd by Rick Motxeg 27/11/96

 7
. .
. ,
: >
. ...
. .. ... ./#
.. I
...
. . .,..:.
. ,...
i.
.. >
'
: i
,., ,'*..'
.,, . ,...
i

-
.
,
.. .,
.
..
. ."
.. .
. . .~.-
8 I .
(_..
. ,
P
h r
. I
5
L
E I
 -r w
. _ . -
'.. . I..
. , ..
,.,.. I :
8 . YI
e
.:I. .
6
, .
. . . .
I
5.
*.J
 -r 8
n
0
n
D '
 Mat Flexural StrenPth Retention

-+E

\ -
*3709
5075

0 100 200 300 400 5ou 600

Exposure Time (days)

Normalized to 33 Weight % Glass Prepared by Rick Mn~zcg 2611 1/90

 \

Mat Tensile StrenPth

2s

23

21
I
-
-+E
3709
-+5075
--

200 300 4 0 5 0 am
Exposure Time (days)
 Strength Retention
3
$I

. .
. : ..

.
. ..
..L.'. .

. .

A i