2010 Structures Congress © 2010 ASCE 2702

DYNAMIC ANALYSIS AND DESIGN OF LARGE COMPRESSOR FOUNDATIONS IN

HIGH SEISMIC ZONE

Song F. Jan, Ph.D1 and Sheng C. Wu, Ph.D2

Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

1
Principal Engineer, Bechtel Corporation, 3000 Post Oak blvd., Houston, Texas 77056,
sjan@bechtel.com
2
Principal Engineer, Bechtel Corporation, 3000 Post Oak blvd., Houston, Texas 77056,
swu@bechtel.com

ABSTRACT

Very large heavy centrifugal compressors are required in the design process for petrochemical,
liquefied natural gas (LNG) plants, and power plant facilities. In a LNG plant, it requires
multiple compressor units that contain methane, ethylene and propane compressors for
converting natural gas to LNG by cooling the liquid temperature to -270oF. The heavy
compressor vibrating machines are supported on concrete table top pedestal that consists of
concrete slabs, beams, columns and mat foundation. For plant site located in a high seismic area,
the dynamic analysis using the response spectrum method is applied to calculate the seismic
force for design of the machine anchorage and the tabletop pedestal/ foundation. Also, the
steady-state vibration analyses of the table top pedestal are performed to ensure that during the
normal operating condition, the resultant maximum dynamic displacement (or velocity)
amplitudes are within the limiting amplitudes set by machine manufactures and industrial
standards. Both concrete cracked and un-cracked properties are considered in the analysis.
Concrete members and connections are designed to comply with the seismic provisions in
Chapter 21 of ACI 318-08 to ensure that the concrete pedestal will behave in a ductile manner
during severe seismic events.

This paper presents the design of a typical concrete table top structure supporting two
compressor units. The plant is located in high seismic area. The presentation includes foundation
sizing criteria and design loads, finite element modeling of the concrete table top structure,
dynamic steady state vibration analysis due to machine imbalances, vibration response limit
criteria, seismic response spectra analysis, and evaluating seismic design forces based on the
static and dynamic analysis results. Due to heavy weight machines and deck, the P- ∆ effect is
considered in the static design. The reinforced concrete structural structural members are
designed in accordance with ACI 318-08 seismic provisions.

1. INTRODUCTION

Designing a concrete framed structure (also referred to as Table Top Pedestal) that supports
multiple units of heavy rotating machines is a great challenge task. For example, in a liquefied
nature gas (LNG) plant, it requires multiple compressor units of methane, ethylene and propane
compressors in order to convert natural gas to LNG by cooling the liquid temperature to -270oF.
Each machine units consists of several compressor and turbine skids with different machine

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2703

operating frequencies. The allowable vibration limits for the compressor machines with high
operating speeds are extremely small (less than 1 mil or 0.001 inches). For soil supported
foundation, the effects of upper and lower bound soil properties are evaluated for static and
dynamic design of concrete pedestal. For a plant site located in a high seismic area, dynamic
response spectrum analysis is required to calculate the seismic force for design of machine
anchorage and the pedestal/ foundation. The following sections will present the design process
from preliminary sizing to final design of a typical a typical table top pedestal and foundation as
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

shown in Figure 1. The compressor building houses two trains of methane compressors, two
trains of ethylene compressors and two trains of propane compressors. In order to minimize
vibration response amplitudes, the methane, ethylene and propane compressors are supported on
separate pedestal structures and mat foundations. This paper will present the design procedures
for Methane Compressor pedestal.

2. MACHINE DATA

Two trains of methane compressors are supported on one common table top pedestal. Each train
consists of two compressors and two gas turbines. :

• Compressor operating speed = 104.5 cps

Total weight of compressors = 522 kips
• Gas turbine operating speed = 64.7 cps
Total weight of turbines = 370 kips

Total weight of machine for two trains = 2 x (522 + 370) = 1784 kips

3. GEOTECHNICAL DATA

The compressor table top is supported on soil foundation. The soil parameters are as follows:

• Wet unit weight of soil = 118 pcf

• Modulus of subgrade reaction = 60 k/ft3
• Poisson ratio = 0.35
• Allowable bearing pressure:
Substained loads: 2500 psf
Substained loads with seismic or wind loads: 3300psf (1/3 increase)
• Shear wave velocity (Site Class D):
Lower bound = 600 ft/sec
Upper bound = 1200 ft/sec
• Soil system damping: 10% (Conservative)

4. PRELIMINARY SIZING

In the initial design phase, the required information are foundation layout drawing, vendor
machine data and soil/pile bearing capacity. The machine data includes general arrangement
showing machine location, outline dimension of machine base and anchor bolt layout, foundation

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2704

bolt specification, weight and location of the center of gravity for the combined machine
assembly and for each componet.

Based on the above information, the preliminary selected table top structure and mat foundation
sizes are as follows:

•
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

Mat size: 93ft x 164 ft x 4ft ( 9200 kips)

• Deck size: 87ft x 128ft x 3.3 ft ( 5500 kips)
• Columns: 40 – 3.3 ft square x 18 ft length (1200 kips)

Total structural weight including mat =15900 kips

Check sizing criteria:

• Wdeck/Wequip = 5500/1784 = 3.06 > 1.0 (ok)

• Wmat/(Wdeck+Wequip) = 9200/(5500+1784) = 1.26 > 1.0 (ok)
Wtotal structural/Wequip = 15900/1784 = 8.9 > 3.0 (ok)
• Check column dimension:
Column unsupported length L = 20 ft
Minimum column dimension h = L/8 =2.5ft ( > 2.0ft)
Use of 3.3ft square column is ok.
• Check soil bearing for dead loads only:
Pbearing = (Wtotal structural+ Wequip)/Amat = (15900+1784)/(93 x 164) = 1.16 ksf
Pbearing < 0.5 x Allowable bearing pressure = 1.25 ksf (ok)

5. FINITE ELEMENT MODELING

The table top structure is modeled as a 3-dimensional model as shown in Figure 2. It consists of
1230 plate elements for deck slab, 2400 plate elements for mat and 80 line elements for 40
columns. The soil stiffness is represented by Winker spring element.

GT STRUDL computer program is used for static and dynamic analysis.

6. STEADY STATE VIBRATION ANALYSIS

The dynamic loads due to machine imbalance are applied on the machine supports at the deck.
These machine unbalance forces are provided by equipment vendors, and are applied at the
machine anchorage locations at base of equipment skid. If the unbalanced force is not available,
it may be estimated as follows:

Fo = me e ω2

Where me = rotor mass;

ω =machine operating speed (radian/sec).
e = rotor eccentricity, obtained from vendor or to be estimated.

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2705

For start- up or coast-down conditions, the unbalanced forces should be modified as:

Fs = ( fs/fo )2 Fo

Where fs = frequency during start or coast-down

fo = normal operating frequency
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

The unbalanced forces obtained from venders for each train are:

• Two compressors = 8.8 kips

• Two gas turbines = 2.2 kips

Total unbalanced forces for two trains = 2 x (8.8+2.2) = 22kips

The unbalanced forcing function may be applied as a harmonic load as:

F(t) = Fosin(ωt + φ)

Where F(t) = Value of the forcing function at any instant of time “t”
Fo = Peak value of the forcing function
ω = Frequency of the forcing function
φ = Phase angle

The dynamic analysis for structures subjected on harmonic loads may be performed in either
time domain or frequency domain. For the compressor foundation, the analysis will be
performed in frequency domain since the peak steady state response is required for harmonic
loads applied at one or more forcing frequencies.

In GT STRUDL program, the amplitude of the harmonic forcing function is specified as:

Fo = a ω2 +b ω + c (where a, b and c are constants)

Under normal operating condition, dynamic analyses are performed for both lower and upper
bound soil stiffness with un-cracked concrete section property. Critical damping ratios are 0.04
and 0.10 for concrete structure and soil system, respectively. The effect of concrete cracking on
dynamic responses is also investigated. Cracked concrete property is conservatively based on
section 10.10.4.1 of ACI 318-08 by reducing the moment of inertia of concrete section. The
ratios of moment of inertia between cracked and un-cracked sections are 0.7 and 0.25 for
concrete deck and column, respectively. Mat foundation is considered as un-cracked as soil
supported.

The peak-to-peak displacement amplitudes from the dynamic analysis results are summarized in
Table 1.

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2706

All fundamental frequencies are well below the machine operating frequencies of 64.7cps and
104.5cps. No resonance will occur during normal operating.

The maximum peak-to-peak displacement response is 0.09 mils at the fixed base condition. It is
well below the allowable limit of 0.4 mils as specified by equipment manufacture. It is noted
that concrete cracking has little effect on the machine vibration response.
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

If the allowable limit of dynamic response is not available from manufacture, the following
industrial standards can be followed:

• Machine vibration, Part 1, ISO 10816-1:1995(E)

• General limits of displacement amplitude for a particular of vibration by Richart (1962)
• Criteria for vibrations of rotating machinery by Blake (1964)

Start-up and coast-down conditions are also investigated. The dynamic response is not critical
under these conditions and will not be included in this paper for simplicity.

7. RESPONSE SPECTRUM ANALYSIS

Based on ASCE 7-05, the structure is classified as occupancy category III. Therefore, modal
response spectrum analysis per section 12.9 of the ASCE code is required. The seismic design
parameters are as follows:

Soil Site Class “D”

Occupancy Category III
Seismic Design Category “D”
Occupancy Important Factor of structure I = 1.25
Response modification factor: (Table 15.4-1 of ASCE 7-05)
Special reinforced concrete moment frame R = 8
Mapped Acceleration parameters:
At short period (0.2s) Ss= 1.12
At one second period S1= 0.48
Design Spectral Acceleration parameters:
At short period (0.2s) SDS= 0.79
At 1- second period SD1= 0.49
Critical Damping Ratios:
Concrete structural damping 0.04
Soil system damping 0.10

Development of the design response spectrum curve for 5% damping is presented in section
11.4.5 of ASCE 7-05. Spectral curves for other damping values are required due to variations in
structural and soil damping values. Newmark (1973) and Riddell (1976) established an
amplification-damping relationship. This procedure can be used to develop response spectrum
curves for all damping ratios less than 1.0. It is used to develop the response spectrum curves
shown in Figure 3 for seismic response analysis of the compressor foundation.

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2707

Two orthogonal excitations are applied in the response spectrum analysis. As for machine
vibration analysis in previous section, the modal response spectrum analysis is also performed
for both lower and upper bound soil stiffness. Both cracked and un-cracked concrete sections are
considered. The dynamic response of displacements and base shears are summarized in Table 2.

The maximum amplitude of displacement is 0.87 inches for lower bound soil property with
considering concrete cracking effect. The height of the pedestal is 23 feet. The allowable story
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

drift based on Table 12.12-1 of ASCE 7-5 is:

∆ = 0.015hsx =0.015 x 23 x 12 = 4.14 in. > 0.87 in. OK

Per Section 12.9.4 of ASCE 7-05, if the base shear from response spectrum analysis is less than
85% of base shear calculated by using equivalent lateral force procedure, the member forces
from response spectrum analysis should be multiplied by 0.85(V/Vt):

Where
V = the equivalent lateral force procedure base shear
Vt = the base shear from modal response analysis

The seismic base shear per equivalent lateral force procedure:

V = Cs W

Where Cs = seismic response coefficient determined as:

Cs = SDS /(R/I)

For special reinforced concrete moment frame:

Cs = 0.79/ (8/1.25) = 0.123

From table 2, the maximum acceleration at deck is 0.97g. The corresponding inelastic
acceleration is:

ainelastic = 0.97/(8/1.25) = 0.152g

The value of inelastic acceleration is greater than acceleration value calculated from the
equivalent lateral force procedure. Therefore, the results from response spectra analysis will be
used in the deign of tabletop structure without scaling the member forces.

Based on the dynamic response presented in Table 2, it is concluded that:

• The effect of concrete cracking on total base shears and deck accelerations is negligible.
• The effect of concrete cracking may increase the amplitude of displacement up to 75%.
• The effect of soil stiffness on total base shears and deck accelerations is less than 10%.

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2708

• The fixed base condition has about 15% lower base shear and deck acceleration than the
condition for soil-structure interaction (SSI).
• The soil stiffness has significant effect on displacement amplitude. The lower bound soil
stiffness may have 60% more displacement than fixed base.
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

8. STATIC ANALYS AND DESIGN

8.1 Loads

The following loads are normally considered in the design of compressor table top structures:

Dead load (DL): Including the weight of structure, mat and equipment.
Live loads (LL);
Thermal loads (T);
Seismic loads (E);
Dynamic loads during normal operating (DY);
Machine loads:
a. Normal torque (DYT): obtained from vendor or estimated as
Torque (ft-lbs) = (7040) (power rating in kw)/ RPM
b. Short Circuit Torque (SCT):
Do not combine with wind or seismic loads.

Note: Wind load is less than seismic load. Therefore, wind load is not included in this paper.

The calculated seismic inelastic acceleration from response spectrum analysis is 0.152g at deck
level. It is conservatively applied as equivalent static load to the whole structure for concrete
design. In addition, the vertical seismic load effect Ev = 0.2SDSDL is also considered per section
12.14.3.12 of ASCE 7-05. The seismic forces in two horizontal orthogonal directions are
combined as 100% in one direction plus 30% in other direction per section 12.4.3.1 of ASCE 7-
05.

8.2 Load Combinations

The load combination for the compressor framed structure is in accordance with ASCE 7-05.
The normal torque load and the thermal load associated with the pipe and machine expansion
load are grouped with dead load in the load combinations

U = 1.4DL
U = 1.2(DL+T+DYT) + 1.6(LL+DY+ SCT)
U = 1.2(DL+T+DYT) + 1.0(LL+DY) + 1.0E
U = 0.9DL + 1.0E

8.3 Reinforced concrete Design

8.3.1 Special Concrete Moment Frame

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2709

Finite element analysis has been perform for evaluating the second order effect of P-∆ effect
(including P-δ) on lateral displacement of the structure. Due to the large column cross-section of
the pedestal, the P- ∆ effect is negligible.

In the static finite element analysis for concrete design, all structural members are considered as
un-cracked sections.
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

The design of concrete structure is in accordance with Chapter 21 of ACI 318-08 as special
moment frames with seismic design category (SDC) D.

The concrete design results are:

• Columns: 3.33 ft square, 28 - #9 rebar

• Deck: 87ft x 128ft x 3.3 ft , # 9 @ 8” top/bottom rebar
• Mat: 93ft x 164 ft x 4ft, #9 @ 12” top/bottom rebar

The maximum net baring pressure under static dead and live loads is 1.85 ksf. It is less than the
allowable net bearing pressure of 2.5 ksf. The maximum net baring pressure under combination
of seismic and static load case is 3.10 ksf which is within the allowable pressure of 3.3ksf.

8.3.2 Intermediate Concrete Moment Frame

The concrete pedestal may be designed as an intermediate concrete moment frame structure.
The response modification factor R is 3 for intermediate frame with height less than 50 feet. The
corresponding inelastic acceleration will be:

ainelastic = 0.97/(3/1.25) = 0.4g

The bearing pressure would be increased by a factor of 2.6 under 0.4g design acceleration, and
would be above the allowable limit. The size of mat foundation needs to be increased to
accommodate the increased design accelerations. Due to the available space limitation in plant
layout, this design option is not adopted

9 CONCLUSIONS AND RECOMMENDATIONS

Based on the discussions in the previous sections, the following conclusions are provided:

• Concrete cracking has little effect on the machine vibration response.

• It is conservative to use the fixed base boundary condition for evaluating the machine
vibration response of high-speed machine foundation.
• The effect of concrete cracking on total seismic base shears and deck accelerations is
negligible.
• The effect of concrete cracking may increase the amplitude of displacement up to 75%
under seismic load.

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2710

• The effect of soil stiffness on total seismic base shears and deck accelerations is less than
10%.
• The soil-structure interaction (SSI) condition has about 15% higher seismic base shear
and deck acceleration than the fixed base condition.
• The soil stiffness has significant effect on displacement amplitude under seismic load.
The lower bound soil stiffness may have 60% more displacement than fixed base.
•
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

The P- ∆ effect is negligible due to the large column cross-section of the pedestal.
• Design of special moment frame for compressor foundation may be required due to the
space limitation in plant.

10. REFERENCES

ASCE 7-05. (2005). “Minimum Design Loads for Buildings and other Structures,” American
Society of Civil Engineers.
ACI 318-08. (2008) “Building Code Requirements for Structural Concrete (ACI 318-09) and
Commentary”, American Concrete Institute.
Blake, M. P. (1964). “New vibration Standards for Maintenance,” Hydrocarbon Processing and
Petroleum Refiner, Vol. 43, No. 1 (Jan.), pp. 111-114.
GTSTRUDL User reference Manual (2007). Georgia Institute of Technology.
ISO 10816-1:1995(E). “Mechanical Vibration, Part 1” International Organization for Standards.
Newmark, N. M., et al. (1973). “Seismic Design Spectra for Nuclear Power Plants,” Journal of
the Power Division, ASCE, vol. 99, No. P02.
Richart, F. E., Jr. (1962) “Foundation Vibrations,” Trans. ASCE, Vol. 127, part 1, pp. 863-898.
Riddell, R., and Newmark, N. M. (1976). “Stastistical Study of Earthquake Response Spectra,”
Second Chilean conference on Seismology and Earthquake Engineering, Santiago, Chile,
pp. B2-1 to B2-15.

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2711

Table 1. Machine Vibration Analysis Response

Horizontal - X Horizontal - Z Vertical - Y
Peak-to-Peak
Frequency Mass Frequency Mass Frequency Mass Displacement
(cps) (%) (cps) (%) (cps) (%) (mils)
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

Lower Bound Soil

Uncracked section 3.17 74.5 2.91 63.5 4.50 90.9 0.067

Lower Bound Soil

Cracked section 2.75 66.0 2.53 55.6 4.42 86.6 0.072

Upper Bound Soil

Uncracked section 3.91 53.0 3.59 42.6 8.51 76.1 0.071

Upper Bound Soil

Cracked section 3.14 49.2 2.88 38.3 8.17 69.7 0.073

Fixed Base 22.4 24.8

Uncracked section 4.33 87.1 3.97 64.7 23.5 11.7 0.08
26.8 20.6
Fixed Base 14.9 12.7
Cracked section 3.34 88.0 3.06 64.8 18.8 11.6 0.09
20.1 13.2

Table 2. Response Spectrum Analysis Results

Horizontal - X Horizontal - Z
Base shear Displacement Acceleration Base shear Displacement Acceleration
(kips) (inches) (g) (kips) (inches) (g)
Lower Bound Soil
Uncracked section 7213 0.54 0.89 6568 0.61 0.92

Lower Bound Soil

Cracked section 7153 0.83 0.92 6556 0.87 0.97

Upper Bound Soil

Uncracked section 6721 0.45 0.90 6134 0.48 0.94

Upper Bound Soil

Cracked section 6693 0.77 0.85 6087 0.66 0.91

FIXED BASE
Unracked section 6136 0.38 0.82 5561 0.33 0.87

FIXED BASE
Cracked section 6176 0.64 0.82 5586 0.55 0.87

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2712
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

Figure 1. Typical LNG Compressor foundation Layout

Figure 2. Finite Element Model

Structures Congress 2010

 2010 Structures Congress © 2010 ASCE 2713

Ground Response Spectra

1.20

2%

1.00 5%
Downloaded from ascelibrary.org by MISSOURI, UNIV OF/COLUMBIA on 03/15/13. Copyright ASCE. For personal use only; all rights reserved.

10%
20%

0.80
Acceleration (G)

0.60

0.40

0.20

0.00
0.1 1.0 10.0 100.0 1000.0

Frequencies (CPS)

Figure 3. Ground Response Spectra

Structures Congress 2010