You are on page 1of 6

Transactions of the 17th International Conference on Paper # K14-1

Structural Mechanics in Reactor Technology (SMiRT 17)


Prague, Czech Republic, August 17 –22, 2003

Dynamic Analysis of Large Steel Tanks


Alejandro P. Asfura1), Basilio N. Sumodobila2) , Farzin R. Beigi2)
1)
APA Consulting, Moraga, California, USA
2)
ABS Consulting, Oakland, California, USA

ABSTRACT

Dynamic non-linear fluid-structure interaction analyses were performed for two large diameter steel tanks with
floating roofs. The diameter of the tanks varied between sixty and one hundred meters with a height of approximately
twenty meters. The tanks’ wall and roof were modeled with plate elements, the soil with equivalent non-linear spring
elements, and the fluid with non-linear fluid elements. Hoop and buckling stresses in the tanks’ wall, uplifting of the
tank bottom, and sloshing height were calculated. Stresses and sloshing heights were compared to the same quantities
given by the API [1] and New Zealand recommendations [2] for the design of tanks.

KEY WORDS: tank, fluid-structure interaction, sloshing, non-linear, dynamic analysis.

INTRODUCTION

The most commonly used design code/recommendations for the seismic design of flat bottom tanks are the API
[1] and New Zealand recommendations [2] (referred to as “NZ code” here on after in this paper). They are relatively
simple codes and their application is expected to result in successful designs. This paper presents the comparison
between the application of these two codes and the results obtained by a more elaborate analytical methodology. Even
though it is clearly understood that engineering analysis methods, no matter how sophisticated they are, have also
implicit simplifications and assumptions, they will give results that can be considered more accurate for a specific case
than the application of a general code. This is true especially for localized responses.
Non-linear dynamic analyses were performed for two large steel tanks. Selected results from these analyses
were compared to the quantities obtained by the application of the design codes mentioned before. This comparison
will give an indication of the accuracy or conservatism of these codes in the calculation of the selected results.

TANKS CHARACTERISTICS

The two tanks included in this study have steel plate walls with flexible flat bottoms and flexible floating roofs.
The first tank (Tank 1) has a diameter of 92.3 meters, a wall height of 21.4 meters and a maximum fluid height of 20.0
meters. Its foundation consists of a reinforced concrete ring. The second tank (Tank 2) has a diameter of 60.5 meters, a
wall height of 19.8 meters and a maximum fluid height of 18.0 meters. Its foundation consists of a reinforced concrete
pad on piles.

SOIL CHARACTERISTICS

The foundation soil for Tank 1 consists of a 10-meter thick soil layer with shear wave velocity of about 200 mps
overlaying a half space of firm soil with shear wave velocity of about 500 mps. Tank 2 is founded on deep soil with
shear wave velocities ranging from 175 to 200 mps.

SEISMIC ENVIRONMENT

The maximum surface ground accelerations at the site of Tank 1 are 0.66g in the horizontal direction and 0.40 in
the vertical direction. For Tank 2, the maximum ground acceleration is 0.30g in the horizontal direction and 0.20 in the
vertical direction. Using the acceleration response spectra defined for these sites, acceleration and displacement time
histories were developed to match the spectra. The energy content of the seismic input for Tank 1 is in the 3.0 hz to 8.0
hz frequency range. For Tank 2, the energy content is in the 2.0 hz to 6.0 hz frequency range.

FINITE ELEMENT MODEL

Each tank’s wall, bottom plate, and floating roof were modeled using plate elements with beam elements for
stiffeners. The pad and ring foundations were modeled with plate and beam elements, respectively. The fluid inside the
tanks was modeled using non-linear fluid elements (ANSYS element Type FLUID80). This element is incompressible
and it is free to move relative to the tanks’ shells in the vertical and tangent directions. This fluid element is also free to
move relative to the floating roofs in the horizontal direction.

1
The tanks are not connected to their foundation, so they can slide and uplift, thus the contact between the bottom
of the tank and the foundations was done using a non-linear contact-friction element (ANSYS element Type
CONTACT52). This element acts only in compression and provides the sliding resisting forces between the tank
bottom and the foundation.
Since Tank 1 is founded on a concrete ring; most of the tank bottom is in direct contact with the soil. The
contact between the tank bottom and the soil and the soil itself were modeled by a combination of elements that allow
for uplift and friction (combination of ANSYS elements Type CONTACT52 and COMBIN40). For Tank 2, which is
supported on a concrete pad on piles, the soil was modeled by linear equivalent horizontal and vertical soil springs
(ANSYS element Type COMBIN40). They do not allow for the uplifting or sliding of the pad. The interface between
the tank bottom and the pad was modeled by non-linear contact-friction elements. The stiffness constants of these
springs included the effects of the piles and the soil and were estimated using soil-pile interaction analysis using
computer code SASSI (Reference 3).

A typical tank model is shown in Figures 1, 2, and 3.

Figure 1: Tank finite element model – Shell and Fluid

Figure 2: Tank shell and base plate

2
Figure 3: Floating roof elements

GRAVITY ANALYSIS

To verify the models, gravity analyses were performed for both tanks. The pressure obtained using the
mathematical model was compared to the static pressure for the liquid-floating roof weight. Table 1 lists the
comparison of the maximum pressure. As expected, the effect of the floating roof on the pressure is negligible.

Table 1: Comparison of maximum gravity pressures calculated at center of gravity of bottom element

Tank Gravity Analysis (ton/m2) Static Pressure (ton/m2)


Tank 1 17.8 17.7
Tank 2 15.8 15.8

MODAL ANALYSIS

The fundamental frequencies of the tanks, which mainly correspond to sloshing and impulsive modes, are listed
in Table 2.

Table 2: Main frequencies

Tank Sloshing mode (hz) Impulsive mode (hz)


Tank 1 0.15 1.61
Tank 2 0.22 1.95

DYNAMIC ANALYSIS

Dynamic non-linear time history analyses were performed for the two tanks using the non-linear finite element
code ANSYS [4]. In this paper, only some key results from the dynamic analysis are presented and then compared with
the results obtained by using the procedures in the API and New Zealand codes. The results shown here are the
maximum hoop stress in the tank wall, elephant-foot buckling stress, slosh height, and tank wall uplift.
Table 3 presents the maximum wall uplift displacements for both tanks. Figures 4 and 5 show the wall uplifting
time histories for Tanks 1 and 2, respectively.

Table 3: Non–linear analyses results – Shell Uplift

Tank 1 Tank 2
Maximum uplift height (m) 0.004 0.055

3
Tank 1 - Uplift Time History

0.0045

0.0040

0.0035

0.0030

0.0025

Uplift, m
0.0020

0.0015

0.0010

0.0005

0.0000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00
Time, sec

Figure 4: Uplift time history, Tank 1

Tank 2 - Uplift Time History

6.00E-02

5.00E-02

4.00E-02
Uplift, m

3.00E-02

2.00E-02

1.00E-02

0.00E+00
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00
Time, sec

Figure 5: Uplift time history, Tank 2

CODE ANALYSIS

Using the API and NZ codes for the design of tanks, hoop stress, elephant-foot buckling stress, and slosh height
were calculated and compared to the analytical results. The established site-specific response spectra for both sites were
used as the input motion. For the API method, the product ZIC1 was set to the response spectrum ordinate
corresponding to the impulsive frequency of the tanks. Per API, the spectrum for the factor ZIC1 corresponds to a
damping coefficient of 2 percent of critical. Since API does not include a method for calculation of tank impulsive
frequency, this frequency was obtained using the method in the NZ code, which includes the soil effects. Similarly, the
factor ZIC2 in the API method was set equal to the spectrum ordinate corresponding to the first sloshing period, as
calculated by the API method. The spectrum for calculation of factor ZIC2 corresponds to a damping coefficient of 0.5
percent of critical.
Since the site-specific horizontal and vertical spectrums were also used for the NZ code method, the product of
αβΑP and α’βΑP defined in that code are set equal to 1.0. In the NZ method, the spectra for calculation of impulsive
and convective loads correspond to 10 and 0.5 percent of critical damping, respectively. The spectrum for vertical
modes of vibration corresponds to 5 percent of critical. Figures 6, 7, and 8 show the comparison of the results of
rigorous analysis of Tanks 1 and 2 to those using API and NZ codes.

4
Hoop Stress

Hoop Stress (Ton/m2)


40,000
30,000
20,000
10,000
0
Tank 1 Tank 2
Analysis 31,400 31,000
API 650 33,500 29,800
NZ 30,100 27,500

Figure 6: Comparison of Tanks’ Hoop Stresses

Buckling Stress
Buckling Stress (Ton/m2)

1,500

1,000

500

0
Tank 1 Tank 2
Analysis 402 864
API 650 547 1,062
NZ 861 1,360

Figure 7: Comparison of Tanks’ Buckling Stresses

Maximum Slosh Height

0.60
Slosh Height (m)

0.40

0.20

0.00
Tank 1 Tank 2
Analysis 0.24 0.31
API 650 0.41 0.33
NZ 0.49 0.42

Figure 8: Comparison of Tanks’ Slosh Height

5
CONCLUSION

For the two tanks studied in this paper, the hoop stresses calculated using dynamic non-linear time history
analyses and those calculated using the approaches given in the API and NZ codes are similar differing only by about
10%. The buckling stresses and the sloshing height calculated using the API and NZ codes are higher than the values
calculated with the more rigorous method. The buckling stresses are overestimated by as much as 114% using the NZ
method, and by 36% using the API method over that obtained by the rigorous analysis. Similarly, the wave height is
overestimated by a maximum of 104% from NZ method, and by 71% from API method over the analysis results.
The comparisons shown here, even though corresponds to a very limited number of cases, indicate that the
approaches in the API and NZ codes give reasonable seismic demands and are adequate for the seismic design of large
tanks.

REFERENCES

1. American Petroleum Institute, Welded Steel Tanks for Oil Storage, API Standard 650, 10th Edition, Washington
D.C., 1998.
2. National Society for Earthquake Engineering, Seismic Design of Storage Tanks, Recommendations of a Study
Group of the New Zealand National Society for Earthquake Engineering, December 1986.
3. Computer Code SASSI. A System for Analysis of Soil-Structure Interaction. Geotechnical Engineering Division,
Civil Engineering Department, University of California, Berkeley, 1999.
4. Computer Code ANSYS. A General Purpose Finite Element Program. ANSYS Inc.

You might also like