Professional Documents
Culture Documents
Nicolae ZEMTEV*
AMEC NUCLEAR RO
e-mail: nicolae.zemtev@amecnuclear.ro
The calculation presented here was part of a benchmark exercise initiated by the International Atomic
Energy Agency (IAEA). The purpose of the benchmark was to understand how several nuclear
equipments and structures were affected during a strong earthquake and to calibrate simulation
technologies examining their ability to represent the observed behavior and identifying main
parameters influencing the response. The report details the approach used to predict the observed
buckling modes of a vertical water tank, during a strong earthquake. The calculations include modal,
spectral and time-history analysis using two finite element models. The water inside the tank was
modeled using specific Ansys finite elements as described below. Thus, the numerical models
succeeded in capturing not only the water mass, but also its behavior. The calculated results show the
predicted tank behavior is similar to the real one, and the fluid-structure interaction was accurately
modeled.
1. INTRODUCTION
The present report summarizes and discusses the results of the analyses performed by AMEC as part of a
benchmark exercise. The intent was to predict the behavior of a vertical storage tank subjected to a strong
earthquake.
Figures 1.11.3 below show the effects of the earthquake on the actual tank. These included:
-
Damage to the foundation and fixing bolts due to severe uplift of the tank base
459
Nicolae ZEMTEV
460
461
Nicolae ZEMTEV
462
463
Nicolae ZEMTEV
464
Figure 2.8. Acceleration Time Histories Measured On-Site during the Earthquake
Stress, [MPa]
450
400
350
300
250
200
150
100
50
0
0.00%
2.00%
4.00%
6.00%
8.00%
10.00%
12.00%
14.00%
16.00%
Elongation
Figure 2.9. Material Properties (Median Values)
18.00%
20.00%
465
2.4 BOUNDARY CONDITIONS
Fluid and shell coincident nodes at fluid / tank boundary are coupled in the directions normal to the
interfaces. Thus, lateral displacements of the fluid relative to the tank wall and bottom are permitted.
For modal analysis (Model 1), the SHELL63 nodes at the bottom of the tank were restrained in all directions
(displacements + rotations).
Model 2 has a rigid foundation with its nodes restrained in all directions. Contact elements between the tank
bottom and the foundation allow for the development of friction forces (friction coefficient 0.8).
Nicolae ZEMTEV
466
467
For comparison, moment capacity values were estimated by hand calculations using the approach in [1],
considering two cases: 1) when vertical seismic accelerations are oriented upwards, and 2) when vertical
seismic accelerations are oriented downwards. This included the calculation of fluid hold-down forces on the
tank bottom from the pressure distribution resulted from the finite element model. Note that the following
assumptions were made:
-
The maximum uplift of the tank base was 90 mm, which is the maximum elongation of bolt before
break point; this accounts for the evulsion of fixing bolt no. 8.
The failure mode at the tank base was considered elephant-foot buckling
Table 3.1 below gives a comparison between the total overturning moment at tank base and the moment
capacity calculated according to [1]. It is interesting to note that the overturning moment at the tank base due
to seismic movement is 1.87 and, respectively, 1.47 greater than the two computed moment capacities. These
values are consistent with usual values of inelastic energy absorption factors calculated for these structures
and suggests that part of the energy was dissipated through plastic deformations (confirms that the failure
mode at the tank base was elephant-foot buckling).
Table 3.1. Comparison of Demand and Capacity (Overturning Moment)
Case
Vertical accelerations upwards
Vertical accelerations downwards
Overturning
moment, MR
[MNm]
71.6
71.6
Moment capacity,
MC [MNm]
M R / MC
38.3
48.8
1.87
1.47
The results show a maximum equivalent plastic strain of 0.018 at Level 1, near Point 2 (see Figure 3.4),
which is consistent with the observation that partial elephant-foot buckling occurred in the north-west side of
the tank.
Nicolae ZEMTEV
468
At upper levels the results show large maximum horizontal displacements (between 110 mm and 160 mm) as
compared to lower levels (up to 3092 mm at level 4). This in turn demonstrates large elastic deformations at
levels 6, 7, which may suggest diamond type buckling occurred all around the tank, as observed in the field.
Other useful results are:
-
Vertical stresses in the shell near the tank base are greater than 140 MPa
Vertical stresses in the shell at upper levels are, with a few exceptions, greater than 25 MPa
be
p R
= CL 1
ye t sb
S1 + ( ye / 248MPa )
1
1
1. 5
S1 + 1
1.12 + S1
(1)
where
CL =
0.605E t
(R / t sb )
(2)
in which S1 = R/(400tsb), tsb is the tank wall thickness, R is the tank radius, p is the tank internal pressure,
and ye is the effective yield stress of the tank material.
The diamond buckling capacity of the cylindrical shell, bd, under combined axial bending and internal
pressure can be essentially lower bound estimated from:
bd = CL
(3)
where CL is the "classical" shell buckling capacity (equation (2)), is a capacity reduction factor due to
imperfection and is a plasticity reduction factor.
NASA SP-8007 document [4] defines different values for depending upon whether the overall loading on
the cylindrical shell is uniform axial compression or bending, and also incorporates an influence for internal
pressure. From NASA SP-8007:
NASA = +
0.605
(4)
axial = 1 0.901 (1 e )
(5)
bend = 1 0.731 (1 e )
(6)
1 R
16 t sb
(7)
where is an increase factor for internal pressure. We used the following equation to estimate :
for s1 1.0:
= 0.22
where
(8)
469
p
s1 =
Et
R
t sb
=1
0.45
+ 0.18
(9)
if 0.55
if 0.55 < 1.6
(10)
where:
= CL / y
(11)
From equations (1) through (11) above, it can be seen that the elephant-foot buckling capacity decreases with
the increase of pressure, and the diamond buckling capacity increases with the increase of pressure.
Therefore, it can be concluded that at locations near the tank base, where the pressure has maximum values,
the tank wall is more susceptible to elephant-foot buckling, while at the upper levels, where there is less or
no internal pressure, diamond type buckling should occur. This conclusion is also consistent with
observations made on site.
For example, near the tank bottom, where maximum pressures vary between 0.14 MPa and 0.24 MPa, the
elephant-foot buckling stress ranges from 32 MPa to 85 MPa, while above water, where there is virtually
zero pressure, the diamond buckling stress has a lower value: 24.5 MPa (diamond buckling occurs first).
The vertical stresses in the tank shell given in Section 3.3 above have values greater than the buckling
capacities computed here. Therefore, the results show that elephant-foot buckling occurs near the tank base
throughout the circumference of the tank wall, while diamond buckling occurs above water levels.
These results are close to what was observed on site.
4. DISCUSSIONS
The following conclusions can be formulated from the above results:
-
The values of the vertical (compressive) stresses near the tank bottom, are all greater than the
"elephant-foot" buckling axial stress capacities given in Section 3.4. This would suggest that this
type of buckling occurs all around the tank, near its base.
The maximum equivalent plastic strain is 0.018 near the tank base (see Figure 3.4), which is
consistent with the observation that partial elephant-foot buckling occurred in north-west side of the
tank.
The values of the vertical (compressive) stresses above the water level are all greater or close to the
value for diamond buckling given in Section 3. 4. This would suggest that this type of buckling
occurs all around the tank, which is consistent with observations on site
Also, at the upper levels, the results of the time history analysis model show large maximum
horizontal displacements (between 110 mm and 160 mm) as compared to lower levels (up to 3092
mm at mid height). This in turn demonstrates large elastic deformations above the water level, which
may suggest that diamond type buckling occurred all around the tank, as observed in the field.
The main difficulty of the time history analysis was to develop a realistic finite element model that does not
require a long solving time. The problem was only partly resolved, by limiting the number of elements in
exchange for some of the results accuracy.
Nicolae ZEMTEV
470
Another aspect that needs to be considered is the flexibility of the foundation, which may influence the
results and should be accurately modeled.
REFERENCES
1. EPRI , Methodology for Developing Seismic Fragilities, EPRI TR-103959 Research Project, R. P. Kennedy, J. W. Reed, 1994.
2. AMEC Nuclear RO, ThSpec Version 1.6 / 2007.
3. Ansys Inc , Ansys Structural Analysis Guide, Ansys Release 11.0, 2007.
4. NASA , Buckling of Thin Walled Circular Cilinders, NASA SP-8007, 1965.
5. AMEC NUCLEAR RO, Integrated Management Manual man01-99.qa, Ed.1, Rev 2, April, 2011, Engineering Integrated
Management System Manual Developed to be in Compliance with the Requirements of: IAEA Safety Series no. GS-R-3 & GSR-3.1(2006), CNCAN-NMC-02, 04, 05, 12(2003), CAN 3-N286-2/2000,10CFR50 Appendix B, 10 CFR 830.120 Quality
Assurance, ASME-NCA-4000/1998, SR EN ISO 9001 (2008) and SR EN ISO 14001 (2005.)