You are on page 1of 4

International CAE Conference 2014

27-28 October. 2014

Seismic analysis of a liquid storage tank used


in wine industry: a FEM-based approach
Denis Benasciuttia, Luciano Moroa, Nicola Cimentib
a

Universit di Udine, Dip. Ingegneria Elettrica Gestionale e Meccanica (DIEGM), Udine, Italy

Gortani s.r.l., Amaro (Udine), Italy

Email: denis.benasciutti@uniud.it
Web: www.diegm.uniud.it

Summary
This paper illustrates an example of seismic analysis of a thin wall cylindrical tank used in the wine
industry. A FEM-based approach, based on response spectrum analysis, is used to simulate the
seismic response of the tank/liquid system to horizontal earthquake action. The tank/liquid system is
modeled by shell elements and displacement-based fluid elements, respectively. Simulations results
are compared with estimations given by Eurocode 8 (Part 4), according to two different approaches: a
simplified lumped dynamic model by Malhotra, a FEM-based static analysis with the hydrodynamic
pressure computed by Eurocode 8. The different approaches are shown to provide a general
agreement for the shear and overtuning moment at the tank base. Elastic and elasto-plastic buckling of
the tank wall is finally assessed with the stress values calculated by the previous approaches.

Keywords
liquid tank, seismic analysis, ANSYS, Eurocode 8

Introduction
Thin wall metallic tanks are used in the wine industry for wine fermentation and conservation.
Typical damage of the tank wall during an earthquake is due to elastic ("diamond shape") or elastoplastic ("elephant foot") buckling. In Italy, the seismic assessment of liquid tank has become more
important after some recent earthquakes, as those in the Emilia region in 2012 (see Fig. 1(a)).
The European standard for tank seismic design is EN 1998-4:2006 Eurocode 8 (Part 4) [1], which is
supported by other codes defining the seismic ground actions (for example, the Italian reference is the
Ministerial Decree January 14th 2008 "Technical rules for constructions" [2]). In this work, the results
by Eurocode 8 are compared with FEM-based simulations with ANSYS software, which applies a
response spectrum analysis to a liquid/tank finite element model. The different approaches are
compared by considering the shear and overtuning moment at the tank base. The tank structural
integrity against elastic and elastic-plastic buckling is also assessed. The main goal of this study is to

International CAE Conference 2014

27-28 October. 2014

evaluate the applicability of the FEM-based approach as a possible substitute of Eurocode 8,


especially for those configurations that are outside the range of applicability of the design code.

(a)

(b)

Figure 1: (a) "elephant foot" damage after earthquake in Emilia region (Italy), May 2012; (b)
geometry of the liquid tank analysed in this study

The case study of a cylindrical tank subjected to horizontal earthquake action is discussed [3]. The
tank is fully anchored at its base, it has diameter 2R=8 m, height H=7.3 m and three courses of
thickness 3, 2.5, 2 mm (from bottom), see Fig. 1(b). The material is AISI 304 stainless steel. The tank
is located in the city of Corropoli (TE), Italy, classified as seismic zone 2.

Seismic analysis by FEM: response spectrum analysis


Figure 2(a) shows the finite element model used in the seismic analysis. The tank is modeled by 4nodes shell elements and the liquid is represented by 8-nodes fluid elements (called FLUID80 in
ANSYS [4]). The fluid element has a displacement-based formulation, where the fluid is
characterized by a "fluid elastic (bulk) modulus" Ef (the value Ef =2240 MPa used in simulations is
calibrated based on the values of static pressure computed in a preliminary static analysis). This
element is particularly suited to model fluid/structure interaction and fluid sloshing. The fluid
elements are not directly attached to the shell elements at the tank wall. Instead, they have separate
and coincident nodes that are coupled only along the direction normal to the interface, so to allow
relative movements in the tangential and vertical directions. Similarly, fluid element nodes at tank
base are constrained only along the vertical direction, while they can slip in the horizontal direction.
Due to symmetry, only one half of the tank is necessary. On the other hand, for a tank subjected to
horizontal excitation in one direction only, the seismic behaviour is symmetric about a vertical plane
containing the diameter parallel to the excitation direction ("cos-type" vibration modes). Preliminary
simulations have confirmed the correctness of this assumption, which means that no vibration modes
are lost due to symmetry.
The numerical simulation applies a response spectrum modal analysis with an acceleration spectrum
defined by the Italian regulation "Technical rules for constructions". The modal analysis adopts the
"matrix condensation" technique (based on Guyan method) the only one supported by the FLUID80
element of ANSYS where the finite element model is reduced to a specific set of Master Degrees of
Freedom (DOF), which are used to approximate the mass matrix [M]. The Master DOFs are selected
at the vertical Z direction on the free surface of the liquid and in X direction (the earthquake
direction), where the liquid/tank system is expected to vibrate. Fig. 2(b) shows an example of modal
response for the first impulsive mode at fn = 9.41 Hz (colour map refers to the vertical displacement).

International CAE Conference 2014

27-28 October. 2014

(a)

(b)

Figure 2: (a) the fluid/tank finite element; (b) modal response at fn = 9.41 Hz

Seismic analysis by Eurocode 8


The Eurocode 8 Part 4 (EC8-P4) suggests two different approaches to estimate the seismic response
of the liquid/tank system. The first is a simplified dynamic model (proposed by Malhotra [5]), which
provides analytical expressions to estimate the first natural periods Timp and Tcon of the convective and
impulsive system responses, as well as the shear and overtuning moment at the tank base.
Alternatively, EC8-P4 gives analytical expressions to compute the hydrodynamic pressure for the
liquid, which can then be applied to a finite element model of the tank to perform a static structural
simulation. Base shear and moment are then calculated as output of finite element simulations.

Approach

Method

Shear (kN)

Moment (kNm)

EC8-4

analytical (Malhotra)

1017

2853

EC8-4

hydrodynamic pressure applied to


FEM static analysis

1117

2222

FEM

modal combination ("SRSS") rule

867

2567

Table1: Shear and overtuning moment at the tank base, calculated by different methods

A comparison of natural frequencies calculated by different methods (analytical and numerical) is


presented in [3]. Table 1 provides, instead, a comparison of the shear and overtuning moment at the
tank base, calculated by different approaches. A general agreement is observed.

Structural integrity: elastic and elasto-plastic buckling


The stress distribution calculated by the previous approaches are used to evaluate the stability of the
tank wall near and above the base for failure modes related to elastic and elasto-plastic buckling. The
analytical expressions for the buckling strength of anchored tanks can be found in [1] and [3]; they are
also summarized by the design chart in Fig. 3, taken from the literature [6].
Elastic buckling generally depends on the geometrical imperfections of the tank wall and on the
internal pressure (which has a stabilizing effect). Instead, elasto-plastic buckling normally occurs
close to the tank base (see the "elephant foot" damage in Fig. 1(a)), due to a combination of vertical

International CAE Conference 2014

27-28 October. 2014

compressive stress and high hoop stress close to the yield limit. An empirical equation is available in
Eurocode 8 (based on Rotter's model [7]), where is shown that the buckling strength decreases with an
increasing internal pressure. The buckling strength of the tank wall is the lowest between elastic and
elasto-plastic strength, see Fig. 2. In the case study analysed in this paper, the buckling strength of the
tank wall was higher than the axial stress calculated by all the approaches examined.

Figure 2: Elastic and elasto-plastic buckling strength of anchored steel cylindrical tanks [6]

Conclusions
This work compared different approaches for the seismic analysis of liquid tanks used in wine
industry. All approaches provide comparable results in terms of base shear and overtuning moment,
although they differ in respect to advantages and disadvantages. The methods suggested by EC-P4 are
relatively simple; however their applicability is restricted to a specified range of geometries (e.g.
ratios H/R within the range 0.33). Instead, the proposed FEM-based approach that models explicitly
both tank and liquid is more flexible, as it can be applied to any type of liquid/tank configuration.

References
[1] EN 1998-4:2006 Eurocode 8 Design of structures for earthquake resistance Part 4: Silos, tanks
and pipelines.
[2] D.M. 14 gennaio 2008 - Nuove Norme Tecniche per le Costruzioni (NTC).
[3] Bearzi D., Benasciutti D., Cimenti N., Moro L.: "Verifica sismica di serbatoi per l'industria
enologica: normativa tecnica e modellazione numerica", Proc. of 43 Conference of the "Italian
Association for Stress Analysis" (AIAS), September 9-12, 2014, Rimini (BO), Italy. (in Italian)
[4] ANSYS User Manual, Release 14.0
[5] Malhotra P.K., Wenk T., Wieland M.: "Simple procedure for seismic analysis of liquid-storage
tanks", Structural Engineering International, No. 3, 2000, pp. 197201.
[6] Hamdan F.H.: "Seismic behaviour of cylindrical steel liquid storage tanks", Journal of
Constructional Steel Research, No. 53, 2000, pp. 307333.
[7] Rotter J.M.: "Local collapse of axially compressed pressurized thin steel cylinders", Journal of
Structural Engineering, No. 116(7), 1990, pp. 1955-1970.

You might also like