Experimental and Numerical Analysis of the Influence of Fluid–Structure Interactions on the Dynamic Characteristics of a Flexible Tank

Abstract

This study evaluated the effect of two-way fluid-structure interactions (FSIs) on the dynamic characteristics of flexible storage liquid tanks. A hybrid approach, combining the finite volume method (FVM) and the finite element method (FEM), denoted as FVM/FEM, was used to model the response of a flexible water tank under seismic loading. The fluid domain was simulated using FVM, while the structural domain was represented using FEM. A two-dimensional interaction equation was solved at the contact surface between an elastic tank wall and the fluid by tuning the relaxation parameter and convergence conditions. The proposed FVM/FEM model provides fundamental insights for modeling interactions two-way FSI. The model enables evaluating the effect of considering two-way FSI on the dynamic characteristics of a tank flexible to a rigid tank wall. The accuracy of the coupling FVM/FEM method was confirmed through a comparative experiment with previous studies and design code, including the quantities of natural frequency, liquid sloshing, and hydrodynamic pressure. The results showed that the frequency of flexible tank walls differed from that of rigid walls, especially the hydrodynamic pressure of liquid motion acting on the tank. The peak hydrodynamic pressure of the fluid acting on thick-walled when considering FSI is 38.2 kPa, deviating only 0.2% from the ACI standards (38.12 kPa) or 5.47% with EC8 (36.11 kPa). However, the study shows that when considering the two-way interaction for thin-walled tanks, this deviation from ACI and EC8 increases significantly to 12.2% and 16.8%, respectively. From that, the numerical results show a "threshold value" is revealed to distinguish flexible or rigid tanks. If the tank stiffness exceeds this threshold, then the thicker the tank, the lower the hydrodynamic pressure. However, if the tank stiffness is lesser, vice versa. A good agreement is observed between the numerical, analytical, published, and experimental data.