This presentation summarizes the methodology and results from a recent seismic evaluation of a safety-class fire water storage tank. Tanks such as this are common at both DOE and NRC regulated nuclear facilities and their safety function following earthquake is relied upon for various design scenarios. The evaluation was initiated out of a concern for overstress and the potential loss of contents due to induced seismic anchor motions. Seismic-induced differential movement between a tank and its draw-off piping is a commonly overlooked failure mode but can introduce nozzle vulnerability that could challenge the pressure boundary of the tank and thus compromise tank inventory. Differential movement can be especially important in situations such as: (a) draw-off piping and tank supported by separate foundations; (b) soil site introducing soil-structure interaction effects; (c) inelastic response of the tank and/or its anchorage; and/or (d) fluid-structure interaction from tank contents. An analytical fluid-soil-structure interaction (FSSI) study is performed on an example configuration of tank /nozzle / draw-off piping to characterize the dynamic response behaviors which affect seismic demands on the tank nozzle.

The FSSI study utilizes a 3D finite element (FE) model, based on insights from hand calculations and simplified analyses. The tank itself is modeled with linear shell elements, and is supported on its concrete ring foundation and a continuum soil domain, both of which are modeled with solid elements. Soil is modeled with layers having equivalent-linear strain-compatible properties consistent with a design-level seismic hazard. The bearing support of the tank (in compression) is modeled using contact surfaces between the tank base and the ring foundation / soil. The tank anchorage to its ring foundation is modeled via non-linear spring elements representative of the elasto-plastic behavior of cast-in-place anchor bolts for realistic representation of tank rocking and uplift. Lagrangian fluid elements are used to represent the water contained in the tank, and coupled to the structural model. A viscoelasticity approach is utilized to model the nearly frequency-independent hysteretic response of the soil, concrete, and tank shell. Draw-off piping is modeled with shell elements at the tank connection, and with equivalent beam elements elsewhere. The 3D FE model is analyzed using dynamic response history analysis via explicit time domain integration, with ground motion time histories applied at the base of the soil profile. Sensitivity analyses are also performed to assess the relative significance of different configurations and details.

Three phenomena are found to drive the peak tank stresses around the nozzle and the maximum positive nozzle moment: fluid mass convection, pressure impulses within the fluid, and tank uplift. Fluid convection, fluid impulse, and tank uplift are interrelated behaviors, and peak demands occur when peaks in each of the contributing response behaviors coincide. Each of these behaviors contribute to differential movement between the tank and the supports for the draw-off piping, which is characterized by both a vertical uplift and a translational displacement (along the longitudinal axis of the pipe), both of which contribute to demands on the tank wall.