The commissioning and construction of new nuclear power plants in the United States has dwindled over the past 30 years despite significant innovation in reactor technology. This is partially due to the ever-increasing seismic hazard estimates, which increases the demand on and risk to nuclear power plant structures.
Seismic base isolation is a mature technology which introduces a laterally-flexible and vertically-stiff layer between the foundation and superstructure to significantly reduce the seismic response of the structure, systems, and components therein. Such devices have also been noted to concentrate the displacement response in one plane, reduce higher-mode participation, and provide damping to protect against excessive displacements, all of which aid in increasing safety margins for seismically-isolated nuclear structures. Despite numerous studies analyzing the applicability of seismic base isolation to nuclear power plant structures, some of which are discussed herein, no seismically-isolated nuclear plant has been constructed in the United States.
This study presents a time-domain procedure for analyzing the performance of seismically-isolated nuclear structures in response to design-basis earthquake events using ALE3D. The simulations serve as a parametric study to assess the effects of soil column type, seismic isolation model, superstructure mesh, and ground motion selection on global displacements, rotations, and accelerations, as well as internal floor accelerations. Explicit modeling of the soil columns and superstructures enables detailed analysis of soil-structure interaction. The soil columns analyzed have constant properties over the height of the finite element soil mesh and include rock, soft rock, and stiff soil sites, as well as a "no soil" case for comparison. Four separate 3-dimensional seismic isolation bearing models were coded into ALE3D and validated. These include models for friction pendulum, triple friction pendulum, simplified lead rubber, and robust lead rubber bearings. Lastly, two superstructure finite element meshes were considered: a cylindrical plant design meant to represent a typical conceptual design for advanced reactors, and a rectangular plant design meant to represent an advanced boiling water reactor. The ground motions considered include 30 three-component time history records scaled to meet the seismic hazard for the Diablo Canyon nuclear plant. Every combination of soil column, isolator model, and superstructure were subjected to a subset of three of the harshest ground motions, termed the "basic motions", and the combinations which included the rectangular plant design atop the rock soil column were subjected to all 30 motions.
The results of the various simulations including accelerations in the soil columns and superstructures as well as displacements and rotations in the isolators and superstructures are presented. The results suggest three possible effects: an isolator-type effect, a soil-type effect, and a slenderness effect. The isolator-type effect refers to significant increases in vertical soil acceleration amplifications, isolator uplift/tension, and global rotations including torsion and overturning for friction bearings in comparison to elastomeric bearings. Additionally it is noted that inclusion of lead plug softening has the effect of increasing peak lateral isolator deformations, especially for the ground motions that naturally induce high-amplitude deformations in the bearings. These results suggest that uplift/tension may be troublesome in high-seismic areas and the use of restrainers should be analyzed as a possible solution. Furthermore, these results reinforce the lateral design displacement estimate procedures for seismically-isolated nuclear structures in ASCE4-11.
The results prove that explicit inclusion of the soil column is necessary for proper response characterization and the chosen soil properties greatly affect the efficacy of seismic isolation designs. The soil-type effect comes from observations of comparative simulations which show that, in general, peak isolator uplift/tension and deformation, as well as peak global displacements and rotations including torsion and overturning increase as the soil column becomes less-stiff, regardless of the isolator model or superstructure considered. These results suggest that although seismic isolation can be effective for structures atop a variety of soil columns, it is imperative that a single isolator design only be considered applicable to a corresponding soil column unless extensive analyses prove otherwise for a specific case.
Differences in peak response parameters between the two superstructures point to a possible slenderness effect. Specifically, the isolator deformations as well as the global displacements and rotations are observed to increase for the cylindrical superstructure in comparison to the rectangular superstructure cases utilizing the same ground motion, soil column, and isolator model. Should further research reaffirm this effect, a practical limit could be set for superstructure slenderness.
