Seismic performance assessment of a structure requires either physical tests or numerical simulations. Physical tests are used when the level of understanding on the behaviour of a structural element is not sufficient enough to develop a numerical model. The results from physical tests are used to develop or calibrate numerical models. Due to the scale of civil engineering structures, such as buildings and bridges, the physical tests are not often used in a structural system level. While a few laboratories in the world can test full scale structures, tremendous resources are required to run full-scale experiments. Thus, numerical simulation is more often used for system-level seismic performance assessment. For high-rise buildings, long-span bridges, or safety critical infrastructures such as nuclear power plants, numerical simulations are probably the only method to evaluate the safety of the structures against earthquake events.
Numerical simulations for seismic performance assessment requires numerical model which can reproduce the behaviour of actual structural elements. At the current level of technical development, however, there are still large uncertainties in the predeicted seismic performance of structures. For example, many researchers and engineers in the world participated in the recent blind prediction contests held in Japan and in the U.S. In the contests, detailed data on material properties, structural design, and input ground motions were provided. Even with all the detailed actual data, the variation of predictions was very large. The large variation in the predicted responses may result from adopted numerical models, modelling assumptions, or seismic performance assessment methods.
To make a breakthrough in the seismic performance assessment method, the Seismic Resilience Group in University of Toronto is actively developing an Advanced Simulation Framework which draws on several decades of expertise in numerical modelling of reinforced concrete structures, unique experimental facilities, tremendous computing power offered by SciNet supercomputer, and integration of these elements through hybrid (experiment-analysis) simulation method.
Modelling of Structural Components
A numerical element of a structural component is the basic building block of a numerical model of a structural system. For reliable seismic performance assessment of a structural system, the numerical element should be able to replicate the behaviour of a structural element, such as stiffness, strength, and energy dissipation, when subjected to cyclic loads up to failure. In general, the development of a numerical model is an iterative process which involves experiments, development of theory based on the experimental results, implementation of a numerical model based on the theory, verification of the numerical model against theoretical solution, and validation of the model against experimental results. The process is repeated until the theory and the numerical model can fully explain the behaviour of a structural element. When new design or materials are introduced, this process is repeated to further develop or calibrate the numerical models.
Seismic Resilience Group at university of Toronto continuously focus on improving the numerical model for different structural components through extensive laboratory testing and numerical simulations. For example, Professors Collins and Bentz have developed theories and numerical models for reinforced concrete structures since few decades. The analysis programs based on the theories are one of the most recognized programs in the world. The programs developed in University of Toronto include VecTor Programs, Response-2000, and Membrane-2012. The programs can fully capture shear deformation and failure, crack development and propagation, bond-slip behaviour of reinforcement bars, temperature effects, pre-stress tendon loss, creep, etc. Furthermore, for experimental validation, full scale models of structural components can be built in Structural Testing Facility and various experiments can be conducted.
The soil-structure-interaction (SSI) plays a critical role in structure’s response, when a structure is massive and stiff. Due to the huge computational cost in SSI analysis, the SSI has been neglected or simplified with lumped nonlinear springs. For a safety-critical structure, the soil-foundation system is assumed as equivalent-linear material. Professor Kwon is currently developing a numerical method in which soil and foundation system can be realistically represented in the seismic performance assessment of a structure. His research group is currently focusing on 1) identifying applicability of nonlinear or equilvalent-linear analysis method based on large-scale field test result, and 2) development of simplified yet accurate time-domain SSI analysis method.
Advanced Simulation Framework
For system level seismic performance assessment, all the components should be seamlessly integrated and the interaction between components should be fully considered. The structural components could be in different computer programs such as VecTor programs, Response 2000, OpenSees, Abaqus, etc., or physical specimens. The numerical models may be analyzed with a desktop computer or supercomputer. The physical specimens may be located in University of Toronto’s Structures Testing Facility or in a laboratory at a different institution.
These diverse structural elements can be integrated in a hybrid simulation framework that Professor Kwon has developed for the past years. The Seismic Resilience Group is currently putting large efforts to integrate currently available resources (in-house programs, testing facilities, and supercomputing facility) for most reliable seismic performance assessment. In addition to running nonlinear assessment, the integrated model will be used for real-time or geographically distributed simulation. This research has culminated in the development of UT10 Simulator which tests up to ten uni-axial specimens, such as braces and friction/yielding dampers in a hybrid manner. The UT10 Simulator consists of a loading frame, servo controlled hydraulic jacks, and control programs. The UT10 Simulator can test up to ten specimens with ±800 kN force capacity, or up to five specimens with ±1,600 kN capacity. Interface program, NICON, is used to interface the controller program with an integration software.
In addition, UT-SIM, an open framework for integrated multi-platform simulations for structural resilience has been created. UT-SIM enables multi-scale numerical simulation by harnessing unique powers of several different computer programs in combination with physical testing at geographically distributed facilities around the world.