Model selection for performance-based earthquake engineering of bridges

Performance Based Earthquake Engineering (PBEE) is becoming an increasingly attractive alternative to traditional bridge design practice. The PBEE framework enables engineers to predict the seismic performance of a structure at multiple levels, including: force and deformation demands, expected level of damage, and likely repair costs. For bridges, these performance levels can be predicted with numerically intensive nonlinear models of the soil, foundation and structure. However, it is preferable in practice to use simpler methods to model a bridge system.; The objectives of this dissertation were to develop a calibrated numerical model of a reinforced concrete bridge on drilled shafts from three experiments (shake table, centrifuge, and pseudo-static component tests), and to use this model to evaluate the accuracy of a variety of simpler bridge modeling strategies within a performance-based framework. This research resulted in (1) the generation of experimental data from a series of shaking table tests, (2) the characterization of the shake-table specimen using response measurements and system identification methodologies, (3) the development of a calibrated numerical model of a reinforced concrete bridge on a drilled shaft foundation (prototype model), and (4) the evaluation (within a performance-based framework) of the accuracy and precision of simpler foundation and structural modeling strategies.; The accuracy and precision of demand and damage estimates from 12 simpler modeling strategies were evaluated by comparing them to the prototype model using suites of 40 near-field and 30 far-field excitations. For stiff, dry sand, the two-span bridge system was accurately and precisely modeled using inelastic column elements and fixing the columns at the estimated depth of maximum moment within the pile (three column diameters for this bridge system). Furthermore, the accuracy and precision of the demand and damage estimates were insensitive to moderate changes in the depth of fixity, enabling the fixity depth to be estimated based on empirical equations rather than by using a nonlinear soil-foundation model.