A methodology has been developed to quantify the simulation uncertainty of a computational model calibrated against test data. All test data used in the study undergoes an experimental uncertainty analysis. The modeling software ROCETS is used and its structure is explained. The way the model was calibrated is presented. Next, a general simulation uncertainty analysis methodology is shown that is valid for calibrated models. Finally the ROCETS calibrated model and its simulation uncertainty are calculated using the general methodology and compared to a second set of comparison test data. The simulation uncertainty analysis methodology developed and implemented can be used for any modeling with a calibrated model. The methodology works well for a process of incremental testing and recalibration of the model whenever new test data is available.

Numerous reinforced concrete buildings vulnerable to earthquake induced collapse have been constructed in seismic zones prior to the 1970s. A major contributor to building collapse is the loss of axial load carrying capacity in non-seismically detailed columns. Experimental investigations have shown that non-seismically detailed columns will only experience axial failure after shear failure and subsequent lateral shear strength degradation have occurred. Therefore, column shear failure and degrading behavior must be modeled accurately before axial collapse algorithms can be properly implemented. Furthermore, accurate modeling of the degrading lateral-load behavior of columns is needed if lateral load sharing between structural elements is to be assessed with reasonable accuracy during seismic analyses. A calibrated analytical model was developed that is capable of estimating the lateral strength degrading behavior of RC columns prone to shear failure. Existing analytical models poorly approximate nonlinear column behavior and require several nonphysical damage parameters to be defined. In contrast, the proposed calibrated model provides the engineering community with a valuable tool that only requires the input of column material and geometric properties to simulate column behavior up to loss of lateral strength. In developing the model, a database of RC columns was compiled. Parameters extracted from database column-tests were scrutinized for trends and regression models relating damage parameters to column physical properties and boundary conditions were produced. The regression models were implemented in the degrading analytical framework that was developed in this project. Two reinforced concrete columns exhibiting significant inelastic deformations prior to failing in shear were tested in support of the analytical work. A newly developed Vision System was used to track a grid of targets on the column face with a resolution of three-thousands of an inch. Surface column deformations were measured to further the understanding of the fundamental changes in column behavior that accompany shear and axial failure and validate the proposed analytical model. This research provides the engineering community with an analytical tool that can be used to perform nonlinear dynamic analysis of buildings that are at risk of collapse and help engineers improve retrofit techniques. Further insight into shear behavior attained through this project is an important step toward the development of better shear and axial degradation models for reinforced concrete columns. / text