Abstract:
This dissertation investigated progressive collapse analysis of three-dimensional (3D) reinforced concrete (RC) frames that are optimized for carrying structural loads by introducing a unique simultaneous multi-column removal load path. While the concept of removing a key structural element from a predefined location is typically used as a means of introducing structural damage for progressive collapse analysis, recent studies challenge that this approach is inadequate to describe structural responses generated from extreme events, such as explosions. The investigation includes formulating an integrated computational framework that incorporates a self-training machine learning algorithm. This algorithm is used to train the largest machine learning models of 3D RC frames containing more than 600 optimized structures to predict the posterior based on the trained priors. The efficiency of the computational framework was shown by conducting a comprehensive study on the optimization and behavior of structures considering both static and dynamic time-history loadings, alternative load path due to progressive collapse, and second order (P–delta) effects. Nonlinearity of materials was considered using plastic hinge models. A more detailed constraint handling is presented compared to the state-of-the-art with implementation of a new multilevel nonlinear penalty function. The detailed constraint handling is envisioned to allow unrestricted section dimensions and reinforcing steel details to ensure that system solutions will meet both structural integrity and constructability requirements of the American Concrete Institute and the progressive collapse requirements of the Unified Facilities Criteria.
