Computational modeling of progressive collapse in reinforced concrete frame structures.

摘要

Progressive collapse of structures during severe loading caused by earthquakes, blasts, and other effects causes catastrophic loss of life. Such collapse is typically caused by the inability of the structural system to redistribute its loads following the failure of one or more structural members to carry their gravity loads. In reinforced concrete (RC) structures, the loss of gravity load-carrying capacity in columns has been observed to trigger a chain of collapse events. This is especially true for structures built according to older building code provisions and thus possessing non-ductile reinforcement details. The evaluation of the vulnerability of these buildings to collapse in the event of an earthquake, and its expected improvement as a result of applying proposed seismic retrofit measures, e.g., using fiber-reinforced polymer (FRP) composites; is an important prerequisite in policy decisions and emergency preparedness. The objective of this dissertation is to develop simulation tools for progressive collapse assessment.; The methodology of the research presented in this dissertation is divided into the development of simulation tools on the component and system levels. Component-level developments refer to modeling the behavior of seismically-deficient RC columns, and establishing criteria for their collapse and removal from the computational finite element (FE) model of the structural system. System-level developments refer to modeling the mechanics of removing a structural element from the FE model during the simulation.; Computational component models are developed and experimentally-calibrated for the distribution of confining stresses in fiber-discretized cross-sections of RC columns and for the confinement-sensitive constitutive material behavior of three common seismically-deficient details: (i) inadequately-confined core concrete; (ii) buckling-prone longitudinal reinforcement; and (iii) insufficiently-developed lap-splices. Thus, a pseudo-solid modeling approach is pursued at the higher computational efficiency of uniaxial material models. Cross-section damage indices are formulated by aggregating the effects of hysteretic damage from the constituent fibers according to their respective constitutive models; and used to identify the collapse limit-state and to establish removal criteria for the class of RC columns dominated by axial-flexure interaction. An existing analytical model from the literature is used to identify the collapse limit-state and to establish removal criteria for the class of RC columns dominated by shear-axial interaction. The developed computational models are implemented in an extensible form using an object-oriented software framework. The computational models are used to simulate a number of previous experimental studies on as-built and FRP-retrofitted RC columns with seismically-deficient details. The results of the simulation exhibit the ability to predict not only the as-built response but also the effects and utility limit of the FRP retrofit.; An analytical formulation is developed for the time-dependent problem of sudden removal of a structural element from a FE model using the principles of dynamic force equilibrium. An automatic element removal algorithm is analytically formulated, computationally implemented, and numerically tested for robustness using an idealized benchmark problem. This formulation is extended to include a simplified account of collision between the collapsed columns and the rest of the structure.; The dissertation concludes by presenting demonstration applications of the developed progressive collapse simulation tools using two test-bed structural systems representing older one- and five-story RC-frame buildings partially infilled with unreinforced masonry (URM) walls. The applications include: (i) the deterministic assessment of progressive collapse response (i.e., the interaction between the URM wall and the RC frames, element collapse mode and sequence,