Modeling and simulation of explosion effectiveness as a function of blast and crowd characteristics

Suicide bombing has become one of the most lethal and favorite modus operandi of terrorist organizations around the world. On average, there is a suicide bombing attack every six days somewhere in the world. While various attempts have been made to assess the impact of explosions on structures and military personnel, little has been done on modeling the impact of a blast wave on a crowd in civilian settings. The assessment of an explosion's effect on a crowd can lead to better management of disasters, triage of patients, locating blast victims under the debris, development of protective gear, and safe distance recommendations to reduce the casualties. The overall goal of this work is to predict the magnitude of injuries and lethality on humans from a blast-wave with various explosive and crowd characteristics, and to compare, contrast, and analyze the performance of explosive and injury models against the real-life data of suicide bombing incidents. This thesis introduces BlastSim---a physics based stationary multi-agent simulation platform to model and simulate a suicide bombing event. The agents are constrained by the physical characteristics and mechanics of the blast wave. The BlastSim is programmed to test, analyze, and validate the results of different model combinations under various conditions with different sets of parameters, such as the crowd and explosive characteristics, blockage and human shields, fragmentation and the bomber's position, in 2-dimensional and 3-dimensional environments. The suicide bombing event can be re-created for forensic analysis. The proposed model combinations show a significant performance---the Harold Brode explosive model with Catherine Lee injury model using the blockage stands out consistently to be the best with an overall cumulative accuracy of 87.6%. When comparing against actual data, overall, prediction accuracy can be increased by 71% using this model combination. The J. Clutter with Reflection explosive model using Charles Stewart injury model with blockage works best for confined-space incidents with an accuracy of 80%. Blockage in a crowd can increase the accuracy by 17% for all models. Line-of-sight with an attacker, rushing towards an exit, announcing the threat of a suicide bombing, sitting inside a vehicle or building, and standing closer to a wall or a rigid surface were found to be the most lethal choices both during and after an attack. The findings can have implications for emergency response and counter terrorism.