Extreme loading conditions generated by blasts that result from terrorist attacks can cause devastating consequences for structures and their occupants. In steel frame structures, the damage and failure of a structural column from a blast load can result in a progressive collapse and catastrophic failure of the entire structure. The objectives of this dissertation were to develop experimental methodologies, analysis strategies and threat assessment tools that can be used to mitigate blast hazards and predict damage in structural steel frame structures. Typically, guidelines and methodologies are developed from conclusions drawn via field testing with live explosives, but due to the harsh environment created by explosives, collecting reliable data is problematic. In an effort to provide key answers to questions related to how structures behave during these types of explosions, the University of California, San Diego Blast Simulator was developed. The Blast Simulator consists of ultra-high speed actuators that accelerate/decelerate impacting modules toward the specimen in a controlled manner. Utilizing the transfer of momentum and energy of masses at high velocities, blast-like loads are applied to the structure without the use of explosive materials. Since no fireball is generated, specimen behavior can be observed for the full duration of the loading using high-speed photography. For this dissertation, experiments were conducted with the Blast Simulator to study the effects of impulsive loading on typical steel W-Shapes. The objectives of the dissertation were met through an experimental and computational effort that included the characterization of column behavior, development of a testing protocol for loading of columns in both weak and strong axis directions, validation and implementation of numerical models to predict loads delivered to specimens as well as column behavior and generation of relations pertaining to Simulator operation. Finally, the validated finite element models were incorporated into the development of a fast running, threat assessment tool using an artificial neural network that can predict the residual capacity of a blast damaged column. The fast running model allows for the analysis of steel structures in situations where a time consuming computational model may not be efficient
