Publications

Publications

Computer Simulation of Impact and Collapse of New York World Trade Center North Tower on 9/11

A 3D computer simulation of the World Trade Center North Tower's destruction, modeling the initial impact, fire-induced strength reduction, and progressive collapse. The simulation uses the AUTODYN-3D program and advanced numerical techniques to analyze the entire catastrophic event.

CFD Model Velocity Profiles to Test Data for Inlet Nozzle Region of Nuclear PWR Fuel Assemblies

This study investigates the impact of different inlet nozzle designs on the axial velocity profiles within a PWR fuel assembly. The research, which utilized Computational Fluid Dynamics (CFD) models and validated the findings against experimental air test data, demonstrates how specific nozzle geometries can be optimized to reduce steep velocity gradients, thereby mitigating fuel rod vibration and fretting-wear damage.

Underwater Implosions of Large Format Photo-multiplier Tubes

This study investigates the failure modes of large-format semihemispherical photo-multiplier tubes (PMTs) under high hydrostatic pressure, a crucial aspect for deep-water neutrino detectors. The research focuses on understanding the shock wave generated by a PMT implosion, comparing experimental data to hydrodynamic simulations, and providing recommendations to mitigate the risk of a chain reaction failure in large detector arrays.

Application of CFD for Modeling Flows in Feed-Water Pipelines

This study used a Computational Fluid Dynamics (CFD) model to analyze the flow patterns and velocity profiles in various PWR pipeline configurations. The research demonstrates how geometrical features like elbows, valves, and branches can alter the flow structure and axial velocity distribution, which can differ significantly from straight pipe flow.

Least Squares Finite Element Interpolations for Analysis of Model Free Problem-Solving

This paper introduces a new application of the Finite Element Method (FEM) for solving model-free problems, such as handwritten number recognition. The FEM-based approach is shown to be a resource-efficient and local alternative to global deep learning methods.