HIGH FIDELITY DESIGN, CHARACTERIZATION AND OPTIMIZATION OF PURDUE UNIVERSITY SUBCRITICAL PILE THROUGH A BENCHMARKED OPENMC FRAMEWORK

This thesis presents a detailed and comprehensive analysis of Purdue University Subcritical Pile system using Monte Carlo-based simulations through the OpenMC code. The primary objective of the study is to design the subcritical pile, benchmark the model with experimental and analytical measurements, characterize the neutron and gamma radiation fields within the pile and explore optimized configurations to enhance flux distribution and overall pile performance. Multiple geometric and source arrangements, including standard, optimized, and hybrid versions, were modeled and evaluated to assess their impact on spatial flux uniformity and energy spectrum.

The results reveal that the Hybrid Optimized Version-4 configuration offers the most significant improvement in the pile’s neutronics performance, producing up to 2.5 times higher total neutron flux, over 3 times higher epithermal flux, and approximately 1.5 times greater gamma flux compared to the baseline configuration. Spectral analysis across various neutron energy groups for example, thermal and epithermal energy groups demonstrate accurate modeling of neutron behavior, while the gamma flux distribution highlights the importance of photon characterization in nuclear energy environments.

Furthermore, this study emphasizes the future potential of using simulation-generated datasets to train machine learning and deep learning algorithms for real-time neutron flux prediction. The integration of physics-based simulations with data-driven techniques can overcome limitations of traditional experimental measurements and provide a robust predictive framework. The insights from this work can also guide experimental designs, detector calibration, material testing strategies, and development of next-generation computational neutronics tools.

Overall, this research establishes a solid foundation for advancing both the simulation fidelity and experimental utility of a subcritical systems and sets the goal for future interdisciplinary work combining nuclear physics, computational modeling, and artificial intelligence.