A General-use Metaheuristic Optimization Framework for Isotope Production Target Design at the High Flux Isotope Reactor: Case Study of Tungsten-188

Few advanced optimization methods exist for isotope production (IP) campaigns at Oak Ridge National Laboratory's (ORNL) High Flux Isotope Reactor (HFIR), leading to years of conservative and historical approaches with minimal innovation. Given the growing demand for new and existing radioisotopes, which is beginning to challenge HFIR's capacity, this work explores the development and integration of a metaheuristic (MH) optimization framework to improve target designs and irradiation strategies. Specifically, the framework focuses on maximizing the specific activity (SA), a critical production metric, of tungsten-188 (W-188), a routinely HFIR produced isotope. The approach utilizes Gnowee, a Python-based MH algorithm, integrated with Monte Carlo N-Particle version 6 (MCNP6) and Oak Ridge Isotope Generation (ORIGEN) codes for simulation and evaluation of various target designs and irradiation scenarios. To manage the computational demands of the full HFIR model, a novel simplified model is proposed alongside the complex model, streamlining the framework before more accurate SA predictions. Variables explored for practical implementation include irradiation location, number of cycles, and the number of samples in the target. Additional innovative modifications include the incorporation of an aluminum-based spacer material, variation in the thickness of tungsten samples and spacers, and the use of spacers made from other materials that could amplify the neutron flux in the target and/or are co-producing secondary targets. Validation against experimental data confirms the full model's accuracy in predicting the SA. Furthermore, optimizations show that thousands of simplified model candidates require a similar amount of time as a single full model simulation. By comparing the results of the two models, the framework's ability to explore the design space fully and converge on top-performing candidates is verified. Results demonstrate over 30% SA increases through practical modifications and up to 80% with theoretical innovations, affirming the framework's efficacy. This achievement establishes a versatile, general-use, computationally efficient optimization tool applicable not only to this test case isotope, but potentially other IP campaigns at HFIR and state-of-the-art facilities alike, thereby advancing isotope production capabilities.