SUMMARY

In reactor physics, the topic of depletion involves modeling time-dependent material compositions. As a nuclear system, such as a power plant, operates, isotopes decay and are transmuted according to reaction rates, leading to a constantly changing nuclide inventory. Modern analysis tools employ a quasi-steady state approach, where steady state transport solutions are coupling with depletion routines. These depletion routines use a fixed set of reaction rates to determine the updated compositions after some length of time. By assuming constant reaction rates, coupling schemes do not accurately model the time dependence of reaction rates needed to represent the time-evolution of compositions. To remedy this, prohibitively small time steps must be used, or else numerical instabilities and oscillations may arise.

This project will develop a hybrid approach using reduced order methods to deplete at sub-intervals between high-fidelity simulations. The reduced order simulations will obtain reaction rates representative of the true spectrum over time without requiring expensive transport solutions at small time scales. Accurate representation of reaction rates will improve the fidelity of the depletion sequence, allowing longer depletion intervals and fewer expensive high-fidelity simulations. A perturbation-theory based reduced order method will be used to predict the change in neutron flux and reconstruct reaction rates at the sub-interval.

This hybrid coupling will be implemented using a custom framework that will hold a reactor model in memory. With this framework, both high fidelity and reduced-order methods to fully represent the problem in a consistent manner. Interfaces to individual programs will be developed to efficiently pass information using minimal disk operations. The framework will be compared to reference cases using identical depletion schemes and with increased temporal resolution.