Predictive modeling of coupled multi-physics systems: II. Illustrative application to reactor physics

• We applied the PMCMPS methodology to a paradigm neutron diffusion model.
• We underscore the main steps in applying PMCMPS to treat very large coupled systems.
• PMCMPS reduces the uncertainties in the optimally predicted responses and model parameters.
• PMCMPS is for sequentially treating coupled systems that cannot be treated simultaneously.

This work presents paradigm applications to reactor physics of the innovative mathematical methodology for “predictive modeling of coupled multi-physics systems (PMCMPS)” developed by Cacuci (2014). This methodology enables the assimilation of experimental and computational information and computes optimally predicted responses and model parameters with reduced predicted uncertainties, taking fully into account the coupling terms between the multi-physics systems, but using only the computational resources that would be needed to perform predictive modeling on each system separately. The paradigm examples presented in this work are based on a simple neutron diffusion model, chosen so as to enable closed-form solutions with clear physical interpretations. These paradigm examples also illustrate the computational efficiency of the PMCMPS, which enables the assimilation of additional experimental information, with a minimal increase in computational resources, to reduce the uncertainties in predicted responses and best-estimate values for uncertain model parameters, thus illustrating how very large systems can be treated without loss of information in a sequential rather than simultaneous manner.