In engineering design, optimization methods are typically based on trial-and-error design and local search approaches. In nuclear reactor design, conservative approaches are directly linked to safety margin in operation under anticipated and unanticipated off-normal scenarios. These approaches require iterative experimentation and analyses (time) to locate the optimum. Moreover, the actual response surface (design solution within design space) may have multiple, regional maxima and minima. The search process may only yield a local optimum, not an absolute optimum. Thus, these practices in reactor engineering need to be changed when applied to next generation nuclear systems. Design conservatism must be reduced in search of efficiency with respect to performance, while meeting regulatory (licensing) requirements.
Recently the speaker and students developed and demonstrated a new systematic design optimization methodology. Three key elements (tools), Computational Fluid Dynamics (CFD), Artificial Neural Network (ANN), and Response Surface Methodology (RSM) were integrated into the design engineering process to limit, reduce, and eliminate past conservative methodologies and to improve the optimization process. To test this methodology, the Printed Circuit Heat Exchanger (PCHE) was selected as a Case Study since the PCHE is being considered as the Next Generation Nuclear Plant's (NGNP) Intermediate Heat Exchanger (IHX). Further, a thermally efficient IHX is key to an optimal NGNP balance-of-plant.