Emulation-Based Uncertainty Quantification of a Fire Dynamics Simulation

SmartUQ and Westinghouse

This paper explores the use of emulators (also known as meta or surrogate model) for computational fluid dynamics (CFD) of fire scenarios at nuclear power plants. The computational burden of CFD limits the uncertainty analysis to point estimates of critical points. For example, using the publicly available Fire Dynamics Simulator for uncertainty quantification can take from a few hours to multiple days to run. Developing an accurate emulator that can rapidly generate response surfaces and predict uncertainty estimates would improve plant risk estimates in critical areas.

This article will be published in the fall of 2017 at the American Nuclear Society (ANS) Probabilistic Safety Assessment (PSA).