NRAP is using reduced-order models (ROMs) as a simplified but very efficient computational approach for reproducing predictions from detailed process models over the range of conditions of interest. NRAP is developing and assessing various approaches to the development of ROMs, ranging from simple look-up tables (LUTs) to response surfaces to surrogate models based on artificial intelligence techniques. ROMs are baselined relative to the process models, and the ROMs are subsequently integrated into the system models.
PSUADE (Problem Solving environment for Uncertainty Analysis and Design Exploration) is a software toolkit to facilitate the use of output from detailed process models to develop response surfaces; it was developed by Lawrence Livermore National Laboratory. https://computation.llnl.gov/casc/uncertainty_quantification/