b"Accelerated technologyModeling reduces size, time, and cost of experiments, increases development throughexperiment effectiveness, saves millions of dollars by avoiding unnecessary new extrapolation andexperiments, and improves new technology time-to-deployment.validation methods D uring the design of any experiment, the designer must assess how close the experiment reproduces the physics compared to the full-size installation the experiment attempts to represent. To assess this, scaling theory involving expert judgment is needed. One challenge of this step is to bound the error prediction of a model when applied to a facility yet to be built, starting from the comparison of the model prediction with a set of experimental results. This problem is known as the validation extrapolation step. Three areas of research PROJECT NUMBER:to advance the state-of-the-art in this crucial step and reduce reliance on expert 21A1050-014FP opinion are dynamic system scaling, physics-guided coverage mapping, and a nonlinear generalization of the representativity concept.TOTAL APPROVED AMOUNT: $1,455,000 over 3 years This project matured the physics-guided coverage mapping and representativity scaling methodologies. The project also developed a framework in which these PRINCIPAL INVESTIGATOR:methodologies can be combined with data informed physical models and artificial Mohammad Abdo intelligence-driven optimizers to inform experiment design and code validation. The CO-INVESTIGATORS: development of the scaling methodologies and of data informed physical models Aaron Epiney, INL were handled as two independent tasks. As a demonstration application, existing Botros Hanna, INL Transient Reactor Test facility fuel experiment data was used to uncover gaps in Charles Folsom, INL existing experiments and propose additional experiments.Christopher Ritter, INL Models of the experiments were created and perturbed according to user-defined Congjian Wang, INL distributions reflecting best estimates and uncertainties. The theories were analyzed Cristian Rabiti, INL to accomplish three primary objectives:Daniel Wachs, INLRamon Yoshiura, INL 1. Evaluate the applicability of a prototype model to a complete target power plant.Hany Abdel-Khalik, Purdue University 2. Extend the theories to handle time dependent, nonlinear models, and the simultaneous consideration of multiple experiments.COLLABORATOR: 3. Propose new experiments necessary to encompass the entire design space of Purdue University normal conditions for the full target scenario.Separate advancements of the algorithms were implemented within the Risk Analysis and Virtual Environment platform, after which common components were abstracted into a unified class. Expansions were made to each theory in pursuit of these objectives. For instance, the representativity factor was employed at every time step, which enabled monitoring of the dynamic evolution of the representativity factor. Additionally, the utilization of the Markov-Chain Monte Carlo technique was integrated to facilitate a more comprehensive parameter calibration process.In the case of physics-guided coverage mapping, the Purdue team used regressors like K-nearest neighbors to extend the theory's applicability to transient effects involving multiple responses. Also, the Markov-Chain Monte Carlo approach was used within the transformed or phase space to conduct data calibration.22"