A comprehensive framework to improve predictions by integrating inverse uncertainty quantification and quantitative validation

The Best Estimate plus Uncertainty (BEPU) approach for nuclear systems modeling and simulation requires that uncertainties in Quantity-of-Interest (QoI) predictions must be quantified in order to prove that the investigated design stays within acceptance criteria. Sources of uncertainties include: -1): parameter uncertainty from design variables or calibration parameters; -2) experimental uncertainty due to measurement noises; -3) model uncertainty caused by missing physics and numerical approximation errors; and -4) code uncertainty as the computer code is not known at every input especially when the code is so expensive to run that surrogate models are required. Failing to account for any of these uncertainties will result in biased predictions. In this paper, we propose a comprehensive framework to integrate results from model calibration and validation to provide robust predictions so that all these sources of uncertainties can be taken into consideration. Inverse Uncertainty Quantification (UQ) is firstly performed to quantify parameter uncertainties based on experimental data. The inverse UQ process takes into account uncertainties from model, code and measurement simultaneously. In the subsequent validation step, the inversely quantified parameter uncertainties are propagated through the computer code to produce QoI predictions with uncertainties, which will be compared with validation data. We will use a quantitative validation metric based on Bayesian hypothesis testing to assess the accuracy of the calibrated models. Such quantitative validation metric (called the Bayes factor) is then used to form weighting factors to combine the prior and posterior knowledge of the parameter uncertainties in a Bayesian model averaging process to prevent over-confidence in the code predictions. In this way, future predictions will be able to integrate the results from inverse UQ and validation to account for all available sources of uncertainties. (authors)