A new strategy for validation of hydrocode predictions through the Bayesian analysis of radiographic data is presented. The Bayesian approach allows one to incorporate prior knowledge about the structure of the objects being analyzed and provides the foundation for assessing the reliability of the results. We propose using the hydrocode prediction at radiographic time as the initial or default object model. The object model is altered from its default in a minimal way to match the available radiographs in the Bayesian sense. A full understanding of the degree of validity of the final model relies on the ability to explore and characterize the uncertainty in the model, a relatively new feature in Bayesian analysis. We suggest that a physics-based validation of hydrocodes themselves may require some, if not all, of the basic concepts presented here to infer aspects of the underlying physics models from hydrodynamic experiments.
Keywords: hydrocode validation, physics based code validation, radiography, Bayesian analysis, Bayes Inference Engine, reliability exploration, adjoint differentiation
Get full paper (pdf, 56 KB)
Return to publication list
Send e-mail to author at kmh@hansonhub.com