A Method for Parameter Optimization of Reactive Flow Continuum Models by Sequential Quadratic Programming

This paper presents a procedure for performing parameter optimization of ignition and growth continuum models for high explosive systems. Continuum modeling of reactive flow in high explosive systems can yield highly accurate predictions of experimental observations. However, the numerical parameter optimization that is needed to establish these predictions generally requires many evaluations of a model, and in the case of continuum reactive flow models, each evaluation requires long run times. In terms of keeping the total number of function evaluations low, the variable metric sequential quadratic programming approach is at present the most efficient general nonlinear local optimization method available. In this investigation, the variable metric sequential quadratic programming code NLQPEB is coupled to the two-dimensional high rate continuum modeling programs CALE and DYNA2D. An example parameter optimization of an ignition and growth model for the HMX-based high explosive PAX2A is presented. The optimization procedure adjusts the ignition and growth reaction rate model parameters in order to locally minimize the difference between calculated and experimental pressure histories from Manganin gauge experiments.