Identification of continuous granular flow models

Granular materials may exhibit different pattern forming behaviors, depending on the average energy per grain. Various granular flow PDE models exist, each capturing different behaviors of the physical phenomenon. In the present work we investigate the model and parameter identification problem of different continuous granular flow models as an encapsulated optimization problem. The identification problem is then split in a series of inverse problems. For the discrimination of the different models, the Fisher information matrix is used and different optimality criteria are discussed. Basic concepts of algorithmic differentiation (AD), which is used for the computation of the sensitivity matrix, are also given. The PDEs are discretized by the finite element method.