Adjoint sensitivity analysis for mathematical models of coupled nonlinear physical processes

The importance of sensitivity analysis of numerical models in studies beyond direct simulations is highlighted. In models with a large number of sensitivity parameters, the adjoint approach, in contrast to a direct parameter perturbation approach, provides an efficient algorithm. The application of adjoint sensitivity analysis for the coupled, nonlinear problems of multiphase multicomponent flow of fluid mixtures and heat transfer is demonstrated here. Adjoint sensitivity equations applicable to a generic mathematical model (linear or nonlinear) have been derived. An illustrative case study involving the performance assessment of a potential nuclear waste repository site is presented.