An adaptive multiscale galerkin method for the simulation of continuous mixture separation processes

Abstract:
The paper is concerned with the adaptive model size reduction for multicomponent mixture processes. Modelling the dynamics of such processes using a continuous distribution function to represent the mixture's composition leads to distributed differential-algebraic systems. These systems have to be discretized in time and in the distributed variable. A fully adaptive discretization scheme is developed using a Runge-Kutta time discretization in combination with a Wavelet-Galerkin discretization. The time step adaptivity is controlled by an error estimator based on step doubling. The error estimator for the spatial discretization error is based on a linearized high resolution model. The proposed technique is an alternative to pseudocomponent lumping methods which are currently used to reduce the model complexity of multicomponent mixture processes. With this novel approach the lack of error control and adaptivity in pseudocomponent lumping can be resolved.

Please use the Online Form to request an electronic copy