An adaptive multigrid method for steady-state simulation of petroleum mixture separation processes

On the basis of improvements in the analytical techniques for petroleum fraction characterization, the available detail in compositional information will strongly increase in the future. The inclusion of this detailed characterization on a molecular level seems to be the only way to fundamentally enhance the predictive capabilities of models for petroleum processes. An algorithm is presented that allows the solution of such models without actually solving the model in terms of single-component balances. Instead, the model is formulated using a distribution function to represent the composition. Using an adaptive wavelet - Galerkin discretization in combination with a multigrid approach, the algorithm is capable of solving the model with high accuracy without actually solving the high-dimensional nonlinear model equations. By satisfaction of the needs for scalable model formulation and an error controlled solution at high compositional detail, the approach presented seems to provide a promising alternative to currently employed pseudocomponent methods.