A fast and accurate solver for the general rate model of column liquid chromatography

Practical application of the general rate model is often hindered by computational complexity and by unknown parameter values. Model simplifications and parameter correlations are not always applicable, and repeated model solutions are required for parameter estimation and process optimization. We have hence developed and implemented a fast and accurate solver for the general rate model with various adsorption isotherms.Several state of the art scientific computing techniques have been combined for maximal solver performance: (1) the model equations are spatially discretized with finite volumes and the weighted essentially non-oscillatory (WENO) method. (2) A non-commercial solver with variable stepwidth and order is applied for time integration. (3) The internal linear solver module is replaced by customized code that is based on domain decomposition and can be executed on parallel computers.We demonstrate the speed and accuracy of our solver with numerical examples. The WENO method significantly improves accuracy even on rather coarse grids, and the solver runtime scales linearly with relevant grid sizes and processor numbers. Code parallelization is essential for utilizing the full power of multiprocessor and multicore technology in modern personal computers. The general rate model of a strongly non-linear two component system is solved in few seconds.