A combined sensitivity analysis and model reduction workflow for the simulation of cake filtration

The macroscopic modeling of cake filtration involves a lot of parameters. Therefore it is important to know, how the solution depends on the parameters. Having sensitivity information available, we can use the information in combination with model reduction for the simulation of cake filtration. Two techniques, which use the sensitivity information in the POD snapshot matrix are considered. They differ in the choice of the weighting of the different entries of the snapshot matrix. In one technique the weights are changed for every new parameter configuration. This leads to higher computational costs, since for every new parameter configuration the POD modes are computed. To overcome this drawback, we combine the technique with interpolation. This reduces the computational costs up to 5 ti mes.