Hard to say what causes this problem.
I tried to reproduce this using a 3 layer pseudo-1D-model (layer 1-2 HDPE and clay, following description above, all layers horizontally), plus a 2 m layer below (with standard FEFLOW parameters). top BC is head = 24.2 m, bottom BC = 20.2 (no-Flow BC else)
The resulting flux brings a nearly perfect match with the analytic solution (deviation below 0.001 %). Do you think my model is representative?
regards,
Alex