Moving boundaries seem to be a problem with the current implementation of the conduction model. Can it be addressed in the current conduction method or does the other "classical" approach work better?