Long simulation time (thousands of years) crashes due to mass transport computation issues

Hey. I am trying to run a variably saturated, mass transport model. The goal is to recreate glacial / interglacial groundwater simulations of saltwater intrusion. however, the model will always end up stuck in a loop where an area gets excessively concentrated areas (different areas in the transition zone each time), where I end up with a tiny area that has impossible concentrations, as well as negative concentrations at the landward side. I have tried to make a smaller mesh, and to make the time series of the sea level rise and fall in constant steps to reduce computation costs. I guess my question is, if anyone has prior experience with this type of issues, how to make a transient state model not to loop due to mass transport issues, and how to make such a model run for thousands of years, without it actually taking thousands of years. Adam

Hi Adam, Density-driven flow problems are hard to solve, but not impossible. Negative concentrations are due to the poor discretization level and dispersivity values assigned to the model. If I were you, I would do the model in steps. Does your model run with constant head values (not tidal effects)? If no, this should be fixed first. You can take a look on the self-paced course below. There is a section covering some information about numerical oscillations and their connections with the mesh size and dispersivity. https://www.theacademybydhi.com/course-sessions/feflow---getting-started-with-groundwater-quality-modelling-11600061-153 Best regards Carlos Rivera