Hi Leo, thanks a lot for your comments. There are in fact some strange things going on at the upper boundary close to the no-flow boundary and the final boundary conditions is still ongoing work. In the current setup I think the infiltration flux is low enough and there seems to be no fully saturated wetting front.
However, the issues I described also happen when the infiltration flux is set to zero, so I assume the (current) problem is not caused by the boundaries. Unfortunately integration timesteps are getting very low (< 1s) which makes running the simulation for a while to get a stationary solution infeasible for me. Best regards, Samuel On Wed, 2019-03-06 at 15:27 +0100, Stadler, Leopold wrote: > Hi Samuel, > > I haven't simulated flow in the unsaturated zone in the last years. > But so far I can remeber, it is hard to find good boundary > conditions. > > So I have some small comments on the boundary conditions on the top. > If you have a no-flow condition in the middle for water and gas in > combination > with a high infiltation rate on both sides, this can lead to a fully > saturated wetting > front where the gas can not escape. > > One solution is to check if the infitration rate is higher than the > infiltraion > capacity of your soil. If so, it is often better to switch to a > Dirichlet > condition with (fully saturated + pressure). > > Looking on your inital SW_profile, the saturation above the layer > (z=9m) is high, > so a low capillary pressure looks realistic. In general it is easier > to let the problem > settle for a while to get realistic initial conditions. > > Paraview is not the best tool for looking at your results since it > interpolates also > data that is constant inside your scv. You should imangine a scv > arround every > node. > > If you have a fully saturated clay layer above the coarser sand, the > air can be traped > in the corser material. There are some nice movies "Water moevement > in soil" on > youtube. > > Kind regards, > Leo > > > Samuel Scherrer <[email protected]> hat am 6. > März 2019 um 14:36 geschrieben: > > > > > > Hi, > > > > I'm currently trying to set up a 2D model for water, air, and heat > flow > > in the unsaturated zone with a clay layer between two sand layers. > > > > The upper boundary is partly no-flow, partly fixed air pressure, > > temperature, and water infiltration flux. > > The side boundaries are no-flow boundaries. > > The groundwater table is inside the domain, 1 m above the bottom > > boundary, and the bottom boundary has a constant temperature and > gas > > content in water (since only the liquid phase is present) and it is > a > > no-flow boundary for air. > > > > The formulation I'm using is air pressure - water saturation. > > > > As a initial condition I'm using hydrostatic air pressure and a > water > > saturation calculated from hydrostatic capillary pressure, initial > > temperature is the same as the boundaries. > > > > I'm using the Box discretization. > > > > The clay layer extends from 5m to 9m and the spatial parameters are > > assigned such that vertices at 5m or at 9m still belong to the clay > > layer (function isClay_ in problem.hh and spatialparams.hh). > > > > When I run this setup I get high spikes in capillary pressure at > the > > boundaries between clay and sand, even in the initial output file > (*- > > 00000.vtu, shown in pc_initial.png). I also use the gnuplot > interface > > to plot the initial capillary pressure and saturation vs depth > along > > x=0 using the same method as for setting the initial conditions > (i.e. > > using the method initialSwitchVariable_(pos) for getting the > saturation > > and calculating the capillary pressure using the material law). > > In this plot no capillary pressure discontinuities are visible. > > > > In the first output this is only visible at the upper clay-sand > > interface, where the capillary pressure is much lower than it > should > > be, but over time a very high pc develops at to lower clay-sand > > interface(up to 1000 bar, see pc_5min.png and pc_final.png). > > > > During inspection of the vtk-output with Paraview I also saw that > the > > porosity of the clay only reaches to slightly below 9m, while the > > initial water saturation has the values it should have in clay up > to > > slightly above 9m (see porosity.png and > initial_water_saturation.png). > > The porosity is set using porosityAtPos in spatialparams.hh, the > > initial water saturation with initialAtPos. > > > > I figured it might be either due to my poor understanding of the > Box > > method, or might be related to the discontinuous water saturation > (see > > initial_Sw_profile.png). > > > > Does anyone know what my problem is or what the solution could be? > > > > Kind regards, > > Samuel > > _______________________________________________ > > Dumux mailing list > > [email protected] > > https://listserv.uni-stuttgart.de/mailman/listinfo/dumux > Im Auftrag > > Dr.-Ing. Leopold Stadler > > _______________________________________________ > Dumux mailing list > [email protected] > https://listserv.uni-stuttgart.de/mailman/listinfo/dumux > _______________________________________________ Dumux mailing list [email protected] https://listserv.uni-stuttgart.de/mailman/listinfo/dumux
