Unsaturated flow - flow boundaries (flux)

Hello,

I have a question related to the flux boundaries conditions in unsaturated flow. I have performed a 3D unsaturated model, with lateral flux boundary ( inflow ). After simulation when I am checking the budget analyser, I notify that water is also going out of these boundaries ?!? ... The simulation results are not respecting the boundary restrictions ? why does water coming out, when inflow boundaries are imposed ? How can I manage it ?

I used the default parameters for calculation : Richard Head-based standart form, and so on ...

Thank you for your help.

Boris

Boris,

It's hard to tell where the reason is without seeing the model. The budget analysis does a backward calculation of the boundary flows based on the resulting head distribution as a check of the resulting conditions (not just an integration of the boundary conditions that have been set). So there might be different reasons for an inflow shown:
1) If you have an extremely high number of boundary nodes with flows near 0, the summation of values at all these nodes could accumulate small errors.
2) If the solution is not stable and the head distribution near the boundary condition is not correct, the budget analysis can give strange results. In that case, this is a hint to have a more detailed look at model stability, e.g., by placing observation points near the boundary and looking at the behaviour of head/pressure.

Thank you Peter for your fast reply.

I have checked the model, and when switching it in the saturated confined form, the boundaries conditions are respected and the solution very stable. Now, my model is an highly low-permeability heterogeneous one. K-field parameters are distribuated between 1E-5 and 1E-8 m/s.
The porosity (unsaturated parameter) is also set as heterogeneous, and very low [0.05 and 0.09].
The initial pressures are ranged between : - 0.1 PSI and  + 1.0
The saturations from 0.06 to 1 --- the model is highly unsaturated (~ 1/3 )

Should I used instead the default Neumann Conditions the Gradient-type conditions ?
Maybe Can I perfom changes in the unsaturated settings using the Mixed head-saturation form or variable substitution form ?

Thanks for reply

Boris

Hello Boris,

A potential source of error for unsaturated models is insufficient vertical resolution. What are the approximate dimensions of your model domain; in particular, how many elements are available to resolve its vertical extent?

Volker

Hello Volker,

Effectively I was thinking about this ... It is an eleven layers model. With a thickness of one or two meter on the border where the conditions are imposed. Maybe should I think to get larger elements. Also my influx are time varying. The variation is quite huge.

Boris

Boris -- is it one to two meters per layer, or one to two meters total domain height?

In the first case, the vertical resolution would be almost certainly too coarse if you use any of the nonlinear pressure-saturation functions such as van Genuchten.

In the second case, the resolution might be sufficient (~10 to 20 cm per layer) but we should still take a look at the parameters, especially the 'A' parameter if it's the van Genuchten model.

Volker