Recommendation of .WEL speed up model

[Gonzalo Eduardo Parada]

Hi everyone,

I'm developing a MFUSG model that includes several drain cells which runs in about 30s. However, when I incorporate pumping wells into the system, the simulation time increases up to 20 minutes.

I would like to ask about your experience in reducing simulation times when adding pumping wells, I know that some optimizations might be done through options of the .WEL, .CLN, .SMS, .OC packages, but I would like to hear about your experiences

At this stage I still dont have a final mesh to use, so I could integrate those recommendations.

Thank you! 

[Lorenzo Lanzafame]

Hi Gonzago, 

The long simulation times are likely due to numerical convergence issues, caused by wells attempting to pump from dry cells. This often happens with multi-layer wells. In the .wel package, pumping rates are set for shallow layers, but the model struggles to target them because the saturated thicknesses are very small or non-existent.

One possible solution is to enable automatic pumping rate reduction. The exact setting depends on the graphical interface you're using. 

Another solution is to adjust the elevations of the screened intervals—or the layers from which the wells are pumping—to avoid drawing from dry or poorly saturated layers. When I encounter these issues, I typically run a simulation with the wells as-is, extract the water table elevation, set this as the new top of the screens (I use GW Vistas, which recreates the .wel package directly from the absolute elevations of the screens), and then rerun the simulation.

Regards

Lorenzo