Model Timestep Question


I currently have a model that I am running. After running about an hour, the total estimated time to complete was down to around 36 hours with a time step of around 3 seconds. I left it overnight and the estimated time went up to 77 hours with a timestep of 1 second or less.

Background for my model: I am running temperature and water quality with withdraw/return, hydraulic boundaries, and flow conditions.

I am only running one model at the moment. I am using 11 out of 12 available cores.

Is this normal or is there something wrong with my setup.


Hello Echen,
The EFDC+ code estimates the time step based on the Courant-Friedrichs-Lewy (CFL) condition and other criteria. It may be something unusual in the hydrodynamic behavior relies on the water balance or flow balance that causes the time step decreasing. Without the model we cannot tell exactly what it is. However, we recommend you starting with a diagnosis by plotting the time series of the time step to find the moment when the time step started to decline. The you can also display the 2DH view of the CFL-time step since that moment.