Efficient time stepping with great variability in pumping/shutin period durations
Dear colleagues,
I am history matching a pumping test that consists of two 3day flow periods each followed by a 10d shutin periods. The flow rates were not constant during the pumping periods and I've had to discretize the pumping periods (see picture). I am running EWASG because of some level of methane in the aquifer. The model runs well so the time steps become fairly large with automatic time stepping and I miss many details during the pumping periods. If I reduce DELTMX to capture these details, the simulation takes much longer than ought to be (many un-needed time steps during the build up periods). If I specify the times at which rates change as printout times, I get these details but the size of the OUTPUT file explodes. Any smart solution? 
1 reply
-
Yannick,
I would try using the actual pumping history with MOP(12) =2. In this way you should at least replicate the cumulative production during each time step.
I would use MOP(16)<4 to make the convergence harder. This could affect in particular the pumping periods where pumping and rate changes could require shorter time steps to converge allowing a more detailed reproduction of actual pumping history.
During build-up periods convergence could be smoother allowing longer time steps.
Even a smaller convergence tolerance could have a similar effect.
Alfredo