ok, I think I get it. You're probably using cauchy nodes to simulate net evaporation from the lake surface. But, you should be able to compute that rate knowing the evap rate and area of the lake and that should be equal to the cauchy discharge. A more realistic and numerically stable approach would be to either add the net evap rate to the abstraction rate for the well or simulate net evap using negative recharge at the lake surface in conjunction with the abstraction well.
Pete