Time rounding in feflow?

I noticed in a model I am working on that if I have a boundary condition that is supposed to change at a specific date, it can change early.  For example, I enter a time varying function where head goes from 1400 to 1100m at 100 days using a constant curve function.  When I run the model to 99.999 days I find that the change in head is already included at some point in the latter stages of the transient run (not sure exactly when, sorry).  Is this a rounding error in feflow and if so what is the tolerance for error?  I need to be able to capture the moment when that boundary change is implemented so I can change timesteps accordingly to minimize instabilities with such extreme head changes.

Martin

In case the 'constant' option is used for time series, the steps internally are not applied at once. For stability reasons, and to avoid the time step to drop too much, there is an internal ramping. Typically, this is small enough to not affect the results.