running simulation with a time step > 1e6 s

[AntoineMazuyer]

Describe the bug
I have noticed some problems when trying to run simulations with dt > 1e6 seconds. GEOSX comes back to a 1e6 seconds time step. 1e6 seconds is 10 days, so this limit is not very convenient when dealing with reservoir simulations.

To Reproduce
Steps to reproduce the behavior:

1. Clone the develop branch
2. Try to run a TPFA simulation with a time step > 1e6

Expected behavior
The time step inputed should be taken into account

Additional context
@klevzoff and I have noticed that there is an hardcoded value here

[AntoineMazuyer]

I have tried to increase this value up to 1e9. Now I have other problem. When the current time becomes higher than 1e9, the time step change from its original value to 1e9, whichever it value was before.

[cssherman]

The fix for this is real quick - I'll add it to the existing pull request for the event manager mpi fix (#136).

[klevzoff]

1e9 is hardcoded here: https://github.com/GEOSX/GEOSX/blob/9f55428ad550a4a06dcc4e16ec907a223c3d92a0/src/components/core/src/managers/Events/PeriodicEvent.cpp#L186
We should probably get rid of both constants. Simulation time scales are physics-dependent and can be arbitrary.

[cssherman]

This one is a bit different, since it is setting the forecast (integer number of cycles until the event is expected to execute), not the time-step.

[klevzoff]

My bad, this is fine then

[cssherman]

We should also set the event start/end time appropriately in the EventBase. It may be useful to have this as a special argument to the documentation node @rrsettgast .

[AntoineMazuyer]

Thanks for your answers, for me it's ok.

I will close the issue once the PR is merged

[rrsettgast]

@cssherman @AntoineMazuyer Is this done?

[cssherman]

Yes, this should be working now.