Posted Tue, 10 May 2016 13:38:21 GMT by Karanka
Hi, I am trying to simulate an unconfined model characterized by 4 layers with 2 Hydraulic head BC, 1 Cauchy BC and 5 multilayers wells: the simulation is too slow!

What could be the problem? and what I could trying to change?


Posted Tue, 10 May 2016 13:45:43 GMT by Björn Kaiser
Do you run a transient model? If yes, does the time-step size increase or is the time-step size relatively small (e.g. < 0.1 d) ? You may check the time-step behavior in the Time-Step History. In the main menu, please click on [b]View[/b] and then on [b]Charts[/b]. After that, please open the [b]Time-Step History[/b].
Posted Tue, 10 May 2016 15:16:54 GMT by Karanka
Hi, thank you for the reply! yes, it is a transient model, I set the initial time step equal to 1 day but I saw from the time steps chart that the actual value is below 1 (about 1e-07d!!!!).
Posted Tue, 10 May 2016 15:34:30 GMT by Björn Kaiser
Small time-step sizes by means of several orders of magnitude behind the decimal separator as predicted by the automatic predictor-corrector time-stepping scheme is an indication for numerical instabilities. A multitude of possible reasons exist. I suggest to contact the <a href="">FEFLOW Support</a>.
Posted Wed, 11 May 2016 07:22:08 GMT by Rapheul Liu CUMT
I have encountered similar problem, but after I changed the error tolerance value to a relatively larger one, the model ran faster and no such tiny time step came up. However, it is true that the model is suffering from a numerical instability situation that maybe we should check the model definition.  Regards

You must be signed in to post in this forum.