[CIG-LONG] [Gale] EulerDeform: coordinate out of bounds when increasing model resolution

[Guillaume.Duclaux at csiro.au]

Hi Matthieu and Walter,

Thanks guys for your advices!

I tried to fix different time steps but the model was still crashing due to EulerDeform problems...

I found that the problem came from the pressure remesher part of EulerDeform. I realised at the same time that my model had problems with nodal pressure. After looking in the CIG archieve, I realise I didn't set properly the minimumYieldStress for a yielding rheology component (it has to be equal to the cohesion - am I right Walter?).

The model starts with a strong Nodal pressure contrast in the upper part. I guess it behaves like this because it doesn't start under isostatic equilibrium.

Removing the Pressure remesher part in EulerDeform looks to avoid the crash without fixing the timestep.

Cheers

Guilly




Dr Guillaume Duclaux
Structural Geologist/Modeller
CSIRO Exploration and Mining
Visiting address: 6G13, ARRC, 26 Dick Perry Avenue, Kensington, Western Australia, 6151
Postal address: PO Box 1130, Bentley, Western Australia, 6102, Australia
Ph: + 61 8 6436 8728  Fax: + 61 8 6436 8555  Web: www.em.csiro.au

-----Original Message-----
From: Matthieu Quinquis [mailto:matthieu.quinquis at ngu.no]
Sent: Monday, 17 November 2008 10:40 PM
To: Duclaux, Guillaume (E&M, Kensington)
Cc: walter at geodynamics.org; CIG-LONG at geodynamics.org
Subject: Re: [CIG-LONG] [Gale] EulerDeform: coordinate out of bounds whenincreasing model resolution

Hi Guillaume,

I also had some problems when using the Euler deform. The only way I
found to get the model to run without problems is to define myself the
time step. Of course the problem here is that you have to find manually
the best time step.

In the models I ran when letting the code calculate the time step, I
also often got into a "bimodal time step", one long followed by a short
time step. So I had to define the time step my self.

>From what I see of your problem changing the number of CPUs or the
number of particles won't change a thing. Unfortunately, I believe you
have to decrease the time step further. But I may be wrong.

Cheers,

Matthieu