[FMDenaro]

Originally Posted by mprinkey

Yeah, I don't know anything about the NITA solver other than what you do. I think it is normalization of the pressure residual for the MG solver that is the culprit. If you turn up the verbosity on the AMG solver, it should show iteration by iteration what the residual is. From that, it should be possible to determine exactly how those residuals relate to the mass defect norm. As I said, this requires getting into the weeds.

ok, but don't you think it should not work at all given this wrong BC.s?
This is an extreme case we are talking about, and only a stupid end-user would set this case for an incompressible flow but imagine some more sophisticated problems where inflow and outflow BC.S are prescribed in such a way not to match the compatibility relation. References for the NITA in FLuent call for the Projection method and I suppose the implementation is based on that..

[mprinkey]

Originally Posted by FMDenaro

ok, but don't you think it should not work at all given this wrong BC.s?
This is an extreme case we are talking about, and only a stupid end-user would set this case for an incompressible flow but imagine some more sophisticated problems where inflow and outflow BC.S are prescribed in such a way not to match the compatibility relation. References for the NITA in FLuent call for the Projection method and I suppose the implementation is based on that..

One would hope, but in reality, it is very hard to just know that the system is singular, right? It isn't like we can just find the determinant of A quickly and check. And a lot of the systems that we regularly solve are ALMOST singular, so there is a numerical fuzziness rolled in here too--what is the epsilon on det(A) that makes it singular or not?

There has been a Darwinian process involved in getting CFD in general and commercial codes in particular to their present state. The schemes are used because they (mostly) work. Iterative solvers don't offer hints at matrix singularity that direct solver do, but direct solvers don't do well on huge sparse matrices. So, we get minimum error "solutions" from iterative solvers. I think there could be pre-processing tests applied to sum up imposed fluxes on connected domains without outlets. Heck, I think if you have a mass inlet somewhere, the code should throw a warning that no outlet exists, because there must be an imbalance somewhere unless a mass sink has been applied to precisely match the influx.

[Rayman]

Dear All

I really and badly quick need your help. Can anybody let me know how I can create a profile file. I did much search but in vain. Someone told me I should write this (but, where should I write this and how in detail):

((pressprofile transient 2 0)
(time 0 36000)
(p 1.503e7 0))

Thanks.
-----------------------
I write my problem down shortly:
I have a porous cylindrical rock saturated with gas under pressure 2185 psi. Outside the cylinder is gas with the same pressure at time zero. As time passes, the pressure outside the cylinder is reducing with time from 1.5e7 pascal to zero in 36000 seconds. As the objective, I wanna see how the pressure inside the cylinder varies.