Hypre Fails with Navier-Stokes! - Can Anyone Try This?

Numerical methods and mathematical models of Elmer
Post Reply
kishpishar
Posts: 54
Joined: 17 Jun 2015, 10:04
Antispam: Yes

Hypre Fails with Navier-Stokes! - Can Anyone Try This?

Post by kishpishar »

Dear Elmer Users,

I have been trying to solve some flow modeling problems with Elmer's Navier-Stokes solver by using Hypre as the Linear System Solver, with no luck so far. To simplify the discussion, I am attaching a 2D model problem - i.e., the classical backward step. Water enters through the narrow end with a velocity of 1.0 m/s and exits through the wider end. The domain is 1.2 m long and the narrow and wider ends are 0.1 and 0.2 m, respectively. The turbulence is modeled using the k-Epsilon model.

What I notice is that, while Elmer's own BiCGStab solver in parallel works flawlessly (and produces the same results as a serial run), Hypre's BiCGStab solver fails to converge to the right solution.

The case and a few results are attached.

Can anyone using Hypre please help me with this?

Thx & Regds
kumar
Attachments
Velocity-BiCGStab_from_Hypre-2Proc.jpg
(19.7 KiB) Not downloaded yet
Velocity-BiCGStab_from_Elmer-2Proc.jpg
(23.66 KiB) Not downloaded yet
BackStep-hypre.tar.gz
(503.84 KiB) Downloaded 284 times
Post Reply