Basic question

Questions and remarks about Code_Saturne usage
Forum rules
Please read the forum usage recommendations before posting.
Post Reply
Alburquenque
Posts: 5
Joined: Thu Mar 28, 2019 7:49 pm

Basic question

Post by Alburquenque » Mon Apr 15, 2019 3:20 am

hello, I'm using cs to simulate a flow on a sphere, I've had some problems for the simulation since the cases diverge, I'm occupying the k-epsilon turbulence model for this.
my question is. It is possible to simulate the flow with a fixed speed?, in which option do I determine this speed? in profiles? .

I have occupied temporary steps of 0.5, 0.08, 0.05, 0.008 and 0.005, in all the result diverges.

The flow speed is 9.5 m / s.

Greetings.

my English is very bad, sorry for this.

Antech
Posts: 119
Joined: Wed Jun 10, 2015 10:02 am

Re: Basic question

Post by Antech » Mon Apr 15, 2019 7:58 am

Hello.
Code Saturne uses SIMPLE-like, not fully-coupled solver. For this type of solvers, solution stability often becomes a problem (regardless the particular program you use). If you work with gas and only need static results, try reference time step 0.0001 s, and max timestep multiplier 100. If you need dynamic results, try to reduce the time step further, for example, to the same 0.0001 s and look at maximum Courant (CFL) numbers you get, they should not exceed 10 (absolute maximum is 50 but the calculation will likely diverge). In static cases, start with Upwind scheme (see equations dialog in GUI), then restart with SOLU with blend of 0.8...1.0 for velocity/temperature and Upwind for turbulent quantities. Select "Least squares method over partial extended cell neighbourhood" for the gradient reconstruction (in global numerics dialog). IMHO, It's rare that other options will help you to stabilize the solution. Maybe the mesh improvement will also help. I found that, although stable with simple meshes, Saturn doesn't want to calculate my cases with inflation layers (4.0...5.0.5 versions, k-omega SST). Multigrid solver doesn't even work with them (deverges for Wall distance or pressure), other solvers work (version 4.0) but the solution diverges: velocities reach cosmic levels at geometry corners and it never stabilizes. I suspected the mesh issues but I found this behaviour regardless of mesh quality in recent calculations (sorry, don't have enough time to create the test case for the forum now).

Post Reply