Bad Convergence on Complete Aircraft Analisys

arthurdiasBR · October 6, 2015, 21:04

I'm a new user of Star-ccm+ Trying to simulate a canard Type aircraft in the following flow:
P = 69820 Pa
T = 268.3380 K
V = 87.45556 m/s
M = 0.2663
Rho= 0.9046 kg/m³

my set up:
three-dim
steady
segregated flow
(Green-Gauss)
constant density
Turbulent
K-omega
All y+ treatment

For some reason I cant get a 1e-4 convergence, not even 1e-3 in the residuals.

My mesh is an Unstructured mesh (Tetra) with a prism layer with y+ =32
Here are some pics from the mesh:

does anybody has some tip for me so I can improve the convergence?

me3840 · October 7, 2015, 19:18

Your mesh is way, way too coarse. There's no wake refinement at all.

Why are you using an ANSYS mesher and a tet grid?

arthurdiasBR · October 7, 2015, 21:13

i'm using a mesh with 6.6 million cells approximatelly, it is for an article in canard airplanes regarding their aerodynamics in the general aviation.
I've never dealt with cfd meshes before as so I was told that the mesh itself was "good". I put 22 layers of prisms for boundary layer calculation wich gives me y+=32. The thing is that I don't know how to make another type of mesh on ICEM other than tet mesh with prisms. I've notice that even though I smooth the mesh till 0.38 minimum quality I keep getting bad quality after building the prisms layers. I'm short on time so I think that any tip regarding geting a good quality mesh (Tet with prisms) will make it work.

My computational resourses are somewhat limited to a i7,24gb, 12 cores computer. For now i'm studying the relation between the size of the domain in the calculation of lift and drag which is very difficult because the solver doesn't give me good convergence.

me3840 · October 7, 2015, 21:41

You're posting in a STAR-CCM+ forum, but you seem to be using ANSYS products, which I don't really understand. STAR-CCM+ has excellent easy-to-use meshers. But for a situation like this you should really be using a hex/trim grid or a polymesh. Tets are very diffusive. I guess the good part for you is tets are relatively memory cheap.

6.6M cells is pretty coarse for aero. With your CPU limitations I would cut the number of prisms in half and spend more on wake refinement. Lower the growth rate around the aircraft and do wake refinement. It should be very easy to see where your total pressure gradients are too coarse if you do some cutplanes.

You are very limited on memory resources, you can probably max out at 16M or so cells.

If by quality you mean skewness, 0.38 is very good quality for Fluent's skewness criterion. You can probably put your worst cell at 0.75 or so and be fine.

If that's STAR-CCM+'s cell quality criterion, you can probably lower it to 0.1 and be fine.

October 6, 2015, 21:04	Bad Convergence on Complete Aircraft Analisys	#1
arthurdiasBR New Member Join Date: Oct 2015 Posts: 5 Rep Power: 10	I'm a new user of Star-ccm+ Trying to simulate a canard Type aircraft in the following flow: P = 69820 Pa T = 268.3380 K V = 87.45556 m/s M = 0.2663 Rho= 0.9046 kg/m³ my set up: three-dim steady segregated flow (Green-Gauss) constant density Turbulent K-omega All y+ treatment For some reason I cant get a 1e-4 convergence, not even 1e-3 in the residuals. My mesh is an Unstructured mesh (Tetra) with a prism layer with y+ =32 Here are some pics from the mesh: does anybody has some tip for me so I can improve the convergence?

October 7, 2015, 21:13		#3
arthurdiasBR New Member Join Date: Oct 2015 Posts: 5 Rep Power: 10	i'm using a mesh with 6.6 million cells approximatelly, it is for an article in canard airplanes regarding their aerodynamics in the general aviation. I've never dealt with cfd meshes before as so I was told that the mesh itself was "good". I put 22 layers of prisms for boundary layer calculation wich gives me y+=32. The thing is that I don't know how to make another type of mesh on ICEM other than tet mesh with prisms. I've notice that even though I smooth the mesh till 0.38 minimum quality I keep getting bad quality after building the prisms layers. I'm short on time so I think that any tip regarding geting a good quality mesh (Tet with prisms) will make it work. My computational resourses are somewhat limited to a i7,24gb, 12 cores computer. For now i'm studying the relation between the size of the domain in the calculation of lift and drag which is very difficult because the solver doesn't give me good convergence. Last edited by arthurdiasBR; October 7, 2015 at 21:16. Reason: adding more technical info

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
good convergence but bad display	zaynah04	OpenFOAM	17	January 10, 2013 01:00
Problems with convergence with an easy system	franzdrs	Main CFD Forum	0	June 15, 2009 18:17
increasing mesh quality is leading to poor convergence	tippo	CFX	2	May 5, 2009 10:55
too bad convergence	Davoche	Main CFD Forum	2	November 20, 2005 05:08
Problems of Duns Codes!	Martin J	Main CFD Forum	8	August 14, 2003 23:19

October 7, 2015, 19:18		#2
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	Your mesh is way, way too coarse. There's no wake refinement at all. Why are you using an ANSYS mesher and a tet grid?

October 7, 2015, 21:41		#4
me3840 Senior Member Join Date: Nov 2010 Location: USA Posts: 1,232 Rep Power: 24	You're posting in a STAR-CCM+ forum, but you seem to be using ANSYS products, which I don't really understand. STAR-CCM+ has excellent easy-to-use meshers. But for a situation like this you should really be using a hex/trim grid or a polymesh. Tets are very diffusive. I guess the good part for you is tets are relatively memory cheap. 6.6M cells is pretty coarse for aero. With your CPU limitations I would cut the number of prisms in half and spend more on wake refinement. Lower the growth rate around the aircraft and do wake refinement. It should be very easy to see where your total pressure gradients are too coarse if you do some cutplanes. You are very limited on memory resources, you can probably max out at 16M or so cells. If by quality you mean skewness, 0.38 is very good quality for Fluent's skewness criterion. You can probably put your worst cell at 0.75 or so and be fine. If that's STAR-CCM+'s cell quality criterion, you can probably lower it to 0.1 and be fine.