|
[Sponsors] |
June 7, 2017, 16:16 |
Convergent criteria
|
#1 |
New Member
anonymous
Join Date: May 2017
Posts: 6
Rep Power: 8 |
Hello everyone
I am doing a Cfd analyse of a nozzle and i got a convergent of the solution of order -3 and it doesn't change the residual value even if i leave it for a high number of iteration.... Do you think the solution is wrong?? Thank you |
|
June 7, 2017, 16:50 |
|
#2 |
New Member
Justin
Join Date: Jan 2015
Posts: 28
Rep Power: 11 |
That depends on several things. A lot of things.
I recently ran a handful of cfd simulations with Fluent of con-di nozzles. I ran my simulations as steady, coupled pressure based solver, with all discretizations set to second order upwind. My residuals would decrease to 1e-5 or lower after roughly 1000 iterations but at some point, the residuals stopped decreasing and essentially remained constant. My simulations would not reach convergence to 1e-6. Depending on the nozzle pressure ratio, my simulations would converge to some value between 1e-5 and 1e-6 and stay constant. My recommendations would be to better initialize the solution and monitor mass flow rate between inlet and outlet to see if they're close to the same (my simulations showed a difference of 0.001 kg/s between inlet and outlet flow rates). |
|
June 7, 2017, 17:39 |
|
#3 |
New Member
anonymous
Join Date: May 2017
Posts: 6
Rep Power: 8 |
Thank you for the reply
I also monitor the mass flow inlet and outlet and it basically remain constant( very very small fluctuation around the 0 value) but the thing that I'm worried about is that the continuity residual which is oscillating with an order of -3 while other parameters like k is Order -4 and omerga -3 |
|
June 7, 2017, 17:59 |
|
#4 |
New Member
Justin
Join Date: Jan 2015
Posts: 28
Rep Power: 11 |
Below is a convergence history for one of my simulations
All of my nozzles were rectangular single expansion ramp nozzles which are inherently unsteady (the external jet oscillates). That is what causes the erratic oscillations of residuals. Notice my continuity residual was also the highest. I ran my simulation out to 3000 iterations. I considered my simulations converged after seeing that residuals stopped decreasing and mass flow at inlet and outlet were essentially the same (1.3949 and 1.3959 kg/s, respectively). That being said, I initialized first with hybrid initialization and then full multi-grid initialization. This allowed me to obtain a solution faster. Hope this helps! Last edited by DA6righthand; June 8, 2017 at 11:11. |
|
June 8, 2017, 11:12 |
|
#5 |
New Member
Justin
Join Date: Jan 2015
Posts: 28
Rep Power: 11 |
I edited the picture. I don't think it was showing up correctly before
|
|
June 8, 2017, 12:06 |
|
#6 |
Senior Member
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,779
Rep Power: 71 |
What about the order of magnitude of the initial residuals?
|
|
June 9, 2017, 05:00 |
|
#7 |
New Member
anonymous
Join Date: May 2017
Posts: 6
Rep Power: 8 |
I got convergent.....I just changed one BC....I initialize my case with an hybrid initialization and the order of residuals are of the order of -7
So if I got a residual plot like the above one...it is not wrong right??? |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Error criteria on coarsest level in Multigrid method | Sufyan | Main CFD Forum | 1 | January 9, 2017 05:36 |
hard to convergent with SU2 4.0 for Euler adjoint | Xianguu | SU2 | 1 | July 14, 2015 17:09 |
Convergent nozzle and preesure of steam | pranabjyoti | CFX | 7 | March 10, 2011 19:23 |
What value shall I set for the Convergence criteria? | steventay | CFX | 7 | May 14, 2010 12:44 |
How can I set the converge criteria properly? | Alina | FLUENT | 3 | September 7, 2007 16:55 |