
[Sponsors] 
December 7, 2014, 13:54 
mesh study with high residuals  redo?

#1 
Member
Join Date: Mar 2013
Posts: 42
Rep Power: 5 
I've done a mesh study and taken significant time/money to come up with an answer. but i later found out my model was not convereged due to higher residual values and needing more inner iterations. When i had done the study, i had good agreement between my drag coefficient to a max difference of .003 between meshs 7 mil and 12 mil.
even though my drag values were off from the expected values from the physical test. can i consider the mesh study conclusive or would i have to redo the test with a more converged solution to prove the values don't change. currently i'm working to refine the timestep and inner iteration count, but i would rather not have to redo the mesh study if i can assume the solution remains conclusive, but only more accurate with the decreased residual values. 

December 7, 2014, 14:10 

#2  
Senior Member
Troy Snyder
Join Date: Jul 2009
Location: Akron, OH
Posts: 135
Rep Power: 10 
Quote:
Use of residuals is only one measure of solution convergence. It is a measure of how well the discretized PDEs are satisfied. However, as you are concerned with drag, you should monitor drag during your computations as well. The solution could then be considered converged when the residuals fall below some tolerance (say 10e3 for continuity and 10e4 for momentum) and the drag value is also converged within a tolerance acceptable for you particular problem. The mesh study is not conclusive if the solutions on each mesh are not converged based on the above criteria. 

December 7, 2014, 16:10 

#3 
Member
Join Date: Mar 2013
Posts: 42
Rep Power: 5 
so with a higher accuracy of solution to PDE's (lower residual values), can affect the spatial solution convergence ie mesh relationship solution?


December 7, 2014, 16:23 

#4 
Senior Member
Troy Snyder
Join Date: Jul 2009
Location: Akron, OH
Posts: 135
Rep Power: 10 
Correct. Any conclusions you draw in moving from a coarse mesh to a finer mesh may be erroneous if the individual solutions are not "converged".


December 7, 2014, 16:29 

#5 
Member
Join Date: Mar 2013
Posts: 42
Rep Power: 5 
aww shucks... thanks for the help though!


Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Moving mesh  Niklas Wikstrom (Wikstrom)  OpenFOAM Running, Solving & CFD  122  June 15, 2014 06:20 
Mesh motion with Translation & Rotation  Doginal  CFX  2  January 12, 2014 07:21 
Visualizing mesh regions with high courant number?  Endel  OpenFOAM  2  August 20, 2013 03:57 
ANSYS Meshing Mem Usage Seems High for 7 million Element Mesh  jonny_b  ANSYS Meshing & Geometry  2  August 15, 2011 08:39 
Need for Mesh Independence Study  nickninevah  Main CFD Forum  6  October 15, 2010 17:25 