CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   What exactly is a sensitivity analysis? (http://www.cfd-online.com/Forums/cfx/67542-what-exactly-sensitivity-analysis.html)

Josh August 18, 2009 12:41

What exactly is a sensitivity analysis?
 
Hi again -

These forums are of great help. Thanks to everyone.

I have noticed in many thread replies, particularly those from Glenn Horrocks, that a "sensitivity analysis" is a good way to ensure the accuracy of the mesh, timesteps, and various other parameters.

I have read all of the Help file topics regarding sensitivity analyses. Although it is mentioned often with regards to finding good values for various parameters, I cannot seem to find a detailed explanation of what a sensitivity analysis is or how it is done.

Additionally to the Help file, I have searched these forums and other online resources extensively to find good examples of sensitivity analyses/checks.

I have determined that a sensitivity analysis involves the following (this is the extent of my knowledge):
  • A sensitivity analysis helps determine
    • the mesh size (e.g. how large the boundary layers are)
    • the timestep size (although adaptive timestepping can also work)
    • convergence criteria
    • other calculated variables
  • Adjusting the maximum value of certain criteria, such as the timestep size, and observing the results (e.g. compute time) in the Solver
  • Observing errors based on the above parameters
  • I think it involves monitor points, but I could be wrong
Any help as to how to complete a general sensitivity analysis, or perhaps a good reference, would be greatly appreciated.

Thanks!

jmcatelen August 18, 2009 16:14

Validation
 
Check the following: ASME V&V 20.
Good luck.

ghorrocks August 18, 2009 18:46

One of these days I will write a FAQ on sensitivity analysis. Won't be for a while things are a bit busy at the moment. In the meantime "Computational Fluid Dynamics" by Roache is the classic reference on CFD accuracy. The JFE editorial policy on CFD results is based on Roache's work and is a good summary - http://journaltool.asme.org/Template...umAccuracy.pdf

Josh August 19, 2009 09:18

Excellent resources. Thanks so much.


All times are GMT -4. The time now is 01:23.