Appropriate convergence residuals?
 

Dear all,

I am simulating thermal stratification in pipe.
The problem is while the residual below 1e-5, the simulation result doesn’t match experiment result well. But if I enlarge the residual of the energy equation(for example 1e-4), I can get a similar result with experiment.
Is anyone encounter the same situation with me. The problem is?

Thanks in advance!

You would have to post some results to show what is changing between the two convergence levels.

Also this may help:
http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

Mesh independence, time step Independence or any parameter in numerical method does not guarantee that your results must match with the experimental data. For example, if you get the good results with coarse mesh and bad results with fine mesh does not mean that fine mesh results are wrong. What is required in numerical methods that you should follow the best practice such as you should make sure that your solution is independent of : mesh , time step , domain size, or any other parameter which you think have effect on the accuracy of solution.

Just assume, if we get the right solution; with any mesh, time step and turbulence model, which is overlapping with experimental data, then there shouldn't be any need of latest research in CFD and even we wouldn't had this sort of discussion any more in CFD-ONLINE

I had a vacation last week. Thanks for ghorrocks and Far 's reply. I'll have a try about your suggest.:)

pls do tell us what is outcome. :P