CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   under relaxation factor (http://www.cfd-online.com/Forums/fluent/106179-under-relaxation-factor.html)

ehsanakrami August 21, 2012 09:59

under relaxation factor
 
hi guys , a question can we change under relaxation factors while fluent is solving ? the Convergence is right ?

misagh August 22, 2012 07:41

hi
during iteration it is not possible.but you can cancel iteration and adjust the urf and then iterate again.
this method is mostly used when the residuals do not change severly after so many iterations say 500-600 iterations,here the problem is converged although it has not reached the specified residuals.

ehsanakrami August 22, 2012 12:07

yes , i mean cancel it and change factors . but when we decrease th factors sudenlly the residuals get down . is the reasult correct in this way ?

diamondx August 22, 2012 14:01

when you see that residual are not stable, it is better to lower the urf.
Start by a very low urf, then while iterate as it gets better , give them higher value.

misagh August 22, 2012 16:04

dear friends
i don't think lowering the urf be a good way of solving the problem.THE BEST THING IS DEFAULT VALUES.if it has oscillation change it or when the residuals does not decrease and is taking a straight line you can lower it down.BUT USE THE MAXIMUM VALUE FOR URF IF IT IS POSSIBLE.
note that the results(after convergence) are dependent to the urf .but if changes are less than 5-10 % don't care about it.you had better check your numerical values with an experimental or theoretical one.
finally i found the link below really useful:
www.bakker.org/dartmouth06/engs150/05-solv.ppt
best regards.

ehsanakrami August 23, 2012 03:54

my residuals are shown in blow . i decrease URF step by step . the places that it suddenly get down , the URF decreases . for example each step i decrease it 0.01 . my result it correct ?

http://picture.hostesr.com/pic/6b8c8d762da1.jpg

misagh August 23, 2012 05:07

dear ehsan,
it's really hard to evaluate the accuracy of your problem without being aware of the details of it.the best thing is to check your results(e.g. the Nuselt Number for constant heat flux is 4.63 OR in a pipe the velocity of centerline must be 2 times of mean velocity OR for laminar flow in a tube in fully developed region f=64/Re ...)
but your residuals have oscillation and my sixth sense(!) threw me into doubt about the accuracy of your solution.i don't know maybe the number of mesh cells are not enough(evaluate your mesh independence)
any way the best thing is to check your results.
regards


All times are GMT -4. The time now is 09:21.