CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   how to ensure that your modelling is correct (http://www.cfd-online.com/Forums/main/7219-how-ensure-your-modelling-correct.html)

rvndr February 27, 2004 08:11

how to ensure that your modelling is correct
 
Hi

I am a beginner and want to know the answer for this questions ?Suppose you are modelling some problem in CFD, which has numerical solution or experimental result. In this case you will compare your CFD result with the experimental result.

On the other hand, now suppose you are modelling some problem which has no experimental result. Then in this case how will you ensure that your CFD modelling result is correct ?

one more thing is if you already have experimental result then what is the need for going CFD modelling again ?

This questions may not sounds well but you have to bare with me as I am a beginner.Thanks in advance

rvndr


Jarmo Monttinen February 27, 2004 11:54

Re: how to ensure that your modelling is correct
 
Well... if you do not have experimental results to compare with, go with other computational results (=verification). If this is not available either, find a case that is similar to yours, and for which there is data available. If you can make reasonable comparison with other data and then show that your own CFD-results are converged (ie. perform grid convergence study etc) your data could be ok.

As about the need for CFD... I could come up with various funny reasons, but on the serious side setting up a CFD computation is cheaper and quicker than setting up an experiment. And there are many other reasons, I'm sure we could fill the whole message board discussing these...

-- Jarmo

chinthakindi February 27, 2004 12:45

Re: how to ensure that your modelling is correct
 
Hi rvndr,it is very good question, in the begining all will get such a doubt. my answer for ur question is, see if u want to do expt it is involving much complexity ,like inaccuracy with instruments, lot of expenses and time. if u take the numerical modeling, it is more accurate and involving of less risk. to test ur models, there is no need of expt results, u can test by doing DNS of the small portion of ur computational domain.again doing DNS for the entire domain is involving of large computational power of CPU and time, u can avoid this by selecting small domain for doing DNS , this results is sifficient enough to test ur results,, prior to this as Jarmo told,ur model will be verified even with grid convergence results.. i hope ur doubt is clarified

alex February 27, 2004 14:47

Re: how to ensure that your modelling is correct
 
fascinating:)

rvndr February 28, 2004 01:01

Re: how to ensure that your modelling is correct
 
hi jarmo,

Thanks for the reply. Could you please give me some examples for "other computational results " & "case that is similar to yours".

Thanks again in advance

rvndr

Jonas Holdeman February 28, 2004 19:48

Re: how to ensure that your modelling is correct
 
Responding to your question, "if you already have experimental result then what is the need for going CFD modelling again ?"

If the experimental results are correct and your computations agree with the experimental results, there is still information to be gained from computation. There are flow properties at points that were not instrumented, additional properties not measured under experimental conditions, flow properties under different conditions and different geometries. Access to instrumented wind tunnels or flow chambers may be expensive, time consumming, or not available near you.

With a good program and a desk top computer or workstation, you have access to computed flow measurements from anywhere in the world at no additional capital outlay.

rvndr February 28, 2004 23:43

Re: how to ensure that your modelling is correct
 
Hi, Jonas Holdeman thanks for the reply. It cleared my boubts fully.

regards rvndr

Jarmo Monttinen March 1, 2004 13:21

Re: how to ensure that your modelling is correct
 
It all depends on what type of problem you are solving... Just an example: you wish to compute the flow over an aspect-ratio 4-8 wing, Mach about 0.85. You could for example compute the ONERA M6 wing at M = 0.8395 for which there is plenty of data available.

Hope this helps.

-- Jarmo


All times are GMT -4. The time now is 10:51.