hi I would like to know the tips for learning a new package. prasat

Hi,

All CFD vendors offer tutorials for quick starts.Being familiar with pre and post processors is just a matter of time. At least you need to know first how to define

-Grid generation -definition of variables to be solved -relaxation factors for variables -no of outer iterations -model parameters -monitoring variables and their locations

If CFD background is not good enough read the manuals or some well known text books. Visit the website of specific CFD code to learn from the experiences of its users. This is the real trick, actually.

Good Luck Nuray

Lokay if suppose i have leanred some easy tutorials cases from some packages. I do not think that it is enough to start applying the package for a new probelm. And suppose I do not the answer to this problem.How should i got about it and check whether I have done it correct or not. Prasat

Being familiar with CFD packages and checking the result of CFD code applications are completely different matters. Please ask your questions specifically not to get unnecessary comments on the subject.

What i mean to say.After applying the packages for few cases how should i go for complex cases. Even if I did somehow how can I convince myself that what I am doing is correct or not.

You've put your finger on the problem:

How much is enough?

Some things I've seen in practice:

1. Run several problems similar to the one you want to solve - but make them problems where you 'know' the answer. The idea is to become more familiar with the code you're using.

2. Repeat #1 using another CFD code (commercial or home grown).

3. Hire a consultant who has published in the area.

4. Run or have run at least a few physical experiments that are relevant to your problem. If they're not 'exactly' your problem, simulate the experiments using CFD to see how the simulation does. If it's different, study the differences until you know why.

This is horribly expensive, in time and in money, but the cost of failed parts or products can also be enormous. Good luck!

Nuray, This thing with CFD is always start simple. Work through tutorials usually supplied by the CFD software vendor (if that is what you are using). The biuld a very simple model (not necessarily related to your final problem). Get that to work (regardless of the results), gaining confidence as a user of the code. Then model a simple case where you have known results (ideally related to your final problem). This should give you confidence in the code, how it performs, what parameters you need to be looking at to get the best performance out of the code (often contacting the CFD vendors for advice - they usually are willing to provide and have seen similar problems before). If your final problem is complex, then build it up a piece at a time ie test the complex physics on a simple geometry - keep run times down so you can maximise your progress; run the complex geometry first with simple physics - if there are too many unknowns and your model crashes, then how will you diagnose the problem. Finaly bring ewverything together. Then pray a little, cross your fingers ! As said previously, a lot of work is required, especially when undertaking a thorough investigation into a problem and there is no substitute for experience when using a CFD code. Good Luck Bob

CFD codes solve mass, momentum and energy balances. All CFD enginners check those balances before delivering the results to the customers. Therefore you need to calculate source terms of inputs and outputs of your application.

Let me give you an example how I check a radiative heat transfer analysis. Summation of source terms with in the medium must equal to heat loss from from all walls.

Some tips for heat transfer analysis:

If conduction is the mode of heat transfer you see linear temperature profile. If there is a source within the medium you see parabolic temp. profile If convective prevails temperature profile follows flow pattern. If you see linear temperature profile that means something goes wrong. At high temperature applications radiation is dominant mode of heat transfer. Steep temperature profile is expected. You can calculate adiabatic flame temperature. If your maximum temperature exceeds that value that means something goes wrong.

As you see these are not strong CFD background just undergraduate knowledge. If you know the basics you can be easily sure what you are doing.

Yours Nuray