[swathi koora]

hello

I am trying to learn CFD from scratch and I have started to learn the theory,meanwhile I was doing a basic laminar flow problem but I was not able to get fully developed flow in parabolic shape.

MY QUERY
1.If our simulation does not meet analytical or experimental results,how to understand the mistake.
In general ,what are the ways to understand if our mistake is due to mesh or numerical schemes or descritization schemes?

solution will get converged but it will not meet the analytical results,how to debug the error in the simulation and overcome those errors

I want to learn CFD in a very clear way

please throw some light on the above query!!


regards
koora swathi

[vinerm]

It is not CFD but Fluid Dynamics that one needs to learn. CFD is driven by Fluid Dynamics. If your concept of FD are clear, then CFD is only about learning a tool and its features. However, without learning FD, you would end up being a tool user, which is not good. As far as CFD is concerned, it is a tool, hence, like any other tool, you learn by experience. There is no clear path. Focus on the phenomena important to your work or your interest, learn more about that, do simulation related to that. Since in your field of work or interest, you would have a pretty good idea about what you are expecting, it becomes easy to spot the issues. Solving those could be a challenge but that's where experience comes handy. So, I'd suggest start learning more about your field of interest, say, turbulence, or multiphase, or reacting flows, etc. as well as about industry those are relevant to, say, oil & gas, construction, energy, metal, auto, etc.

[swathi koora]

Thnaks for the response
I am fine with what you have said,I am pretty clear with FD concepts
But when I try to solve the problem in ansys, my results will not match with analytical solution , am clear as far as FD of my problem is concerened,so my QUERY IS

HOW TO KNOW IF THE ERROR IN THE SIMULATION IS DUE TO MESH OR NUMERICAL SCHEMES OR DISCRETIZATION SCHEMES OR HOW TO RELATE THE ERROR IN CFD WITH FD?
HOW TO IDENTIFY AND SOLVE THEM?

THANKYOU

[vinerm]

It depends on the type of error. Though the types of error in CFD are clearly defined their reason is not always clear. If by error you imply difference from experimental data, then first thing to look at, before even looking at the mesh or physical or numerical setup, are the assumptions made for the CFD model. For a new physical setup, grid sensitivity analysis is always recommended to ensure that grid does not have much influence on the results. Once these tests are done, then the only remaining aspects are physical setup, which in a way is related to assumptions and models along with their validity, and numerical setup. Numerical setup does affect the accuracy because this solves the matrices but cannot guarantee it because even with higher order models you can get bad results despite good convergence. So, most important part is physical setup. You need to learn about the validity of each model and the phenomena these models are applicable to.

I don't know what kind of model you are working on, however, if currently your results are not matching with the theory, then do note that theory has more assumptions that CFD has.