Using Verification and Validation Techniques to Evaluate the Accuracy of CFD Results

Verification and validation (aka "V & V") is a huge topic on which, literally, whole books have been written. These days, production CFD solvers are large and complex, so you can count on bugs being in there somewhere. If that's the case, how can one evaluate the accuracy of CFD results?

You can't check everything, but don't let that discourage you; you don't have to check everything. There are many steps which you can take to assure yourself that the CFD solver you are using is capable of giving you the results you need at the fidelity you require.

While many of these techniques can be applied to any solver, some of them require access to the source code and a compiler (or two). If you are using a solver that does not come with the source code, just skip over these techniques.

Pressure contours on the Eglin wing with a pylon and a generic finned store

You may be thinking that this is not something that you as a user of CFD need to worry about. If so, I encourage you to read more on the importance of verification and validation . The road to CFD hell is paved with the bones of careless CFD users...or something like that.

Error in a CFD case rises as mesh is refined--not right!

With that out of the way, our first stop deals with the dangers of compiler errors. . This is an area that most CFD users never worry about, but it has caused me problems more than once. CFD solvers are complex collections of software iterating on equations which describe dynamic non-linear systems. It doesn't take much of an error to produce a wildly divergent result sometimes. Make sure it doesn't happen to you .

Another powerful tool in the CFD verification and validation toolkit is the Method of Manufactured Solutions (MMS). In fact, the line plot above is from an MMS application where the solver in question had a coding error. This error was not apparent from just looking at the results of regular applications. Using MMS, however, it was possible to not only determine that it was present, but also that it was something about one of the boundaries that was causing the problem.

If you have a question about any of the above topics, or another topic that you'd like to see added, please contact me and let me know

When you're ready return to the Innovative CFD home page.

New! Comments

Have your say about what you just read! Leave a comment in the box below.