Is CFD Software Accurate?

[Anonymous]

Yes, when used properly, it can closely replicate actual physical test results.

As a CFD consultant, I often benchmark the software against known test results.  Consider the following example from a whitepaper I came across recently . . .

This setup produces a large recirculation region in the space, as shown below.

Looks good, but are these CFD results plausible? I checked both Autodesk CFD and Ansys Fluent against the test results from the whitepaper.

Both programs showed very good correlation using just the typical "out-of-the-box" solver settings with a reasonable mesh density applied.

[Anonymous]

Excellent! @Anonymous 

[Anonymous]

Very nice! Good job 🙂 
@Anonymous 

[Anonymous]

Of course, this is only for one application and set of physics . . . incompressible adiabatic flow of a turbulent jet entering a large room.  Accuracy will vary for other applications (e.g., turbomachinery).

It may be possible to get even better correlation by tweaking mesh and solver settings.  However, caution must be used since the accuracy of the physical test (e.g., instrument calibration) cannot always be validated from the source.

Meshing plays a critical role here.  For example, look at the results below for OpenFOAM when an improper mesh is used.  This is not a knock on OpenFOAM since it is quite capable of handling this problem as well given the proper inputs. 

[hamed.sangin]

Hi 

i want compare my model

how can i access to this model and compare to mine ?