[loook]

Hi all,



I am running a simulation of a Stirling engine using LES approach. I have moving meshes in the cylinder. I would like to compare the results with RANS data. However, since the mesh is constantly changing I do not know how to average the velocity field for a certain piston position. Due to numerical errors it is not possible to get exactly the same piston position for different cycles. Does anyone know how this could be done?


Thanks
loook

[Svetlana]

Do you need average velocity in a point or on a line?

I think it might need programming, such as:

1. Export the data for all time steps.
2. Import the data in your script.
3. Get the time steps which are the closest to your required piston position.
4. For each pair t_i and t_i+1, interpolate to get the exact position which you need.
5. Average on the data sets which were obtained by interpolation.

I think this can be simplified a lot if you not only export the velocity data at every time step, but also export the piston displacement at every time step. Then step 3 will be much easier to implement.

[vinerm]

Hi

Averaging over moving meshes do not make much sense since the domain size is changing. If RANS simulation was transient, then you should compare instantaneous results. If RANS was steady-state, which could have been done at multiple piston positions, then you should compare results at piston positions. As you mentioned, positions are difficult to match, but difference can be reduced as much as possible.

As Svetlana mentioned, it would help to know if you want averaging at a point or line or time-averaging over whole domain for certain time duration because time-steps for URANS and LES are different. In that case, solution (was) to enable averaging in Fluent during run-time and save data files at the times where URANS data is available.

More information from your end might help with further suggestions to resolve the issue.