[letzel]

Dear FOAMers,

consider the task of calculating atmospheric LES gust factors near an obstacle, where the gust factor G is locally and instantaneously defined as the ratio of vg_local and vg_ref. Both vg_local and vg_ref are scalars, only the velocity magnitude is considered. The timestep of the LES is about 0.1 s.

• vg_local is the local gust velocity at a given distance from the surface of the obstacle, i.e. it is a 2D object defined in 3D space, only defined on a surface whose wall distance from the obstacle patch has a given value.

• vg_ref is the gust velocity measured at a nearby reference location, i.e. it is a 0D object defined at a single location in 3D space.

At present, the the problem is simplified, and only the following steps are performed:

1. Use probes to record a one hour time-series of mag(U) both at the object's surface (only few selected locations) and at the reference location. Output to disc and postprocess them outside of OpenFOAM as follows. Time series frequency is about 10 Hz.
2. Calculate a moving average of mag(U) both at the object's surface and at the reference location. Averaging time dta is on the order of a few seconds.
3. Calculate a moving maximum of 2. both at the object's surface and at the reference location. Averaging time dtb is on the order of a few minutes. This gives instantaneous values of vg_local and vg_ref. Calculate instantaneous gust factors G=vg_local/vg_ref.
4. Output the results of 3. as time-series of G - one time-series for each of the selected locations.

Now, I wonder whether it is possible to extend this procedure to calculate G not only at few selected locations near the obstacle but on an entire surface with a given distance from the obstacle. This would require modification and optimization of my procedure.

I invite you to share your experience in how to approach this task. I hope to do as many steps as possible on-the-fly inside OpenFOAM. Some preliminary thoughts:

• In controlDict, one can define a sampledSurface. Can this be defined by a given distance from a given patch? Following calculations would need to be performed only on this surface.
• Likewise, one can define a sampledSet containing only one point - the reference location - for which the same calculations would need to be performed.
• Is it then possible to calculate a moving average, a moving maximum only for the sampledSurface and sampledSet?
• Finally, is it possible to output time-series of the ratio of the two moving maxima?

• For the simple moving averages, I think it is unavoidable to keep mag(U) for the previous ... many time steps in memory, in order to update the average at every new timestep of the simulation.
• For the moving maxima, only the value and the time when it occurred should be sufficient.

I am not sure whether I could describe the problem clearly enough. So, any feedback would be welcome, suggestions how to solve as well as requests to clarify unclear aspects.

Best regards,
Marcus