You have a transient model, and you'd like to see some time averaged values for some part of the model. If the model has a fixed grid, then you can directly use the Calculator Function "TempMean()" to compute a Temporal Mean of a variable, on a part.
However, what if you are using an AMR grid? Or a changing connectivity domain? What if the location where you are interested in calculating the Temporal Mean is fixed, but other portions of the model are changing? Well, initially you might think you are stuck, as your attempts to use TempMean() return an error about having a changing connectivity grid. Stuck... well not quite. There is potential method to allow you to capture the temporal mean...
The main thorn here is the changing connectivity. We can get around this via the following method. Let's assume you want to see the temporal mean on a Clip through the domain (in a location where the geometry does not change (but the grid might)):
a. Create a Clip plane at a timestep representative of perhaps the finer grid resolution (i.e. not a coarse representation).
b. Ideally, do not have any variables activated.
c. Select just that part, at the current timestep, and go to File -- > Export -- > Geometric Entities. Choose EnSight Case Gold Format, and change the timestep range to be ONLY the current timestep. This will write out that Clip Plane, for a single timestep to Case Gold file. (If you had variables activated, you perhaps should edit the .case file and remove out any specification of variables for this clip plane.)
d. Read this Clip Plane model back in as a second case (Case Add, not Case Replace).
e. Use CaseMap() function to map the desired variable onto this Clip Plane.
f. Now, you can use TempMean() on this CaseMapped variable of the Clip Plane in the second case.
Caveats with this method:
a. This method interpolates the values onto this fixed grid clip plane. Using a coarse representation here is not ideal. If you have an AMR grid, best to choose the finest representation of the grid to help reduce interpolation errors.
b. This method can work for an IsoSurface as well. In addition, you could do the same with a model part too (writing it out/read back in again), if that part does not move in space.
c. This method does not work well in areas where the domain boundaries (and therefore the clip extents) change.
d. You will need EnSight Standard (or HPC) to achieve this, as it requires multiple cases, as well as generic EnSight format reader. To be clear, this will NOT work with EnSight Desktop for Converge, or EnSight Standard for Converge.