How Transient Statistics Works in CFX
 

Hi,
Sorry for the naive question. Are transient statistics functions like Trnavg or Trnmax computed at the end of the specified "Iteration End" or Trnavg is a kind of moving average that already exists during the run, but does not have the value that we want/expect it to have?

I am computing a unit vector which is the direction of time-averaged WSS vector say (q). Then I compute the time-averaged value of the q component of the WSS vector (WSS.q). However, q itself is only available at the end of the simulation. How then can CFX record for each time step the value of WSS.q, which it needs to be able to compute the time-averaged value of WSS.q.

Thanks in advance.

It computes them from Start to End as specified in the setup. That is, it is not a running average with a width equal to End - Start.

Once the software passes over End, the statistics remain frozen until they are written to the results file (or backup file).

A moving window running average requires storage of the variable from Start to End, and removal of 1 storage level and addition of another storage level at the calculation progresses.

Hi Opaque, Thanks for your message. Does it mean that CFX will ignore all those variables which are somehow related to trnavg/trnmax/trnmin (directly or indirectly) and computes them as soon as it reaches the end of solution (End), just before writing the results?
Thanks

It computes them as it goes but only within the Start->End window. Once the End of the window is reached (the solution may still continue) the transient statistics are frozen.

It cannot wait to compute any transient statistics later because the data from previous timesteps/iterations is not stored anywhere in memory.

Thanks a lot Opaque, So it seems that it cannot give me what I expect to compute, yes? I am computing a unit vector which is the direction of time-averaged WSS vector say (q). Then I compute the time-averaged value of the q component of the WSS vector (WSS.q). However, q itself is only available at the end of the simulation. How then can CFX record for each time step the value of WSS.q, which it needs to be able to compute the time-averaged value of WSS.q.

Do you want WSS.q at a single point or the distribution across the entire surface?

Thanks for your attention. Basically, it is a distribution over a surface. Thanks.

Why is not q available during the run ? If you request the statistics of Wall Shear Stress (assuming is available for transient stats), the solver will compute it at each time step within the statistics window and average it as request.

Try and see if it works. It should.

Hi Opaque, thanks so much. Of course q is available during the run, but it does not have the right value. q must average of one Start-End period, and then be used to compute WSS.q. So the right value of q is only available at the end of the Start-End period. Now how can it be used to compute WSS.q during the run, including Start-End period?
Thanks

Hmm...even if I'n not 100% sure to have understood the problem, I might provide you with a solution.
If you need something like a transient average over a time period of a variable that is avalaible at each time step, you can do this in cfxpost by using script files.(In CFXPost Session -> Play Session)
There is the Ansys power syntax (a programming language based on pearl) that allowas you to write script files that will evaluate variable and functions at each time step and write the results to .csv file.

E.g. You want the value of q as areaAverage over Plane P1 for each time step: then it would work like in the attached .cse file (just remove the .txt ending to play it in post. Had to add .txt to be able to upload it)
You can write there any expression for post process evaluation you want. And the time average can then be determined in excel by post processing the .csv file.

Is this maybe what you were looking for?