# Periodic oscillating transient solution: Phase average

 Register Blogs Members List Search Today's Posts Mark Forums Read

 May 24, 2022, 08:50 Periodic oscillating transient solution: Phase average #1 New Member   Jordi Griera Join Date: Apr 2022 Posts: 1 Rep Power: 0 I am working on a multielement airfoil, a wing with a flap where the flap is set at a high deflection angle, which implies detached flow. Using URANS (SST model) the transient simulation reaches, after some time, a pseudo-steady state solution where the coeficients are periodically oscillating. This is due to a periodic generation of rolling vortices in the separated flow region. Let's suppose (for easy math) that these oscillation period is 48 timesteps and after 2000 timesteps the simulation reached this periodic-oscillating state. I would like to have 4 solutions (velocity fields, pressure fields, etc) that are the average of 20 oscillations but for a phase of the oscillation (0deg, 90 deg, 180deg and 270deg). So for the 0deg would be the average of the timesteps: 2000, 2048, 2096,..., 2912. for the 90deg would be the average of the timesteps: 2012, 2060, 2108,..., 2924. and the same for 180deg and 270deg. In ansys CFX when setting the Transient Statistics in CFX-Pre it just allows to set a start iteration and stop iteration. My questions is: Exist the possibility to generate 4 different Transient Statistics files starting at different timesteps and averaging the solutions every N (48) timesteps? I know I could save all the previous mentioned timesteps as Transient Results with a Timestep List and do this averaging once the simulation has finished, but this means a lot of memory (As I really want to do 16 phases in 20 oscillations in a ~17M cells simulation saving 11 fields), so this is why I would like to know whether this could be done on the fly. Thanks in advance

 May 26, 2022, 03:53 #2 Senior Member   Gert-Jan Join Date: Oct 2012 Location: Europe Posts: 1,814 Rep Power: 27 Not sure if this is possible. I would ask ANSYS. But if you want to save all the timesteps, then why you not save .trn-files with only the variable that you want to analyse (velocity fields, pressure fields, etc). These files are quite limited in size, not eating your complete hard disk.

 May 26, 2022, 09:56 #3 Senior Member   Join Date: Jun 2009 Posts: 1,770 Rep Power: 31 As Gert-Jan said, very unlikely. You can try it in a simple setup (tutorial), and take a look at the output file. If I recall correctly, it will mention something along with duplicate/multiple statistics with different ranges, and what the final selected range would be. That is, the multiple setup is reduced to a single statistic for a given variable. Now, an alternative I have not tried myself, but it may work is to copy a "variable of interest" to an additional variable and have different settings for each additional variable. Likely will not scale in memory, but it may work for you needs. MyPressure1 = Pressure MyPressure2 = Pressure .. Then different settings for MyPressure1, MyPressure2, etc.. __________________ Note: I do not answer CFD questions by PM. CFD questions should be posted on the forum.

 Tags cfx-pre setup, phase averaging, time average