Time average results in Transient CFX Simulation
 

Hi my friends, I am doing a transient simulation using CFX and my question is how can I get a time average flow field of simulation. I tried to use the Arithmetic Average in CFX Output Control. But the results seems not right. Who can help me out? Thank you very much.

BalanceChen

Did you use the transient statistics option? What makes you think the arithmetic average did not work?

Yes, I use the transient statistics option.

I am doing a transient simulation of turbomachinery and the inflow is not uniform circumferentially. The instantaneous total pressure at the rotor inlet is distorted as it should be, but the arithmetic average not. Please refer to the attachment for more information. Thank you very much~

good work. well done

I have just found the trouble spot~

As for the transient turbomachinery simulation, a time-average results is time averaged over passage but not over the whole machine. I made a mistake to average over the whole machine to destroy the circumferential distortion.

makes any difference ?

time avrage
 

hello,

I am doing transient problem and I am gonna do time avrage for velocity . can you tell me how can I do it?

Best regards

Just activate "Transient Statistics" under the "output control->TrnStats" in CFX Pre.

I dont know how can i find avrage result in post cfd.

You just select in CFX Post the "Velocity.TrnAvg" Variable to be displayed (on a Plane for example). Then you will see the time averaged Velocity.
All time averaged Variables are named like this "Variablename.TrnAvg"

Thanks
 

thank you so much.

But if I want the average velocity in one specific period, how can I do? Thank you

Hello, you got to specify this time period (start averaging time - end averaging time).

If you want a time average for a specified period only you will have to determine the start and end iteration during your calculation run and specify them as start and end iteration in CFX Pre for the definition of the Transient statistics....
Unfortunately I never discovered an option to specify a start and end TIME...only iteration loops.

Heat flux time average in CFX
 

Hi,

I want to find out the time average Heat flux on a surface in CFX. I have activated Arithmetic avg in trn stats in CFX Pre but there is no sign of Heat flux.Trnavg in the result variables. Can somebody please help me with this issue?

Thanks

No change in transient results in CFX
 

I am doing the supersonic flow over a forward facing step and
it's a transient simulation. Actually it converges after 50-60 iterations. (the momentum components).
The problem is when i go to CFD post and check the results I don't see any difference between the results from the very written result (i have them written every 0.1 s) till the 2 seconds that is the end of simulation. I have tried very fine mesh and regular mesh and I know the mesh is fine. The boundary conditions are followed as the problem descrpition. my timestep is 1e-4.
My friend got reasonable results with the same mesh size and BC in OpenFoam and saw the shocks developing overtime and detach.
Mine, on the other hand, just shows the detachment of the shock from the very first step till the last one as if it is steady!!!!
I have attached the pictures for 0, 0.1 (first), 0.5, 1, and 2.
I would appreciate the help
Best,
Hesam

You have too much dissipation in your model and it is causing the small flow features to disappear. You will need:
* Finer mesh (do a mesh sensitivity study)
* Finer time step (do a time step sensitivity study)
* High order differencing (High Res is probably best)
* Dynamic mesh refinement at the shocks would help

Hi, I haven't specify it in pre, is there any way to do it in cfx post?

Mina,

I dont think that is possible. I had same problem and I ended up resolving the case.
By the way is there anyway to define the model in a way that it does the transient averaging only over a portion of the total simulation time?

Thanks,
Ali

well if you start and end the averaging in transient result it should average over the time specified, isn't it ?

You can always create an expression with the step function to get a non-zero result only during the time you want. then try an integration
it is a gross solution, but a solution anyway

The Only way that i know is to set it in pre:

Outline>Output Control>Trn stats

then define the start iteration list.(for example from iteration number 2500 or so).

but if you don't do it in pre i suppose you have to do it with matlab.

hello，Thomas，
I want to know how to use cfx cel and expression to get averaging in transient result ,can you help me!

thank you very much

This is an old post that i came across while searching for how to get time averaged statistics. I can't find how/where to do this. Can anyone help.
Also I haven't created transient statistics in cfx-pre.

@monkey1 Hello, I'm trying to plot the time average for a simulation in CFX but I didn't activate the "Transient Statistics" at the Pre stage, as you explained here. Is there any other way to get this activated in order obtain the plot after the simulation has already ended, i.e. at the Post stage? I need the whole time average of some variables... All I need to get is the plot, so if there is any way to do it, I'll just do with it.

Thanks in advance to any helpful reply you may provide!

As far as I can see, there are 2 ways for time averaged results:
1) Create a Monitor point with time averaging (Max, Min, Avg, etc) - this will be saved to the *.res file for every time step and can be extracted via monitor in solver manager or cfx5mondata command.
2) Create a *.trn file every n iteration with the desired variables.

In case you have saved some *.trn files, you can loop through them with a script and extract the desired data. The averaging has to be done manually in Matlab for example.
Alternate way: create a plot in Post with the time on one axis. Caution: Hitting apply let's Post loop through all *.trn files, which might take long.

If you don't have any *.trn files and have no monitor points set up accordingly, you only have the results of the last time step in your *.res file.

when your run is finished you can extract the values for each time step (as long as you have the .trn files) by using a perl script as -Maxim- said, to import the values to e.g. excel and calculated there the time averages.
An example for a script file can be found under
http://www.cfd-online.com/Forums/cfx...tml#post385850

Thank you both -Maxim- & monkey1. I will try and apply your recommendations.

How can I do what you recommended in your answer, -Maxim-? In which stage do I create the monitor? In Pre setup, in the Solver or at the Post stage?

How can I activate the cfx5mondata command and in which stage do I do that?

How about this... I have the last .res files and some .trn but not all of them. If I run another simulation, just varying the Reynolds number of the flow, in transient regime, saving all .trn files and activating the .TrnAvg statistics in Pre, using as the entry values of the new run the last .res file created, that would provide a continuity to the simulation process with just a different reynolds number and, hence, provide more data. If I do all that, will the new .res only contain the info related to this new run or will it use the previous run extra-polating through it to use this data? All that to make sure I could get to that previous info.

I don't know if the last paragraph makes sense to you...

Monitorpoints are set in cfx-pre under output control and evaluated during the run. You can only see the results in the solver manager and from there they can be exported to a csv file.

If you start a new simulation...then why trying to "provide continuity"?...especially when you change basic flow settings. A new simulation run is a new simulation run and when you activate the transient statistics there, you will get them for this run.
An extrapolation as you thought of would only be possible if you don't have any big changes in your solution during the simulation.
I think it would then be better to use your results and few trn files to extract the required values with a perl script and calculate the statistics from there, even if the statisical basis is not fully given.

1) create monitor points incl. averaging etc in Pre. You can set monitor points on expressions. Consult the manual or search this forum if you don't know how.

2) regarding cfx5mondata, the manual is explaining is pretty well. Or search this forum. Here I wrote some sample syntax:
http://www.cfd-online.com/Forums/cfx...tml#post603695

3) I don't think CFX interpolates/creates averaged variables if you continue with a new setup. I think the best way would be to set up a new run with all the monitor points etc, let it run for a few iterations until you get a couple of *.trn files and then stop to get the *.res file. Then test your scripts/monitor points. If result is good, do you long calculation.

Good luck

dear BalanceChen,I am also doing a transient simulation of turbomachinery with inflow total pressure distortion circumferentially, and I also meet the problem that the arithmetic average total pressure contour doesn't vary circumferentially.Have you solved this problem?
what do you mean by the time averaged over passage or the whole machine? And where could I set up the averaged domain?
It's a post long time ago, wish you could see my confusion

You have not completely described your problem.

You have a total pressure distortion in front (as an inlet) of a turbomachinery component, correct?

On which frame is the distortion, and on which frame is the component? If both are in the stationary frame, you should see the circumferential variation. However, if the distortion is moving relative to the component frame, you will only see the variation in one frame, but not in the other.

Thanks first, Opaque!
There are 3 domain in my simulation.First, it is a stationary domain which consists of inlet boundary condition,and steady total pressure distortion is in this domain.Second ,it is a rotating domain which has rotor, and I check the countor whether total pressure changes or not circumferentially. Third, it is stationary domain which has stator.
domain which stator is in it.

Are you looking at the contour on the stationary side of the sliding interface, or the rotating side?

Are you looking at the "Stationary Frame Total Pressure" or "Total Pressure" variable?