[beauty]

dears,

I am solving a time-dependent problem using fluent. Fluent help says "You can set the Sampling Interval such that Data Sampling for Time Statistics can be performed at the specified frequency". However, I am puzzled by the relation between sampling interval and sampling frequency. I wonder whether the following equation is right.
sampling frequency=1(time step sizesampling interval)

In addtion, how can I set an appropriate Sampling Interval, according to sampling frequency in the experiment?

Thanks for your help!

beauty

[enayath]

Hello Beauty!

Did you find the answer of your question?

I have the same question and wondering if can provide me the answer.

Thank you,
H

[LuckyTran]

There is no sampling frequency option. There is only a sampling interval.

Sampling Interval = 1 means sample every time-step
Sampling Interval = 2 means sample every other time-step

To achieve a certain sampling frequency (with units 1/s), you must choose the sampling interval and time-step size accordingly.

The relation between frequency, sampling interval, and time-step size given was correct.
frequency = 1/(step-size x sampling interval)

[enayath]

Thank you LuckyTran!

[srv537]

Originally Posted by LuckyTran

There is no sampling frequency option. There is only a sampling interval.

Sampling Interval = 1 means sample every time-step
Sampling Interval = 2 means sample every other time-step

To achieve a certain sampling frequency (with units 1/s), you must choose the sampling interval and time-step size accordingly.

The relation between frequency, sampling interval, and time-step size given was correct.
frequency = 1/(step-size x sampling interval)

sample interval = 200 means it will calculate average value of all 200 time steps.
is it right?
i am using delta_t = 10^-6, could you please suggest me what sampling interval should i use?
thank you

[sbaffini]

It means it will collect data each 200 time steps. Hence, after the first 200 time steps, your statistics will coincide with the instantaneous field saved at the 200th time steps.

To do what you wrote you need sample interval = 1. This, btw, is always your best choice if you don't know what you're doing.

[srv537]

Originally Posted by sbaffini

It means it will collect data each 200 time steps. Hence, after the first 200 time steps, your statistics will coincide with the instantaneous field saved at the 200th time steps.

To do what you wrote you need sample interval = 1. This, btw, is always your best choice if you don't know what you're doing.

if i select select sample interval =1 or 10 then i should save Data and case file for every 1 or 10 time step respectively.
if sample interval is 3 and autosave of data and case file is 100 then i wont get u', v' and w' at any time step.
am i right? if not then please explain.

thank you

[sbaffini]

I never tried but i suspect that it simply works such that when you save you just get the last set of collected samples. For example:

Autosave each 100 time steps
Sample interval 30 time steps

Then, i guess, at the first autosave after 100 dt the statistics saved in your file are relative to:

Time 0 (not sure here)
Time 0 + 30 dt
Time 0 + 60 dt
Time 0 + 90 dt

Still, obviously, i can't see any reason to do something like this.

In theory, the sample interval should be the minimum to achieve sufficient statistical independence between the samples. In practice, 1 works just great if you collect enough samples.

[The_seeker]

Hi,
I am running a transient simulation with time step size of 0.0002 sec. The total simulation time is 30 sec. The data needs to be collected from 10-30 seconds. So, what I do is.... I calulate the flow till 10 seconds and then I turn on the data sampling with sampling interval of 1 from 10 to 30 sec. Is it the correct way of getting time averaged solution from 10 to 30 sec.

[LuckyTran]

Yes, that is how to do it.

[Yi Zhao]

Originally Posted by LuckyTran

Yes, that is how to do it.

Hi
I am running a large eddy simulation with time step size of 0.002 sec. The total simulation time is 4 sec. The number of iterations is 2000. I turn on the Data Sampling with Sampling Interval of 10.
I would like to know if the time-averaged velocity data saved in the last iteration step is representative of the time-averaged velocity of the entire flow process? If not, how do I get the time-averaged velocity that characterizes the entire flow process for comparison with the RANS results?
Thank you!

[LuckyTran]

The average that you get at the very last time step is the running average starting with the time-step that you enabled data sampling (i.e. average of 200 samples) and this could be the time-average that you get in RANS....


However, you set a sampling interval of 10 and not 1 and your time-step size is rather large and your number of time-steps is rather low. Sampling theory applies and small samples may not represent the population.

[Yi Zhao]

Originally Posted by LuckyTran

The average that you get at the very last time step is the running average starting with the time-step that you enabled data sampling (i.e. average of 200 samples) and this could be the time-average that you get in RANS....


However, you set a sampling interval of 10 and not 1 and your time-step size is rather large and your number of time-steps is rather low. Sampling theory applies and small samples may not represent the population.

Thank you LuckyTran！
The time step size is obtained by making the Courant number less than 1, of order 10e-3. It takes 3.1 sec for the airflows to flow over the entire computational domain in the model, so I set the number of iteration steps to 2000 (i.e. 2000*0.002=4 sec), is it right?
If I set Sampling Interval to 1 and enable Data Sampling, will the average of the last time step (2000th sample) obtained be representative of the total and compare with the results of RANS? If I set iteration 50 time steps in Autosave to save a .dat file, is the average value obtained from the last (40th) .dat file the same as the average value obtained from the 2000th sample of Sampling Interval?
Also, I would like to ask if the square of Mean RMSE Velocity X can be considered as the time average of u'u'?
Thank you again.

[maria94]

for your last question, this could help: https://www.afs.enea.it/project/nept...ug/node462.htm