Disturbances in drag monitors after switching on 'Data Sampling for Time statistics'

shashanktiwari619 · August 25, 2017, 03:39

I am running a LES simulation for a flow over bluff body where in I monitor drag and lift coefficient w.r.t time. I switch on the 'Data Sampling for Time statistics' after my converge fairly well which is at around 1 secs (my time step is 1e-5). I have observed that the value of drag coefficient monitor show a lot of oscillation around the previously obtained time data of drag coefficient after I switch on the 'Data Sampling'. why is it so?? I have attached the time vs drag coefficient plot where I switch on the time statics after around 1.8 secs and the oscillations are clearly visible.
I get that switching on data statistics gives RMS and mean values of all variables for every time step specified. But why should it effect the drag monitor values??Am I missing out in understanding something here??

LuckyTran · August 25, 2017, 16:01

There shouldn't be any relation. Probably your problem was going to oscillate anyway. Can you prove it was caused by switching on data sampling? I.e. if you turn data sampling off does it go away? If you had not turned on data sampling, what result do you get? Did you also do something else?

shashanktiwari619 · August 26, 2017, 05:58

I tried running the simulations by turning off the statistics and attached are the plots that I have got.
The only two changes I did after initial run were 1. switching on the Data Sampling for time statistic and 2. Switching on Extrapolate variables.

LuckyTran · August 26, 2017, 08:53

I don't know the exact procedure but when you use extrapolate variables, Fuent predicts the values of the time-step (extrapolates) and then it uses that as the initial guess. Normally, the final values from the previous time-step are used as an initial guess.

Extrapolate variables is a separate option from time sampling and it does affect the solution.

Next time, be honest to yourself about what you did.

shashanktiwari619 · August 27, 2017, 03:34

Thanks for your reply LuckyTran.
I got it that this is because of the Extrapolate Variables that I have switched on.
My problem is as the plots change due to switching on the extrapolate variables, an FFT of the time domain signal shows that the frequency changes too.
In terms of Strouhal Number for Re 300 initially I get a value of 0.07 which does not match with reported value of 0.137 in literature. Whereas when I do the FFT analysis of the time data I obtain after switching on the Extrapolate Variables option gives me a value of 0.134 which is very much in agreement to previously reported values in literature. Why is there such a difference in St?
Which one of the two St should I consider and why?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Parachute drag data	Abuzar Adid	Main CFD Forum	3	July 27, 2013 08:33
"Data sampling for Time Statistics"	Madhukar Rapaka	FLUENT	2	June 20, 2011 12:28
Data sampling for time statistics	Muhammad Shakaib	FLUENT	0	October 3, 2006 00:33
Sampling of statistical data	Boerge	Main CFD Forum	0	July 25, 2006 05:40
Data sampling for time statistics with RANS	Flav	FLUENT	0	June 6, 2001 05:32

August 25, 2017, 16:01		#2
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,761 Rep Power: 66	There shouldn't be any relation. Probably your problem was going to oscillate anyway. Can you prove it was caused by switching on data sampling? I.e. if you turn data sampling off does it go away? If you had not turned on data sampling, what result do you get? Did you also do something else?

August 26, 2017, 08:53		#4
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,761 Rep Power: 66	I don't know the exact procedure but when you use extrapolate variables, Fuent predicts the values of the time-step (extrapolates) and then it uses that as the initial guess. Normally, the final values from the previous time-step are used as an initial guess. Extrapolate variables is a separate option from time sampling and it does affect the solution. Next time, be honest to yourself about what you did.

August 27, 2017, 03:34		#5
shashanktiwari619 New Member Shashank Join Date: Sep 2014 Posts: 28 Rep Power: 12	Thanks for your reply LuckyTran. I got it that this is because of the Extrapolate Variables that I have switched on. My problem is as the plots change due to switching on the extrapolate variables, an FFT of the time domain signal shows that the frequency changes too. In terms of Strouhal Number for Re 300 initially I get a value of 0.07 which does not match with reported value of 0.137 in literature. Whereas when I do the FFT analysis of the time data I obtain after switching on the Extrapolate Variables option gives me a value of 0.134 which is very much in agreement to previously reported values in literature. Why is there such a difference in St? Which one of the two St should I consider and why?