
[Sponsors] 
June 5, 2013, 08:27 
Turbulent Heat Flux <u'T'>

#1 
Senior Member
n/a
Join Date: Sep 2009
Posts: 198
Rep Power: 7 
Hello Foamers... I made a fairly simple modification to my solver to enable the computation of uT. Given that I am interested in the timeaveraged values of this variable, I added it to the controlDict file to allow for this. Now, I understand what mean of uT is, just <uT>. However, for the prime2Mean, what would that be for uT? Is it <u'u'T'T'> ? And if so, how do I get the turbulent heat flux, <u'T'> from <u'u'T'T'>. I know there is another way using <u'T'>=<uT><u><T>, which is an approximation. But how can I compute the <u'T'> directly in OpenFOAM?


June 10, 2013, 06:49 

#2 
Senior Member
Mohsen KiaMansouri
Join Date: Jan 2010
Location: CFD Lab
Posts: 112
Rep Power: 6 
Dear deji
I used the same method as yours i.e. adding the uT variable in my solver and activating the averaging function for <uT> in controlDict of LES and then computing <u'T'> using <u'T'>=<uT><u><T> . I don't know the physical meaning of prime2Mean (Variance) for u'T' and the relationship between them. If you find something, please post it in here so that everybody can use it Thanks 

June 10, 2013, 11:57 

#3 
Senior Member
n/a
Join Date: Sep 2009
Posts: 198
Rep Power: 7 
Thanks for the response. I've actually solved the problem; the definition of <u'T'> that I described to be an "approximation" actually is not one if the averaging time is "long".


June 10, 2013, 14:24 

#4  
Senior Member
Mohsen KiaMansouri
Join Date: Jan 2010
Location: CFD Lab
Posts: 112
Rep Power: 6 
Quote:
I will consider the convergence time as the time which the flow sweeps my whole domain 3 times from inlet to outlet. (for the cases with inflow, NOT cyclic B.C.). (Although in my experience 2 Times would be sufficient and the results does not change after that) After that, I will turn on the averaging function in LES to run. I will consider the averaging time the same as convergence time i.e. the time which the flow sweeps my whole domain 3 times from inlet to outlet. after that I will postprocess my results. Is it the same approach that you usually use?
__________________
Best Regards Mohsen (Don't forget the Like button! ) 

June 10, 2013, 16:13 

#5 
Senior Member
n/a
Join Date: Sep 2009
Posts: 198
Rep Power: 7 
Typically, in forced convection turbulent boundary flow, to attain statistical convergence of 2nd order statistics (u'u',u'T','u'v', etc), it is "recommended" the flow sweep the computational domain 1520 times. Some refer to this a flowthrough times. However, since my research pertains to buoyancy affected wall bounded turbulent flow, I have discovered the flow needs roughly 3040 flow through times to acquire statistical convergence of some 2nd order statistics. Thus, it greatly depends upon the kind of turbulent flow you are computing in order to properly "ascertain" how much flowthrough times is required to achieve 2nd order convergence. Remember to always compare the 2nd order quantities as opposed to the 1st order quantities to determine this!


June 10, 2013, 17:44 

#6  
Member
Ping
Join Date: Dec 2011
Posts: 62
Rep Power: 4 
Quote:
You can calculate <u'T'> directly if you check the source code how OF calculate <u'u'>, <u'v'>, etc. (i.e. UPrime2Mean). If you are still interested in it, I can find the code for you. Ping 

August 20, 2013, 09:59 

#7 
New Member
Join Date: May 2011
Posts: 6
Rep Power: 5 
Hi ripperjack,
it would be very nice, if you could provide the source code to calulate the turbulent fluxes. Tobi 

August 20, 2013, 10:23 

#8 
Senior Member
n/a
Join Date: Sep 2009
Posts: 198
Rep Power: 7 
It is actually fairly easy to compute the turbulent heat fluxes. A minor modification to the source code to calculate uT is all that's need. With the variable uTmean available, <u'T'> can be computed in a postprocessing step using <u'T'>=<uT><u><T>, assuming the averaging time is "long enough".
Cheers Deji 

August 20, 2013, 11:57 

#9 
New Member
Join Date: May 2011
Posts: 6
Rep Power: 5 
Hi Deji,
is there also a way to calculate the fluxes at runtime? Cheers Tobi 

August 20, 2013, 12:15 

#10 
Senior Member
n/a
Join Date: Sep 2009
Posts: 198
Rep Power: 7 
I do believe there is a way to compute the fluxes at runtime. You'll have to modify the code to enable the calculation of <uT> ,<u> and <T> at runtime, instead of using the fieldAverage function object for this. I think it ought to be straightforward.


November 13, 2013, 14:26 

#11 
New Member
Simon Emhardt
Join Date: Feb 2013
Posts: 9
Rep Power: 3 
is there anyone who can provide the modification of the solver for computing uT?


November 13, 2013, 14:32 

#12  
Member
Ping
Join Date: Dec 2011
Posts: 62
Rep Power: 4 
Quote:
BTW: you can get reference at this link: Why Uprime2mean is minus in some grids? Regards, 

November 13, 2013, 14:40 

#13 
Senior Member
n/a
Join Date: Sep 2009
Posts: 198
Rep Power: 7 
is there anyone who can provide the modification of the solver for computing uT?
It is actually a rather simple modification that is required to the code being utilized for the computation of the turbulent heat flux. You ought to compute uT in your flow solver and utilized it to calculate <u'T'> as a postprocessing step. 

January 23, 2014, 07:02 

#14  
Member
Peter
Join Date: Nov 2011
Posts: 45
Rep Power: 5 
Quote:
I want to calculate <u'T'> too. I'm not so sure how to add uT in my solver. Could you please post your code here? Just the "adding uT" part would be fine. It would be very helpful for me and others who try to do the same thing. Regards, Peter 

Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Turbulent Heat Flux Question  deji  OpenFOAM  4  May 22, 2013 03:18 
Heat flux and Heat transfer coefficient  sakil2k3  FLUENT  3  May 9, 2013 22:42 
heat flux distribution  gtheo  CFX  2  April 27, 2011 04:36 
how to impose experimental dat as boundary conditi  Rogerio Fernandes Brito  FLUENT  14  November 25, 2008 06:47 
Wall heat flux  cujo  CFX  3  November 6, 2002 04:43 