
[Sponsors] 
April 15, 2013, 08:20 
Manual calculation of Heat Transfer Coefficient

#1 
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 
hey there,
i know this subject was discussed for many times ... but i didn't find the answer in one of this thousands threats. of course the calculation for the HTC for laminar flow and steady state is: HTC = areaInt(wall heat flux)@<interface> / (area()@<interface>*(areaAve(T)@<interface>tbulk)  i set tbulk to my inlet temperature  the wall heat flux and area are for <interface> side 1 and <interface> side 2 equal but i have a difference for the areaAve(T)<interface> side 1 and <interface> side 2 so my question is: what is the definition of T_wall ? is it the fluid side or solid side ... anyway if i try both i get different values to the "direct" HTC calculation (areaAve(HTC)@<interface> side 1) ... what could be the reason for that? could be the reason that iam calculating conservative variables instead of hybrid values ? how is cfx calculating the HTC with conservative or hybrid ? thanks in advance! other b.c.: grid: ggi fluid: air interface: fluidsolid 

April 15, 2013, 18:58 

#2 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,803
Rep Power: 85 
At a solidfluid interface the Twall parameter comes from the solid domain.


April 16, 2013, 01:33 

#3 
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 
yeah that was my first impression, too ... and of course ... if i calculate the HTC with the temperature of the solid domain i will get a "better" result.
"better" means that the value is closer to the HTC value which i got from CFX with the CEL command areAve(Wall Heat Transfer Coefficient)@<interface> But i won't get the exact value ... i tried to calculate the solid temp. with hybrid values but the result will not change ... actually i can not understand why i get different values if i calculate the HTC manually instead of using the CEL command ... :/ In the CFX help i only found this explanation: "For laminar flow modelling the wall temperature is just the local fluid temperature at the vertex adjacent to the wall." ... i have no idea what they suggest with that Last edited by TrII4d; April 16, 2013 at 03:03. 

April 16, 2013, 19:34 

#4 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,803
Rep Power: 85 
Be careful which variables you are using. The default conservative variables give you the condition assumed to exist over the entire control volume  sort of an average. The centroid of the control volume lies a little way into the domain, so represents the conditions a little way into the fluid or solid side from the domain. Hybrid variables can give you temperature of the interface, but you cannot do CEL expressions on hybrid variables.
"For laminar flow modelling the wall temperature is just the local fluid temperature at the vertex adjacent to the wall." is just a statement of the laminar heat transfer boundary condition. In laminar flow the control volume at the boundary is assumed to have the wall temperature. In turbulent flow the wall functions can mean there is a jump in temperature at the wall to account for the laminar sublayer which is not modelled directly. 

April 17, 2013, 01:25 

#5  
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 
hi ghorrocks,
first of all ... thanks for ur great support! Quote:
that's interesting ... i didn't know that ... this will be a problem because i'm using a perl script for doing my post processing Quote:
my problem is to find the CEL function which calculates the "real" wall temperature ... 

April 17, 2013, 07:51 

#6 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,803
Rep Power: 85 
Let me clarify the laminar comment  the face of the control volume which forms part of the interface has the wall temperature. The control volume will have whatever thermal conductivity says the region that far away from the wall will have. So if the boundary is a solid/fluid interface then the fluid side and solid side control volumes will be slightly different as they represent control volumes slightly into their respective domains.
The CFX doco talks about a variabel called 'Surface temperature' but it does not seem to be listed in the available variables. 

April 17, 2013, 09:13 

#7 
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 
hmm ... where exactly did u find the variable 'surface temperature' in the docu?
can u tell me the version, name of the book and side pls (f.e. cfx 14.0 CFXsolver theory guide, page 5556) actually its very sad ... CFX is calculating the HTC but u r not able to find the equation in the theory guide to understand what CFX is doing 

August 29, 2013, 10:15 
Summary

#8 
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 
I did some small studies for a plate, cube and cylinder last week ...
the results are:  calculating HTC with the formula in the first post and with the temperature (conservative) from the solid side it gives the same value as CFX (for hexa and tetra mesh)  in a first study i used Tw=const. as BC ... in a second study i used q/(dt)=const. .... both give the result above  i didn't check the calculation with hybrid values (cause of time problems)  nevertheless if i calculate HTC for a complex geometry in a complex field size the difference are sometimes round <10% ... sometimes its <1% but i don't know why .... if there are any other studies pls let me know ... thanks! 

August 29, 2013, 10:53 

#9 
Senior Member
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,098
Rep Power: 19 
Not sure if this is related to your problem, but the last time I used CFX for heat transfer simulations I had to activate the expert parameter "Tbulk for HTC".
Otherwise, CFX takes a temperature value somewhere in the flow as the local reference Tbulk temperature, which is inconsistent with the usual formula for heat transfer coefficients. 

August 29, 2013, 18:35 

#10 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 10,803
Rep Power: 85 
Note this Tbulk for HTC expert parameter just affects the value reported for post processing. It does not change the heat transfer modelled in the solver at all  so it just changes the reference temperature used to report the HTC in the post processor.


August 30, 2013, 02:05 

#11 
Senior Member
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,098
Rep Power: 19 
I didnt mean to imply anything else. Sorry for being a little unclear.
I just thought that when comparing heat transfer coefficients manually calculated from wall temperatures and heat fluxes to the ones calculated by CFX, the Tbulk for HTC option is the better choice. 

August 30, 2013, 02:20 

#12 
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 

August 30, 2013, 03:36 

#13 
Senior Member
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,098
Rep Power: 19 
So you were already aware of this and used the expert parameter?
Then I have some doubt about the expression you use for calculating the average heat transfer coefficient manually. Instead of: areaInt(wall heat flux)@<interface> / (area()@<interface>*(areaAve(T)@<interface>tbulk) try to use: areaAve(wall heat flux/(TemperatureTbulk))@<interface> 

August 30, 2013, 03:49 

#14 
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 
yeah i already used it ...
your equation is still the same ... iam calculating with the heat flux and you are calculating with the heat flux density ... both is possible and both give the same result 

August 30, 2013, 05:01 

#15 
Senior Member
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,098
Rep Power: 19 
They dont give the same result if the wall temperature or heat flux is not constant. I just tried that in CFX.
My alternative expression recovers the area average of the heat transfer coefficient in this case. The other expression does not. 

August 30, 2013, 05:29 

#16 
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 
okay this can be possible ... but in my case i always calculated steady state .... so there were only three options for BC and i chose two of them (Tw=const. or Q=konst.) ... but your statement is an interesting information ...


August 30, 2013, 05:35 

#17 
Senior Member
Alex
Join Date: Jun 2012
Location: Germany
Posts: 1,098
Rep Power: 19 

August 30, 2013, 05:39 

#18 
Member
Join Date: Nov 2011
Location: Germany
Posts: 40
Rep Power: 5 
oh ok ... interesting study ... now i got your point ... i will try your alternative equation for the complex geometry (because in this case Q is not constant in space) ... hopefully i can give feedback by next week !


Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Heat transfer coefficient  what is waht  Stan  FLUENT  20  March 11, 2014 11:42 
strange wall heat transfer coefficient  Lea  FLUENT  2  December 18, 2012 10:23 
Distribution of surface heat transfer coefficient along the axial direcion of a tube  andred  FLUENT  0  November 16, 2010 22:13 
How can I increase Heat Transfer at Domain Interf?  B.Simon  CFX  3  October 28, 2008 19:53 
TwoPhase Buoyant Flow Issue  Miguel Baritto  CFX  4  August 31, 2006 12:02 