|
[Sponsors] |
Fully developed temperature profile for laminar/turbulent flow in FLUENT |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
April 16, 2012, 08:49 |
Fully developed temperature profile for laminar/turbulent flow in FLUENT
|
#1 |
New Member
Learner
Join Date: Nov 2011
Location: Ingolstadt
Posts: 27
Rep Power: 15 |
Hello FLUENT users,
I am new to heat transfer and am trying to simulate a simple case in FLUENT. The flow is between 2 parallel plates (so it is an internal flow). My task is to analyse the HTC (heat transfer coefficient) against the first cell height. So, basically it means that I need to check what mesh is best suited. I am trying to analyse both laminar and turbulent cases (only with komegaSST model). I use FLUENT 6.4. In laminar case (and also in turbulent), I have a problem. The velocity Profile is fully developed but thermal profile according to Le = 0,034*Re*Dh*Pr (from Lienhard and Lienhard (2008)) is never achieved. My geometrie includes a 5m length of the plates and a distance of 0,05m between them. So Dh = 0,1m 1) For the laminar case, I have analytical solutions with me. The results from simulations seem to match the results from analytical solutions with an error of 3% (this happens even though the temperature profile is not developed!!!!). Could some1 explain the reason for this? P S: However, I use the T_bulk as the reference temperature for all calculations in laminar case. Even if I consider the T_centreline, I get results with 20% error. This I understand is pretty normal. 2) For the turbulent case, I have analytical solutions too. But here, the velocity profile and the thermal profile dont seem to develop at all while infact I should be able to achieve the fully developed profile earlier than the laminar case due to turbulence. Could any1 clarify if I am performing the simulations correctly? Thanks in advance....... Raghu |
|
April 16, 2012, 15:18 |
|
#2 |
Member
banty
Join Date: Mar 2012
Posts: 52
Rep Power: 14 |
What are the boundary condition u r applying.
|
|
April 17, 2012, 03:35 |
|
#3 |
New Member
Learner
Join Date: Nov 2011
Location: Ingolstadt
Posts: 27
Rep Power: 15 |
Thanks for the reply.
Initially I simulate with the working fluid as air and then I also do simulation with water as the working fluid. I apply a velocity inlet of 1 m/s and an inlet temperature of 283K. outlet is a pressure outlet with 0Pa wall is a temperature of 10 W/m2-K and 293K (I simulate both the constant wall temp and constant heat flux case.) I also feel there maybe some differences because of the geometry. I have a geometry of 3m length, 0.05m height and a small width of 0.001m. I assume this assumption of 0.001m as the z-coordinate serves the purpose of a simulating a 2D flow. Basically I would like to simulate a 2D flow. (This is because I need to migrate to OpenFOAM if I am sure about the results and the setup) I can attach a sample .cas if necessary or send it to your email directly (the file size may be a little too large to upload in this forum) |
|
April 17, 2012, 04:46 |
|
#4 |
New Member
Learner
Join Date: Nov 2011
Location: Ingolstadt
Posts: 27
Rep Power: 15 |
Update:
I understand that I have not enough length of the domian to obtain a fully developed profile for the turbulent case. TO obtain this, should I first develop the velocity profile and then run the energy equation or is it possible to develop them simulataneously in FLUENT? The entrance length according to my calculation is ~16.6m. I assume its best to write a UDF. |
|
April 17, 2012, 09:09 |
|
#5 | |
Member
banty
Join Date: Mar 2012
Posts: 52
Rep Power: 14 |
Geometry is fine if u are trying 2D simulation but make sure that there is only one cell in z-direction and apply symmetry boundary condition to z-direction boundarys.
U can check the range of Z-Velocity after simulation. it should be in range of 10^-10 to be compatiable with 2D result. Quote:
|
||
April 17, 2012, 09:10 |
|
#6 |
Member
banty
Join Date: Mar 2012
Posts: 52
Rep Power: 14 |
Yes we can solve both energy and momentum eq. simultaneously.
|
|
April 18, 2012, 09:34 |
|
#7 |
New Member
Learner
Join Date: Nov 2011
Location: Ingolstadt
Posts: 27
Rep Power: 15 |
hi Banty,
Thanks for ur instant reply. a simple Q, according to calculations in laminar flow, laminar entrance length for velocity as mentioned in the first post. Is there a mesh-dependency on fully developed velocity profile? Here one has to understand that I am uisng a simultaneously developing flow and not inputting a fully developed velocity profile at the inlet. The above Q s coz, for a 3m length of the pipe for air as working fluid and v = 0.1m/s, I'm not able to achieve a fully developed velocity profile even at the outlet. the profile is still varying. But for a very fine mesh and a first cell about 0.1mm, I get the velocity profile developed. has mesh got to do with development of velocity profile. according to my understanding, NO. |
|
April 18, 2012, 10:02 |
|
#8 |
Member
banty
Join Date: Mar 2012
Posts: 52
Rep Power: 14 |
Hi,
entrance length does not depend upon the mesh physically but numerical error certainly does. so for any simulation, grid independence should be checked. |
|
April 25, 2012, 04:45 |
|
#9 |
New Member
Learner
Join Date: Nov 2011
Location: Ingolstadt
Posts: 27
Rep Power: 15 |
Hi banty,
I just happened to find the solution to achieve a full developed temperature profile in FLUENT... Use Periodic boundary conditions from FLUENT. Since I'm simulating incompressible flow, its easy for me...... Now I have a new Q, In fully developed turbulent flow, when I use air as the working fluid, everything is perfekt, but when I use water as the working fluid (all other settings remaining same), 1) for constant wall temperature: the solution is diverging 2) for constant heat flux: Im getting unphysical solutions, heat transfer coefficient is too high than analytical solutions. This problem is due to the case setup , or do I need to have a better boundary layer to simulate higher density fluid? Raghu |
|
April 25, 2012, 04:48 |
|
#10 |
New Member
Learner
Join Date: Nov 2011
Location: Ingolstadt
Posts: 27
Rep Power: 15 |
hi banty,
I have done grid independence check using GCI method. But I now need to know for what values of y+ do i get a good result. My case is a simple 2D one and this proves as a basis for further complex geometries..... But understanding this is utmost priority... Thanks |
|
November 7, 2012, 10:57 |
|
#11 |
New Member
Ayoub
Join Date: Nov 2012
Posts: 9
Rep Power: 14 |
Hi
I need to reach fully develop flow in microchannel(D=100 micrometer,L=1mm) q=500,Re=100 to 1000,Tin=300,working fluid is water. i can not reach thermal fully developed,i don't know why even when i apply Le=.05 Re D Pr would you please show me a solution? |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Udf for a fully developed velocity profile atinlet | philip meppen | Fluent UDF and Scheme Programming | 10 | November 3, 2015 16:56 |
time dependent BC with fully developed velocity profile | susheil | FLUENT | 6 | December 20, 2012 06:41 |
Nusselt numbers for fully developed laminar flow | Danro | Siemens | 8 | July 19, 2012 08:22 |
Fully Developed Flow in Star-cd | SMM | STAR-CD | 0 | September 5, 2011 23:08 |
Writing an expression for fully developed flow! | Usman | CFX | 12 | December 20, 2007 12:26 |