|
[Sponsors] |
Ansys Fluent UDF - for Data Center Air flow management |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
December 20, 2010, 01:00 |
Ansys Fluent UDF - for Data Center Air flow management
|
#1 |
New Member
|
i to everyone,
I am doing a masters level project in the field of Air Flow management is a Data Center using CFD. I have done a modeling and meshing for this problem. I have created a .msh file. the only problem that I am facing is in order to apply the boundary condition in Fluent. I am working on Ansys Fluent 12.0. The cold air is sucked in from the front end of the Rack, in which all the servers are mounted (19inch std racks of 42U) the air picks up the heat from the servers and is thrown out from the rear end. The hot air naturally goes up and again goes back to computer room air conditioner unit (CRAC'S) and becomes cool and the cycle continues. The problem is how to apply boundary conditions at the rack inlet and outlet? one solution is writing a separate UDF for it? UDF should be such that it should keep the mass flow rate of air same at inlet and outlet of Rack and it should take up the heat while it is travelling over a rack (say a 5KW IT load/Rack of heat generation) Should I go for writing a UDF? or is there any other way of specifying BC's? please help me out guys.. I am working on this project for 5months and stuck up at this stage. Thanks do reply to this blog Kedar |
|
October 4, 2011, 05:12 |
|
#2 |
New Member
Aruna Krishantha
Join Date: Oct 2011
Location: Sri Lanka
Posts: 5
Rep Power: 15 |
I also face in your problem since 2 month can u help me
|
|
October 4, 2011, 05:17 |
Data center modeling
|
#3 |
New Member
Aruna Krishantha
Join Date: Oct 2011
Location: Sri Lanka
Posts: 5
Rep Power: 15 |
Dear Sr.
I am a undergraduate in University of Moratuwa,Sri Lanka.I am doing my final project which models data center energy management by using CFD(Ansys fluent).I also faced in your problem.Can U help me? .I believe U complied in your project.I am waiting for your request Aruna(arunakma@gmail.com) |
|
October 20, 2011, 21:34 |
|
#4 |
New Member
Tom M
Join Date: Oct 2011
Posts: 7
Rep Power: 15 |
You do not need a UDF for this. You can simply use a mass flow inlet BC with the correct mass flow and a pressure outlet BC. Fluent will automatically conserve mass by varying the pressures in the system. As far as modeling the heat loads, just enable the energy equation and set the boundary temperatures of your electronics or specify a heat flux on their geometry.
|
|
October 21, 2011, 05:45 |
Cfd
|
#5 |
New Member
Aruna Krishantha
Join Date: Oct 2011
Location: Sri Lanka
Posts: 5
Rep Power: 15 |
Thanks for U replay
I modeled according to your method. but I need to flow air through the racks for getting correct value, I tried but I can not mode until now, please help me anyone.... |
|
October 21, 2011, 14:08 |
|
#6 |
New Member
Tom M
Join Date: Oct 2011
Posts: 7
Rep Power: 15 |
Your response doesn't give any more insight into your problem. I can't help if you don't tell me anything new. If you use these BC's and model the rack geometry (likely in 3D), there shouldn't be any issues. It is a simple steady flow problem. Use the regular incompressible air properties and add a K-Epsilon turbulence model if your velocities indicate the need.
|
|
October 21, 2011, 15:29 |
Cfd
|
#7 |
New Member
Aruna Krishantha
Join Date: Oct 2011
Location: Sri Lanka
Posts: 5
Rep Power: 15 |
in my model,there are 32 racks which are placed in rows, 8 racks are placed in single row. there are four rows,inlets placed alternate rows ,air comes through the inlet and it should be flowed through the racks to adjacent rows that mean I need to select boundary condition of the all racks, specially I need to select heat surface.I model space of the data center then I cant select boundary condition surface of the racks according to above mention requirement
|
|
October 21, 2011, 15:46 |
|
#8 |
New Member
Tom M
Join Date: Oct 2011
Posts: 7
Rep Power: 15 |
do not model the racks as solids, just use their profiles to define the edge of your fluid boundary. Label each profile so you know what you're assigning BC's to and then in fluent, apply an appropriate heat flux for each rack (W/m^2) if you know it. If not, apply a constant surface temperature for each based on your best guess of rack temp. The BC you're looking for is simply "wall" to model the rack profiles.
|
|
October 23, 2011, 10:00 |
Cfd
|
#9 |
New Member
Aruna Krishantha
Join Date: Oct 2011
Location: Sri Lanka
Posts: 5
Rep Power: 15 |
Thanks for your reply
I can not understand the use of the profile.can U explain with more details |
|
December 20, 2011, 21:36 |
|
#10 |
New Member
Ruoxun Wu
Join Date: Dec 2011
Posts: 2
Rep Power: 0 |
Excuse me!Have you solved this problem?I'm also doing similar master project,but haven't built the model yet!Would you like to give me some advice?Thanks!
|
|
December 20, 2011, 21:38 |
|
#11 | |
New Member
Ruoxun Wu
Join Date: Dec 2011
Posts: 2
Rep Power: 0 |
Quote:
|
||
November 13, 2013, 01:15 |
|
#12 |
New Member
|
Hello;
I have tried what you have specified here. But it still doesn't work properly. I am getting the air flow inside the rack. But it doesn't catch up the heat and goes out as desired. Do we have to model the fans which are present at the rear side of rack? Kindly reply Thanks and Regards |
|
November 13, 2013, 02:23 |
|
#13 | |
Senior Member
Join Date: Feb 2010
Posts: 164
Rep Power: 17 |
Try porous media for the racks, and "Fan" boundary condition for rack inlet.
Quote:
|
||
Tags |
ansys fluent udf |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
A Problem of Fluent Interpreted UDF: parse error | knight | Fluent UDF and Scheme Programming | 25 | August 16, 2018 11:26 |
To define Moist Air in Fluent | JW | FLUENT | 7 | April 29, 2015 06:18 |
Fluent (ANSYS12) UDF Help, PLEASE! Urgent! | ANITIX87 | FLUENT | 21 | April 12, 2011 04:55 |
To define Moist Air in Fluent | JW | FLUENT | 1 | April 15, 2006 18:31 |
UDF in Fluent to Match Mass Flow at Pressure Outlet | Jonas Larsson | Main CFD Forum | 1 | April 29, 1999 11:44 |