CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS (http://www.cfd-online.com/Forums/ansys/)
-   -   Out of my depth with CFX and ANSYS (http://www.cfd-online.com/Forums/ansys/83549-out-my-depth-cfx-ansys.html)

 tomcatbobby January 3, 2011 09:03

Out of my depth with CFX and ANSYS

Hi

I'm hoping someone will be able to help with a project I am trying to run. I want to model heat transfer in a room. I have my own geometry but I cant get that to work, thats another problem for the moment. Within the room there is essentially a water pipe absorbing/cooling the air.

To begin with, can someone just confirm whether I am on the right lines in my methods to model this. I was thinking to use fluent or cfx to model the fluid flow in the pipe which would give me an effective thermal conductivity (which I could assume constant) for the pipe in the room which I would model using CFX?

I'm pretty new to all this so apologies if this isn't clear. Can answer any questions if it helps. I've done many tutorials on CFX and used FLUENT before but it doesnt seem to help that much when you have a problem of your own!?

Thanks for help

 ComputerGuy January 3, 2011 20:48

Bobby,
If I understand correctly:
1. You have fluid flow (let's say cooling water) within a pipe
2. There is heat transfer from the fluid to the pipe, whose walls have a known thermal conductivity
3. The pipe is exposed to air (or whatever fluid) within a room
4. There is natural or forced convection from the outer wall of the pipe to the air
If that description is correct, then this is no problem at all to model. The few things that make this challenging are:
1. The grid will have to be relatively fine on both the external and internal surface of the pipe. Depending on the size of the room and the pipe, it could be large model (cell count)
2. You'll need fluid properties as a function of temperature for the air in order to see a natural convection current set up
3. You'll need to define the boundary conditions of both the pipe, as well as the boundaries on the room -- these include temperature, pressure, or flow conditions.
Let us know if we can help more. I can't speak for CFX, but I'm certain that Fluent can do this.

 tomcatbobby January 4, 2011 06:17

Quote:
 Originally Posted by ComputerGuy (Post 289140) The grid will have to be relatively fine on both the external and internal surface of the pipe. Depending on the size of the room and the pipe, it could be large model (cell count) You'll need fluid properties as a function of temperature for the air in order to see a natural convection current set up Let us know if we can help more. I can't speak for CFX, but I'm certain that Fluent can do this.
Thanks for getting back to me. I think I will have some more questions. I am currently learning/trying to model natural air convection within a room and a heater.

1-it is a model 1646x1646x1646 (mm)
2-can I not enable buoyancy within the model and apply -g in the z direction. I did a tutorial using this method. Where would I obtain such data?

Slight adaption to the problem/another scenario. Modelling just a room with a heater in it that has a given heat flux. I think I have managed to set this up in CFX. Chosen a domain with air in it, set an initial T for the air. If the walls are adiabatic, and I run it for long enough, it should just get hotter and hotter right? My issues is what convergence criteria/monitors would I set for a problem like this? How long would it need to be run for?

Thanks

Thanks

 ComputerGuy January 4, 2011 22:09

If you're modeling natural convection, turn on the Boussinesq model for more rapid convergence (than with density as f(T,P)). See the following:

Steps in Solving Buoyancy-Driven Flow Problems

It'll only be valid for relatively small changes in temperature, however (i.e., thermal expansion coefficient * dT << 1).

For the case you describe, with adiabatic walls and a constant internal heat source, the room will continue to get warmer and warmer:

in - out + generation = accumulation
0 - 0 + heat flux = temperature rise

Boussinesq will break down (be invalid) over very large temperature swings, but you can always get the density of air as a function of temperature and put it in the materials panel. It'll work the same. See the link for for detail.

What you're describing is inherently unsteady (the temperature never plateaus), and thus you'll have to run for an infinite period of time. If you're only interested in how the room warms and the velocity of the air within the room changes, choose a time step described in the link above (see equation 13.2-22) and run the simulation in transient mode. Stop whenever you get bored!

If you have some heat loss through the walls, or flow out of the domain such that you might achieve a steady state, simply monitor the total heat flux out of the domain and the heat flux out of your heater. The sum will be zero at steady state. The standard convergence criteria will do. The Fluent manual suggests solving a "steady state" solution as a large number of transient steps; in the link provided >5000 time steps is suggested.

ComputerGuy

 tomcatbobby January 10, 2011 16:11

Hi

Got simulation working in CFX which is good. The choice of an appropriate timestep confused me as I dont know the time constant and have never quite understood length scales?

The simulation is not right though as within the setup I cannot deselect the heater 'face' as a wall. This is a problem as it includes it in the boundary condition for adiabatic wall? Any ideas?

Also, when in the CFX post, I would like to view a video of the velocity vectors coloured by temperature to observe the natural convection in the room over time but Im not able to select this?

I'll post the last two questions in the CFX forum as well as you said you can't speak for it as much as FLUENT.

Thanks for you help

 All times are GMT -4. The time now is 04:06.