CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Solve for two or more "Temperatures" (http://www.cfd-online.com/Forums/cfx/26230-solve-two-more-temperatures.html)

Rui August 7, 2008 19:07

Solve for two or more "Temperatures"
 
Hi,

I want to simulate a transient flow with heat transfer. I want to solve for different energy boundary conditions, but, as the geometry and momentum boundary conditions are always the same, the flow will also be the same (I assume the viscosity is independent of the temperature). If this was a steady state simulation, I think I could use an Initial Values File with the flow results and just solve for the energy equation with the different energy boundary conditions. But I can't do this for transient simulations. So, I thought if it would be possible to set some additional variables (T1, T2, etc) and transport equations and solve (just once) for the flow and for these variables as if they were different temperatures, with different boundary conditions. The boundary conditions for the energy would be given temperatures at the inlet and given temperatures or heat flow at the walls. I think, at least theoretically, this could also be done with additional variables. But then I'm not so sure CFX deals numerically with the temperature as it deals with an additional variable ... actually I think it doesn't, I think CFX solves for the enthalpy and not for the temperature (am I right?), but as I'm dealing with an incompressible flow, I just want to solve the Thermal Energy equation and all the properties (Cp, k, ro, etc) are constant ... it should be the same ... or not? What's your opinion about this? Do you think it could be done? I must add that I'm considering this option because just the flow simulation takes some days, and it would be a waste of time to be obtaining the same flow results.

Thanks

Rui

(P.S.: Good to be posting here after a long time away from CFX)


Rogerio Fernandes Brito August 8, 2008 07:53

Re: Solve for two or more "Temperatures"
 
Dear Mr. Ruy,

what is the geometry of study?

Rui August 8, 2008 08:51

Re: Solve for two or more "Temperatures"
 
It's a geometric geometry :) It has an Inlet, an Outlet, a Symmetry plane and Walls. But what I've asked can be applied regardless of the specific geometry, can't it?

Rui

Rogerio Fernandes Brito August 8, 2008 09:54

Re: Solve for two or more "Temperatures"
 
I will think about your geometric geometry.

Kind,

Rui August 8, 2008 11:34

Re: Solve for two or more "Temperatures"
 
I'm geometrically thankful

What I asked was if CFX would produce the same results solving for an additional variable with a transport equation and solving for the temperature, if the same initial and boundary conditions were applied (to the additional variable and to the temperature), and so if I could solve for an additional variable pretending it was a different temperature. And this has nothing to do with the geometry. It would, or it wouldn't, work whether the geometry is a cube, a cylinder, or even the weirdest shape you may have in mind.

Rui

brunoc August 9, 2008 10:16

Re: Solve for two or more "Temperatures"
 
Hi Rui,

I believe you're assuming your flow is already steady state and changes in time and negligible. In this case, you don't need to create an extra variable. You can just choose to not solve the momentum/turbulence equations using expert parameters. This works for transient analysis as well.

You can even use a steady state result as an input for your momentum/turbulence variables, but, again, solve just your energy equation.


Rogerio Fernandes Brito August 9, 2008 10:39

Re: Solve for two or more "Temperatures"
 
Hi Sir brunoc,

How to choose to not solve the momentum/turbulence equations using expert parameters? I know the path, but which is the parameter? f or t?

Thanks.

brunoc August 9, 2008 11:05

Re: Solve for two or more "Temperatures"
 
Rogerio, here they are:

EXPERT PARAMETERS: solve fluids = f solve turbulence = f END

And it is 'f' for FALSE.

CycLone August 9, 2008 15:31

Re: Solve for two or more "Temperatures"
 
Hi Rui,

If you fluid properties are not dependent on temperature and your solving for "Thermal Energy", CFX does indeed solve for Temperature and the formulation is identical to that of a scalar variable solved using the "Transport Equation" option. You can check the theory documentation to verify this for yourself.

To do what you want, just set up an AV for each temperature field and specify the diffusivity of the AV accordingly.

Note that if the AV solution is linear (i.e. the diffusivity of the AV and the flow solution are independent of the AV value), you can solve 1 AV equation for each boundary and create a new solution by taking a linear combination of the AV's. This may save you a lot of time, depending on what it is you are trying to do.

-CycLone

Rui August 9, 2008 16:33

Re: Solve for two or more "Temperatures"
 
Thanks for your reply, brunoc

But the flow itself is transient.

Rui August 9, 2008 19:58

Re: Solve for two or more "Temperatures"
 
Hi CycLone,

Thanks a lot for your reply. That's what I thought. But I still wonder if CFX deals numerically with the Temperature in the same way as it deals with an AV. I have to take a closer look at the documentation ... but not everything is there. I think the best I can do is to run a simulation solving for the flow, for the Temperature and for an AV (with the same boundary conditions as imposed for the energy equation) and check if the Temperature and the AV solutions are the same.

About your point of linear combination of the AVs, this sounds quite interesting. Yeah, the thermal conductivity (AV diffusivity) is constant and the flow solution is independent of the temperature (AV value). But I didn't really understand what you meant by solving one AV equation for each boundary, and create a new solution by linear combination. If I have an Inlet, where I want to tests 6 different temperatures (Tin1 ... Tin6), and two walls, WallA and WallB, with 6 different temperatures each (TwA1 ... TwA6 and TwB1 ... TwB6), this would give 6^3 = 216 combinations and would require this number of simulations. Is this what you mean?: The fluid temperature at any point at any instant will be a linear combination of the temperature of the three boundary conditions (T=a+b*Tin+c*TwA+d*TwB)? And I just have to get 4 solutions, performing only 4 simulations, and find the values of a, b, c and d for each point and instant?

But it won't work on my problem, because there is an exothermic reaction happening in my fluid and the heat released isn't linear dependent of the temperature. But your point is really interesting.<hr> If you wanna know what I'm trying to do, here is a small description: I'm trying to do simulations of mould filling (the geometry, for now, is always the same) with a reactive resin. The resin's viscosity, in reality, varies with the Temperature and with the Degree of Cure of the resin (which I have to model as an AV). But I don't have information about how this variation is, so I have to assume it's small and impose a constant viscosity. Because the density is also constant, the flow solution won't depend of the temperature. Then I want to test different temperatures at the inlet and different temperatures or heat flows at the walls. But I also want to use resins made of different proportions of the 2 main components (which results in different rates of cure reaction, and different rates of heat release by the reaction). And also test two different reaction models. So, there's a lot of combinations to test. But as I'm gonna assume the viscosity and density, the geometry, and the momentum boundary conditions are always constant, the flow solution will also be always the same. I'll just get different Temperature and Degree of Cure solutions.

Sorry for the long post

Rui


CycLone August 11, 2008 09:58

Re: Solve for two or more "Temperatures"
 
Hi Rui,

"If I have an Inlet, where I want to tests 6 different temperatures (Tin1 ... Tin6), and two walls, WallA and WallB, with 6 different temperatures each (TwA1 ... TwA6 and TwB1 ... TwB6), this would give 6^3 = 216 combinations and would require this number of simulations. Is this what you mean?: The fluid temperature at any point at any instant will be a linear combination of the temperature of the three boundary conditions (T=a+b*Tin+c*TwA+d*TwB)? And I just have to get 4 solutions, performing only 4 simulations, and find the values of a, b, c and d for each point and instant?"

This is exactly what I mean. While the fluid flow may be non-linear, the transport of your AV, or energy in this case, is linear and therefore you don't need to run every case.

However, I don't think problem you have described is linear due to the reaction. The source term from the reaction adds a dependancy on the other components.

As for the numerical treatment of temperature or an AV, they are identical, so long as you are solving the Thermal Energy equation and not Total Energy.

-CycLone

HekLeR September 9, 2008 21:58

Re: Solve for two or more "Temperatures"
 
If specific heat and density are constant then the answer is yes.


All times are GMT -4. The time now is 18:35.