[jpina]

Dear all,

I’m carrying out a simulation with a simulation very alike to the following one:



The inlet is a pressure inlet. The red zone is initialy patched with a polymer at 523K and the blue zone is initially air.

The point is to see how the polymer enters to the cavity and “freezes” due to the heat transfer with the walls.

In the paper I am using:

http://www.google.es/url?sa=t&rct=j&...HbYVnchyWdRUFQ

They say:

“A heat transfer coefficient is defined to be h=5000W/m^2/K at the wall”.

And they set the wall temperature at different values.

My doubt is:

- Is it possible in ANSYS Fluent to define a heat transfer coefficient and a temperature of a wall? As far as I have seen, the “Fixed temperature” doesn’t allow me to set the h value. And “convective temperature” uses a free stream temperature and a h with the exterior, which is not what I want!

Moreover:

- For “copying” the paper simulation, should I make solid zones in ANSYS Design Modeler for the walls, mesh them and carry out a conjugate heat transfer? I have tried it without much success.
- Or should I use “Named Selection” for calling Wall1 and Wall2 the two edges?

I will appreciate your answer, I really need to simulate this for going further in my studies…

Thanks a lot.

[vinerm]

A convection coefficient is meaningless without a reference temperature, in most cases free stream being reference. At the inside, CFD calculates h, hence, condition is applied on the outside of a wall. A zero thickness wall only implies infinite conductivity, however, it is still a wall with inside and outside. So, if h is for outside, temperature has to be for free stream outside. If I use wall temperature, then to calculate heat flux, I'd require

[jpina]

Dear vinerm,

your message seems to be unfinished

The paper I linked says to use a h=5000W/m^2/K, which is the h between the wall and the fluid. How could I set this?

[vinerm]

I'd require temperature of external stream, which is unavailable. Therefore, with h it is always external free stream temperature that is applied and wall temperature is calculated to make sure heat energy is balanced across wall

[vinerm]

This has to be between wall and external fluid. If I already know h between wall and fluid for which I'm solving my car, then, I don't need to run a case as far as thermal energy is concerned.

[jpina]

Maybe the paper doesn't state it, but they suppose that the external temperature is around 298K?

However, if i set the convective BC I will have h with the external zone and external temperature.

Where would the h of fluid/solid of 5000 W/m^2/K and wall temperature be?

[vinerm]

5000 is external h.

[jpina]

First of all, thanks for your replies.

So, I understand that I should "Name" the two edges as wall. Use "convective heat transfer BC", set the h=5000W/m^2/K and the free stream temperature would be the wall temperature or the "ambient" temperature?

My concern is that, by using this, we are not setting the wall temperature to a fixed value like the paper. Moreover, the 5000W/m^2/K I understand in the paper that is the h between solid and fluid. Not between solid and external "air"...

[vinerm]

Free stream temperature is always fluid temperature and that's what every CFD tool requires. If they set wall temperature, then that would be different simulation and not same as convection condition. Possibly, they did multiple simulations and in one particular case, they used fixed temperature

[jpina]

Originally Posted by vinerm

This has to be between wall and external fluid. If I already know h between wall and fluid for which I'm solving my car, then, I don't need to run a case as far as thermal energy is concerned.

In the paper they say that the choose of the h=5000W/m^2/K is based in experimental results. So I understand that this is the h between wall and fluid. The interest is to see when the fluid (a polymer) freezes.

[vinerm]

Fluent or CFX or any other tool won't allow user to apply h on internal side. All that can be done is - use h value to apply heat flux. How are they modeling freezing? Did they use enthalpy-porosity method or some other way?

[jpina]

Originally Posted by vinerm

Free stream temperature is always fluid temperature and that's what every CFD tool requires. If they set wall temperature, then that would be different simulation and not same as convection condition. Possibly, they did multiple simulations and in one particular case, they used fixed temperature

You mean that "free stream temperature" is the fluid temperature of the polymer? Or of the "air" outside?

[vinerm]

It is always of the fluid that is NOT being modeled because for fluid being modeled we expect thermal field to be predicted by tool

[jpina]

I'm reading another paper of the same author and they say:

"At the wall,

a constant heat transfer coefficient is used for the thermal boundary condition. The value of the coefficient was chosen so that simulated replications matched the experimental values using reverse engineering. This procedure was chosen due to lack of experimental data for the microscale heat transfer coefficient."

"The magnitude of this effect depends on how fast heat is transferred from the hot polymer to the mold wall. The most common way to describe this heat transfer is to use a heat-transfer coefficient, saying that the heat flow is proportional to the temperature difference between the polymer and the wall."

So, I understand it different from you. 5000 W/m^2/K is the h between solid and simulated fluid (the polymer). But Fluent convective condition doesn't allow to fix the h, instead it calculates it itself...

I want to set the h between fluid and solid and fix the wall temperature.

[vinerm]

Fluent will not allow that directly. However, you can modify it using UDF written in either C or Scheme. However, a sanity check is required. If we fix temperature of wall as well as h, essentially, I'm defining profile of fluid flow which may not be in sync with conservation equation. A simpler method could be to define constant temperature of wall and try different Re number flows of polymer. h is function of only Re and Pr. So, by modifying these two, you can obtain required h

[jpina]

Dear vinerm,

and how could I modify Re?

Viscosity, density and velocity are not variables, they are what I want to simulate.

Related to Pr number, Cp, viscosity and conductivity are also what I want, I don't think I can change this values for obtaining a given h... Well, I could modify the conductivity in order to obtain a given h, but this makes me doubt:

Is the same to set a given h than to modify the conductivity of the polymer for obtaining a given h?

[vinerm]

So, you've got to right point. If you specify h, essentially, you're not using the properties you specify in material panel because h is function of those. It is possible that in the paper, they apply h and some only thermal energy equation to model solidification. Let me go through paper in detail

[jpina]

Thanks a lot Vinerm, it is really important for me. The PhD I'm starting is related to this and I should be able to at least replicate this paper before introducing what is different in my situation.

I'm also open to talk to you via Skype if you think it is necessary and then report the obtained results in this thread for future doubts.

[jpina]

In this link there's the thesis:

https://www.google.es/url?url=https:...1mZRAHeYmkWy-w

If you go to pág. 72 (paper C, 3.3.2. Boundary conditions) you'll see:



In this paper, however, the T of the wall is considered variable. For me it would be enough to consider it fixed, like the first paper I linked in this post. (They are from the same author).

[vinerm]

This is how convection coefficient condition is implemented in CFD tools. There is nothing wrong with this. However, they are using a different coupling described in next section and that coupling can be implemented in Fluent using C UDF.