[Lionel Trébuchon]

Hello!

I read through your thread and was very glad for the good questions asked and the nice answers. I am a student in 3rd year and new to CFX and simulating.

REAL PROBLEM
I have to model a cavity, which is a short cylinder with a halfsphere to close one end, and a circle face on the other end, with a little opening. The cavity is supposed to behave like a black body and absorbes all light that comes in by the little entry. One can compute the incoming light with monte-carlo ray tracing methods, but this is not my question.
To the opening is attached a Compound parabolic concentrator (CPC), which is used to focus the light which is reflected towards it by a solar reflector park .

The whole construction is thus on a tower, light comes in from "below".


GEOMETRICAL AND BOUNDARY MODELLING
Out of preliminary calculations of reynolds number, the flow is laminar, but some results I had could point out turbulence, however for now: laminar.
I don't model no Solid for now, only one big liquid domain that is the air inside the construction, to which surfaces I apply the boundary conditions.
To simplify, the concentrator is an adiabatic wall which has one opening into the outside air (modelled as a big parallelepiped).

The cavity is known to operate at 1200 [K], temperature reached thanks to the incoming light flux, so i model it as a 1200K wall. Some parts of the wall are adiabatic, others slightly cooled, but this is not my concern here.

The whole construction tilts down with an angle of 30° from the horizontal. (modelled by putting an angle of 30° on the gravity vector) Buyoancy is thus activated and an important factor in the modelisation, as it could drive heated air up into the CPC and even out into the "wild".



CALCULATING THE HEAT LOSS
My problem is very simple and I am sorry for that.

I would like to calculate the heat loss provoked by the incoming air that heats up because of the hot walls in the cavity, then flows out due to buoyancy and other factors such as wind (not there yet), and the air in general that is heated up, because all the air that is heated is lost for the purpose of this construction.

So I would like to monitor a heat loss.



QUESTION:

I wonder how to calculate the lost heat by air that flows out in front of the CPC opening.
To simplify, one could say that i just want to calculate the heat loss due to out-flowing air of a cylinder, open on one side, wall at 1200 K on the other side, with a certain tilt angle towards the horizontal.

*** How can one calculate the heat loss due to outflowing air? ***

In front of this opening I added a big parallelepiped of air, so that my simulation is more exact when assuming the boundary conditions air temperature (in this case ambient temp, 293 [K])

I cannot set the walls of that cylinder as an opening (which they technically are) if I wanted to use "wall heat flux", CFX doesn't seem to accept that.
=> So I thought to set them as wall with a very high heat transfer coefficient. Maybe my understanding of thermodynamics is too bad here, but is it right just to set a veeeery high heat transfer coefficient (which represents watts transmitted per surface area per temperature difference)? And let CFX realise that "Oh, as soon as air hotter than 293 [K] touches this wall, heat diffuses throught it and thus I can use my "wall heat flux" function?

This cannot be really right, as the heat transfer coefficient exactly is defined by how this temperature diffuses:
HTC = areaInt(wall heat flux)@<interface> / (area()@<interface>*(areaAve(T)@<interface>-tbulk)

=> Could one set the opening faces of the parallelepiped as "opening" and calculate the heat loss by making an analysis of the volumetric air flow out of the volume times enthalpy, or temperature etc...?

The problem is I cannot find such variables/expressions/functions...




I know you will be able to answer very fast, sorry for the stupid question!

Friendly greetings,
Lionel Trébuchon

(In amazement for the quality of this forum)

[ghorrocks]

First of all - this post has nothing to do with the thread you posted it on. I have moved it to a new thread. Please start a new thread for new questions.

Can you post an image of what you are modelling? I do not follow your description of what you are modelling.

But it sounds like you want to know the heat loss from the super-hot gizmo in the middle of the thing. It sounds like convection and radiation are the key heat transfer issues here (and maybe conduction, I do not know from the information you have supplied so far).

To model convection you need:
* A gravity vector defined
* A fluid which responds to the gravity vector with temperature. With temperatures over this range you are going to need ideal gas and possibly variable properties/real gas.
* A flow setup where cold flow can come in somewhere and hot flow can go somewhere else.

To model radiation you can:
* If the surroundings are roughly even temperature (hopefully close to ambient) then a simple s*e*(T^4-Ta^4) radiative heat transfer condition will work
* If the surroundings have a range of temepertures, or if emissivity, reflectivity or other complicating factors are present then you will need a radiation model.

[Lionel Trébuchon]

Thanks a lot for your very fast response!

Your questions:
Gravity defined, ideal gas for now, cold as well as hot flow can go in and out of the front of the Compound Parabolic Concentrator (CPC)
No radiation needed.

Here comes an image - sorry.
The image is a CFD-Post image, the model uses vertical symmetry to win time. That's why the cut through the central plane.
On this image you see my modelled air mass. In front of the mushroom like form you see the cavity receiver, wich ideally absorbs all light that comes into it. This phenomenon doesn't concern me for now. The cavity is the one with a assumed steady uniform wall temperature of 1200 K.
Then, "attached" to it, the parabolic compound concentrator, which has been modelled with the perfect length for my purposes. Adiabatic wall for now.
In front of all this, I model some exterior air by a big parallelepiped. One could have used a cylinder, I wanted to use the parallelepiped so as to easilier set wind in case I model wind later, but if you think it would be better to have something else to model the air, I would be glad for any explanation!
This air-modelling block is quite bad, in the next step I will model air all around the construction, but I have my issues with the design modeller, so I keep this for now.

What I would like to approximate is the convective heat loss into the inner of the (heated) cavity (which might or might not flow into the CPC and then out into the surrounding).
So my question is 2 questions:
Main concern: how to calculate the heat lost into the air (this will not be used in the application).
Second question (less important): how would one calculate how much heat flows out of the opening in front of the CPC?

In this model for instance, with the boundary conditions of the block of air set to "wall" with 1 000 000 heat transfer coefficient, I found a wall heat flux of 8 [W] (value was expected little, but maybe not that little). However, what interests me more is the total heat lost into the air.

Next post shows a picture of the computed air flow.

[Lionel Trébuchon]

And here a picture of the flows:

Thank you so much for the help! I've been stuck during a time now and am very very glad that there is this forum!

Lionel

[ghorrocks]

Can you put labels on the image to show what all the regions you are referring to are? It might be obvious to you but it is not obvious to me.

I am also puzzled by your statement that radiation is not important. Why do you say that?

How to calculate the heat lost into the air? areaInt(Wall Heat Flux)@SurfaceName will give the total heat into the air.

Heat through the opening? Put a plane through it in CFD-Post and use the function calculator to integrate the heat flow over that plane.

[Lionel Trébuchon]

Hello Ghorrocks!

Again thanks a lot for the answers! I am really really glad.

1) Yes I know the problem. I will upload it once I can. On the first picture I posted above you can see that the element on the left side of the picture, the "cap" so to say, is at "red" temperature (which corresponds to 1200 K here). Maybe this helps understanding the rest. Except this "red" hot cap, all the other boundaries are adiabatic walls. (on this very picture I modelled it wrong, I set the boundary temperature of those walls to 300 K, but I reviewed my decision).

In the image underneath you see a cut through a similar geometry (not mine). The cavity is hot and loses heat in several fashions. One of them is the convective loss inside ("interior"). As you pointed out we can also analyse radiation losses. Again, I am only interested in the inside of the cavity and especially the convective heat losses.

2) You are right. I thought that radiation wouldn't give too much heat into the air, but apparently yes. How can you activate multiple heat transfer factors in a model? I guess it will use Planck's body law of radiation, which is perfect for me as I assume a black body.

3) Here is all my problem. If I make an areaInt(Wall Heat Flux)@Hot_Cap_1200K, how can cfx tell i want to know the heat flux that is transported by the air immediatly at it's boundary with the heated cap? All I have is 1200 K and air convecting from it. I don't really understand if/how areaInt(Wall Heat Flux) works on this?
I know this is a bad question but I am not sure if it yields the right result... I didn't really find something likely in the tutorials, that is why I ask.
Also, as you just said, this kind of "heat flux" doesn't account for the radiation I guess. But as I would prefer to separate both, it is not bad I think.

[Lionel Trébuchon]

4) Okay thanks a lot for the advice! I have put a plane through a certain location as shown in the image. I have questions however. One of them is: how can you define that your calculations on the plane don't account for all the plane, but only for its cut with the mesh (the mesh is only present in the air volume, the plane I have as an image stretches outside my mesh)?

[Lionel Trébuchon]

5) In the end, I will want to make an analysis of several data points found for the heat transfer with several tilt angles models and also several geometries.
That is why an automated way of finding results is important, and that is why point 3) is important.
As far as I can judge, putting a plane through the body and evaluating heat fluxes there all are part of post processing, and I will want to run a .cfx on a cluster. I haven't really understood if using a workbench with that makes any sense. Therefore I would need a way to have the values I search for in the output file.

But maybe you can give me the very satisfying answer that Wall Heat Flux applied onto the heated cavity walls yields exactly what I am searching for?

Thanks!
Friendly regards, Lionel

[ghorrocks]

From memory there are variables for wall heat flux (which includes all heat transfer mechanisms), and also convective heat flux, radiative heat flux for specific forms of heat. You can work out how much is each type of heat transfer with these variables.

Make the slice a cut rather than sampling and it will be only the actual modelled region.

[Lionel Trébuchon]

Hello!

I will follow your advice once I think that my simulation is actually consistent.
My problem of late is that I am trying to find a way that there are no velocity gradients in that front parallelipipedical "box" you can see on the pictures (always bottom right hand side).
Hello dear Doginal!

I write to you because I saw your replies on some threads that were similar to the problems I encounter

I have troubles conditionning this box, with actually is just a box of aire whose sides are openings. The only reason for its existence is that I could set my boundary conditions with more certainty: I assume that at the sides of this box, air will have a temperature of 293 [K] and a pressure of 1 [atm].
To model the sides, I consequently thought it was right to set as boundary condition "entrainment" and temp to 293, and relative pressure to 1 atm. I chose to "tick" the pressure options, and to specify "static pressure". I guess that thanks to that the program knows that I am deifining the static pressure as 1 atm at this boundary?

However, no matter what solution model I use (k-epsilon, laminar etc..) and no matter what temperature I set for my cap, there is always high velocity gradients on the sides of my front air box.

1) Is my use of entrainment right? Does the "tick" on static pressure make any sense? I saw some threads about the topic and it seems to be just the right thing to do. But then I don't understand those air velocity gradients!

2) Do you maybe know why there always are velocity gradients especially at the boundaries of my air box? Even when I set the cap as standard conditions? Meaning that actually no air should flow at all, as I haven't specified any entring velocity (wind).
And the values of the velocity are quite considerable, in the range of several [m/s]...

3) As you suggested, I am also wondering about the influence of intern heat radiation upon the result. I might set it "on" later, but first I need to be sure that my simulations actually mean something.

I am so glad for your help, you are really very nice and I appreciate that you put some time and effort into this!
Friendly regards,

[Lionel Trébuchon]

Back again!

Another question I came upon is the definition of the density for the air. This however has nothing to do with the setting of the boundary conditions, I must confess.
1) I calculate with an ideal gas model. However, when defining the buoyancy, you are also asked for a ref. buoyancy density. I guess this one is to say which density there would be at standard conditions. So I set 1.2754 [kg/m^3] but I am of course wondering: how does the computer know what standard conditions are? There is no option for reference temperature for instance.
2) Also I wonder if you could tell me what you think of the assumption of ideal gas properties for air? I imagine that at a certain temperature (especially the ones that should be achieved in my model) this assumption becomes quite weak.

Thanks again!
Friendly greetings,
Lionel