Heat transfer coefficient

[wjpowers]

I am modeling a 3D building. it is embedded in a cube of air on the outside and sitting atop a layer of soil.

I need to be able to model the thermal properties of the walls, roof, floor, and windows.

My choices are to include in the mesh the associated thicknesses and associate the heat conductivity for each body;
or to try to employ a heat transfer coefficient.

It would be much easier to employ the heat transfer coefficient.
If I do use the heat transfer coefficient on the boundary of these interior surfaces, I believe I have to also include a value for the external temperature.
However, the temperature outside these surfaces are generally unknown.
For example, the air temperature and soil temperatures just on the other side of the various surfaces are not known.

In this case, what ought I to do?

Thanks,

Bill Powers
White, SD USA

[petroo]

Hi Bill,

(I seem to remember that I gave an answer to a similar question / problem case not long ago, but can't find it in my own posts list, so alas ...)

Not trying to draw you away from Elmer but just to point to a potential alternative: For me, the problem that you describe seems better fitted for other simulation systems than Elmer. A complete house as a simulation object has other needs than a clear-cut and somewhat simple physical situation with rather restricted structural complexity and interaction schemas. On the long run you would have to take into account such things as moisture effects, localized sun irradtion effects, air intrusion by wind pressure, heating schemas, effects of room temperature dependet venting and much more. I do not see how you would introduce them into an Elmer simulation without exaggerated efforts yet not necessarily meeting reality.

For such a task I'd suggest taking a look at the EnergPlus system of the US Department of Energy (http://apps1.eere.energy.gov/buildings/energyplus/). It is a free-of-charge program system like Elmer and directly designed to model buildings dynamically with all their peculiarities and practical catches and snags. It is by no means as detailed as Elmer can be, but, imho, takes the right simulative shortcuts to efficiently treat buildings in a practical, yet very detailed manner.

Regards,

Peter

[wjpowers]

Peter:

It was to a question I put to you back in August, I think, that you are referring to.

At that time you said that I could have an internal boundary condition using the heat transfer coefficient.

My problem is, however, what to use for the external temperature input.
It appears to be required for using the heat transfer coefficient. But, as I indicated, one doesn't know this temperature since it is a problem variable.

I looked somewhat at the building code that you suggest. It is similar to a code that I had written using ASHRAE manual (and others) when I first began looking at this problem. My code was a simple one and undoubtedly much simpler than the DOE code. But I was getting unrealistic results for the losses to ground. Indeed, if the deep temperature of the ground is warmer than the internal temperatures of the building (actually it's a winter greenhouse), then the earth would be supplying heat to the building, which is not likely. So I knew I had to do a better job modeling the ground-building interface.

It is true that all the effects required for a full simulation would be difficult using Elmer. I'm still uncertain, e.g., how well Elmer will do on forced and natural convection losses, which can be important. My initial interest in using Elmer was to test the efficacy of surrounding the perimeter of the greenhouse with an insulating material down to 4 feet in the ground. I think that this kind of question is going to be difficult to answer with the DOE code, but I could be wrong.

So, I would still be interested in knowing how I deal with the external temperature issue for the heat transfer coefficient. But I will look more deeply into using the DOE code since it will likely in the long run serve as a better model for the overall greenhouse problem. I just hope that it can provide me with each heat loss component.

Thanks,

bill powers
White, SD USA

[raback]

Hi Bill

Certainly I know less of this application field than the other Peter but still some points:

1) You could omit the need for heat transfer coefficients by taking into account few meters of the ground as well. You could use larger and larger elements the deeper you go so the overall added cost might not be that large. And I guess you don't have freezing and snow as we do here in Finland which complicated things quite a bit... Particularly, if you want to study phenomena that break the 1D temperature profile under the building then I would say FEM is your tool.

2) The convection introduces another time-scale in seconds which is poorly compatible with studies over months. You should the use RANS turbulence models, but I would think that here reduced dimensional models would be even more attractive.

-Peter

[petroo]

Hello Bill,

Indeed, if the deep temperature of the ground is warmer than the internal temperatures of the building (actually it's a winter greenhouse), then the earth would be supplying heat to the building, which is not likely. So I knew I had to do a better job modeling the ground-building interface.

Not having anticipated this type of "building" I now agree that Elmer might be the better choice for simulating it: EnergyPlus is definitely more targeting the simulation of complicated buildings with their many zone-wise differentiated interactions that human inhabitants require and initiate. Compared to that a greenhouse is somewhat simpler, yet more demanding if you need to take into account the far-field effects like transient temperature profiles in the ground and such.

... interest in using Elmer was to test the efficacy of surrounding the perimeter of the greenhouse with an insulating material down to 4 feet in the ground.

Just a question from a practical physicist (the one with the shovel in his hand) on this behalf: Why bothering to dig down the 4 ft? Why not lay out the insulation on the surface (or just some cm deeper) which should have the same effect and even create a somewhat larger warmth reservoir for the greenhouse floor to suck heat from?

But I will look more deeply into using the DOE code since it will likely in the long run serve as a better model for the overall greenhouse problem.

From my somewhat limited knowledge of the internal structures of the codes I would decline this approach now, as your greenhouse is more like a simple volume with border elements quite fit for Elmer treatment. I would regard EnergyPlus as oriented towards a different order of interior interaction complexities that requires a somewhat coarser look on the tinier details in order not to get lost on the myriads of individual differential interactions.

Btw., as I intend to investigate long term heat deposition and recovery potentials by deep bores in a more hobby-like fashion I'd be interested in some executive summary of your work once you arrive at some greenhouse heating conclusions!

Regards,

Peter

[wjpowers]

Peter:

I'm happy to see that you think Elmer might be a better fit for the project, although I have downloaded and installed EnergyPlus. What is neat about it is that there is a possibility to use Google Sketchup to create the building and surface properties. I have not used this code but my 13 year old son has created some incredible (and fantastic -- as in fantasy) complexes with it. Unfortunately, Goggle Sketchup is not available in Linux (lest you used wine, which I have not had a great deal of success with).

First, regarding your suggestion of simply covering the top surface of soil surrounding the greenhouse, I think is an intriguing idea.
From the little that I have been able to glean from the literature, it seems that the most important part of the ground-building interaction is very near the surface. This is, in part, what an Elmer simulation could confirm. What I think an Elmer simulation might tell us is how deeply any insulation needs to be. Just laying down insulation on the ground will help keep the earth beneath the insulation warmer, but will do little to insulate it from the colder surrounding earth, whereas, burying the insulation will not keep the surrounding earth warmer, but will keep the earth beneath the greenhouse warmer.

From a practical point of view (always important for one strapped by finances, time, and manpower), laying or burying in shallow ground the insulation on top of the ground is far easier, although with machinery traffic it would likely break up over time. So I like the idea. Most greenhouses could probably avoid heavy machinery traffic in the area immediately surrounding the greenhouse. This would not be the case for many high tunnels, where they may use moderately heavy machinery to till the soil prior to planting.

So I'm glad that you make this suggestion because it is one that Elmer can be used to examine and one that I have not heard previously addressed. I have actually constructed a small winter greenhouse about 4 years ago. I received a small US Federal grant to make in situ measurements of the greenhouse and attempt to develop models to describe their behavior. I am about to begin my second year of measurements. As a physicist, however, I am never happy with just measurements. That's what led me to Elmer.

I am, however, still waiting (with patient hope) for the answer to my question. If I use a heat transfer coefficient, the heat flow at the boundary is modeled linearly with the difference between the "external temperature" and the temperature on the interior side of the boundary. Elmer requires, as far as I know, for the user to supply the "external temperature." But when the boundary is an interior boundary, the "external temperature" is a variable. What, then, should I do? That's the question.

Finally, regarding your "hobby" research, by a deep bore I assume you mean a deep hole, perhaps like that produced for oil excavation, although I don't imagine you have much of that in Finland. What kind of heat deposition do you have in mind? Is this something more along the lines of geothermal applications?

Thanks for your assistance,

bill powers
White, SD USA

[raback]

Hi Bill

There are two Peters here. I'm in Finland and the one with the greenhouse is in Germany.

But to get back to your question: you should really model some depth of the the ground as well. There is no other way of knowing the temperature distribution in the ground. You could do this analytically in 1D assuming sinusoidal temperature, for example. As a result you would get some kind of impedance condition for temperature. However, I see no reason for this approach when working with fem.

-Peter

[wjpowers]

Peter-in-Finland:

I agree and have always intended to model some depth of the soil.
My intention was to assign a deep temperature earth temperature, say at 3 meters, and have the code determine the vertical temperature distribution.

My reason for wanting to use the heat transfer coefficient can be aptly demonstrated here.
Suppose the greenhouse sits on a cement slab.
There are two approaches.
I could zone up the cement and indicate the physical properties of the cement.
But I was thinking that a far simpler way of doing this is to describe the interface between the soil body and the floor of the greenhouse was to use a heat transfer coefficient at this interface.

I understand others to be saying that this is kind of internal boundary is permitted in Elmer.
The problem I have is that the heat flux at the interface is proportional to the difference between the "external temperature" and the "internal temperature."
As I understand it, Elmer requires that one indicate what the value of the "external temperature" is.

But if I model the soil by a slab of soil with a known temperature at "infinite" depth, I would not a priori know the "external temperature."

So what do I do? That's the question.

Thanks,

Bill Powers
white, sd usa

[Juha]

Hi,

i'm maybe off the track here, but if i understood correctly, you are really asking how to model the walls (and
ceiling and floor) without meshing those separately, only the boundary elements between inside and outside
air or whatever ? If so, i think you need to break the direct diffusion in the mesh by doubling the boundaries,
then you can define interaction with the two surfaces with heat transfer coefficients where the 'external'
temperature is taken from the from the 'inside' for the 'outside' part of the mesh and vice versa. Have a look
at the fem/tests/HeatGap test case:

http://elmerfem.svn.sourceforge.net/vie ... ts/HeatGap

Regards, Juha

[wjpowers]

Juha:

I've looked at this HeatGap problem and I don't quite see what is going on.
I ran ElmerGrid on the grid file and looked at the mesh.
It almost looks like every cell is both a boundary of type 1 and 3.
The exterior appears to be boundary 1.

Can you say something more about this mesh?

Thanks,

bill