help: building application in fluent!

[quiri]

Hello everyone,

I am new to the forum and only for a couple of weeks I have started to use the fluent.
I have to study the propagation of air flows within a historical building, which happen by temperature difference (and therefore pressure) between two points of the building itself.
To this end I have built a 3d model that includes my building, and on the outside I have placed an air box that intersects the ground and that would like to represent the atmosphere around the building (and I will take care to make sure that it is big enough because the walls of that box do not affect the motion of the fluid).

I wanted to ask you what is the best configuration for the boundary conditions in this case?
in particular the walls of the above box, I must place them as "wall" and give a temperature equal to the atmospheric one, or is there any option that can better simulate the fact that such surfaces should in fact be transparent to the flow (in or out) of air?

Thank you for your courtesy.

ps: Incidentally, I'm making mesh with icem cdf (which I think I will be able to use, helping me with the tutorials present in the software) and I thought I would make a tetrahedral mesh (it seems to me to understand that it's the easiest thing).

 

[sail]

Hello everyone,

I am new to the forum and only for a couple of weeks I have started to use the fluent.
I have to study the propagation of air flows within a historical building, which happen by temperature difference (and therefore pressure) between two points of the building itself.
To this end I have built a 3d model that includes my building, and on the outside I have placed an air box that intersects the ground and that would like to represent the atmosphere around the building (and I will take care to make sure that it is big enough because the walls of that box do not affect the motion of the fluid).

I wanted to ask you what is the best configuration for the boundary conditions in this case?
in particular the walls of the above box, I must place them as "wall" and give a temperature equal to the atmospheric one, or is there any option that can better simulate the fact that such surfaces should in fact be transparent to the flow (in or out) of air?

Thank you for your courtesy.

ps: Incidentally, I'm making mesh with icem cdf (which I think I will be able to use, helping me with the tutorials present in the software) and I thought I would make a tetrahedral mesh (it seems to me to understand that it's the easiest thing).

mmm. If you just need to know what happens to windows and closed doors maybe it is better to simulate only the interior of the building.

However wall as a boundary condition, from how I understood you structured your domain is not correct. symmetry would probably be more appropriate.
Maybe post some clarifying image if you can.

 

[quiri]

Unfortunately my building, in its real configuration, does not have doors and windows, so I think it is inexpensive to consider what happens in relation to the outside air.

As you recommended, I attach images that I hope will clarify the problem.
http://dl.dropbox.com/u/49599479/zisa/elaborati.jpghttp://dl.dropbox.com/u/49599479/zisa/imm001.jpghttp://dl.dropbox.com/u/49599479/zisa/imm002.jpgI do not know if it is correct and I have to check the effects, but a colleague with some extra experience in fluent, told me that I could simulate the situation in two ways:

1 - applying the outflow condition to the control volume walls, which should behave as a free wall

2 - applying the wall condition to such walls, but taking care to build such a very large volume of control, so that it does not affect with what happens around the building.

According to him the first condition is more expensive from the computational point of view, while the second is to be preferred from this point of view, provided, in the areas distant from the building, a very wide mesh.


Here we come to the second problem, that of mesh (I hope this is not the wrong section of the forum to talk about it).
I have abandoned the icem in favor of gambit as it is said to be more compatible with fluent. Maybe it is not so, but in any case now I have acquired a certain mastery with the latter, so I will continue with this.

the problem is what kind of mesh do for my problem. I read something about the Fluent Guide. I predict that, compared to the geometry that I present in the attached images, I adopted a simplified model in which all the curves of the vaults and arches are rectified (and then maybe reintroduced them later).

to what I have understood, in general a regular hexedral mesh is to be preferred compared to a tetrahedra because it involves a lower computational expense for subsequent processing in fluent. Moreover it seems to me that well fits a geometry formed totally by parallelepipeds with angles of 90° like mine. the only problem that would seem to have this type of mesh is that of the difficulty of construction for complex geometries, but after so much sweat I managed to find a simple way to meshare my volume with a very regular mesh.

However, I also read that a structured hexaedral mesh is to be preferred in case it is built in order to somehow indulge the expected flow (so for example in a pipeline with pressure fluid), while it is better an unstructured tetrahedral mesh for more complex motions as it is probably what develops within my model.
as I have already said, in fact, the motion in my building must happen for temperature difference, so they will mainly generate convective motions in which the turbulent component will not be negligible at all.

therefore the doubt remains: hexedric or tetrahedra?



I hope I wasn't too long, I just wanted to explain my problem well. apologize for all the likely errors in the message; I still have confused and imprecise ideas about how many things work in the world of cfd, so please correct any conceptual errors.

Thanks again for your availability, I hope to have news as soon as possible!

 

[sail]

Unfortunately my building, in its real configuration, does not have doors and windows, so I think it is inexpensive to consider what happens in relation to the outside air.

As you recommended, I attach images that I hope will clarify the problem.
http://dl.dropbox.com/u/49599479/zisa/elaborati.jpghttp://dl.dropbox.com/u/49599479/zisa/imm001.jpghttp://dl.dropbox.com/u/49599479/zisa/imm002.jpgI do not know if it is correct and I have to check the effects, but a colleague with some extra experience in fluent, told me that I could simulate the situation in two ways:

1 - applying the outflow condition to the control volume walls, which should behave as a free wall

2 - applying the wall condition to such walls, but taking care to build such a very large volume of control, so that it does not affect with what happens around the building.

According to him the first condition is more expensive from the computational point of view, while the second is to be preferred from this point of view, provided, in the areas distant from the building, a very wide mesh.


Here we come to the second problem, that of mesh (I hope this is not the wrong section of the forum to talk about it).
I have abandoned the icem in favor of gambit as it is said to be more compatible with fluent. Maybe it is not so, but in any case now I have acquired a certain mastery with the latter, so I will continue with this.

the problem is what kind of mesh do for my problem. I read something about the Fluent Guide. I predict that, compared to the geometry that I present in the attached images, I adopted a simplified model in which all the curves of the vaults and arches are rectified (and then maybe reintroduced them later).

to what I have understood, in general a regular hexedral mesh is to be preferred compared to a tetrahedra because it involves a lower computational expense for subsequent processing in fluent. Moreover it seems to me that well fits a geometry formed totally by parallelepipeds with angles of 90° like mine. the only problem that would seem to have this type of mesh is that of the difficulty of construction for complex geometries, but after so much sweat I managed to find a simple way to meshare my volume with a very regular mesh.

However, I also read that a structured hexaedral mesh is to be preferred in case it is built in order to somehow indulge the expected flow (so for example in a pipeline with pressure fluid), while it is better an unstructured tetrahedral mesh for more complex motions as it is probably what develops within my model.
as I have already said, in fact, the motion in my building must happen for temperature difference, so they will mainly generate convective motions in which the turbulent component will not be negligible at all.

therefore the doubt remains: hexedric or tetrahedra?



I hope I wasn't too long, I just wanted to explain my problem well. apologize for all the likely errors in the message; I still have confused and imprecise ideas about how many things work in the world of cfd, so please correct any conceptual errors.

Thanks again for your availability, I hope to have news as soon as possible!

I gave an eye to the pictures you posted.

we start from the type of meshatura: The hexaedrical cells intrinsically give a greater order of accuracy of the solution even if they are not aligned to the flow (as long as they maintain reasonable aspect ratio) at the same number of cells. against tetraherical cells are easier to use.

the images you posted reveal a great complexity of the object from meshare, and combined with a phenomenon not just banal to model you immediately tell you that to have something reasonably accurate you will fall into values of cells between 6 and 20 million per eye. so make sure you have computational power accordingly.

for sw both gambit and icem are perfectly compatible with fluent. the only thing that changes is the type of approach considered: gambit uses philosophy bottom up i.e. you create lines first, then faces, and finally volumes. with icem parts from the volume, split it into blocks and apply them to your surfaces.

personally hate gambit, but it is a matter of personal taste, I saw mesh excellent created with this sw.

as regards bc:

I can share the option of putting symmetries, not wall! far away.

all outflow makes me twist my nose a bit because I fear that without an inlet you may have mass preservation problems, and even put a puppet inlet with inlet velocity = 0 is not a good thing.

another option could be, if you can consider your compressible fluid and use a density based solutor, putting all the atmosphere as pressure far field. In this way you would have the advantage of eliminating compressibility errors and imposing a realistic bc. On the other hand, you may have numerical instability, especially in the early stages of simulation, due to very small differences.

Basically, as you see, it's not a very simple problem to solve and I've never banged our heads so much. However it will be interesting to follow your progress here on the forum :biggrin:

Have you tried to read fluent manual carefully? is done well and often contains guidelines for various applications

Happy birthday.