[aleisia]

Hi guys,
I'm trying to solve a species transport problem, with the gaseous contaminant not included in fluent material database. The whole model goes like this, you can also find it from the attachments.
Some certain amount of contaminant was released to air, at a given rate, unit being (kg/(m^3s)), the total dosage and release time are known. The we are trying to see if the contaminant can penetrate into a surrounding facility with 6 offices in it. There are doors and windows in each office, I give each door and window a very small uniform crack around it, say, 1mm crack. There is a windmachine at the northwest corner (wind velocity given), and three fans at the ceiling (wind velocity given too), a trailer next to the building was trying to clean the contaminant in the air and supply clean and pure air inside the facility to maintain an overpressure to prevent any penetration of gaseous contaminant.
Here is how I set the boundary conditions:
1. Fan BC.at the inlet of windmachine and fans, interior BC at the outlet of them. Direction of the wind was checked to be sure the wind goes in the direction I want it to.
2. Model the source block as SOLID, and a velocity inlet BC on the six surfaces of the source block (near the windmachine). I'll explain later why I didn't set the block as fluid and a source term.
3. Outflow BC at the trailer's inlet (trailer was modeled as a solid block)
4. Velocity inlet BC at the windpipe end surfaces in different offices, the pipes are set as SOLID.
5. Doors and windows are WALL, and cracks are set as interior BC.
I've so many questions about this model:
1. I built this model in gambit, I started with creating many volumes, and created those "doors" and "windows" and "cracks" by spliting different volumes with each other, so I have the openings left. each of the six offices, windmachine, fans, trailer, pipes, source are split from the whole domain. Since the cracks are really small, I created some volumes surrounding these window and door cracks just to get a much finer grid around them, of'coz these volumes are split from each office, and I didn't set their surface BCs so as a default their surfaces will be INTERIOR, just the way I want them to be.
Then when I mesh the model, I started with each surrounding volume of the window, creating finer grid, then the volumes around the door cracks. After that, I mesh the six offices one by one, with a coarser grid, I checked my mesh, skewness is below 0.97, so it doesn't give me any warning about the skewness. I then mesh the left whole domain with a even coarser grid(just to save some computation time). I always use tet/hybrid to mesh my model.
I can see, for example, at the interface of the surrounding volume a window, the mesh size of the surrounding volume and that of the whole domain is different, and the latter large tetrahedron doesn't seem to have a multigrid nice transition of grid size. I'm a little worried about this, is this OK? Is the way I model this model in gambit OK? I mean, do I have to set the surfaces between surrounding volumes of windows and doors with offices and whole domain as INTERFACE so that FLUENT can connect the surfaces for me and make the grid size transition smoothly?
2. About Outflow BC at the trailer's inlet. I read fluent help file, and I know that outflow can correct the mass balance of the domain and has no impact on upstream flow, but in my model, the outflow should be balancing the velocity inlet BC through the windpipes in the offices, right? But clearly I have some contaminant in the air, and at the inlet of trailer, but I set it as pure air at the inlet of windpipes in the offices. So what about the species conservation? Does the "correct mass balance" mean that outflow BC only takes care of the air mass balance between trailer inlet and windpipe inlet so that the trailer doesn't "inhale" any contaminant into it? Or do I have to set a surface source sink at the inlet to represent the "sucking" of contaminant?
3.About the source, I set the surfaces as velocity inlet instead of setting the volume as fluid and a source term, because I don't know how to set UDF source terms. I actually tried to do that, and I can set it as source term and set the mass term as, say, 0.1kg/m^3s (choose CONSTANT) and species term as the same number 0.1kg/m^3s. I don't care much about the momentum input this release of substance brings along, so leave it alone. But according to what I found from other discussions, it seems that setting like this doesn't work, and I have to do it through UDF. I don't know if I'm getting this picture right, so you know, please tell me.
4. In my model, I found that I have to do transient simulation of this process. So do I have to run a steady-state simulation to get a good initial guess then move to transient or I can start with transient?
5. about the under-relaxation factor. I choose SIMPLEC as solver, and second-order upwind for all properties except STANDARD for pressure. I choose 1 for pressure under-relation factor (from I found about SIMPLEC), and all other factors are in default, say 0.7 for momentum, except that I set 0.0001 for my contaminant species. My question is: can I set, say 10^(-10) for my contaminant species as the under-relaxation factor? Of'coz I will start with 1 in the beginning ,and monitor the continuity to make sure it goes down to a lower level and keeps steady there, then decrease the factor, and wait till it decreases to another lower level and doesn't change much there, the do it over and over... Same thing for the momentum under-relaxation factor. Is there any lower limit of this factor? Am I doing it right? I'm worried about this.
6. Since the properties of my contaminant material are not in the fluent database, I created it. The properties I found from different online sources and literature are: molecular weight, isobaric specific heat Cp, viscosity, diffusivity and density. When I tried to choose these five properties, for diffusivity, there are only two: diffusivity in solid or UDS diffusivity, I had no choice but to choose the latter, even though I don't know what it is. then I defined this material, add it to air (air is in the bottom of the mixture I defined, as it is the bulk fluid) and I defined this gaseous mixture. then I looked at the properties of this mixture, I found that somehow fluent set a thermal conductivity (0.045W/mK) for me and a diffusivity 2.088*10^(-5), while my real contaminant diffusivity in air is 5.52*10^(-6), which I already input as UDS diffusivity. Anyway, I changed this mixture diffusivity back to 5.52*10^(-6) m^2/s and saved this material. Am I doing this right? or I have to define this UDS transport equation? I've no idea about it right now.
7. about getting a finer grid around cracks by creating surrounding volumes just to enhance the meshing, am I doing it right in gambit? if you have any suggestions, please comment. what about the size difference between the surrounding volumes and say, the whole domain? can I just neglect that? after all, mesh the whole domain with a fine mesh will significantly increase my computation time. Is there some other way to work this out?

I've asked so many questions and shared some of my trials with you, thank you for your patience to read this post through, if you could help me figure my problems, I would really appreciate it! I've learned so much from many of the discussions from cfdonline, this is my first time to post here.
Any comment will be great!

Thanks.

[aleisia]

Also later, in my transient analysis, I want to set this source in a way like this:
turn it on for 1s, then switch it off. There are two ways to do this, first is, to run the simulation to 1s (with a time interval ,say,0.01s) and stop it, change the BC to 0m/s and carry on.
second is, write a udf source macro for my source, but I dont know how to, if you could share with me a good source code, I would be greatly indebted!
My email is aleisia@hotmail.com

Thanks!

[amirmohebbi]

dear all
hi
I am new in fluent
how i can define all of my mesh zone stagnation air
and mass flow inlet another flow like that fuel
then i have to the mixing rate of them
please help me?
thanks for your attention
amir mohebbi