CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Mass imbalance causing no convergenc: domain IS supposed to be positively pressurised (http://www.cfd-online.com/Forums/fluent/99711-mass-imbalance-causing-no-convergenc-domain-supposed-positively-pressurised.html)

kingjewel1 April 10, 2012 06:24

Mass imbalance causing no convergenc: domain IS supposed to be positively pressurised
 
I'm having trouble converging a case of a building ventilation system which is positively pressured hence the mass imbalance. More air goes in than comes out the vents. In reality building leakage takes care of the rest.

But what can I do in Fluent to sort this out and allow the case to converge?:confused:

LuckyTran April 10, 2012 22:57

Quote:

Originally Posted by kingjewel1 (Post 353941)
I'm having trouble converging a case of a building ventilation system which is positively pressured hence the mass imbalance. More air goes in than comes out the vents. In reality building leakage takes care of the rest.

But what can I do in Fluent to sort this out and allow the case to converge?:confused:

What did you expect? You are trying to simulate something that is physically impossible (more mass going in than out).

You need to account for the mass leakage. Or reduce your mass coming in, either way you'll never get them to balance exactly.

You need to relax your boundary conditions a bit. CFD is only a tool to help you understand the problem, don't become a slave to it. Make engineering approximations! Use a pressure outlet or something that makes sense (more sense than having more mass coming in than leaving). You already knew that in real life flow leakage takes care of the mass imbalance, now what can you do in the CFD to take care of it?

kingjewel1 April 11, 2012 04:38

Quote:

Originally Posted by LuckyTran (Post 354093)
What did you expect? You are trying to simulate something that is physically impossible (more mass going in than out).

You need to account for the mass leakage. Or reduce your mass coming in, either way you'll never get them to balance exactly.

You need to relax your boundary conditions a bit. CFD is only a tool to help you understand the problem, don't become a slave to it. Make engineering approximations! Use a pressure outlet or something that makes sense (more sense than having more mass coming in than leaving). You already knew that in real life flow leakage takes care of the mass imbalance, now what can you do in the CFD to take care of it?

My initial reaction was to make the outlets pressure outlets and so allow any amount of fluid extraction. It just seems wrong wrong not to put some sort of windows in, door cracks etc- What do you think?

LuckyTran April 11, 2012 08:03

Quote:

Originally Posted by kingjewel1 (Post 354128)
My initial reaction was to make the outlets pressure outlets and so allow any amount of fluid extraction. It just seems wrong wrong not to put some sort of windows in, door cracks etc- What do you think?

That's what I mean by engineering assumptions. Do you really need all that? Will it make any difference? Yes. How much? Just think about all the other grand assumptions you are making, starting with, how accurately do you even know your mass flow rate in the first place? Or the size of the domain. Or heck, even the fluid properties themselves. Lastly, unless you are doing DNS, you already must admit that you are modeling most of the fluid behavior. Isn't that just as wrong as not putting in cracks and windows?

Don't overcomplicate things. Give the CFD what it needs and get on with your work/life. Unless you really want to know what flow through cracks looks like. How much do you care and how much are you willing to do?

kingjewel1 April 11, 2012 18:04

Quote:

Originally Posted by LuckyTran (Post 354168)
That's what I mean by engineering assumptions. Do you really need all that? Will it make any difference? Yes. How much? Just think about all the other grand assumptions you are making, starting with, how accurately do you even know your mass flow rate in the first place? Or the size of the domain. Or heck, even the fluid properties themselves. Lastly, unless you are doing DNS, you already must admit that you are modeling most of the fluid behavior. Isn't that just as wrong as not putting in cracks and windows?

Don't overcomplicate things. Give the CFD what it needs and get on with your work/life. Unless you really want to know what flow through cracks looks like. How much do you care and how much are you willing to do?

So are you saying that it's not really worth any extra effort because the result is still just a rough outline of reality anyway?

Then why do I do grid independence studies if I'm going to end up converging to a solution that's probably BS to start with?

And hence it's extremely difficult to draw any legitimate conclusions from this type of CFD?:(

LuckyTran April 11, 2012 22:13

Quote:

Originally Posted by kingjewel1 (Post 354260)
So are you saying that it's not really worth any extra effort because the result is still just a rough outline of reality anyway?

Then why do I do grid independence studies if I'm going to end up converging to a solution that's probably BS to start with?

And hence it's extremely difficult to draw any legitimate conclusions from this type of CFD?:(

I'm not saying that it is a rough outline of reality (although in reality it is). I'm saying, adding these extra fine little details would not change your solution (much). So do you really want to put the extra effort into changing your solution by a few %? If you do, go ahead. If not, then why not just accept that there is a small error? From the beginning, the fluid model is only a model (we have already made many grand assumptions before even starting CFD, we have already made them when we called matter a fluid).

We make these assumptions and simplifications in order to get rid of the unnecessary details so that we can perform analysis and get meaningful results. Now, if you can keep all the details and still get the analysis done, then more power to you.

I am not saying that it is difficult to make legitimate assumptions. I am saying the opposite; it is much simpler to make legitimate conclusions by making assumptions! So why make your life more difficult?


All times are GMT -4. The time now is 12:35.