Hello everyone,

I'm modelling the rise in temperature of a 'sealed' air tank, from 300K to 450K, over about 20 minutes.

In reality, the tank is not 100% sealed, but according to theoretical calculations, the velocity magnitude through the leakage zone is very small compared to the rest of the tank (flow driven by free convection).

Is there a simple way to account for the leakage without geometrically representing the leakage zones? (i.e. pressure remains constant in the tank, even if the temperature increases and the tank is 'sealed'.) A mass sink maybe? any successful experience?

Thanks for your input. /JB

You can use DEFINE_PROFILE UDF (UDF manual, Chapter 2, 2.3.13 DEFINE_PROFILE ) Chouki

Thanks Chouki.

I will try to implement something with a UDF on a simpler model, and see how it behaves (I'm not familiar with UDFs.)

/JB

Hi, I feel you need not require any special care to maintain constant pressure inside the tank. For natural convection problems with density formulation by Boussinesq approximation the pressure variations remain almost the same. Regards,

Thanks Joe.

I went through the user's guide, and now I'm just totally confused:

How can Rho=Pop/((R/Mw)xT) for Incompressible Ideal Gas flows?? That would mean that if I have a 100% sealed tank, and I heat it up, the overall density will drop... M[kg]=rho[kg/m3]xV[m3], right?? If the volume of the tank is constant, that means a fraction of the mass initially contained in my volume will just disapear as the temperature increases?

Thanks for your insight.

Ok, I just found in the user's guide that a typical application example of the floating operating pressure is the heating of a gas in a closed domain. So I guess it is just not a good idea to simulate that kind of problems with anything else than floating OP...

New question: If we come back to my heated up tank problem: Joe mentionned the Boussinesq model in his post. Unfortunately the temperature variation inside my model (even at steady state) are too high to use Boussinesq. From what I read the only other option is to perform a transient analysis. What happens if I use the incompressible ideal gas model in a closed volume that I heat up? Does someone know where the mass will go? I really lack of theoretical background to understand what's happening here... Is it OK to 'cheat' using this setup to simulate a mass sink? Fluent won't crash, but still...(my system will heat up at a quite slow rate, which means that the actual leak rate should be very small, so I guess the mass sink could be either applied at a very specific location or in the whole domain)

Thanks! /JB