CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   buoyancy driven flow (

adam January 31, 2005 10:15

buoyancy driven flow
Hi, Does anybody know how to treat density when modeling flow with high difference of altitude (order of 800 m)? I have pressure inlet and pressure outlet BC's.The fluid is air. The hydrostatic pressure must be removed from the static pressure at the inlet and otulet and the changes of pressure with altitude have to be taken into account after getting the pressure field. Is it the same case with density? For incompressible ideal gas the density will depend on temperature only (the operating pressure is constatnt). Should I let fluent to calculate the density field in such a way and then just calculate the field again with pressure containing the hydrostatic term to see the impact of height? Is it ok to do it like that? Best regards

Evan Rosenbaum January 31, 2005 14:12

Re: buoyancy driven flow
Instead of using incompressible ideal gas, try ideal gas. I think that should account for pressure, although I've never actually tried such a problem.

adam February 1, 2005 04:46

Re: buoyancy driven flow
thanks for respond, Ideal gas will be even worse... If I used ideal gas, fluent will calculate the density according to current pressure (density=(operating pressure+local pressure)/(R/M*T). This will take into account the local pressure which is lower due to the assumption of removed hydrostatic pressure and thus the density will be also much lower. Then the density will have impact on sollution of the Boussinesq term in momentum equation. I am just not sure wether ideal gas or incmopressible ideal gas will better fit to the problem. It is sad that I cannot find much about it in the UG. regards adam

giorgio February 10, 2005 13:26

Re: buoyancy driven flow

I performed a similar simulation. If you set the operating pressure to zero you obtain that local pressure=absolute pressure, the density should be calculated in a rigth way. But after that I had a lot of troubles with the turbulence values, TKE especially.



adam February 11, 2005 04:44

Re: buoyancy driven flow
Hi Giorgio, Yes, you're right but did you have large difference in altitute? If you do, you need to redefine the pressure due to the hydrostatic head. The redefined static pressure is p's=ps-rho0*g*h. It doesn't matter if it is absolute or gauge - it is lower and the calculated density is also lower. Probably it is necessary to define my own formula for density calculation by UDF. Regards

giorgio February 11, 2005 05:55

Re: buoyancy driven flow
Hi Adam,

in my simulations the height of the domain was 1 km. The formula you've writted is valid only if rho is constant. In the hydrostatic approximation you have to integrate the hydrostatic law from the ground to the top of your domain. Inside the integral form you have rho=rho(pressure(z), temperature(z)) so calculate this integral and you get a vertical pressure profile p=p(z) with z the altitude. Build a pressure profile file to import in fluent. I applied this profile at the outlet (pressure outlet) while at the inlet I had set the inlet as velocity inlet.

It should work


adam February 14, 2005 05:11

Re: buoyancy driven flow
Hello Giorgio, I have Pressure Inlet and pressure Outlet BC's. The pressure inlet is vertical ca 600m high and pressure outlet is horisontal at the top. Fluent calculates velocity at the pressure inlet from Bernoulli's equation (for incompressible). If I would have difference in inputs between p. inlet and p. outlet, a flow would appear, even there nothing flows in reality. If I calculate the integral you said I will get p = p0*exp(-g*x/(RT)). Therefore pressure to be input at the boundary must be constant: ps = pop - p0*exp(-g*x/RT). To the obtained pressure field I must add the hydrostatic head p0*exp(-g*x/(RT)). The problem is with density. The density also chanegs with height as rho = rho0*exp(-g*x/RT). If I used incompressible ideal gas or ideal gas for density calculations the density would be calculatedc as rho = pop/(RT) and rho = (pop + p)/(RT) respectively. In both cases it seems to be wrong since in first it does not depend on pressure in the system and in the second p is very low everywhere in the system and density will also be low. I tried to apply rho = rho0*exp(-g*x/RT) for density calculation throuh UDF but it takes ages to do one iteration and difficulties with divergence appeared... So far I cannot see any solution for that... regards Adam

All times are GMT -4. The time now is 00:52.