[Gearb0x]

Hello,

I have to simulate a curved pipe. Here are the caracteristics of the flow :

Incompressible
Re = 40 000
Hydraulic diameter = 0.08m

So we found U characteristic = 15.7m/s

k = 0.3676
epsilon = 0.3328
(with a turbulent intensity of 6, don't remember the characteristical length used to calculate this)

Since I begin in CFD, I've asked help to a professor wich told me that I should impose these BC :

Outlet = Ambient pressure
Inlet = Total pressure
To calculate the total pressure, we used empirical relation for pressure loss and we find pstatic by adding the pressure loss to the outlet pressure and we just have to add rho u² to have total pressure (u characteristic is used for u)

But in all the examples I found over the web, openfoam, ... BC used are always :
Inlet : velocity
Outlet : static pressure

I don't understand why I have to fix different BC for my case. My professor told us that this was more physical and it probably will cause less numerical problems but he is used to work in compressible flows and said he was not familiar with incompressible solver.

Also in openfoam as well as in Fluent, the BC with the total pressure seem to cause problem (convergence in openfoam)

When I fix the BC with the velocity @15.7m/s at the inlet, my total pressure is quite different from the one we calculate with empirical formulas and I don't understand why (We found 102 Pa total pressure and the numerical simulation with velocity inlet give us 500 Pa in fluent)

Can someone help me to solve my problem with the boundary conditions with the total pressure? Our professor told us it could be the mesh ...

Thanks for the help!!

[msbealo]

Did you find the solution to this?

Mark

[santiagomarquezd]

Originally Posted by Gearb0x

Hello,

I have to simulate a curved pipe. Here are the caracteristics of the flow :

Incompressible
Re = 40 000
Hydraulic diameter = 0.08m

So we found U characteristic = 15.7m/s

k = 0.3676
epsilon = 0.3328
(with a turbulent intensity of 6, don't remember the characteristical length used to calculate this)

OK.

Since I begin in CFD, I've asked help to a professor wich told me that I should impose these BC :

Outlet = Ambient pressure
Inlet = Total pressure
To calculate the total pressure, we used empirical relation for pressure loss and we find pstatic by adding the pressure loss to the outlet pressure and we just have to add rho u² to have total pressure (u characteristic is used for u)

It's Ok.

But in all the examples I found over the web, openfoam, ... BC used are always :
Inlet : velocity
Outlet : static pressure

It's usual because you know the volumetric flux or mass flux of your pipe, then setting an static pressure at the outlet you can obtain the pressure losses. Another related problem is to select the correct diameter to have the desired volumetric flux with no more of X pressure loss.

I don't understand why I have to fix different BC for my case. My professor told us that this was more physical and it probably will cause less numerical problems but he is used to work in compressible flows and said he was not familiar with incompressible solver.

Also in openfoam as well as in Fluent, the BC with the total pressure seem to cause problem (convergence in openfoam)

When I fix the BC with the velocity @15.7m/s at the inlet, my total pressure is quite different from the one we calculate with empirical formulas and I don't understand why (We found 102 Pa total pressure and the numerical simulation with velocity inlet give us 500 Pa in fluent)

Can someone help me to solve my problem with the boundary conditions with the total pressure? Our professor told us it could be the mesh ...

Thanks for the help!!

Setting total pressure at inlet allows you to obtain a completely developed profile in all pipe extension, which should match the pressure losses better than a velocity inlet. This is because velocity inlet BC generates a developing zone, near the inlet, which is larger as viscosity increases.
The differences are probably due meshing issues, you need to satisfy y+ parameters in near wall regions in order to use k-epsilon model appropriately (check these Guidelines). Once you have a good mesh then results for pressure-pressure and velocity-pressure should be slightly different for Re outside of laminar region (region of validity of k-epsilon model and short development zone)

Regards.