[Manu4CFD]

For an incompressible steady state problem with density modeled through Boussinesq approximation, Is it okay to use boundary conditions such as “Mass flow inlet” for flow entry and “Velocity-inlet” (with a negative value to specify that flow exits the domain) for flow exit. By using such an approach, if we manage to achieve mass conservation, then will it be okay to change the usual practice of setting boundary conditions such as “Velocity inlet” for flow entry and “pressure outlet” with gauge pressure set to zero for flow exit. Kindly provide me your valuable insights.




[/QUOTE] :
Sometimes a velocity inlet boundary is used where flow exits the physical domain. This approach might be used, for example, when the flow rate through one exit of the domain is known or is to be imposed on the model.



In such cases you must ensure that overall continuity is maintained in the domain.

In the pressure-based solver, when flow exits the domain through a velocity inlet boundary FLUENT uses the boundary condition value for the velocity component normal to the exit flow area. It does not use any other boundary conditions that you have input. Instead, all flow conditions except the normal velocity component are assumed to be those of the upstream cell

[LuckyTran]

Use the outlet velocity only if you specifically need/want to. That is, if you actually want an exit velocity profile to be a specific shape and value then by all means do it. But if you don't have this reason, do not play around with it.



The usual practice of using a pressure outlet is the usual practice for good reason. You need to give BC's for velocity and pressure. When you fix the velocity, you apply the zero gradient condition on pressure... Not only do you fix the outlet velocity, you also allow weird things to happen on pressure.


Imagine you are solving the heat equation and the only boundary conditions is no heat flux at the boundaries. Well that allows the temperature field inside to be almost anything.... In truly incompressible flows this is okay because that's what happens in incompressible flows. But you are using a compressible solver...

[Manu4CFD]

Originally Posted by LuckyTran

Use the outlet velocity only if you specifically need/want to. That is, if you actually want an exit velocity profile to be a specific shape and value then by all means do it. But if you don't have this reason, do not play around with it.

Thanks a lot for your kind reply.
In my case, I am dealing with an incompressible fluid flow and I know the inlet mass flow rate. There is a heat source in my domain and I am using boussinesq approximation for density.
I presume that in order to achieve the mass conservation, the amount of air that comes in to the domain should go out of the domain through the outlet. So, i just calculated the outlet velocity based on the outlet area to get the same mass flow rate out of the domain, as that got in to the domain through the inlet. So, under such a scenario, is the approach to use "mass-flow-inlet" as inlet condition and "velocity-inlet" as outlet condition for an incompressible fluid flow justifiable? I have attached a simple figure to depict my physics for your reference.

[Manu4CFD]

Originally Posted by LuckyTran

The usual practice of using a pressure outlet is the usual practice for good reason. You need to give BC's for velocity and pressure. When you fix the velocity, you apply the zero gradient condition on pressure... Not only do you fix the outlet velocity, you also allow weird things to happen on pressure.

I do agree that in most of the cases the usual practice is the most appropriate manner to setup the simulation cases. But, in my case, as you have understood, i need to specify a certain velocity at the outlet. Hence, I had take a detour from the usual practice and investigate the approach that I had mentioned in this post.

If the velocity is fixed doesn't it mean that the pressure is also fixed. Say, if my outlet velocity is -0.2m/s (negative sign indicates flow exits the domain) and operating pressure is 101325Pa, doesn't it mean that air is flowing out of the domain with a velocity of 0.2m/s at 1atm?

[LuckyTran]

Originally Posted by Manu4CFD

But, in my case, as you have understood, i need to specify a certain velocity at the outlet. Hence, I had take a detour from the usual practice and investigate the approach that I had mentioned in this post.

Nothing you say tells me you actually need a velocity at the outlet.... Especially since you are calculating it based on your inlet!

So no you don't need a velocity outlet. Your idea that mass flow is conserved is a constraint (not a boundary condition) and it will be satisfied since you are solving a continuity equation.

Saying that the velocity is fixed doesn't say anything about pressure. As a corollary, saying the velocity is fixed doesn't say anything about temperature.


Do the right thing. Use a pressure outlet. For incompressible flows, pressure doesn't even matter anyway. Or, there's also an outflow boundary condition you can use that is even more appropriate.

[Manu4CFD]

Originally Posted by LuckyTran

Nothing you say tells me you actually need a velocity at the outlet.... Especially since you are calculating it based on your inlet!

So no you don't need a velocity outlet. Your idea that mass flow is conserved is a constraint (not a boundary condition) and it will be satisfied since you are solving a continuity equation.

Saying that the velocity is fixed doesn't say anything about pressure. As a corollary, saying the velocity is fixed doesn't say anything about temperature.


Do the right thing. Use a pressure outlet. For incompressible flows, pressure doesn't even matter anyway. Or, there's also an outflow boundary condition you can use that is even more appropriate.

Thank you for your kind reply.

As rightly mentioned by you, for an incompressible flow the density change with respect to pressure and temperature is negligible. Hence, the constant density approximation. Since there is a heat source in my simulation, I have used the boussinesq approximation to consider the buoyant forces.

In my study, there is a room with four inlets and four outlets. I need to split the total inlet mass flow rate equally among the four outlets. Which means, say if I have 1 kg/s of total inlet mass flow rate into the domain at 16 Degree Celsius, then through each outlet the mass flow rate going out of the domain should be 0.25kg/s at 24 Degree Celsius. if i apply either of pressure outlet / outflow condition, such a constraint cannot be placed. Hence, I am specifying a velocity-inlet condition with a negative value. In this particular scenario, isn't it justifiable?

[LuckyTran]

There is a flow split outlet boundary condition for when you want a particular % of the flow to go through a particular outlet. Surprise! There is a boundary condition that does exactly what you want that isn't fixing the velocity!

Again, you don't need a velocity outlet!

Again, there are situations where you would want to use a velocity inlet at an outlet and fix all the properties at the outlet. Believe me, I have done this before! But I did it because I wanted to ignore physics and fix all the properties at the outlet. I didn't use a velocity outlet because I wanted to conserve mass and didn't know what to do.

And how do you know the flow at the outlet is a uniform velocity and uniform 24 degrees Celsius? You don't. If you did, you wouldn't be solving a transport equation to solve for the velocity and temperature. Honestly think about this. You are also imposing a uniform velocity profile. What makes you think the velocity should be uniform there? For any given inlet mass flow rate. There are an infinite number of velocity profiles at the outlet that all give the same mass flow rate. What makes you think your uniform velocity of 0.2 m/s is the correct one? hmmm???? Can you prove that to me?


The answer is the velocity profile of the outlet depends on what happens inside the domain. For example, there might be an elephant inside!

[Manu4CFD]

Originally Posted by LuckyTran

There is a flow split outlet boundary condition for when you want a particular % of the flow to go through a particular outlet. Surprise! There is a boundary condition that does exactly what you want that isn't fixing the velocity!

I believe you are referring to the flow rate weighting option available in the outflow boundary condition. I had tried this method for the flow exit. But, it seems to give reverse flow at the flow exits, although eventually the solution converges with the reverse flow warning. While going through this forum, there were a few suggestions to remove the reverse flow phenomenon, for instance, extending the flow domain, so that the flow develops fully. But in my case, i can't extend my computational domain. Meanwhile, when I imposed the velocity outlet condition, interestingly there was no reverse flow warning. Why would it be so?

Originally Posted by LuckyTran

And how do you know the flow at the outlet is a uniform velocity and uniform 24 degrees Celsius? You don't. If you did, you wouldn't be solving a transport equation to solve for the velocity and temperature. Honestly think about this. You are also imposing a uniform velocity profile. What makes you think the velocity should be uniform there? For any given inlet mass flow rate. There are an infinite number of velocity profiles at the outlet that all give the same mass flow rate. What makes you think your uniform velocity of 0.2 m/s is the correct one? hmmm???? Can you prove that to me?

That's an interesting point for me to ponder. Since, l got the reverse flow warning while using the pressure boundary conditions such as pressure-outlet or Outflow, I think the flow is not completely developed while it exits my computational domain. So the velocity profile isn't parabolic, as it is not fully developed. But while imposing a velocity outlet condition there is no reverse flow warning, but incorporating velocity outlet boundary condition would inherently impose a uniform velocity profile as mentioned by you.

Originally Posted by LuckyTran

The answer is the velocity profile of the outlet depends on what happens inside the domain. For example, there might be an elephant inside!

Lol.