[sasanghomi]

Dear friends,

Many say that it is not correct to use velocity inlet as a boundary condition when it comes to compressible flow.
I am wondering what the reason is behind this idea! What problems could be raised if I used velocity inlet for compressible flow simulations?


Best Regards

[evcelica]

As far as I know, there is no problem using velocity inlet for compressible flow.
The only reason I can think that one might say that is that perhaps your mass flow may not be exactly as expected since your density may be different depending on the inlet pressure.
But a boundary condition is a known/defined condition at that boundary, if you know the velocity at that boundary, then that is what you should use as your boundary condition.
I'd be interested in the reasoning of the "many" though, as I don't see a problem with it.

[Opaque]

Perhaps the problem the context and or the wording/interpretation of "correct".

In the appropriate context (given all the boundary conditions of the problem), it is correct to specify the velocity as long as the pressure level is defined somehow; however, it is not useful for engineering purposes or design iterations.

Upstream of the system you may be able to control the pressure/total pressure to deliver a specific mass flow. The velocity level will be a function of such pressure/total pressure and the relationship from pressure/total pressure to velocity via mass flow is not trivial since it also depends on the equation of state.

Since you have not described your outlet boundary condition, the context is incomplete. Now, if you are set (assumption) in using a mass flow outlet, or outlet velocity, the use of specified velocity at the inlet is definitely incorrect since the pressure level would be missing.

[ghorrocks]

To add to the useful comments already on this thread:

If you think about the fundamental purpose of boundary conditions, it is to put a boundary on the modelled domain such that the boundary is representative of how the modelled domain interacts with the rest of the world outside the modelled domain. A correctly implemented boundary condition will mean that the modelled domain reproduces the actually thing you are modelling with sufficient accuracy that the results are useful.

So the most important assessment of whether a boundary condition is suitable or not depends on the accuracy you are looking for. A preliminary design assessment will have very different requirements compared to a precision benchmark solution.

And secondly, if a constant velocity boundary is a reasonable representation of the actual situation to an accuracy you are happy with then it sounds like a good boundary condition.

And finally, don't forget CFD has assumptions and simplifications about a continuum, the applicability of the Navier Stokes equations, approximate geometry due to meshing, numerical approximations due to numerical methods, and many more approximations (for instance the boundary condition of zero gradient of pressure normal to a wall is not correct in some creeping flows and there is no way in CFX to fix that).

[sasanghomi]

I appreciate your thorough replies. In fact, I am simulating a compressible flow inside a domain whose outlet boundary is constrained by static pressure.

So, it seems to me that as long as there is no concern about the mass flow rate, no problem threatens the accuracy of my simulation in terms of boundary conditions.

[Opaque]

In theory, you are correct; however, there is a major assumption in your statement:

no problem threatens the accuracy of my simulation in terms of boundary conditions

Accuracy is only relevant for a converged solution, i.e. it cannot be accurate if it is not converged.

The proper selection of boundary conditions also helps to converge the equations. Some combinations are difficult to satisfy (inlet static pressure + outlet static pressure), or insensitive to minor variation during convergence; therefore, there are different convergence behaviors depending on the combinations of the chosen boundary conditions (usually problem dependent)

For example, the case for velocity inlet + static pressure outlet, the static pressure on the inlet is not bounded during the convergence algorithm while a total pressure inlet will limit the range of static pressures the solution can converge to. The mass flow inlet boundary condition is a bit softer numerically, but indirectly maintains a bound, i.e. regardless of the static pressure value, the mass flow through the system is fixed and it is something it can be converged to (say a pipe flow).

Hope the above helps

[calf.Z]

I am also simulating compressible flows. I want to use the result to compare it with the experiment. The experiment gives the mass flow rate. So should I define a mass flow BC for inlet? If so, I think flowRateInletVelocity is suitable.

And I use pressureInletOutletVelocity in the outlet of velocity. Is it suitable?

Thank you.

[ghorrocks]

Are you using OpenFOAM? Then you should ask the question on that forum.

But in general, you can either define the mass flow rate and check against experiment to see if it gets the pressure drop right, or you can specify the pressures and see if it gets the mass flow rate correct. You only mention the mass flow rate, so if that is all the information you have you have to use that as the boundary condition and you do not have a check on the pressure for accuracy.