Learning Forum

[Marcin Nowak]

Hi,
I am modeling a transient flow through a straight flexible pipe in Fluent. I got the experimental data: time-varying velocity at inlet, time-varying pressure at the inlet, time-varying pressure at the outlet.
I have to decide what fluid boundary condition to use. The time-variance is realized by the UDF define_profile.

I tried with the velocity-inlet and pressure-outlet. I monitor the calculated inlet pressure, and it differs very much from the experimental data - it is almost equal to the imposed pressure at the outlet.
So I tried with the pressure-outlet and pressure-inlet BC, where for the inlet I calculated the total pressure (static pressure+dynamic pressure, anyway dynamic pressure was negligible). In such model I got much discrepancies in calculated velocity at the inlet and measured velocity.

So the question arises, how to properly specify the boundary conditions for this problem.
I found that in the pressure outlet panel, there is an option "target mass flow", where I can specify the time-varying mass flow, determined based on the measured velocity time profile. Is this a proper solution? The measured velocity however, is at the inlet, not at the outlet... and they will not be equal because there is a mass accumulation in the domain.

The best solution for me, will be to impose: at the inlet both the velocity and pressure time profile, and at the outlet, the pressure time profile. However as far as I know there is no possibility to impose both velocity and pressure at the inlet, and I don't know why.

Please guide me with this problem,
Regards

[Rob]

The inlet pressure will vary as you alter the inlet velocity to ensure that much material flows into the system. You could alter the outlet pressure with time too, but I'm not sure what that will prove in your model. 

 

[Marcin Nowak]

I would like to prepare a numerical model that would reliably reflect the experimental data. Which means:
1) impose boundary conditions as pressure-inlet and pressure-outlet, validate using experimental data: velocity time profile and deformations,
or:

2) impose boundary conditions as velocity-inlet and pressure-outlet, validate using experimental data: inlet pressure profile and deformations.

The solid material properties are tuned, using FEM-only analysis.

In the case 1) I receive the proper pressures at walls and proper deformations, but completely different velocities at the inlet.

In the case 2) I receive too low pressure at walls (and at the inlet) and therefore too small deformations. But velocities are proper, due to the imposet inlet-velocity BC.

I think that model 1) is better, but I am looking for the setting which will lead to the solution of all variables coincide with the experiment, i.e. including velocities.

[Rob]

OK, so you also want to model, or impose the deformation of the boundaries? If you're modelling the deformation then velocity inlet and velocity outlet may be suitable. 

[Marcin Nowak]

I want to model the deformations.

"If you're modelling the deformation then velocity inlet and velocity outlet may be suitable. " - You mean, should I try to use the same velocity profile at the inlet and at the outlet?
I dont have however, the velocity measurement at the outlet, only at the inlet, and they may be different due to the deformation and mass accumulation in the domain.

[Rob]

You can't use the same profile at both ends as they will (more or less) mean you have the same amount of fluid in the system. As you're deforming the domain you need more information, so, flow enters, potentially pools and then leaves. The inlet flow and outlet pressure may determine pooling but as both are transient you need to know those values, or potentially assign the motion. 

As CFD has moved from fixed density to compressible flow to deforming mesh the amount of input data has risen. In the "good old days" we just needed an inlet velocity and a decision on turbulence (yes or no)! That's no longer the case. 

[Marcin Nowak]

"The inlet flow and outlet pressure may determine pooling but as both are transient you need to know those values, or potentially assign the motion. " - Yes, I know these values. However when I use them in the model, the determined inlet pressure differs much, from the measured value at the inlet. And therefore, as the determined inlet pressure is different (smaller), the wall pressure is smaller, and deformations are smaller than measured. Of course decreasing solid material stiffness will increase deformations, however the determined inlet pressure still varies.

Maybe the solution, will be,
1) first, model the deformations using pressure-inlet and pressure-outlet, as these BCs properly reflect deformations (but velocities, no)
2) then, impose the calculated deformations from model 1), and impose velocity-inlet and pressure-outlet, to properly reflect the flow field,

What do you think Rob?

[Rob]

I think you need to tread carefully to avoid fixing too many parameters. The experiment behaves as it does because of "something". What different "something" is happening in the CFD? 

As a note, I can advise on settings etc in the software but must take care regarding application & engineering knowledge because the Forum is considered public in legal terms.