[erichu]

Hello everyone,

I am trying to calculate the flow rate of air in a converging nozzle (10mm diameter --> 3 mm diameter, total length of nozzle/lance is 1000mm) using fixed pressures for inlet and outlet (0barg). There is a range with inlet pressures of interest as it can be compared with experimental data, 50mbarg -->2barg. 6 cases are therefore under construction.

I guess that it is not necessary to use a compressible solver for low pressure differences but for higher flows experiments indicated high velocities. However, I would like to stay with one solver for all cases. Currently, rhoSimpleFoam appears to be of interest (steady state/compressible/transonic).

The boundaries (for 1 specific case) are as following:

Pressure
Inlet: totalPressure 161325
Outlet: outletInlet 101325
Wall: zeroGradient

Velocity
Inlet: pressureInletOutletVelocity
Outlet: inletOulet (000)
Wall: fixedValue (000)

Temperature
Inlet: fixedValue 298
Outlet: zeroGradient
Wall: zeroGradient

For turbulence I have used compressible functions as used in tutorial cases but with modified initial values. Upwind schemes are also used.


Now to the problem;

The simulation converges (very slow) for low pressure differences ~max 400mbar but the time step continuity constantly increases for every iteration. Time step local sum error is 90 after reached convergence criteria.

Fo higher pressures I cannot reach convergence and the time step is steadily increasing. In general, I suspect that the pressure is problematic as its residuals decreases very slow, if at all. However, if I turn the transonic option off, the solution steadily but slowly approaches convergence.

I have tried coarse- and fine meshes (structured 4k to 200k cells), checkMesh OK, tried with/without turbulence, different initial conditions and BCs but I have hard time moving forward. The only partly working solution is transonic = no.

Does anyone have any suggestions of how to setup a case like this?
Why is not transonic = yes working?
How can I possibly speed up convergence?
Why is the time step increasing even though the residuals are decreasing?



Thanks!

[erichu]

Update:

Earlier today, I decided to run without the transonic option and used ParaView to look at the changes upon floating point exception as well as the convergence pattern. The slow convergence could be removed by using fewer cells upstreams. Calculations up to 1.5 barg is ok and correlates very good to experimental data. Outlet speed is close to sonic speed at 1.5barg.However, for 2 bar g there is almost no difference from 1.5 barg which I found strange.

If I run 2barg with transonic option activated, OpenFOAM returns floating point exception related to pressure I think. Relaxation factors, mesh are being changed but I see very little improvement.

I think the problem might be near the outlet as there is a sudden pressure drop, see attached image. Does anyone know how to come around this problem?

[fpierart]

Originally Posted by erichu

Update:

Earlier today, I decided to run without the transonic option and used ParaView to look at the changes upon floating point exception as well as the convergence pattern. The slow convergence could be removed by using fewer cells upstreams. Calculations up to 1.5 barg is ok and correlates very good to experimental data. Outlet speed is close to sonic speed at 1.5barg.However, for 2 bar g there is almost no difference from 1.5 barg which I found strange.

If I run 2barg with transonic option activated, OpenFOAM returns floating point exception related to pressure I think. Relaxation factors, mesh are being changed but I see very little improvement.

I think the problem might be near the outlet as there is a sudden pressure drop, see attached image. Does anyone know how to come around this problem?

Hi,

How are you, I am trying to solve a similar case than you is a compressible flow with one inlet and two outlets, but I can do that rhosimpleFoam solve nothing, I try first with the tutorial of squareBend but didnt work, could you send me your case with the transient off just to believe that the rhosimpleFoam can be used, because I solve the same example for a incompressible flow and works perfect and when I made all the changes to run with the compressible solver dosent work.

Thank you in advance.