Dynamic simulation with significantly different behaviour from version 3.0.13 to 3.0.14 and upwards

[andr1976]

I have used the dynamic separator example from colab as a basis for a dynamic separator simulation, both with an inlet valve and without (fixed inflow). Both examples have widely different behaviour, both the liquid level and the dynamic pressure response. The code is available here: https://github.com/andr1976/neqsim-fun/blob/main/dynanamic-separator.py

A few outputs with version 3.0.13:

With v. 3.0.29

Despite explicit setting inital valve opening to 30% it is 100% in v >3.0.13 (it seems the difference occurs in version 3.014 and above).
Also the liquid level drops significantly after v 3.0.13 despite valves are specified with 0.5% valve opening.

@EvenSol did the valve model change from v3.0.14 and up? could this be the reason?

This file does not have an inlet valve https://github.com/andr1976/neqsim-fun/blob/main/ors-pst.py and the behaviour is even more strange. Comparison with Unisim, the v3.0.13 is closer (yet not fully the same)

[EvenSol]

Thank you for reporting this! We are looking into this, and will come back to you on this soon.

Ecem

[andr1976]

It may be something with the valve model, both two-phase and liquid phase. It seems the two phase inlet valve fails to account for sub-critical flow (a qualified guess)

[EvenSol]

Again, thanks for pointing to this bug Anders.

I'm working on this issue in a new update of NeqSim.

I'm testing it here (and as you can see I get significant different results in this update):

https://colab.research.google.com/drive/1R4-BKk4_qysSMXVmI_v2cCGmYYqNzjS2?usp=sharing

Even

[andr1976]

Hi Even

Something is a bit odd when going from SS to dynamic. The flow sudenly drops. I think this may be due to multiphase valve flow not rigorously defined, initially flow is specified, and the valve is sized accordingly, when the dynamic simulation is started the calculated Cv is used to estimate the flow - and this is not 100% self-consistent. Otherwise, it looks as a great improvement.

[EvenSol]

Hi Even

Something is a bit odd when going from SS to dynamic. The flow sudenly drops. I think this may be due to multiphase valve flow not rigorously defined, initially flow is specified, and the valve is sized accordingly, when the dynamic simulation is started the calculated Cv is used to estimate the flow - and this is not 100% self-consistent. Otherwise, it looks as a great improvement.

Thanks for reporting. The steady-state (SS) simulation does not update the flow rate based on the valve opening, meaning that the flow rate is only updated after a dynamic calculation is performed. I have updated my example with a small time step for a dynamic calculation, and this strange behavior is now gone.

The model for the valve has been updated in a recent version. You can find the valve models in NeqSim here: Valve models in neqsim. The current default is ControlValveSizing, which is a very simplified model for gas valves. I will look further into two-phase/multiphase valve simulation.

And also added some examples here and some valve documentation here.

Happy for all suggestions for improvement!