[DarrenC]

Hi All,

I am working on a multiphase flow simulating a multistage oil/gas/water separator. The stages are essentially cascaded in series and operating at different pressures. Each stage is held at its operating pressures and liquid levels by means of restriction orifices and valves.

Due to the complexities of actually simulating the valves I would like to model them using interfaces in CFX. I have tried using interfaces in between the pipeworks where the restriction orifices/valves are. I define them as additional interface -> mass & momentum -> interface model and then i have tried either pressure interface or mass flow interface.

For the pressure interface i define the pressure drop between the interface. I know the operating pressures of both sides so i set this pressure drop (constant) to mimic that.However using this interface i am not able to control the mass flow. I find that if the pressure drop is increased, then the mass flow increases as well, which makes sense as the interface essentially treats everything in betweenits faces as a black box

Then I tried using mass flow interface instead, specified a constant mass flow and let the solver adjust the upstream and downstream pressures.I found that this works, but only to a certain extent. interface between the first stage works. Mass flow converges to the desired flow rates. But in subsequent stages I am losing liquid (through the gas lines in each separator stage which is also connected to each other by way of mass flow interfaces) even though my mass flows are defined in such a way to try to build up the liquid level at each stage. The oil forms a very thin liquid on the gas line walls and escapes each stage from there.

FYI my gas oil and water inlets are all defined as mass flow and my gas,water,and oil outlets are defined as pressure outlets. All interfaces in between the different stages, are defined as mass flow interfaces.

I know what i am describing is not the clearest but it gives you guys an idea of what im trying to achieve. My question is this : Can you realistically model a restriction orifice in the pipeworks between the stages using the interface model? And if so, how is it done so that it gives you the pressure gradient and also mass flow restriction? I am fairly certain it can be done, perhaps using expressions but I dont know where to start.

Also can flow restrictions using mass flow interfaces be defined in the manner that i have described above? My intuition tells me no and you would run into problems due to well posedness of your BCs, but I need to confirm this.

I Hope someone can help me!

Thanks!

DarrenC

[ghorrocks]

You can define the pressure drop to be a function of mass flow rate and vice versa (the mass flow rate to be a function of pressure drop). Then you can model the actual oriface flow condition and if the flow changes the flow conditions at the oriface will change to stay realistic.

The normal way to do this is with a 1D interpolation function, but it can be done as an analytical CEL expression.

And defining the mass flow as a function of pressure drop is likely to be more numerically stable.

[DarrenC]

Hi Glenn,

Thanks for the suggestion and the lightning quick response =). I will check if I can pull up some data on the pressure v mass flow characteristics of the orifice/valves and generate an analytical expression based on them. Hopefully the data sheet for them is available.

As an additional thought, I wonder if it is possible to implement using CEL an expression that adjusts mass flow on the interface based on pressure, and also additional variables (e.g. orifice opening) so that I can also model the opening and closing of lets say a valve? I imagine that it is almost like getting a point solution (mass flow) from a 2d surface function, instead of a 1d curve function.

Cheers
DarrenC

[ghorrocks]

Yes, you can make the function dependent on most flow variables. But if you want to open and close the interface you should use the conditional closing options for that. That will handle stopping the flow better.

[DarrenC]

Thanks Glenn. Ill have a go at it.

[DarrenC]

Hi All,

The last time I asked for help on this topic was about the modelling of a passive orifice, which was quite straightforward as an expression relating mass flow and pressure drop needed to be implemented.

This time im trying to implement something much more complex, a PID control valve, whereby the mass flow changes not only with respect to pressure drop, but also with valve opening. The PID control algorithm will be implemented using CEL expressions where it will change the effective Cv value of the massflow - pressure relation to maintain the pressure across the interface during surges of fluid across the valve. But before I implement and test this, I was just wondering if a constant pressure change additional interface model achieves the same outcome.

I have done some tests with the constant pressure change additional interface model and from what I can see of the pressure traces at the inlet and outlet of the interface, it does resemble the workings of a control valve during a surge of fluid.

The following is the pressure trace at the inlet (red) and outlet (blue) for the interface model. In this simulation I was trying to maintain a pressure difference of approximately 2 bar. A surge in fluid mass flow (essentially a square wave of high mass flow) was introduced for 1s at the 5s mark.



What I find interesting from this is that the solver tries to adjust the pressure upstream and downstream of the system to maintain that pressure change across the interface (especially obvious from the initial spike of pressure). I believe this behaviour is somehow similar to how a control valve works.

Ive tried to search in the cfx theory and guide to determine how this pressure change is enforced in the solver but i could not find anything useful. Anybody has an idea? Appreciate your responses =)

Cheers,
DarrenC