Inducing backflow restrictions in domain interface for multiphase flow

DarrenC · August 1, 2014, 08:30

Hello Everyone

I was just wondering if it is possible to implement backflow restrictions on one fluidic phase only in domain interfaces. My problem is like this.

I have a horizontal pipe and a domain interface in the middle with additional interface model -> pressure change -> XX bar. The domain interface defined this way is supposed to mimic a differential pressure control valve holding the pressure drop at a specified value. Upstream of the domain interface is flowing gas but downstream of the interface is a liquid with a static head. In effect, what I am hoping to achieve is to mimic physically the blowing of the gas across the valve into the liquid phase, driven by the pressure drop induced across it.

Perhaps rather unsurprisingly, because I have modelled the valve this way using a domain interface, the liquid phase is backflowing towards the gas phase. I thought of an idea of trying to restrict the backflow of the liquid phase through the interface to sort of mimic the fact that the valve opening is much smaller than the pipe diameter and hence will probably not backflow upstream of the valve, but im not sure if this is at all possible using domain interfaces.

Has anybody tried to restrict the backflow of fluids through an interface, similar to how inlet and outlet boundaries restrict backflow by placing walls at a proportions of the inlet/outlets? If so can you give some pointers on how it was setup? Appreciate the help!!

Cheers
Darren

JuPa · August 1, 2014, 08:47

I don't think you can physically restrict a phase from flowing back into a domain interface without the use of mass and momentum sinks.

But before you dive into a complicated solution answer the following questions:

1. Is your setup correct? Is your reference pressure correctly set? Are all boundary conditions correct? Have you done a pen and paper calculation using simple Bernoulli principles to calculate the pressure drop in each section and hence have you a ball park figure of what the CFD solution should give you?

2. Are you sure that the simulation is correct, and you're not wrong? I.e. what the simulation is showing you could be representative of what is actually happening. And what you thought would happen isn't actually happening!

3. Is there a simpler way of doing this without domain interfaces? Have you perhaps considered replacing the valve area with a porous medium with some resistance to flow (i.e. the valve resistance).

4. Please provide plenty of pictures.

Lastly if you invest so much time trying to do a work around from actually drawing and meshing the valve, it might be economical to just put up with it and draw/mesh the valve and the horizontal pipe in the first instance.

DarrenC · August 2, 2014, 13:59

Hi Ricochet,

Thank for replying

1. Is your setup correct? Is your reference pressure correctly set? Are all boundary conditions correct? Have you done a pen and paper calculation using simple Bernoulli principles to calculate the pressure drop in each section and hence have you a ball park figure of what the CFD solution should give you?

Yes i believe so. This is just a variation of a much simpler case (only gas phase), I have been running before with good results. Before this I was more interested in getting the macro flow quantities across the valve (i.e a certain mass flow for a certain pressure drop) right and not so much on the particulars of the flow inside the valve etc. Hence a pressure drop domain interface was fine. However since there is liquid holdup downstream of the valve now I believe a domain interface is not adequate anymore as it does not account for the size of the valve orifice.

2. Are you sure that the simulation is correct, and you're not wrong? I.e. what the simulation is showing you could be representative of what is actually happening. And what you thought would happen isn't actually happening!

Quite sure that the simulation is correct, but im still looking into it

3. Is there a simpler way of doing this without domain interfaces? Have you perhaps considered replacing the valve area with a porous medium with some resistance to flow (i.e. the valve resistance).

Hmm this might actually work. Haven't work with porous mediums before but might have a look at this. I guess a problem with using this i can forsee it will it be able to dynamically change the level of porosity, because remember I am actually modelling a control valve here.

4. Please provide plenty of pictures.

Dont have access to my CFD machine right now. Will upload some when I do. Stay Tuned!

Lastly if you invest so much time trying to do a work around from actually drawing and meshing the valve, it might be economical to just put up with it and draw/mesh the valve and the horizontal pipe in the first instance.

Well a few obstacles. We dont actually have the valve dimensions, just its performance characteristics. Also because its a control valve we will be dealing with moving mesh, which is a new area for me, so another hill to climb. Then i forsee that there will be spatial and temporal requirements to have the actual valve there (higher mesh count, finer mesh, smaller time steps) because what I have described so far is only a small part of a larger simulation I am running. But if this workaround is a dead end I might just have to bite the bullet and do it

Cheers
Darren

ghorrocks · August 3, 2014, 07:31

Another question on the fundamentals:

Is the flow on either side of the valve coupled together enough that they need to be run in the same simulation? If they can be decoupled (that is, run the simulation on the downstream side as a stand alone simulation with a boundary condition applied at the valve which mimics all the upstream stuff) then it is easier to apply a boundary which does what you describe - and the simulation is smaller and more tractable.

But if you need to do it as you describe which both side of the valve then just use a momentum source acting on the fluid phase which sets the fluid phase velocity at the valve interface to zero. That should stop back flow in the fluid phase.

DarrenC · August 6, 2014, 22:17

Thanks Glenn. Been busy with another project so this dropped out of the radar for a while. Yes the flow at that region is highly coupled. I think decoupling is doable but Ill have to be careful with it. Your other suggestion seems like a simple way to go around the problem. Will give that a go. Im just swamped with another project at the moment so itll be some time before i will know how this solution goes.

Cheers
Darren

August 1, 2014, 08:30	Inducing backflow restrictions in domain interface for multiphase flow	#1
DarrenC Member Join Date: Sep 2009 Posts: 69 Rep Power: 17	Hello Everyone I was just wondering if it is possible to implement backflow restrictions on one fluidic phase only in domain interfaces. My problem is like this. I have a horizontal pipe and a domain interface in the middle with additional interface model -> pressure change -> XX bar. The domain interface defined this way is supposed to mimic a differential pressure control valve holding the pressure drop at a specified value. Upstream of the domain interface is flowing gas but downstream of the interface is a liquid with a static head. In effect, what I am hoping to achieve is to mimic physically the blowing of the gas across the valve into the liquid phase, driven by the pressure drop induced across it. Perhaps rather unsurprisingly, because I have modelled the valve this way using a domain interface, the liquid phase is backflowing towards the gas phase. I thought of an idea of trying to restrict the backflow of the liquid phase through the interface to sort of mimic the fact that the valve opening is much smaller than the pipe diameter and hence will probably not backflow upstream of the valve, but im not sure if this is at all possible using domain interfaces. Has anybody tried to restrict the backflow of fluids through an interface, similar to how inlet and outlet boundaries restrict backflow by placing walls at a proportions of the inlet/outlets? If so can you give some pointers on how it was setup? Appreciate the help!! Cheers Darren

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
mass flow rate not conserved in turbomachine, interface defined wrong?	wildli	FLUENT	3	September 15, 2022 13:19
mass flow in is not equal to mass flow out	saii	CFX	12	March 19, 2018 06:21
Question about heat transfer coefficient setting for CFX	Anna Tian	CFX	1	June 16, 2013 07:28
Material identification between a flow domain and a porous domain	Ervideiro	CFX	1	June 1, 2010 10:50
fluid flow fundas	ram	Main CFD Forum	5	June 17, 2000 22:31

August 1, 2014, 08:47		#2
JuPa Senior Member Mr CFD Join Date: Jun 2012 Location: Britain Posts: 361 Rep Power: 15	I don't think you can physically restrict a phase from flowing back into a domain interface without the use of mass and momentum sinks. But before you dive into a complicated solution answer the following questions: 1. Is your setup correct? Is your reference pressure correctly set? Are all boundary conditions correct? Have you done a pen and paper calculation using simple Bernoulli principles to calculate the pressure drop in each section and hence have you a ball park figure of what the CFD solution should give you? 2. Are you sure that the simulation is correct, and you're not wrong? I.e. what the simulation is showing you could be representative of what is actually happening. And what you thought would happen isn't actually happening! 3. Is there a simpler way of doing this without domain interfaces? Have you perhaps considered replacing the valve area with a porous medium with some resistance to flow (i.e. the valve resistance). 4. Please provide plenty of pictures. Lastly if you invest so much time trying to do a work around from actually drawing and meshing the valve, it might be economical to just put up with it and draw/mesh the valve and the horizontal pipe in the first instance.

August 2, 2014, 13:59		#3
DarrenC Member Join Date: Sep 2009 Posts: 69 Rep Power: 17	Hi Ricochet, Thank for replying 1. Is your setup correct? Is your reference pressure correctly set? Are all boundary conditions correct? Have you done a pen and paper calculation using simple Bernoulli principles to calculate the pressure drop in each section and hence have you a ball park figure of what the CFD solution should give you? Yes i believe so. This is just a variation of a much simpler case (only gas phase), I have been running before with good results. Before this I was more interested in getting the macro flow quantities across the valve (i.e a certain mass flow for a certain pressure drop) right and not so much on the particulars of the flow inside the valve etc. Hence a pressure drop domain interface was fine. However since there is liquid holdup downstream of the valve now I believe a domain interface is not adequate anymore as it does not account for the size of the valve orifice. 2. Are you sure that the simulation is correct, and you're not wrong? I.e. what the simulation is showing you could be representative of what is actually happening. And what you thought would happen isn't actually happening! Quite sure that the simulation is correct, but im still looking into it 3. Is there a simpler way of doing this without domain interfaces? Have you perhaps considered replacing the valve area with a porous medium with some resistance to flow (i.e. the valve resistance). Hmm this might actually work. Haven't work with porous mediums before but might have a look at this. I guess a problem with using this i can forsee it will it be able to dynamically change the level of porosity, because remember I am actually modelling a control valve here. 4. Please provide plenty of pictures. Dont have access to my CFD machine right now. Will upload some when I do. Stay Tuned! Lastly if you invest so much time trying to do a work around from actually drawing and meshing the valve, it might be economical to just put up with it and draw/mesh the valve and the horizontal pipe in the first instance. Well a few obstacles. We dont actually have the valve dimensions, just its performance characteristics. Also because its a control valve we will be dealing with moving mesh, which is a new area for me, so another hill to climb. Then i forsee that there will be spatial and temporal requirements to have the actual valve there (higher mesh count, finer mesh, smaller time steps) because what I have described so far is only a small part of a larger simulation I am running. But if this workaround is a dead end I might just have to bite the bullet and do it Cheers Darren

August 3, 2014, 07:31		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,870 Rep Power: 144	Another question on the fundamentals: Is the flow on either side of the valve coupled together enough that they need to be run in the same simulation? If they can be decoupled (that is, run the simulation on the downstream side as a stand alone simulation with a boundary condition applied at the valve which mimics all the upstream stuff) then it is easier to apply a boundary which does what you describe - and the simulation is smaller and more tractable. But if you need to do it as you describe which both side of the valve then just use a momentum source acting on the fluid phase which sets the fluid phase velocity at the valve interface to zero. That should stop back flow in the fluid phase.

August 6, 2014, 22:17		#5
DarrenC Member Join Date: Sep 2009 Posts: 69 Rep Power: 17	Thanks Glenn. Been busy with another project so this dropped out of the radar for a while. Yes the flow at that region is highly coupled. I think decoupling is doable but Ill have to be careful with it. Your other suggestion seems like a simple way to go around the problem. Will give that a go. Im just swamped with another project at the moment so itll be some time before i will know how this solution goes. Cheers Darren