Dynamic simulation of valve
 

I want to simulate a marine breakaway coupling which involves a 3D petal like structure, similar to shown here:https://www.youtube.com/watch?v=NARNaiXCqCQ

This coupling is connected between the hoses in offshore fuel transfer system and the hoses are separated when it gets stretched (e.g. if ship floats away because of tides etc) and seals the fuel in the hoses from spilling. Smaller closing time means less spillage. But it also means higher pressure surge experienced by upstream components, and there exists a threshold pressure beyond which pressure should not increase to safeguard upstream components. Objective is to find out minimum time of closing such that maximum pressure during surge still remains below threshold value.

But I thought of first simulating a rectangular channel with a simple valve closing as shown in the image below:
http://imgh.us/valve.png

Any pointers regarding how I go about it?

If you are trying to find the maximum pressure surge for a given closing time then I would not model the closing action of the valve at all. In fact I would not do this with CFD either. I would write a simple 1D solver initialised with a given flow and then slow the flow at the rate you define at one end and see the pressure transients it generates. CFD would appear to be an overkill for this application.

Also: There is probably analytical solutions for this as well, then you won't event need to do a 1D solver.

Hi Glenn, thanks for the reply.

Yes there are some analytical formulae (Joukowsky equation etc) that can be employed if the movement of petal were near-instantaneous and uniform. But here if you watch the video, of the 8 petals, 4 alternate petals close more rapidly and rest 4 slowly. Classical formulae don't seem to yield the results seen in testing of the coupling so we build a healthy markup in the specs for the pressure surge. There is a leakage of the flow through the gaps near rim of petal arrangement (until all of them are touching in final seal position) that needs to be accounted for as well. If what is happening inside and upstream of the valve is understood in detail and if a method to see the effect of different factors (e.g. no. of petals, their shape/orientation, closing sequence and its speed etc) is developed, it will help improve the valve design. Making prototypes of all these cases and testing them is prohibitively expensive. Hence I wanted to first model a simple case and then use this methodology to model the 3D case in CFD.

I had a read through CFX documentation and rigid body motion seems to be a way to do this?

So the purpose of the simulation is to understand and optimise the valve design? I understood from your first post that you were trying to see if it was going to cause excessive pressure spikes in the system.

If you want to model the valve system in detail, then a rigid body simulation would appear to be the way to go. You cannot do contact modelling in rigid body simulations (ie the petals stop when the hit each other) but in this case putting in a hard-coded maximum rotation should be close enough for most applications.

There are some examples of rigid body simulations in the CFX tutorials. They are available on the ANSYS Customer webpage.

The Joukowsky equation can model an effectively instantly closing valve, but you can easily model a valve with an arbitrary closing speed using the 1D approach I described. But I now understand this is not what you are trying to do so it is not relevant.

Yes Glenn, the objective of the exercise is to feel the physics behind the valve action and subsequently optimise the design. 1D simulation is useful and infact there are standard 1D softwares such as VariSim that are widely used in the industry. However they don't give us pointers for improving the valve design. I will have a look at simulating this with rigid body solver and will chime back if I have any queries.