Heat transfer through gap in CFX?
 

I am analysing a shell and tube heat exchanger.
I want to use 'thin interface' option instead of using the 'tube' as solid domain.
So I removed the tubes, which creates gaps between the 2 fluid regions.
Closing the gap everywhere in the model is very time consuming & so I want to analyse with the gap.
I simple ran the simulation, with all other normal setting, to see what kind of error it shows up.
The solver manager shows error message which recommends to switch on "permit no intersection" in the advanced tab of interface.
I switched on that option (also made "true" for 'permit no intersection' option in expert parameter) & again ran the simulation.
This time, the solver completed normally.
BUT THERE IS NO HEAT TRANSFER BETWEEN THE 2 FLUID REGIONS!

So my doubts are
1. Is it really possible to simulate the Heat transfer through gap with some kind of settings?
2. If possible, then what is limit of this gap?

also some doubts on thin interface option

3. If we create thin interface, the thickness will be added to which side of the fluid mode.For example, if there is 2 co-eccentric cylindrical fluid domains and if we add a thin interface of 1mm, then this 1mm will be considered towards inner fluid domain or towards outer fluid domain.

hope i put the questions clearly!
please suggest your opinions on my doubts:)

There is two sections to the GGI interface: First it finds matching elements across the interface, then it couples the fluid conditions across the matching elements.

So if you have a gap then you better make sure the first step is working properly, that is that it correctly finds matching elements across the gap. Have a look at the overlap variable in CFD-Post to see if there are any areas which have not connected.

Then, when the mesh is connected you need to know how the GGI works. It simply couples the flow through the interface element face on one side of the interface with the matching interface element face on the other side. No account is made for a gap, the flows are directly coupled.

To answer your questions:
1) Yes.
2) You can increase the tolerance CFX uses to match up elements on either side of the interface. When you increase it too much you will start matching up the wrong elements.
3) There is no thickness to an interface, so no thickness needs to be added to anything. If you are using thermal resistance or something like that all it does is generate a thermal resistance between the two faces as a function of how thick you say the interface is. But the interface still has no physical thickness.

There is two sections to the GGI interface: First it finds matching elements across the interface, then it couples the fluid conditions across the matching elements.

So if you have a gap then you better make sure the first step is working properly, that is that it correctly finds matching elements across the gap. Have a look at the overlap variable in CFD-Post to see if there are any areas which have not connected.

Then, when the mesh is connected you need to know how the GGI works. It simply couples the flow through the interface element face on one side of the interface with the matching interface element face on the other side. No account is made for a gap, the flows are directly coupled.

To answer your questions:
1) Yes.
2) You can increase the tolerance CFX uses to match up elements on either side of the interface. When you increase it too much you will start matching up the wrong elements.
3) There is no thickness to an interface, so no thickness needs to be added to anything. If you are using thermal resistance or something like that all it does is generate a thermal resistance between the two faces as a function of how thick you say the interface is. But the interface still has no physical thickness.

@ghorrocks
Thanks for the reply :)
I have gone through the help file regarding GGI interface.
It seems that gap is permitted, but permitted to the maximum thickness of 'half the elements size' which is at the interface!

As Glenn suggested, you can also find the following in the documentation: