1- The answer to this question depends on the physical phenomenon of the induction heating process. Do you have any more details about this process to give us?
2- I would start the model with an ideal situation, where you have perfect surface contact with the two material. This is the simplest model, just you create your CAD geometry with the two parts sharing the same dimensions at this contact. In reality of course this will not be so nice, you will either have a metal to metal contact with it's associated thermal resistance, or you can use a thermal interface (such as the paste used for heatsinks) which you can model as a contact resistance in the software.
Thanks Amit for your advice with the contact issue.
With respect to the induction heating process:
1- The heating element (coil) which is made of a conductive material (as copper) will be inserted inside the hollow cylinder;
2- As the current starts flowing inside the coil, a magnetic field will be developed inside and around the coil. When using an alternating current the coil will produce a magnetic field which changes its direction at the same rate of the frequency of the alternating current. When a conductive material is placed within the range of the alternating magnetic field, a voltage will induce in the work-piece; this voltage will results in a flow of electrons (current) which also changing its direction. The resistance of the workpiece to the flow of the electrons will develop the heat .
This is just a quick and simplified idea about the physics of induction heating; I do not like to make coupling between the magnetic and thermal parameters in my simulation. The induction system has a certain power output for example 20 KW, I would like only to represent this value of power in the simulation. I have tried the "heat source" option for the cylinder internal surface with a heat generation rate of 20 KW. Do you think this is sufficient or a right way of representation!
If the coil is close to the surface of the cylinder then I use use surface heat. If it is well inside the material, then I would use volume source. Either way I do not believe it will affect your simulation too much, since it sounds like all the heat will be conducted away into the water anyways.