1) If all you're looking for is thermal results (and nothing structural like displacement or stress) then you'll just need to have the proper thermal conductivity, specific heat, and density. You stated you cannot create your own material; is this because you get an error when you try to create your own material or that you cannot figure out how to do it?
2) Are you trying to simulate conduction through the air? If so then yes; you will need to add in a volume that represents that air and apply the correct material properties to it (also make sure to capture that thermal contact resistance correctly). Any surface that does not have a boundary condition applied to it (or a contact definition) is treated as adiabatic (i.e. no heat flux). However, if all you're trying to do is model surfaces being cooled by air, then just apply an appropriate convection boundary condition.
1) I believe I got the administrator rights and cleared the firrst problem up. It's going to take quite a bit to change all of the materials, though.
2) Does convection (not always but in this instance) rely on gravity and density? Because I was talking to someone I work with who said he would not use gravity for the simulation (and his SW history is much longer than mine) and if this is the case, convection would have no effect.
Convection always relies on gravity and density. The difference between conduction and convection is that convection has fluid flow due to density differentials. The density of the boundary layer of air decreases as the air heats up; gravity then causes the hotter air to rise because of the buoyancy effect. If your heat sink was in the International Space Station (where gravity is essentially zero) or if your fluid had a near zero change in density as it heated up, then there would be no convection (and instead you would have fluid conduction).
A convection boundary conditions is based on an approximation method for modeling true convection. This boundary conditions requires an ambient temperature definition (the temperature of the air) and a convection heat transfer coefficient (which is usually denoted by the symbol 'h'). Typically this approximation is reasonable if you have forced convection (like a fan blowing air onto a heat sink) and can be used for modeling natural convection if you're very careful with how to obtain 'h'.
For a linear static thermal analysis, it would be pointless to use gravity. However, this does not mean convection has no effect! What we have here is a difference between the physical system (the real world object you're trying to model) and the finite element system (what you create to model that real world object). The real world is all multi-physics and non-linear, and the analysis you're creating is single-physics and linear. Now, this doesn't mean that your analysis is meaningless; that is dependent on how good your assumptions are! I suspect the person you talked to simple meant that he would not add a gravity load into the model, not because convection was minimal, but rather because it wouldn't have an effect (since the model doesn't capture fluid flow).
If you can determine that there will be next to no air movement in the physical system (whether forced or natural), then it would be reasonable to model the air with a solid volume. However, if you determine that there will be air movement in the physical system (whether forced or natural), then you need to capture that accelerated cooling effect properly.
I think the reason why you got results that were in the range of thousands of degrees Celsius is because you have improperly defined boundary conditions or a units mistake.
Justin, Shaun's got it here. For thermal analysis - static, thermal conduction is important. Density and specific heat if you go transient. The rest if you go to thermal stress afterwards.
on question 2, same thing, convection coefficient is what you want. if you know the convection is dominated by natural convection (Gravity) use a low value. if it is fan, a high value. take a look at your heat transfer book for values. they vary based on the fluid and fluid velocity usually.
if you want a more accurate solution, IE just inputting the parameters of the fluid, you need to go with flow simulation.
A couple things from our website that might help: