RISAFloor does not give you a way to apply thermal loads to members.  It's mostly geared towards vertical loads (Dead, live, snow, et cetera).  

RISA-3D can certainly apply thermal loads (to members or plates).  The question really then becomes how do you choose to model your composite beam in RISA-3D.  Along with what type of thermal load you are applying.  
Is this just a uniform increase (or decrease) in temperature?  In which case this would mostly result in axial forces or stresses developing in the system.  Or, is there a change in temperatured from the top of the composite slab to the bottom of the steel beam or something along those lines?  In which case there would be a good amount of moment created from the changes in temperature.   

This is a uniform change in temperature, creating axial forces in the members.

It should be much easier to handle that type of thermal load.... RISA-3D has the ability to apply that type of loading to both plates and beams.  So, once you've got your composite beams built in RISA-3D, it should be an easy matter of applying the thermal load to each of the elements.  For beams this is done as a distributed load with a direction of "T".  For plates, this is done as a plate surface load with a direction listed as "T".   

The problem with that is the member's composite section properties do not get transfered into RISA-3D, so, the code checks are performed on a non-composite member.

You'd have to do this to a member that was created in RISA-3D using the technique for modeling composite beams described in the Modeling Tips section of our help file.  That's not a perfect solution, but it is probably the best you're going to get.  And, hopefully, you only have to run this type of analysis on a couple of beams.   

That rather defeats the purpose of using RISAFloor in this case if one has to re-model the composite members and the floor diaphragm in RISA-3D.

RISAFloor calculates the member's composite section properties, so, why not transfer that information into RISA-3D?

Composite beams are complex.  

What should RISA-3D use as the "effective" area of the beam for axial stiffness?  Should this area be different for tension vs. compression?  

Then temperature loads add an extra level of complexity.  Part of the area is due to concrete, part of it is for steel.  What sort of thermal expansion coefficient should you use? Clearly, it would have to be weighted in some way based on the area of steel vs. the area of the concrete, but what method would you use?

What's worse is that because of the different expansion coefficients of the two materials, you will get differential expansion of the concrete and steel. This would induce a bending moment into the beam, wouldn't it?  I cannot think of a reasonable way to model this short of building the composite beam out of plate elments and such per the Modeling Tips section I pointed out in my previous response.  

In the end, RISAFloor is really designed to handle gravity type loading on typical commercial buildings.  It's a great tool for that and can be expanded out into some other building sectors as well.  But, the more you lean towards specialty loading (like thermal loads or bridge loading) the more you're getting away from that program's strong points and the more limitations you will encounter.