I've studied this problem in some depth. From the still confusing description is *sounds* like you are going to use the gas/IR heater to heat the air, of a fluid bed. There is nothing wrong with this at all, but hot air is hot air. There is a question of how effectively the IR heater will heat air vs being absorbed and by the chamber. You *may* need to create a baffle coated with lampblack(or candle soot) to transfer the energy from IR heating element to air efficiently.
The problem with fluid bed is that the same hot air you are use to transfer heat is also used to "jostle" the beans. If the air velocity is very much below the terminal velocity of the beans, then they just sit on the lower grid. If the velocity is much above the terminal velocity they get pegged on the upper grid (or go out the flue). One common design approach here is to create a roasting chamber that widens toward the exit side (funnel shaped) or else only has a small/partial inlet and this causes a decrease in air velocity as the chamber cross-section increases. Without that sort of approach you have almost no adjustment range for the air velocity.
Incidentally the terminal velocity of a green coffee bean is about 3000 ft/min (34mph), and *maybe* a bit higher for peaberries. The Vt drops by about 25% as the beans increase in size and lose mass. So to calculate your fan req you need to provide ~3000ft/min times the cross section of the lower aperature (determines the cu.ft/min, cfm).
So aside from the constrained airflow rates fluid bed has a problem in that the hot air passes over the beans and is then lost. That's a very expensive loss of energy. According to "Coffee Technology" (Sivetz & Browne) you need only to apply 250 BTU / lb of greens for roasting. Large commercial units actually use ~460BTU/lb about 55% efficient. I've calculated that a stir-crazy is ~50% efficient and a popper (fluid bed) ~20% efficient. To increase efficiency you need to increase the bean/air contact time. You can't make the bean mass height very large or else you have too much pressure for the fan to operate and jostle the beans. Note the top diagram here and how it increases the bean-air contact time.
http://www.sivetzcoffee.com/Fluid%20bed.htmA big advantage of fluid bed is that you aren't recirculating the smokey air over the beans, which improves flavor, but OTOH you lose a lot of heat. One of the german companies produced a fluid bed where most of the exit air is re-routed to the combustion chamber where the combustion removes a lot of the smokey resin-y gunk. There used to be some better diagrams but ...
http://www.nepro-vortex.com/nepro.htm FWIW a 22000BTU/hr grill produces 367 BTU/minute so you'd probably like to get the system efficiency up around 40% so you can roast 4 lbs in under 7 minutes. If the efficiency was as low as a popcorn popper you're 4lb target would take almost 14 minutes - too long.
[silly units error corrected in re-edit - see posts below]
