Small, modular Uranium reactor is more practical
Although this is not widely appreciated outside the nuclear engineering community, a "small reactor" (small volume) application like a steel plant or tar-sands oil well heat source strongly favors the neutron physics of a conventional enriched uranium reactor or a liquid-metal uranium/plutonium reactor rather than a thorium thermal breeder reactor. To use thorium effectively, the core has to be large enough to recapture at least 89% of the neutrons emitted from fission-- 40% just to keep the reaction going, another 45% or so to convert enough Th232 to U233, and an extra 4% margin to make up for the unavoidable loss of fuel from the undesired neutron capture reaction (n + U233 > U234). It can be shown from neutron diffusion models that the minimum core size for a high-temperature thorium reactor is at least 3.5 meters in diameter (including the graphite neutron reflector). A reactor of that size, when the shielding and supporting structures are added, barely qualifies as "small modular." Efficient enriched uranium cores and fast neutron liquid metal uranium breeder cores can be made to operate in much smaller volumes, which would be much more portable or modular.
Therefore, while I think that using small modular nuclear reactors is a great idea to replace fossil-fuel heat sources for industrial processes; and I think that thorium breeder reactors are a great idea to replace coal-fired power plants in the near future-- I don't think that those two applications are solved by the same nuclear reactor. We should be developing medium-sized to large (300 to 1000 megawatt thermal) thorium breeder reactors for electricity generation and hydrogen fuels production, and developing small modular uranium reactors for industrial applications. It is a simple matter of neutron diffusion physics.
