couple things, i was equating the fuel cell to a gas tank, the cell holds the hydrogen which is syphoned off and used to power a generator, like how they propose to power cars with hydrogen.
second, not worried about melt downs. current technology for the most part uses fuels with negative thermal coefficients of reactivity. so if the coolant is shut off the reaction actually slows. i'm not sure if thats the case with HEU.
with regards to shielding, i concede the point at the moment because i'm not sure the amount required for the scale of reactor i'm thinking about.
as for cooling, yes for large reactors because of the amount of heat generated the cooling requirements are immense. i suspect however, although i'm not positive that it will scale with the reactors size, dramatically reducing the cooling needed for the scale i'm proposing.