Experimental piles apart, gas-cooled graphite moderated reactors began with the plutonium producing Windscale pile, located in Cumbria in the North West of England at a site now called Sellafield. Because of the urgency with which plutonium was required for military purposes, a simple reactor design was chosen, i.e., an unpressurized core with a once-through coolant. The obvious choice of water as coolant; in use at the time in the USA, was rejected for the more densely populated UK, because of the potential risk of loss of coolant leading to fuel melt-down. Instead, atmospheric air was chosen as the coolant but, in order to remove the large quantity of heat, the aluminium cladding was made with longitudinal fins. A disadvantage of air is its exothermic reaction with graphite. Consequently when a particular Wigner energy release increased the graphite moderator temperature to a point where the heat of reaction overtook the cooling effect, the moderator went on fire. This happened in 1958, eight years after the pile went critical.
When the UK moved towards nuclear power generation at Calder Hall on the same site, it was natural at the design stage in 1952 to opt for gas-cooling again. Here the Magnox reactor used carbon dioxide as coolant, as did the subsequent Advanced Gas-Cooled Reactor. Carbon dioxide is much less reactive than air and its reaction can be inhibited by additives.
Helium is nonreactive and has been favored for high and very high temperature gas-cooled reactors, the forerunner being the European DRAGON experimental project at Winfrith in the South of England. The advantage of non-reactivity is counterbalanced by increased cost, increased leakage and problems of sticking of moving parts. Designs for full scale versions have been proposed in USA, Europe, Japan and the former USSR and considerable developmental work on fuel and core has been carried out. As well as having a graphite core, the fuel is usually in intimate contact with graphite in the form of a matrix of graphite coated spheres or as a pebble bed.
Gas-cooled reactors with secondary Containment, as is now standard, have advantages in safety over liquid-cooled reactors. Firstly, when loss of coolant or of coolant pressure occurs there is little change in reactivity and there is much smaller change in heat transfer. Natural convection is sufficient to prevent melting of the cladding.