How a PHWR CANDU reactor works

Instead of a large vessel, the CANDU has a calandria: a thin-walled horizontal stainless steel cylinder about 8 meters in diameter, crossed from side to side by 380 to 480 pressure tubes of zirconium-niobium alloy. Each tube is about 6 meters long and holds 12 half-meter fuel bundles.

The pressure is carried not by the calandria but by each tube, with walls barely 4 mm thick. That is why no 20 cm steel vessel needs to be forged: the calandria works at almost atmospheric pressure. A concentric calandria tube, with a CO₂ gas gap, insulates each hot tube from the cold moderator.

The calandria is filled with heavy water, D₂O, at about 70 °C. Deuterium slows neutrons almost as well as hydrogen but absorbs them 600 times less: there are neutrons to spare to keep the chain reaction going with natural uranium, 0.7% U-235, unenriched. It is the design's reason for being, conceived in Canada when enrichment was a military secret.

That cold moderator has another virtue: it is a large heat sink independent of the primary circuit. And since it is separate from the coolant, reactivity control is done from there: absorber rods, adjustable light water compartments and, in an emergency, poison injected directly into the moderator.

Inside the pressure tubes flows another mass of heavy water, this time at about 100 bar, entering at 265 °C and leaving at 310 °C without quite boiling. Each tube connects at both ends, through individual pipes called feeders, to headers that gather the flow of hundreds of channels.

Neighboring channels flow in opposite directions, so the circuit traces a figure eight: two pumps and two steam generators in each half. A 700 MWe plant holds about 500 tonnes of heavy water, which costs hundreds of dollars per kilogram; recovering it and keeping it pure is part of the craft of running a CANDU.

On the secondary side everything is ordinary light water: the steam generators are the same as a PWR's, with thousands of U-tubes, and produce steam at about 47 bar and 260 °C, somewhat less than a PWR because the primary is cooler. The steam drives a conventional turbine and is condensed with water from a lake, a reservoir or the sea.

Thermal efficiency is around 30%, a little below the PWR. In exchange the fuel is cheap and abundant: a CANDU uses less mined uranium per kWh than any light water reactor, and can burn uranium recovered from other reactors or thorium.

With natural uranium the reactivity reserve is minimal, so fuel must be replaced continuously rather than every 18 months. Two fueling machines, one on each face of the reactor, latch onto the same channel with the reactor at full power: one pushes in new bundles and the other receives the spent ones, without opening the circuit or losing pressure.

About 15 bundles are changed per day. That is why CANDUs have such high load factors, and also why the international community watches them closely: because the fuel comes out a little at a time and at low burnup, IAEA safeguards count every bundle that goes in and out.

Two independent shutdown systems, with different logic, sensors and physical principles: SDS1 drops 28 cadmium rods from above in under 2 seconds; SDS2 injects gadolinium nitrate at high pressure directly into the moderator. Either one shuts the reactor down on its own. The first CANDUs, such as Pickering, also used fast moderator dump into a lower tank: without moderator there is no reaction.

If cooling failed, the moderator and the shield water surrounding the calandria absorb the decay heat and delay any damage by hours. The known weak point is the positive void coefficient: if the heavy water in the tubes boils, power rises. The shutdown systems are designed precisely to beat that effect.