Coolant void reactivity in a CANDU: A 2-D transport/diffusion comparison

We have analyzed using diffusion and transport theory methods an axially infinite 2-D representation of a CANDU reactor fueled with 37 element bundles. Our studies indicate that a relatively fine mesh is required to ensure that the reactor model is spatially converged for both diffusion and transport calculations. The effect of using the more physical annular boundary condition in transport is very small while it leads to much higher errors in diffusion. The power distribution computed using diffusion theory are more peaked towards the core center than their transport counterpart with differences reaching 6 % near the core/reflector interface. When full cell homogenization is considered, the differences in the coolant void reactivity between a transport and a diffusion calculation show that the diffusion theory simulates relatively well this effect. On the other hand, using the pin model in transport yields much larger differences in coolant void reactivity. (author)