Generalized perturbation theory and stochastics algorithms applied to improve CANDU6 reflectors

Generalized perturbation theory (GPT) is applied to investigate the response of the reactor to reflector modification in CANDU6 reactor. At equilibrium refueling, GPT is used to calculate the variational power values in the 14 reactor control zones. New mixtures composed of steel and heavy water are then introduced in the reflector to flatten the power distribution. Taboo search combined with the genetic algorithms yield to the replacement of a pure heavy water reflector by a composite reflector which could be deployed to improve power distinct.

► We use generalized perturbation theory to evaluate perturbation of the reflector.
► We combined genetic algorithms with the taboo search to find the combination of the reflector.
► As result, we improved the power distribution and we saved more the 36% of total heavy water.