Date of Award

Fall 2011

Degree Type


Degree Name

Master of Applied Science (MASc)


Engineering Physics


Adriaan Buijs


Guy Marleau



Committee Member

Guy Marleau


Due to the strong interest in thorium fuels in CANDU reactors rooting back to 1970’s and 1980’s, four experiments were done in the ZED-2 critical facility at Chalk River Laboratories to test the properties of (Th,Pu)O­2 fuel. The fuel was placed in five bundles with a typical CANDU design, stacked vertically in the center of the core (K0 site) and surrounded by natural uranium fuel.

The simulation of these experiments using the transport code DRAGON coupled with the diffusion code DONJON is presented. DRAGON is initially used to model two lattices and the full cores in 2D. These models are designed to calculate direct/adjoint flux, k and keff values using the collision probability method. Furthermore, the models determine a set of homogenized and condensed cross sections in two energy groups. Subsequently, DONJON is used to model the full core facility in three dimensions. Using the homogenized and condensed macroscopic cross section libraries obtained from the DRAGON models, DONJON is able to calculate the flux alongside with the keff values for the specific cases in two energy groups. The results are then compared to those from the experiments and will further validate the accuracy of the simulations.

Sensitivity and uncertainty results for the infinite lattices and the 2D full core model using DRAGON and SUSD3D code are discussed. The direct and adjoint flux values determined by DRAGON for the lattices and the full core model are used by SUSD3D to calculate the sensitivity profiles for specific reactions of the isotopes present. The sensitivity profiles are then used alongside with the covariance matrices to calculate the uncertainty contribution of nuclear data to criticality. The sensitivity and uncertainty (S/U) results of the 2D model for full core, using the DRAGON/SUSD3D code coupling are then compared with the S/U results of a 3D model of the full core using the code TSUNAMI within the SCALE6 package in reference. The comparisons will show an excellent degree of consistency between the two methods, while reasons for possible differences in the results are also presented.

McMaster University Library