Modeling reactivity devices for advanced CANDU reactors using the code DRAGON

Full core analysis of typical power reactors generally performed uses few group diffusion theory, it is necessary to generate beforehand, using a lattice code, the required few group cross-sections and diffusion coefficients associated with each region in the core.

For the ACR? (Advanced CANDU Reactor), the problem is more complex because these reactors contain vertical reactivity devices that are located between two horizontal fuel bundles. The usual calculation scheme relies in this case on a 2D fuel cell calculation to generate the few group fuel properties and on a 3D supercell calculation for the analysis of the reactivity devices present in the core. Because of its complexity, the supercell calculations have usually been performed using simplified fuel geometries. The development of new geometry features in DRAGON and the availability of faster computers have made it possible to improve the 2D cell and 3D supercell models by using explicitly 3D assemblies of clusters to simulate the reactivity devices in CANDU reactors, including the ACR. These studies will thus improve the fine reactor core results by generating more accurate and appropriate reactor databases.

In this paper, we will review the lattice-cell/supercell calculation procedure using the code DRAGON by introducing a new supercell model. The use of such an explicit 3D geometry implies a very fine spatial mesh discretization that can generate a large number of regions leading to problems that cannot be solved by the collision probability (CP) method. The method of characteristics (MoC) is then the only alternative for such cases. A comparison of results using these two methods will also be presented for 3D models with a coarse mesh discretization.