CAN A DOUBLE BEAM MODEL OF FUEL ASSEMBLIES IMPROVE EVALUATIONS OF THE SEISMIC BEHAVIOR OF A PWR CORE ?

摘要

Analysis of the seismic behavior of a standard core of arnPressurized Water Reactor aims primarily to evaluate thernintegrity of the mixing grids and of the fuel rods.rnFor the purpose of such a mechanical analysis, the fuelrnassemblies are usually assumed to be vertical beams, supportedrnat both ends. Non-linear calculations are performed tornevaluate the impacts between assemblies, at the level of thernmixing grids.rnThis paper proposes an improved approach, recentlyrnimplemented in the CLASH 2.00 computer program, for thernpurpose of evaluating the seismic behavior of PWR cores andrnspent fuel racks for wet storage .rnThis method represents each fuel assembly as a pair of beams:rn· The first represents the skeleton, i.e. the guide tubes, therngrids, the head and the foot. It is relatively rigid andrnlightweight.rn· The second beam models the fuel rods only.rnIts mass is large but its rigidity is comparatively small.rnSprings link the two beams at the level of each mixing grid.rnNon-linear relationships describe these springs by relatingrntheir compression force to average depression.rnFor small deformations, the relationship reflects the totalrnstiffness of the actual springs, at a given grid level.For larger deformations, it can simulate the permanentrndeformations or even the collapse of the springs.rnThis Double Beam Model was first tested on several simplernbenchmarks, to qualify the code according to ISO 9001rnstandards.rnThe model was then compared to the classical Single BeamrnModel.rnThe sensitivity of the seismic response to burn-up could bernexamined only with the Double Beam Model. A fresh fuelrncase took into account a nominal spring pressure on the rods.rnIt was compared with a case featuring totally flat springs, i.e.rnthe situation when the burn-up is significant.rnAll the non-linear calculations presented are based onrnartificial accelerograms. These are randomly generated fromrna relevant response spectrum.rnThe results of a given simulation depend, to a large extent, onrnthe numeric trigger used to randomly generate the artificialrnaccelerograms. Each calculation case was thus evaluated withrn30 different accelerograms, to minimize the uncertainties inrnthe conclusions drawn from this preliminary investigation.