Improvement of Subchannel Scale Analysis Capability of CUPID with Grid-directed Cross Flow and Fuel Rod Models

Abstract

As the recent computing environment increases, the whole core pin-by-pin analysis has been actively performed using coupled subchannel analysis code and neutron transport code. In particular, the revised safety standards, which emphasize realistic consideration of the fuel status in accident analysis including the thermal conductivity degradation, motivate the whole core pin-by-pin analysis to estimate safety margin more accurately.

In order to guarantee accurate prediction of the reactor thermal-hydraulic behaviors using a subchannel analysis, it is necessary to consider the effect of the mixing vane on fluid lateral transfer and therefore, subchannel analysis codes include a model to reproduce the mixing vane directed cross flow. However, the previous study on whole core calculation of APR1400 using CUPID did not consider the effect of mixing vane. For this reason, this study aims to extend the previous study by implementing grid-directed cross flow model as its first objective. Furthermore, the previous work did not include the fuel rod heat transfer model and it was required to implement the fuel model to make CUPID possess the features necessary for the reactor core subchannel analysis. The second objective of this study is, therefore, to improve the fuel rod conduction model and verify it by solving conceptual problems.

CTF (COBRA-TF) code uses the grid-directed cross flow model in the momentum equation to simulate the fluid transfer induced by the mixing vane of a spacer grid. It uses the staggered grid, which defines lateral velocities on the gaps between two subchannels. Accordingly, CTF could simulate the momentum exchange by a mixing vane directly as the location of the mixing vane is matched with the gap where the velocities are defined while CUPID uses collocated grid system and all velocity components are defined at the center of the subchannels. In this case, the momentum exchange by the mixing vane could not be well reproduced as the two opposing directional momentum caused by the mixing vane is canceled out and eventually, no momentum is added in the subchannel center. Therefore, to compensate the canceled momentum, two modifications were proposed: first, the grid-directed cross flow model was implemented into not only the momentum equation, but also into the scalar equations, mass and energy equations. The turbulent mixing coefficient β was replaced by β_default+β^' to take into account the additional vane-induced turbulent mixing. β^' was determined from code-to-code comparison with CUPID and CTF. In addition, the non-uniform effect of the mixing vane directed cross flow in the near of the guide tube and the pressure drop caused by the spacer grid were considered.

Meanwhile, the fuel rod heat conduction equation of CUPID was improved. The rod-centered approach of the default fuel rod model in CUPID was modified to a cell-centered approach for subchannel scale analysis, and the fuel rod was divided into four sections. Divided fuel rods face the subchannels which have different coolant temperature and velocity, so the cladding surface temperature could be different in a single rod. Thus, circumstantial heat conduction occurs across the interfaces among the quarter rods, and the temperature difference gradually decreased at the center of the rod. In this study, the circumferential heat conduction was considered explicitly in order to avoid solving a system of equations which increases computational time and memory usage. Furthermore, the Biasi correlation and CE-1 correlation were added as a CHF calculation model to observe the DNBR distribution.

After the implementation and improvement of the models, the verification of APR1400 single assembly was conducted. For simulation, the calculation results of nTRACER, a neutron transport code was applied as the volumetric heat source of the heat conduction equation. The implemented power distribution was from an assembly which has the larger volumetric heat source than surrounding. Afterward, the verification of the models was followed. It has been confirmed that grid-directed cross flow model made temperature distribution flattened and the fluid velocity stream line rotate. Furthermore, the transverse velocity increased at the support grid position. The code-to-code comparison between CUPID and CTF was performed and the result showed comparable liquid temperatures and axial velocities at the outlet. Afterwards, the fuel rod heat conduction model was verified. The pellet centerline temperature and the volumetric power had cosign-shape, whereas maximum cladding surface temperature and coolant temperature occurred at the outlet of the test section. In addition, the DNBR distribution using CE-1 correlation shows more conservative minimum DNBR than that obtained using Biasi correlation.

Then, the APR1400 whole core was simulated using the verified models. The subchannel type was specified for whole core simulation, and the power distribution was implemented from nTRACER as done in the single assembly calculation. For the whole core simulation which requires efficient memory handling scheme, CUPID parallel processing method was improved. Due to the grid-directed cross flow model, the below was observed

the changed coolant temperature distribution, non-uniform lateral velocity in the vicinity of the guide tube, and the increased flow velocity in the water gap which has higher hydraulic diameter. In addition, the fuel rod heat conduction model was activated for the simulation. For the assembly 23 used for the single assembly calculation, the simulation result of single assembly and the whole assembly were compared and possible reason for the discrepancy was discussed.