Development of a Pin Level Thermal/Hydraulics-Neutronics Coupled Core Simulator for High Fidelity Steady-state and Transient Analyses

Abstract

A pin level reactor core thermal-hydraulics (T/H) code capable of massively parallel execution is developed and coupled with a transport direct whole core calculation (DWCC) code and with a pin-by-pin SP3 code for both steady-state and transient neutronics-T/H analyses. The Efficient Simulator of COre Thermal hydraulics (ESCOT) employs the 4-equation drift-flux model for two phase calculations while the numerical solution is obtained by applying the finite volume method and the Semi-Implicit Method for Pressure Linked Equation Consistent (SIMPLEC) algorithm. Important constitutive models to describe key subchannel phenomena, such as turbulent mixing, pressure drops, vapor generation, liquid-vapor interfacial heat transfer and wall heat transfer, are implemented to ensure the validity of subchannel-scale analyses. The ESCOT code solutions are validated through the simulation of various experiments and the comparison between the predicted quantities. The solutions are assessed also by the comparison with the corresponding results of the other subchannel-scale solvers like COBRA-TF, MATRA and/or CUPID. ESCOT has been successfully employed in steady-state transport DWCC analyses by coupling it with the nTRACER code through a wrapping system. The general coupling technique based on the Picard fixed-point iteration (FPI) has low robustness and the application of relaxation factors leaves too much freedom to the user. Thus, the application of the Anderson Acceleration (AA) as an effort to improve the stability of coupled steady-state calculations is analyzed through a series of 3-dimensional problems solved with increasing complexity starting from a single assembly steady-state problem to a full core depletion problem via checker-board (CB) problems. The convergence behavior is examined in terms of true error reduction by comparing the intermediate fission source distributions with the fully converged reference solution obtained by applying a very tight convergence criterion. It turns out that the number of neutronics-T/H iterations is reduced considerably because the oscillatory behaviors of the local solutions noted in the ordinary FPI can be smoothened. Convergence is reached earlier with AA so that the computing times of the coupled calculations can be reduced by about 25% retaining the solution accuracy. In addition, the improvements in both accuracy and details of the time-dependent coupled analyses are shown through the solution of the Main Steam Line Break (MSLB) accident. This scenario involves a considerable reduction of the inlet coolant temperature of one side of the reactor core which results in significant asymmetry in the radial flow characteristics. Because of this asymmetry, the positive reactivity feedback effect introduced by the decrease of the coolant temperature occurs with strong spatial dependence. For sufficient conservatism, a stuck rod in the cold side is assumed during the reactor trip. Thus, employing the pin level solvers increases the fidelity of the calculated results. Despite the increased performances of transport transient solvers, the computing time is still a burden for the calculation of transients lasting longer than 20sec in simulation time. Therefore, ESCOT has been coupled with a pin-by-pin SP3 based code instead of a DWCC code. The analysis of the Nuclear Energy Agency of the Organization for Economic Cooperation and Development MSLB benchmark is performed by solving the Exercise II problem which does not require system modeling since it provides two sets of core flow boundary conditions. It turns out that the better neutronics and T/H nodalization of the core leads to a higher SCRAM worth which implies a lower maximum return-to-power when it is compared with assembly-wise solvers (< 2%). It is noted also that the mixing effect between the hot and cold sides is constrained only to the first assembly row and the size of the mixing region increases with the core axial level. A dominant axial velocity and a CB-like power shape around the separation between the two sides are the primary reasons for the lack of mixing beyond the first assembly row. Moreover, the better T/H nodalization describes more reliably the coolant behavior around the stuck rod. The use of a pin-by-pin solver allows also to capture the high gradient in pin power inside the assemblies close to the stuck rod at the instance of maximum return-to-power, which was not possible with the conventional assembly-wise solvers. The pin-level coupled neutronics-T/H does not increase the computing time noticeably owing to the parallelized execution capability. This study demonstrates the importance of advancing to pin-wise coupled transient analyses in order to fully understand the core power and temperature behaviors in the severe conditions involving highly distorted flow and power distributions.