Optimization of the Nodalization of Nuclear System Thermal-Hydraulic Code Applied on Primary Loop Benchmark

In best estimate plus uncertainty approach for thermal-hydraulic simulation in nuclear engineering, a crucial step for the qualification of the scenario simulation is the discretization, i.e., the nodalization of nuclear power plants and related integral test facilities (ITFs). Since intermediate break loss-of-coolant accident (IBLOCA) simulation is getting more and more attention in this decade, we focused on the nodalization of an IBLOCA scenario—a primary loop (PKL) I2.2 benchmark delivered by the organization for economic cooperation and development PKL4-project—using the analyses of thermal-hydraulics for leaks and transients (ATHLET) code. This work followed mainly the nodalization methodology of Petruzzi and D'Auria, including both qualitative and quantitative criteria, being divided into three phases for component volume, steady-state, and transient, respectively. The authors used also some specific approaches: (1) for component volume qualification, a volume fractional parameter was introduced, considering not only the relative error of each component but also the volume fraction in the whole system (an 0.2% acceptability level was chosen for this parameter); (2) the experimental data were not used directly as a reference within the nodalization procedure but the calculated results delivered by the most refined nodalization. Based on the estimator of average amplitude in the fast Fourier transform-based method (FFTBM), the convergence, rationality, and an optimized result of nodalization in the simulation of an actual IBLOCA transient benchmark have been judged. After three phases of nodalization qualification, it has been proved that the final nodalization has the necessary degree of convergence for a good reproduction of the benchmark geometry, allowing the proper simulation of involved phenomena. Finally, a middle-refined nodalization was found as being optimal, fulfilling the convergence criteria with a reasonable central processing unit time consumption. The nodalization scheme in this work was not seen as being the single factor influencing the simulated results, but just as a prerequisite to allow further reliable improvements on the models used by ATHLET (aspects not referred to in this particular study). Therefore, the simulated results presented here will match the experimental ones only as general trends; improvements may be further achieved by using new and more precise models (e.g., for critical mass flow, heat transfer, countercurrent flow, etc.) in the system thermal-hydraulic code.