Analysis fom the OECD/NEA PWR Main Steam Line Break (MSLB) Benchmark exercise 3 with the Coupled Code System RELAP5/PANBOX

The main purpose of the computational OECD/NEA PWR MSLB-Benchmark is the evaluation of the prediction capability of advanced code systems by means of a code-to-code comparison. The postulated MSLB-transient is characterized by a strong non-symmetrical core thermal behaviour due to the feedback between neutron kinetics and plant thermal hydraulics. The analysis of such transients with pronounced spatial power distortion represents a considerable challenge for advanced code systems. It is initiated by a break of one main steam line when the reactor TMI-1 is operated at nominal power. High heat removal through the break leads to a strong cooldown rate of the broken loop compared to the intact one. Under such conditions a power increase and a re-criticality of the core despite scram can not be excluded due to the negative reactivity coefficients. The MSLB-Benchmark enfolds three exercises as follows: Exercise 1: integral plant simulation with best-estimate codes using the point kinetics, Exercise 2: multidimensional simulation of the core for given initial and boundary conditions, and Exercise 3: integral plant simulation with coupled, best-estimate codes using 3D-neutron kinetics models. Forschungszentrum Karlsruhe (FZK) and Framatome advanced nuclear power (ANP) Erlangen participated on the MSLB-Benchmark with the code system RELAP5/PANBOX for the Exercise 3: Based on the plant and core models elaborated for Exercise 1 and 2, an integral TMI-1 plant model was elaborated for Exercise 3. Special emphasis was put on the development of a multidimensional core model for the space-time kinetics. Two scenarios, the best-estimate (BE) and the return-to-power (RP) scenario, were investigated. Additional investigations aimed to investigate the influence of the coolant mixing on re-criticality and power increase. Results of these investigations are presented and discussed in this report. It has been demonstrated that RELAP5/PANBOX is capable to simulate complex transient in a reliable way compared to the point kinetics approach. (orig.)