HELIOS/DRAGON/NESTLE codes' simulation of the Gentilly-2 loss of class 4 power event

A loss of electrical power occurred at Gentilly-2 in September of 1995 while the station was operating at full power. There was an unexpectedly rapid core power increase initiated by the drainage of the zone controllers and accelerated by coolant boiling. The core transient was terminated by Shutdown System No 1 (SDS1) tripping when the out-of-core ion chambers exceeded the 10%/sec high rate of power increase trip setpoint at 1.29 sec. This resulted in the station automatically shutting down within 2 sec of event initiation. In the first 2 sec, 26 of the 58 SDS1 and SDS2 in-core flux detectors reached there overpower trip (ROPT) setpoints. The peak reactor power reached approximately 110%FP. Reference 1 presented detailed results of the simulations performed with coupled thermalhydraulics and 3D neutron kinetics codes, SOPHT-G2 and the CERBERUS module of RFSP, and the various adjustments of these codes and plant representation that were needed to obtain the neutronic response observed in 1995. The purposes of this paper are to contrast a simulation prediction of the peak prompt core thermal power transient versus experimental estimate, and to note the impact of spatial discretization approach utilized on the prompt core thermal power transient and the channel power distribution as a function of time. In addition, adequacy of the time-step sizes employed and sensitivity to core's transient thermal-hydraulics conditions are studied. The work presented in this paper has been performed as part of a project sponsored by the Canadian Nuclear Safety Commission (CNSC). The purpose of the project was to gather information and assess the accuracy of best estimate methods using calculation methods and codes developed independently from the CANDU industry. The simulation of the accident was completed using the NESTLE core simulator, employing cross sections generated by the HELIOS lattice physics code, and incremental cross sections generated by the DRAGON lattice physics code [4] based upon HELIOS generated cross sections. Core thermal-hydraulic conditions, zone controller levels, and control device positions as a function of time are all taken from the CERBERUS/SOPHT-G2 simulation of this event. To address sensitivity to spatial discretization treatment, NESTLE core simulator predictions based upon the finite difference method (FDM), nodal expansion method (NM) without utilizing assembly discontinuity factors (ADF), and NM with utilizing ADF will be contrasted. Note that NESTLE Version 5 has the option to solve the two-group neutron diffusion equation via the nodal expansion method, employing a quartic polynomial flux expansion and quadratic transverse leakage representation in solving the 1-D transverse integrated diffusion equation. (author)