In this work, we assess uncertainties that can be expected in computational analyses of boiling water reactor reactivity initiated accidents with the SCANAIR V_7_5 computer program. The work is intended to supplement a similar uncertainty assessment, focussed on pressurized water reactor accidents, which is currently being organized as an international benchmark exercise by the OECD Nuclear Energy Agency. Here, we apply the same methodology as in the international benchmark.
The uncertainty assessment is made by use of Monte Carlo simulations, using an experiment from the Japanese Nuclear Safety Research Reactor as a reference case. This test, FK-1, simulates the conditions expected in reactivity initiated accidents occurring under boiling water reactor cold zero power conditions. Totally seventeen uncertain parameters are considered in our uncertainty and sensitivity analyses with SCANAIR V_7_5: seven parameters are related to the pre-test conditions of the FK-1 test rodlet, three parameters define the test conditions, and seven parameters are related to key thermo-mechanical models in SCANAIR. Calculations are done with both the standard single-phase and the optional two-phase coolant channel model in SCANAIR V_7_5.
By use of order statistics applied to the results from the 200 runs of the Monte Carlo simulations, the 5th and 95th percentiles are calculated for key output parameters versus time. The results show that the modelling of clad-to-water heat transfer beyond cladding dry-out is a major source of uncertainty in the performed analyses. In case cladding dry-out occurs, the models for clad-to-coolant heat transfer have a large impact on the calculated results, not only for cladding temperature, but also for cladding deformations and fuel rod internal gas pressure. This finding is in line with results obtained from the aforementioned benchmark, and it is believed that the conclusion holds also for reactivity initiated accidents occurring under boiling water reactor cold zero power conditions.
Our calculations show that SCANAIR V_7_5 reproduces the behaviour of the FK-1 test rodlet fairly well, provided that clad-to-coolant heat transfer models for the post dry-out regimes are calibrated to the results of this particular test. Measured data for fuel pellet and cladding tube deformations generally fall within the uncertainty band defined by the calculated 5th and 95th percentiles from our uncertainty analyses. Data for the fuel rod internal gas pressure, however, fall outside the calculated uncertainty band during the early part of the test.