Simulation of QUENCH-06 Experiment by MELCOR v2.2 with Uncertainty Analysis

Deterministic integral codes, such as MELCOR and ASTEC, have been developed to predict and characterize severe accident progression in nuclear power plants. Due to the complexity and the mutual interaction of several physical phenomena occurring in severe accident scenarios, the validation of these codes is fundamental. Moreover, considering the limited experimental database in prototypical conditions, sensitivity analyses and quantification of code uncertainties should be carried out.

The present paper describes the assessment of a MELCOR v2.2 input deck of the QUENCH test facility, located at Karlsruhe Institute of Technology, and the QUENCH-06 test, which was selected for code validation. This experiment was aimed at evaluating the effect of subcooled water injection on the hydrogen production and degradation of a pre-oxide pressurized water reactor–like rod bundle.

Having as reference past QUENCH analyses that are available as public technical literature and the experimental data, the authors developed an input deck exploiting several configurations and code features. The nodalization provides a fine representation of the test bundle active region, as well as a detailed definition of the boundary conditions and of the thermal insulation system.

Validation was performed by evaluating the accuracy of the code by comparing, both qualitatively and quantitatively, the MELCOR results for some relevant parameters (such as hydrogen generation, maximum cladding temperature, and oxide scale of the rods) against the experimental data. The results showed that MELCOR can reproduce the experimental data of hydrogen production and cladding oxide thickness in the instrumented bundle positions.

In addition, the study included a sensitivity analysis to test the behavior of different Zircaloy-steam oxidation correlations in the temperature range 1100 to 2200 K, assessing their impact on hydrogen production and degradation predictions. Finally, an uncertainty analysis was carried out to evaluate the dispersion of the figures of merit involved in the simulated phenomenology.