Aerial view of the Fukushima plant area in 1975, showing separation between Units 5 & 6 and the majority of the complex. (National Land Image Information color aerial photographs, Ministry of Land, Infrastructure, Transport, and Tourism.)
Jeff Cardoni (in the Severe Accident Analysis Dept.) presented the paper “MELCOR Simulations of the Severe Accident at the Fukushima 1F3 Reactor” at the 2012 ANS Winter Meeting and Nuclear Technology Expo, which held an embedded topical meeting called “International Meeting on Severe Accident Assessment and Management: Lessons Learned From Fukushima Daiichi.” on November 11–15, 2012, in San Diego, California, where it received the “Outstanding Paper Award.”
In response to the accident at the Fukushima Daiichi nuclear power station in Japan, the U.S. Nuclear Regulatory Commission and Department of Energy jointly sponsored an accident reconstruction study to assess the MELCOR code’s severe accident modeling capability. Project objectives included reconstruction of the accident progressions using computer models and accident data and validating the MELCOR code and the Fukushima models against plant data.
MELCOR appears well-suited for reproducing the thermal-hydraulic response of the severe accident at 1F3. Using the current Sandia MELCOR 2.1 model of 1F3, the code calculates reactor pressure vessel pressures and containment that agree reasonably well with the Tokyo Electric Power Company (TEPCO) data. Given rough agreement with the TEPCO data and the reactor building explosion, MELCOR predicts the degree of core damage in 1F3 and the associated radionuclide release to the environment.
In capturing the principal boundary conditions for the 1F3 simulations, it is apparent that both the quantity and timing of hydrogen generation are important. A rapid hydrogen generation rate increases containment pressures (with safety relief valves open) and allows the simulation of flammable conditions in the reactor building. Numerous alternatives can explain the explosion, even beyond the standby gas treatment system and ex-vessel scenarios that have not yet been successfully modeled. However, there is significant forensic value in investigating each plausible scenario individually. The MELCOR code can be used to identify which scenarios are possible (or even likely), assuming the current 1F3 model and the TEPCO data are in reasonable agreement.
The Fukushima Daiichi accidents highlight some potential areas for MELCOR code improvement. Most importantly, the MELCOR models for lower head failure may need to be extended to provide higher-fidelity simulations of scenarios involving gradual melt-through of vessel penetrations; in such scenarios, the use of the gross-failure model, even in conjunction with the simple penetration failure model, may never predict vessel rupture and may therefore be too optimistic.
Severe accident codes such as MELCOR provide a valuable means of characterizing severe nuclear accidents. Uncertainty and variability ensure that such analyses will render wide variations in predicted accident progressions, and this is a reflection of true variability in the potential responses of complex systems undergoing severe accident conditions.
This paper’s authors included Jeff Cardoni, Randall Gauntt, Donald Kalinich, and Jesse Phillips (all in the Severe Accident Analysis Dept.).