Review of the Fukushima Accident Sequence for the Office for Nuclear Regulation

Jacobsen Analytics performed a review for the UK Office for Nuclear Regulation (ONR) which was aimed at providing a reference discussion of the events of the accident sequences that occurred at the three Fukushima units, as well as the code analyses performed in Japan to understand the sequences and the conclusions that could be drawn from these analyses.

The report prepared by Jacobsen contains:

  1. short overview of BWR containment and reactor systems,
  2. discussion of general features and phenomena expected in severe accident scenarios for BWR units with Mark I containments,
  3. summary of the timelines at Fukushima Daiichi units 1, 2 and 3, and
  4. discussion of the accident analyses performed by TEPCO and NISA using MAAP and MELCOR.

The code analyses presented in the Japanese report suggest there was some degree of core damage at Fukushima Daiichi Units 1, 2 and 3. It also appears (based on these analyses) that vessel failure probably occurred at Fukushima Daiichi Unit 1. For Units 2 and 3. Scenarios with and without vessel failure, subsequent to core meltdown, both appear plausible. The belief that core damage occurred is backed up by the indications of substantial hydrogen generation and combustion at Units 1 and 3. Hydrogen generation is indicative of substantial core degradation and damage; on the scale apparently seen at the Fukushima Daiichi units, hydrogen production would have arisen due to the oxidation of zircaloy at high temperature in a steam environment.

The report prepared by Jacobsen contained discussion of various aspects of the accident as well as plots comparing the measured data at the units to the predictions of the MAAP and MELCOR codes. These comparisons yielded a number of interesting insights. It can be seen that the MAAP and MELCOR analyses are able to reproduce many features of the accident progressions. There are also some differences in the results of the two codes (which may be due to the code models themselves or due to "user" effects: for example, for Unit 2, the two analyses differ noticeably in their calculation of the containment pressurisation. Differences are also seen in the predicted vessel failure times (but less so for core melt timings). Some aspects of the data are not well predicted by either code. For example, there is large increase in drywell pressure in the Unit 2 accident progression that is not well predicted: this may well be due to release of a large amount of (non-condensible) hydrogen that is not predicted by the codes.

Figure 1 presents plant data, MAAP predictions and MELCOR predictions for the drywell (primary containment) pressure at Fukushima Daiichi Unit 2.

Drywell pressure at unit 2

Figure 1: Drywell pressure for Unit 2 accident versus time. (Data taken from Japanese Report and cross-plotted- plant data v. MAAP v. MELCOR)

The performance of safety systems during the accident was also of interest during the preparation of the report. A review of the plant data and the long duration of the accident sequences confirms that the RCIC pump at Unit 2 and the HPCI and RCIC pumps at Unit 3 operated for a very long time. This is interesting since typically Probabilistic Risk Assessments and safety analyses have assumed that this equipment would only operate for around 8 hours in accident conditions (for numerous reasons). The actual running times at Fukushima were much longer. It is believed possible that the role of the operators at these units may have contributed to the long period of operation of these pumps.

Accident management strategies applied at Fuksuhima are also discussed in the report prepared for ONR. Accident management strategies applied at the units included:

  1. the reactors were manually depressurised at Unit 2 and Unit 3 (in the case of Unit 1 it is not known if the reactor was manually depressurised, or whether this may have occurred due to the failure of a valve in an open position),
  2. external freshwater and/or seawater injection was used at all three units, and
  3. the primary containment was vented at Unit 1 and Unit 3, and attempts were made to vent the Unit 2 primary containment but it is not clear if these were effective. A possible explanation for the hydrogen combustion and control problems seen at the Fukushima Daiichi units is that the loss of power may have lead to difficulties isolating the hard pipe vent from the soft pipe vent.

Some useful links are provided below. These include Jacobsen Analytics contributions to ONR's request for submissions of information on the Fukushima accidents. (We believe that our two contributions of information are very interesting to read in hindsight, since based on very sparse information available in March 2011 we were able to correctly identify many features of the accidents that occurred)

References

  1. ONR's final report on the Fukushima accidents. - see in particular Annex L of the report.
  2. Jacobsen Engineering first submission of information to ONR
  3. Jacobsen Engineering second submission of information to ONR

Note: At the time of the original submissions to ONR, the company name was Jacobsen Engineering Ltd