Benchmark Study of the Accident at the Fukushima Daiichi Nuclear Power Plant (BSAF Project) – Phase 1 Summary Report

The Great East Japan earthquake occurred on March 11th 2011 at 14:46 (Japan time zone). At the onset of the earthquake the units at TEPCO’s Fukushima Daiichi Nuclear Power Station (NPS) were successfully shut-down. However, the subsequent tsunami led to a beyond design basis accident. Even though there is not yet direct evidence of core melt , it is believed that the units 1 to 3 experienced severe accidents involving core meltdown, deriving from the total or partial loss of the core cooling capabilities.

 

In 2012, following a proposal from Japan, the Committee on the Safety of Nuclear Installations (CSNI) decided to initiate a joint project to conduct a study of the accident progression for the Fukushima Daiichi NPS units 1-3 accident, called the Benchmark Study of the Accident at the Fukushima Daiichi Nuclear Power Station (BSAF). Sixteen organizations from eight countries (France, Germany, Korea, Russia, Spain, Switzerland, the United States, and Japan) joined the project. The objective of the first phase of the project was to focus on the first 6 days of the accident to:

 

The approach to calculations and results was divided into a common case and a best estimate phase. First, a common case analysis was conducted using an agreed set of boundary conditions based on engineering estimates. The common case identified the effects of main parameters (e.g. initial in-core water level, decay heat) on the prediction of the codes. Also, the large sensitivity of the common case results on the estimated boundary conditions resulted in inconsistencies between the calculations and the measured data. These inconsistencies were attributed to the large uncertainties in the functioning of several safety systems (e.g. RCIC, HPCI, PCV venting, external water injection). Therefore, a best estimate case phase followed where each participant could apply adjustments to the boundary conditions in order to reconcile and harmonize code predictions with actual observed events. The common case represented a way to rationalize the accident and a helpful starting point to develop best estimate scenarios of the accident.

 

This summary report presents the major technical results of the NEA BSAF project, beginning by listing the ascertained events of the Fukushima Daiichi accident and employed codes. The main part of the report deals with the discussion of the results with a brief description of the common case and eventually focusing on the best estimate results. In the discussion of the best estimate results it has been highlighted, as a typical result of SA codes comparison, that when the boundary conditions (e.g. geometry, input values for safety systems) are well known and fixed, all the codes provide comparable agreement of the thermalhydraulics phase, as well as the fuel temperature excursion phase. Agreement in this context refers to RCS pressure and water level transient, fuel temperature and hydrogen generation transients. It has been underlined however that differences are introduced once the core geometry is altered during the relocation process and in the attempt to stabilize the plant with external water injection and venting. The main differences have been attributed in particular to code modeling differences. An attempt to identify the influence of the employed models during core relocation has been performed during the BSAF meeting discussions. The main physical modeling uncertainties were identified regarding the RCS failure at high core temperatures (e.g. penetration failure or creep rupture), computation of the debris surface area once the core changes configuration, creation of possible paths for the debris to move from the core region to the lower head through the core lower structures and core plate, failure mechanisms of the lower head and leak/failure of the containment system. Due to these uncertainties, future work may be needed to include physics insights from new experimental and analytical activities. Also, differences exist during the plant stabilization due to uncertainty of the boundary conditions, in particular regarding venting and external water injection by the fire trucks for all the three units.

 

Regarding the actual progression of the accident in the three units and the expected current plant status, several common observations were made from the computations, such as the prediction of the isolation condenser operation and the ex-vessel scenario in Unit 1; torus room flooding, RCIC selfregulating operation and containment failure in Unit 2; and RCIC employment and first phase of HPCI operation in Unit 3. On the other hand uncertainties exist in the computations regarding the extent of core degradation in the units, the possibility of in-vessel or ex-vessel scenarios in Unit 2 and Unit 3, the effectiveness of alternative water injection in all the units and the actual S/C venting operation in Unit 3. Finally, the description of the debris location and composition that has been estimated from these analyses may help with the development of defueling and debris retrieval technology.