Appendix G
Hydrogen Control in Severe Accidents
This appendix describes regulatory actions to control hydrogen in nuclear plants since the 1979 Three Mile Island accident and why they were insufficient to prevent hydrogen explosions in the Fukushima Daiichi plant.
G.1 REGULATORY ACTIONS FOLLOWING THE THREE MILE ISLAND ACCIDENT
Immediately following the 1979 Three Mile Island accident, the USNRC established a lessons-learned task force to identify and evaluate safety concerns arising from the accident and recommend appropriate changes to licensing requirements and licensing processes for nuclear power plants. The task force made a number of recommendations (USNRC, 1980e), including two recommendations for controlling hydrogen produced by severe core accidents:
1. Provide inerting for all Mark I and Mark II BWR containments.
2. Provide the capability to add a hydrogen recombiner system (for hydrogen control) within a few days after an accident.
The inerting requirement was implemented in December 1981 as the first interim hydrogen rule for Mark I and Mark II reactors.1 Plants that
_________________
1 This resulted in an amendment to 10 CFR § 50.44 requiring inerted atmospheres in BWR Mark I and Mark II containments.
did not already have inerting systems were required to install them, and new plants were required to be equipped with hydrogen inerting systems.
These systems were used to displace air inside the containment with nitrogen to reduce oxygen concentration below 4 percent when the reactor was operating. This change was adopted worldwide, including at the Fukushima Daiichi plant. It has been widely accepted in the nuclear power and combustion communities that inerting resolved the hydrogen issue for plants with Mark I and Mark II containments (USNRC, 1987).
Hydrogen control and equipment survivability became important considerations in other containment designs (PWR plants with ice condenser containments and BWR plants with Mark III containments) that were coming online in the 1980s. In 1985, a rule required that plants having these containments must control combustible gas generated by up to 75 percent metal–water reaction to less than 10 percent hydrogen. New reactor designs were required to consider up to 100 percent metal–water reaction.
Three unresolved (generic) safety issues arose from the Three Mile Island Action Plan and subsequent research on hydrogen combustion inside containments:
• GSI-A48: Hydrogen Control Measures and Effects of Hydrogen Burns on Safety Equipment. Initiated by Three Mile Island Task Force findings and resolved in 1989 with changes to 10 CFR § 50.44 and results of research and testing programs. The exception was the large dry containment systems, which were treated by GSI-121.
• GSI-121: Hydrogen Control for Large, Dry PWR Containments. Initiated by USNRC staff as part of rulemaking for GSI-A48 and resolved in 1992. No new requirements were made for large dry containments and deliberate ignition systems were judged to not be cost-effective.
• GSI-189: Susceptibility of Ice Condenser and Mark III Containments to Early Failure from Hydrogen Combustion During a Severe Accident. Proposed in 2000 in response to industry requests to reconsider 10 CFR § 50.44 and long-standing concerns regarding station blackout leading to inoperable deliberate ignition systems. Resolved in 2007 through the addition of backup power systems.
Preventing containment failure by managing both pressure and thermal loads is critically important. The installation of severe accident–capable vents, availability of backup air and power sources, and revised accident management strategies are all steps that are currently being taken to address this critical issue.
G.2 IMPLICATIONS OF FUKUSHIMA DAIICHI ACCIDENT FOR HYDROGEN CONTROL
The accident at the Fukushima Daiichi Plant demonstrates that inerting primary containment is not sufficient to protect plants against hydrogen explosions. If the containment fails during a severe accident, the hydrogen generated by the metal–water reaction in the damaged reactor core can be released into the reactor building, mix with air, and burn. For this reason, the most effective control strategy is to manage the pressure and thermal loads on containment to prevent its failure. This requires the capability to safely vent hydrogen in a timely fashion with a minimum release of fission products into the environment.
The maximum amount of hydrogen generated in a severe core accident is almost three times the volume of nitrogen present initially in the primary containment. This quantity of hydrogen overwhelms the inerting effect of nitrogen. When the hot hydrogen–nitrogen–steam mixture leaks into the reactor building, the steam will begin to condense, and a flammable mixture will be formed.
The explosions at the Fukushima Daiichi plant significantly degraded the ability of personnel at the plant to mount an effective accident response. Substantial structural damage occurred to the Unit 1, 3, and 4 reactor buildings, and particularly Units 3 and 4, creating concerns about the integrity of their spent fuel pools as well. The explosions also created pathways into the environment for radioactive material leaks from containment. An intact BWR building acts as a filter to trap fission products released from the damaged core during a severe accident. Filtering is effective only if the reactor building remains intact and fission products can be removed by passing the exhaust gas through the filters in the standby gas treatment system.
In the 1980s, researchers at Oak Ridge National Laboratory examined severe accidents in boiling water reactor plants and the mitigating role of reactor buildings (i.e., secondary containment) on fission product releases. Greene (1990) specifically examined the potential for secondary containment failure due to combustion of hydrogen. He noted that reactor buildings have complex structures and relatively low failure overpressures (the pressure resulting from even a low-speed combustion event will substantially exceed the estimated failure pressure of the building outer walls); consequently, combustion of large amounts of hydrogen in a reactor building “would probably challenge the integrity of the secondary containment” (Greene, 1990). Greene identified two key mitigation strategies that focused on maintaining primary containment integrity: primary containment sprays and primary containment venting.
The explosions at the Fukushima Daiichi plant were indeed extremely destructive. The complex structure of the lower part of the reactor buildings is well suited to cause flame acceleration and potentially transition to detonation (see Sidebar 4.1 in Chapter 4). Ironically, having a strong structure with multiple compartments can greatly enhance the damage over a weaker structure—this result, although not intuitive, is now well established (NEA, 2000) and is an important consideration in combustion hazard analysis.
Based on what has been known about hydrogen behavior since 1980, the explosions and damage to reactor buildings at the Fukushima Daiichi plant should not have been surprising. They illustrate in dramatic fashion the importance of hydrogen control in severe accidents. Of course, the first line of defense in controlling hydrogen is to prevent the metal–water reaction in the core from occurring. The second line of defense is to manage the pressure and thermal loads on the containment to prevent failure. These are the primary goals of all accident management strategies. If these actions can be accomplished, then as a secondary result, hydrogen generation, releases, and explosion hazards will be minimized.
The Fukushima Daiichi accident prompted the Nuclear Energy Agency to produce a report on hydrogen generation, transport, and mitigation under severe accident conditions (NEA, 2014). The report summarizes the status of national requirements for hydrogen management and mitigation and computer codes for hydrogen risk assessment. The Natural Resources Defense Council considered a wide range of topics related to hydrogen explosions in severe accidents and issued a report giving their perspective on the issues (Leyse, 2014). These reports were issued just as the present report was being finalized.
________________
