Post-accident hydrogen behaviour in containment has been recognized as a safety issue since hydrogen was released and burned inside containment following the event at Three Mile Island (TMI) in 1979. While the hydrogen burn at TMI was well contained by the reactor building, the events at Fukushima Daiichi in 2011 graphically reminded everyone of the devastating effects of hydrogen combustion.

International guidelines on containment design recognize the need for hydrogen mitigation [1]. Although various means have been used to mitigate hydrogen generated by accident conditions, more and more stations worldwide have implemented passive autocatalytic hydrogen recombiners (PARs) or are in the process of doing so.

It is not so surprising. PARs are a simple and cost-effective means of mitigating the consequences of hydrogen release. PARs are a passive hydrogen removal device designed for use in post-accident conditions where hydrogen may be present in containment. Each PAR consists of an open-ended stainless-steel box with catalyst-coated elements inside, and a cover to protect the elements from water sprays (Figure 1). The catalyst elements convert hydrogen (H2) and oxygen (O2) into water vapour and heat. The heat of reaction creates a natural convective flow through the recombiner, eliminating the need for pumps or fans to transport new hydrogen to the surface of the catalyst, and contributing to hydrogen mixing in containment. The PARs are self-starting in presence of hydrogen, and therefore require no operator action.

Canadian framework

In Canada, the Atomic Energy Control Board (AECB, now Canadian Nuclear Safety Commission) issued a generic action item (GAI) on hydrogen in containment in 1988. In 1991, the AECB published regulatory requirement R-7 requiring all new plants to control the concentration of hydrogen and/or oxygen following an accident to prevent explosion or deflagration inside containment.

In response to the GAI, Canadian utilities embarked on a comprehensive programme to analyze hydrogen generation, distribution in containment, burn consequences and mitigation. Utilities inventoried hydrogen sources, evaluated hydrogen release rates from these sources, developed more and more detailed containment models for predicting hydrogen concentration, reviewed validation of hydrogen distribution calculations based on available literature, and developed plans to implement PARs, even if some stations had previously installed igniters for hydrogen control.

Simultaneously, Atomic Energy of Canada Limited was at the forefront of hydrogen research. Several facilities were built to support research on hydrogen, including vented and unvented combustion, flame speed, acceleration and transition to detonation, effect of isotopic on flammability limits, standing flames, hydrogen mixing, and testing of PARs. The PAR programme was initiated in the early 1980s, building on catalyst technology developed for heavy water production via the liquid phase catalytic exchange (LPCE) process. The catalyst is specifically adapted to wet/humid and potentially cold conditions that could be present in containment. Hundreds of tests on various scale levels were done to adjust the geometry, demonstrate the catalyst resistance to various poisons and quantify several operating parameters (for example, self-start threshold, capacity, body temperature, plume temperature, ignition limits).

The GAI action was followed by specific closure criteria that required utilities to:

  • Consider the entire spectrum of design basis accidents including loss-of-coolant accident (LOCAs) coincident with loss of emergency-core cooling (LOECC) [note: this combination is part of the design basis of existing CANDU reactors.]
  • Carry out conservative and representative assessments of hydrogen mixing and transport:
  • To determine the local hydrogen distributions in critical containment regions, with current and additional hydrogen mitigation measures
  • To identify the envelope of hydrogen combustion in the short term and in the long term so as to:

– Demonstrate, using recently-developed criteria [2], and other justifiable criteria if required, that local pockets of sensitive gas mixtures (combustible clouds) cannot lead upon ignition to deflagration-to-detonation transitions (DDT) (supersonic velocities) and fast deflagrations (sonic velocities) with potentially unacceptable combustion loads in any region of containment

– Demonstrate via well-supported calculations and/or experiments that gas mixtures in the slow deflagration domain do not have, if ignited, consequences detrimental to the containment boundary, and, the hydrogen mitigation system(s)

– Demonstrate via well-supported calculations and/or experiments that potential standing flames do not threaten the survivability of the containment boundary, and the hydrogen mitigation system(s)

– Ensure the survivability and/or functions of post-LOCA hydrogen mitigation system(s) and vulnerable containment boundaries, if any, via equipment qualification and/or augmented protection and mitigation.

  • To evaluate the PARs for their potential to enhance the effectiveness of the short-term and long-term mitigating measures, and reduce the risk from potential hydrogen burns.

The Canadian utilities also joined forces through the CANDU Owners Group (COG) to address common issues. AECL was in turn contracted by COG to produce state-of-the-art reports on hydrogen burn, a PAR qualification summary, a technical basis document for the implementation of PARs, and a summary of PAR test programme in Canadian CANDU reactors, to name a few.

In Canada, single unit CANDU 6 stations have implemented PARs, and multi-unit stations are in the process of doing so. No deadline has been set so far.

Further analyses using the computer simulation code Modular Accident Analysis Program MAAP-CANDU are carried out to assess the effects of severe accidents, and the capability of the current installations to cope with those events.

The benefits of PARs

PARs have several benefits compared to other means of hydrogen control:

  • Passive (no operator, controls or power required)
  • Start depleting hydrogen in non-flammable condition
  • Easy to install
  • Easy in-service inspections.

Candu offers PARs for hydrogen mitigation in nuclear power plants following postulated design basis accident (DBA) or severe accident (SA).

To meet the specific requirements for size and hydrogen depletion rate (capacity), Candu offers three PAR models, which correspond to the small, medium and large sizes respectively. Inside the PAR1, PAR2 and PAR3 housings, there are installed 31, 47 and 67, respectively, flat rectangular catalyst elements (plates).

The catalysts used in the PARs offered by Candu were developed specifically for application in nuclear containments. The catalysts rapidly oxidize hydrogen in the concentration range of interest, are not deactivated by water vapour or steam, and operate over a wide range of temperatures.

The ability to start under wet and cold conditions is essential to ensure PAR start-up in containments with powerful cooling systems such as in CANDU and ice-condenser containments. The PAR catalysts offered by Candu eliminate this concern by using a proprietary wet-proofing procedure. The catalysts operate at temperatures up to 750°C without loss of catalytic activity or their wet-proof characteristics, and are unaffected by high radiation exposures.

This means that if the PAR stops after consuming the initial hydrogen release, it will be able to restart if hydrogen continues to be produced (for example by radiolysis or other delayed sources), even after the containment cools down.

PARs offered by Candu have been rigorously and extensively qualified to meet foreseeable post-accident conditions in CANDU and LWR reactors.

PARs were subjected to cumulative stressors that simulated the operational conditions the recombiner could be exposed to during its lifetime, including accident conditions and pre- and post-accident conditions. The stressors were thermal and radiation aging, chemical contaminants, and seismic loads. Baseline functional tests were performed to determine the PAR’s operational characteristics (for example, self-start thresholds and capacity). Subsequent intermediate and final functional tests were carried out to demonstrate the PAR performance is maintained after exposure to the cumulative stressors.

As part of the qualification process, we subjected the PAR to foreseen pre- and post-accident conditions. Some of the tests performed are described below.

  • Exposures to CANDU and LWR operating conditions were conducted to evaluate the effects of volatile organic compounds (VOCs) present in the containment air. In certain areas of containment, it was found that the catalyst may be temporarily affected at low temperature, requiring more time and/or higher hydrogen concentration and/or higher temperature to start compared with as-new condition. Once the catalyst starts, however, it cleans itself and thereafter performs like new.
  • The effects of hydrogen concentration, pressure and temperature on capacity were evaluated. The capacity of PARs to recombine hydrogen is largely related to the density of the hydrogen molecules at the surface of the catalytic elements. As a result, the capacity rises with increasing hydrogen concentration and pressure, which represents an inherent control mechanism whereby the capacity adapts to the demand. Higher ambient temperatures lead to lower capacity for the same reason, but the effect is minimal.
  • The potential to initiate a hydrogen burn was evaluated. When the hydrogen concentration reaches about 7.5 % vol., the plate may be sufficiently hot to ignite passing hydrogen. Although the installed capacity of PAR should aim to prevent this condition, this behaviour can be viewed as an ultimate ‘automatic breaker protection’. By burning at 7.5 %vol., which is still in the slow-combustion regime, the PAR will prevent burn at even higher concentration where much more violent combustion modes may occur.
  • The implications of water spray on startup and capacity of the PAR were evaluated. We tested the effect of water spray on an operating PAR to verify the potential thermal shock impact on capacity, as well as the effect of a pre-established spray prior to hydrogen release to verify the impact on the self-start of the PAR. The sprays included chemicals present in CANDU and LWR water. No significant effects were observed.
  • The effects of low oxygen on self-start capability. Low oxygen in containment is a condition that may occur in severe accident scenarios after the PAR has reacted much of the oxygen initially available in containment. Small and large scale tests were done to show that PAR can start with as little as 1-2% vol. O2.
  • The effects on PAR performance of poisons that could be airborne post-accident such as iodine, methyl iodide, hydrazine, chlorine, and hydrochloric acid. None had significant effects at expected concentrations in containment.
  • The implications of carbon monoxide (CO) in containment. CO may be generated in severe accidents by a core-concrete reaction. Carbon monoxide is combustible and represents a threat. Fortunately, it was shown that the PAR recombines CO to CO2 at the same time and at about the same molar rate as it recombines H2 to H2O.
  • The effect of fuel aerosols on PAR capacity. We participated in two test series. The first one was the H2PAR test conducted at Cadarache, France. In this test, a 1/5-scale PAR was tested with and without exposure to non-irradiated molten fuel aerosols representative of a PWR. The performance was identical in both cases, demonstrating the PAR was not impacted by the aerosols. The second test was the PHEBUS test FPT3. This test was conducted with irradiated fuel using small catalyst coupons from several suppliers mounted side by side in matchbox-size cells representing a PAR. This test gave a very unusual result, as the catalyst did not respond despite having responded perfectly in several other tests with similar or more demanding conditions. To understand the behaviour of the AECL sample in FPT3, AECL conducted complementary bench tests to explore the effects of catalyst geometry and test apparatus configuration on sample response to hydrogen. Tests were done with two side-by-side cells, each containing an identical catalyst coupon. The tests showed that the coupon that activates first may steal the flow of hydrogen from the coupon in the adjacent cells, preventing those adjacent coupons from reacting. It was also demonstrated through an independent experiment in the REKO-1 facility in Germany that the AECL catalyst coupon used in the PHEBUS FPT3 experiment was likely partially deactivated prior to the experiment, but that once started, it performed like fresh again. Ultimately then, while the PHEBUS FPT3 test was inconclusive for the AECL catalyst, it led to an increased understanding of how a PAR would respond and how a PAR test should be designed. It also illustrated the depth of technological capability behind the PARs offered by Candu.


Implementing PARs is quite easy compared to other engineered systems. The accident scenarios are analysed to determine the event sequence and combustible gas generation. The hydrogen distribution in containment is then calculated to determine the proper number and location of PARs to meet safety requirements.

Placement of recombiners in an existing containment, while being driven by safety objectives, is also subject to operational constraints. Parameters to consider in selecting locations are:

  • Hydrogen source location
  • Potential hydrogen pocket location
  • Easy access for in-service inspection
  • Accessibility for other safety components
  • Potential impact on other safety components (heat and plume generated by PAR under accident conditions)

A walkdown is performed to identify the available locations, noting the space available, type of supports, dimensions of auxiliary support structures that may be needed and verifying clearances from safety-related equipments.

Installation, verification and maintenance of PARs is easy.

The installation is usually limited to anchoring the support on floor, wall or ceiling. Rebar scanning is performed to confirm exact location prior to installation. Occasionally, auxiliary supports need to be designed to interface between the existing structures and the PAR standard supports, while meeting the seismic requirements.

Periodic testing of the catalyst is required to ensure the PAR availability for service, (for example, its capability to self-start at the required station-specific conditions of temperature and hydrogen concentration). To meet this requirement, Candu offers a whole plate tester (WPT) which consists of a heated enclosure into which a single PAR plate is inserted (Figure 2). An H2-air mixture passes over the plate at a controlled flow rate and temperature. The heat of the hydrogen-oxygen recombination reaction causes the catalyst surface temperature to increase. An array of sensors monitors the catalyst temperature increase as a function of time.

In the event that the catalyst does not meet the acceptance criteria, the catalyst plates can be regenerated by oven heating.

In summary

Experience has shown that hydrogen risk in nuclear installation cannot be taken lightly. Means to prevent and mitigate the consequence of hydrogen burn following both design and severe accident conditions are required. The implementation of PARs is a cost-effective and proven approach to the mitigation of the hydrogen risk. To date, over 3000 PARs offered by Candu have been installed in Canada, Finland, Korea, and France and Ukraine (with ALSTOM Power).

Author Info:

Richard Moffett, Candu Energy Inc, 2285 Speakman Drive, Mississauga, Ontario L5K 1B1, Canada

This article was originally published in the August 2012 issue of Nuclear Engineering International

Related Articles
Operational behaviour of passive auto-catalytic hydrogen recombiners


[1] IAEA Design of Reactor Containment Systems for Nuclear Power Plants, NS-G-10, 2004.

[2] OECD State of the Art Report on Flame Acceleration and Deflagration-to-Detonation Transition in Nuclear Safety, NEA/CSNI/R(2000)7), August 2000.