Operational behaviour of passive auto-catalytic hydrogen recombiners

The Fukushima Daiichi accident in March 2011 has had a significant impact on nuclear safety research priorities all over the world. While some topics may have experienced a change in priority, the importance of the investigation on hydrogen combustion and mitigation has been further stressed. In Europe, hydrogen safety issues were long before a high-priority topic, which was broadly dealt with for example in the Severe Accident NETwork (SARNET). Now, the accident in Fukushima has heavily increased the interest in hydrogen mitigation measures, especially in passive auto-catalytic recombiners (PARs).

In PARs, hydrogen and oxygen react chemically at catalytic surfaces to form water vapour. In contrast to the classical combustion, where the hydrogen concentration in air has to exceed a flammability limit of approx. 4 vol.% and ignition energy has to be provided, the catalytic reaction already takes place at ambient temperature and significantly lower hydrogen concentration (~ 1 vol.%). The resulting reaction heat induces a chimney flow which provides self-start and self-feeding of the device. These properties make recombiners a passive safety device, suited for application in nuclear power plants, where operation needs to be assured even in case of a station black-out.

At the first glance, a PAR simply represents a sink of hydrogen and oxygen, and a source of steam and heat. Typical removal rates per PAR unit are in the order of 60 m³ hydrogen per hour, which correlates to a heat contribution of approx. 180 kW. A fleet of typically 40 units are installed inside an light water reactor (LWR) containment. The basic technical ability of PARs to reduce the amount of hydrogen in the containment atmosphere has been proven in many studies over the last two decades.

The working principle of a PAR is simple, but the details of the operational behaviour are rather complex. The quick start of a recombiner depends on the fast heat-up of the catalyst sheets due to the exothermic reaction, supported by the exponential temperature dependency of the reaction rate (Fig. 1). However, water deposited on the catalyst as well as other air-borne substances —which are expected to be numerous during a severe accident—may slow down the initial processes. The catalysts used are not strictly selective for hydrogen and oxygen, and consequently may influence or may be influenced by atmospheric components. Finally, but not concluding, local combustion initiated by hot catalyst sheets at high hydrogen concentrations may lead to a pressure rise inside the containment and significantly change the composition and thermal conditions of the containment atmosphere.

For basic accident assessment, hydrogen removal capacities are in the focus of interest. Consequently, the early numerical models, describing the PAR efficiency based on early performance tests performed in the 1990s, relied on simple parameter correlations, providing an estimate of the amount of hydrogen removed per unit time, dependent on the PAR inlet hydrogen concentration and on the containment pressure. Today, where accident analyses have become more complex—partly as a result of increasing computational power—a far more detailed understanding is needed. To simulate accident sequences taking into account the complex interaction of the recombiner with the atmosphere is a challenging task. PAR operation can dissolve stratified gas layers, transform aerosols into different species, and induce additional thermal fluid dynamic patterns into the local containment thermal hydraulics. These complex interactions require mechanistic models—that is, formulations reflecting the physico-chemical processes—rather than simplified parameter models. As a basis for such model development, well-instrumented experiments with well-defined boundary conditions are mandatory.

Basic understanding with small-scale recombiner tests

Sophisticated models require detailed experimental data. Jülich Research Centre has in recent years significantly enhanced the experimental capabilities. In the REKO facilities, small-scale set-ups are investigated under well-defined conditions. Although these do not represent realistic conditions, they are essential for model development. The basic strategy is to separate the phenomena inside the catalyst section, that is, heat and mass transfer between gas phase and catalyst surface, catalytic reactions, and so on, from the chimney effect which determines the amount of gas to be transferred through the PAR.

In the REKO-3 facility, a set of four instrumented catalyst sheets is mounted inside a flow channel (Fig. 2). The test channel is fed with a well-defined gas mixture including for example hydrogen, air, nitrogen, steam and carbon monoxide. Measurement of the catalyst temperature profile along the catalyst sheets in combination with analysis of the resulting gas mixture behind the sheets provides valuable data for the correlation of the inlet conditions with the resulting catalyst temperature and conversion behaviour at different flow velocities. These data help to develop a reaction model for the catalyst section, even under complex situations such as oxygen starvation or a parallel CO reaction.

The next level of complexity is reached when in the REKO-4 5.5 m³ vessel (Fig. 3), the channel with the instrumented catalyst sheets is equipped with a chimney. Measurements are similar to REKO-3, with the addition of optical flow field measurement by means of Particle Image Velocimetry (PIV). After injection of hydrogen into the vessel, the continuous correlation of the PAR conditions including the throughput with the ambient conditions is enabled.

The full picture: THAI tests

The OECD-THAI (Thermal-hydraulics, Hydrogen, Aerosols, Iodine) project was started in January 2007 under the auspices of the OECD Nuclear Energy Agency, supported by safety organizations, research laboratories and industries from seven European countries and partners from Canada and the Republic of Korea [1]. Different available PAR designs have been investigated in the 60 m³ THAI test vessel in hydrogen-air and hydrogen-air-steam mixtures. The THAI facility (Fig. 4) is operated by Becker Technologies GmbH at Eschborn, Germany, under the sponsorship of the German Federal Ministry of Economics and Technology.

The large volume of the test vessel allows PAR operation with unrestricted natural convection which enables observation of the interaction of the PAR with the vessel atmosphere. Systematic tests have been performed to investigate and quantify the influence of various parameters affecting PAR performance, such as: steam content, oxygen starvation, containment pressure, ignition initiation by PAR, and exposure to fission products. Recombiners of three different suppliers have been tested under realistic conditions with comprehensive instrumentation to measure catalyst temperatures, gas temperatures, and gas concentrations at different locations.

The majority of the tests consisted of two consecutive hydrogen injections inside the test vessel (Fig. 5). The first injection provided details of the PAR start behaviour and the following hydrogen recombination rates, while the second injection period was enhanced until ignition occurred, which gave details on the PAR ignition conditions and the resulting consequence of the hydrogen deflagration in terms of peak pressures and peak temperatures. From the systematic investigation of the different PAR designs, it can be concluded that the ignition limits of all designs lie within around 5.5-7.5 vol.% hydrogen for dry air conditions, for higher steam concentrations around 8-9 vol.% hydrogen, reflecting the effect of steam inertisation on combustion processes.

Two examples for the complexity of the investigations shall be given by describing special test series that were dedicated to investigate unresolved issues related to PAR behaviour in case of a severe accident. The first issue refers to the conversion of metal iodide into gaseous iodine by PAR operation, which is of relevance for the radiological source term. Challenging experimental test conditions are required for these tests, with PAR catalyst temperatures of 800°C-900°C and aerosol concentrations of several g/m³ to obtain significant iodine conversion. Sophisticated instrumentation was installed to monitor aerosol parameters comprising filter stations, photometers, cascade impactors, and gas scrubbers for the measurement of gaseous iodine. The test results reveal mean conversion yields in the range of 1.0-3.0 % for conversion of CsI into gaseous iodine upon passage through an operating PAR at catalyst temperatures of about 800°C.

The second issue referred to the potential poisoning effect by PAR exposure to aerosols (hygroscopic and inert aerosol mixture), iodine (I2) and eventually forming condensate layers on the catalyst plates. Subjected conditions included very low hydrogen concentration for PAR start-up. The instrumentation described above was further extended by two Maypack filter stations for the discrimination of the airborne iodine species.

In the experiment, the PAR was exposed to excessive aerosol concentrations of insoluble tin oxide (SnO2) representing insoluble aerosols, highly hygroscopic and sticky lithium nitrate (LiNO3) solution droplets, steam and iodine, a potential catalyst poison. The results showed that even under such challenging conditions, PAR recombination efficiency remained in the range of 50-70 %. This is comparable to the results of the THAI HR series with similar thermal-hydraulic conditions, but without aerosols and iodine. Also, the onset of recombination occurred at hydrogen concentrations comparable to corresponding tests from the HR series. No major negative effects of aerosols and/or iodine on PAR performance could be observed.

A further important result of the PAR experiments performed in the frame of the OECD-THAI project was that a much higher O2-to-H2 ratio than stoichiometrically required to obtain unimpaired PAR performance. The required oxygen surplus for unimpaired performance is a factor of 2-3 typically. For stoichiometric conditions, hydrogen recombination rates have been observed being reduced to less than 50% compared to unimpaired conditions. These results confirmed precisely the findings from a test series on the effect of oxygen starvation performed in the REKO-3 facility at JÜLICH, demonstrating the valuable benefit of bringing together the results of different experimental facilities. In this sense, the results of the THAI PAR experiments have triggered single-effect research for even greater physical understanding to some specific phenomena. Additional PAR experiments in the THAI facility investigating the onset of recombination at low or even extremely low oxygen concentrations are performed in the frame of the ongoing OECD THAI2 project.

Code validation for accident simulation

In order to apply the experimental findings into real-world use, the results need to be converted into computer programs. The database developed in the frame of the OECD-THAI project is currently extensively used to validate and improve codes. More realistic prediction of the hydrogen distribution inside the containment can be expected; this will have significant impact on future accident simulations.

The combination of detailed and integral experiments enables for the first time the validation of detailed numerical PAR models. As one example, the SPARK code developed by IRSN/France is a promising approach for assessing the ignition behaviour of PARs. The code includes a detailed chemical model for both the catalytic surface reactions and combustion in the gaseous phase. Another example is the REKO-DIREKT code developed at JÜLICH, which is able to calculate detailed information about the PAR operation including the catalyst temperature. Due to its very fast performance, the code may be used with accident codes.

High-resolution 3D calculations with computational fluid dynamics codes can provide a detailed prediction of the hydrogen distribution. In this context, REKO-DIREKT has been successfully coupled with a CFD code. Although at present such detailed simulations are too time consuming for full containment application, they have already been applied to reasonable-size volumes. Recently, Jülich has performed calculations to assess the applicability of PARs in car garages (Fig. 6). For this purpose, a hydrogen release scenario without mitigation measures has been compared to a situation with recombiner installed. The simulation shows that although the fluid dynamic boundary conditions differ significantly between this arrangement and a reactor containment, the recombiner efficiently removes hydrogen. This example demonstrates the successful coupling of a PAR model and a CFD code and gives an idea of analytical possibilities to be available in the near future. Until then, enhancing the understanding of the rather complex phenomena inside a PAR remains a scientific challenge.

Sponsorship of the PAR experiments in JÜLICH and in THAI by the German Federal Ministry of Economy and Technology (BMWi) and by the partners of the THAI OECD project (2007-2009) is gratefully acknowledged.

Ernst-Arndt Reinecke, Jülich Research Centre, 52425 Jülich, Germany; Gerhard Poss, Becker Technologies GmbH, Kölner Strasse 6, 65760 Eschborn, Germany