> Grupo de apoyo: Laboratorio Energético del Hidrógeno

Safety Issues on Nuclear Production of Hydrogen 

José M. Martínez-Val, Jesús Talavera(*), Agustín Alonso
ETSII, Universidad Politécnica de Madrid- Getafe, Spain
(*) Refinería de Puertollano, REPSOL YPF, Spain


3. A brief overview on hydrogen risks in LWR

Hydrogen risks in light water reactors were first formally recognized in 1964 by the competent US Advisory Committee on Reactor Safeguards, ACRS, on its report on the proposed construction of the 1473 MWt Connecticut Yankee PWR. The report alluded to the need for a stronger containment system “if a large fraction of the zircalloy cladding would undergo metal-water reaction in a core meltdown, releasing heat and hydrogen, which could burn adding more heat”.  Since that statement, the study of hydrogen risks has been considered worldwide by regulatory authorities and the industry and a great deal of progress has been achieved to understand and control such risks.  Nowadays it can be said that the knowledge acquired through research and from the analysis of the operating experience is sufficient to guarantee the safety of such power plants from that type of risk.

The early theoretical considerations were soon proven correct. The accident at the Three Mile Island Unit 2 nuclear power plant (March 28th, 1979), involved the oxidation of 45% of the zirconium fuel cladding, generating some 460 kg hydrogen and reaching about 7.9% volume concentration in the containment atmosphere 35% rich in steam. A hydrogen deflagration took place, the pressure peak reached 3 bar, but no major damage was found in the containment (Henrie 1987). Likewise, in the 1986 Chernobyl-4 nuclear accident substantial amounts of hydrogen were also produced by metal-water reactions, hydrogen detonations and deflagrations largely contributed to the disruption of the system.

All these experiences clearly demonstrate that in case of a severe accident, the integrity of the containment building could be threatened by the formation and ignition of flammable hydrogen/air/steam mixtures (IAEA 1991). Violent combustion regimes could lead to the loss of leak-tightness of this ultimate barrier, while stabilised diffusion flames could damage components or systems of essential safety concern. The prediction of the formation and behaviour of such flammable mixtures and their possible combustion processes constitute essential information for the design and implementation of different mitigation devices and procedures (Breitung 2000). As a result of the large resource investment put into experiments, theory and computing, substantial background knowledge, experimental database and powerful analytical tools for simulations have been developed in this area.

On their side, regulatory organizations have been developing appropriate regulations to prevent the hydrogen problem through specific design requirements (10CFRPart50 Appendix A) and to mitigate its consequences in the unlikely event that such metal water reactions are produced. Nevertheless there have not been implemented unanimous solutions.

The hydrogen risk within a Nuclear Power Plant is mainly related to the flammability of hydrogen and, in the event of an accident, the consequent damage to safety equipments and systems or containment buildings.

3.1 Phenomenology of Hydrogen Risk within Nuclear Power Plants.

The first problem to be investigated was, of course, the generation rate of hydrogen from zirconium metal reaction under accident conditions. Baker and Just have studied this reaction in depth in the early sixties. They determined that the reaction rate could be expressed by a parabolic law and determined the corresponding activation energy. Later on, through experiments performed in the LOFT and Phebus facilities, many improvements have been achieved covering a large range of accidental situations.

The hydrogen risk within a Nuclear Power Plant is mainly related to the flammability of hydrogen and, in the event of an accident, the consequent damage to safety equipments and systems or containment buildings. For a hydrogen burn to happen within a facility, hydrogen must escape from the primary system during a loss-of-coolant accident (LOCA) and then be released to the containment environment (containing air (oxygen), steam and other gases) and be mixed and react with them. The features of the environment receiving the released hydrogen will therefore drive the dynamics of the mixing/distribution, combustion and/or mitigation processes taking place.

Two geometrical features can be distinguished: confined enclosures and semi-confined (partially vented) rooms. Some other non-geometrical boundary conditions to be accounted for are: (1) presence of inert or diluting agents (steam, CO2, additives), (2) particles (solid and droplets, bubbles within liquid pools), (3) heat and mass sources/sinks, (4) systems for fire/explosion mitigation, (5) other systems influencing mixing or combustion (fan coolers, sprays) and (6) ignition sources (either deliberate or accidental).

All of these factors combined and the complex geometrical features of the large multi-compartment containment buildings establish the driving mechanisms of hydrogen transport/mixing and reaction (combustion) processes. The thermalhydraulic behaviour of the atmosphere determines the hydrogen gas concentration (NEA 1999). Main phenomena are the establishment of jets at the break surroundings; the buoyancy plumes of light gases, which help hydrogen to rise, the mixing process by natural convection and condensation and the diffusion of the different chemical species. The effect of these phenomena, in the atmosphere evolution, leads to either well-mixed or stratified conditions.

The combustion process could happen when the hydrogen/steam/air mixture reaches the flammability limits and an ignition source appears (NEA 2000). The flame established would propagate if there were enough hydrogen concentration in the cloud. The combustion regime depends on the properties of the mixture, the turbulent creation processes and the characteristics geometrical sizes of the flammable clouds. Three typical combustion regimes can be distinguished in gaseous mixtures: slow subsonic deflagrations (the flame velocity is less than the sound speed in reactants), fast supersonic flames (the flame velocity is less than the sound speed in products, but more than the sound speed in reactants), and detonation (the flame velocity is more than the sound speed in products). The pressure peak is increased according to these regimes up to typically 1, 10 to 40 bar, respectively. The fast combustion modes could be possible on local or global scale depending on the details of the hydrogen/steam source and the thermal-hydraulic behaviour of the containment.

Experimental initiatives have also served as the development basis of predictive models and computational tools. In particular, ISPs suggested by the NEA have been conceived as test exercises for the assessment of the codes predictive capabilities.

3.2.  Severe Accident Analysis and Management

Large efforts from regulatory organizations, NPPs owner operators and research organizations in different countries and institutions have been dedicated to a variety of research programs. Essentials of the hydrogen issue in the nuclear safety field were first addressed soon after the TMI-2 accident in the USA (Camp 1983) and later by the IAEA (IAEA 1991). Specific research projects have been undertaken within the theoretical and experimental analysis of severe accident phenomenology and within multicompartment geometries representative of LWR systems either at small, medium or large scales. The most important hydrogen distribution analysis have been performed as International Standard Problems promoted by the NEA on experimental tests representing typical PWRs: ISP-29 (NEA 1993), ISP-35 (NEA 1994), ISP-47 (NEA 2002), as well as those for specific designs (Lündstrom 1996) and (Blumenfeld 2003). A review of containment thermalhydraulics has been published by the NEA (NEA 1999).

In parallel, analysis of the hydrogen combustion phenomena under different geometries have been explored thoroughly (Berman 1986), (Kanzleiter 1989), (Dorofeev 1996), (Koroll 1996), (Wolf 1999) and compiled recently into (NEA 2000). Concerning hydrogen production during specific phases of the core degradation processes, the CORA, PHEBUS and QUENCH programs have provided very valuable information related to hydrogen generation under such circumstances.

Experimental initiatives have also served as the development basis of predictive models and computational tools. In particular, ISPs suggested by the NEA have been conceived as test exercises for the assessment of the codes predictive capabilities. Historically, two approach pathways have been followed. Traditionally, the first applications involved integral multipurpose general tools (lumped-parameter codes) incorporating models for analysing the combined effects of the large variety of phenomena featured by severe accident scenarios. Lumped parameter codes have been used and further developed even nowadays. COCOSYS (by GRS), CONTAIN and MELCOR (by SNL), GOTHIC (by Westinghouse), MAAP (by EPRI), among others, belong to this family of codes.

 It is because of the growing capacity of the computing systems during the past decade, that an increasing field of application is being devoted to Computational Fluid Dynamics (CFD) codes, like CFX (by ANSYS), FLUENT (by Fluent Inc.), GASFLOW (by FzK), PHOENICS (by Cham Ltd.), STAR-CD, TONUS-CAST3M (by CEA), TRIO (by Framatome), among others. This kind of tools provide a much larger degree of detail of the phenomena involved and in fact stand as the only choice for analysing the detailed combustion under turbulent, flame acceleration or deflagration to detonation transitions regimes within confined geometries, codes such as: COM3D, FLAME3D, DET 3D (by FZK). The very promising capabilities of CFD codes as a complementary reference standard for safety analysis and regulation have been recently discussed by the IAEA (IAEA 2003). In fact, the application of CFD codes has demonstrated to be suitable for severe analysis in the containment building (Breitung 2000) as well as its complementary application with lumped-parameter codes (Alemberti 2000), (Jiménez 2003).

The efficient management of hydrogen risk has been made possible through the extensive and generally well-coordinated research effort conducted in the last two decades. Utilities and regulatory bodies, as end-users, have applied knowledge and progress achieved by the research programs. Actually, risk analysis for commercial plants are required by the regulatory authorities and the safety requirements evolve with the increasing knowledge ever to attain higher safety levels. In this way, the procedural basis for Probabilistic Safety Assessments (PSAs) beyond the DBA was initially introduced in the USA and Western countries through the reference document (NUREG-1150). Concerning hydrogen, the level of risk at both existing and new design plants is being substantially lowered by the implementation of hydrogen mitigation strategies, as introduced below.

Modern managerial practices substitute old engineered safety features that require operator intervention, in line with trends for LWRs generally. Methods of mitigating the potential for hydrogen explosions used or proposed for use within the nuclear industry include the following (IAEA 2001):

Oxygen control:

Pre-inerting: the reactor building is filled with an inert gas (generally nitrogen).

Post-accident inerting: inert gas will be injected into the containment after the accident.

Hydrogen control (by removal):

Electrical recombiners: used at concentrations below 3%, to prevent deflagration.

Catalytic Recombiners (active/passive): operative at very low concentrations.

Igniters: to be used carefully to prevent energetic combustion or detonation.

Hydrogen control (dilution/dispersion):

Mixing: use of fans to achieve a uniform atmosphere with no hydrogen pockets.

Spray systems: decrease temperature and pressure and help mixing.

Containment venting: to the atmosphere (via filters).

The Euratom organized FISA symposia on the European Union Research on Nuclear Safety, which  are held every two years, are worldwide recognized as the ideal fora to present the latest research results and applications in nuclear safety within the European Union.

3.3. Present Situation

The design of the containment is an important parameter for the mitigation strategy selected as optimum for the different LWRs (Arnould 2003). Some of the small BWR containments (Mark I and II) are pre-inerted. Mark III BWR or PWR with ice-condensers are equipped with ignitors. Large-dry PWR containments are being now under consideration. Nevertheless, there is not yet a common consensus among the different countries on the specific technologies to be used for hydrogen hazards mitigation.

The Euratom organized FISA symposia on the European Union Research on Nuclear Safety, which  are held every two years, are worldwide recognized as the ideal fora to present the latest research results and applications in nuclear safety within the European Union. Similar symposia are organized annually in the US and Japan. The latest of the FISA symposia took place in November 2003. Regarding the hydrogen risk and the threat they pose to the containment, the symposium concluded as follows.

"The response of containment structures to loads generated by severe accidents has been a subject of interest for many years. FISA-2003 has included Session 2.2 to treat such aspect; it includes twelve presentations, the one with the larger number of papers. Two main containment challenges are considered: those created by the presence of hydrogen and its deflagration and detonation capacity and the possible containment loss due to sustained, non-coolable corium concrete interaction.

Project HYCOM has added to the many previous research efforts significant experiments in medium and large-scale facilities involving hydrogen combustion regimes ranging from slow to fast turbulent deflagration. It has nevertheless concluded that to reduce conservatism and uncertainties in real plant calculations it is necessary to conduct more investigations on combustion events, mainly under deceleration conditions, and to get involved in a solid validation of computing codes under such cases. On the practical side, the use of catalytic coated thermal insulation elements has been partially explored in project THINCAT, while project PARSOAR has consolidated, in a handbook, the present knowledge on passive autocatalytic recombiners from the point of view of the hydrogen risks".

Volver al comienzo de página

Índice de temas

Índice del Laboratorio del Hidrógeno


Energía sostenible

Portada Índice de temasGrupos de apoyoObjetivosVínculosBibliografíaContactar Buscar


Este portal ha sido subvencionado parcialmente por el
Plan Nacional de Difusión y Divulgación de la Ciencia y la Tecnología, 2003 del Ministerio de Ciencia y Tecnología