Why lower containment pressure does not mean that the atmosphere is under control
Containment spray helps reduce pressure by removing heat and condensing steam. The actual pressure response depends on the overall mass and energy balance within containment. This is an important part of the response: less steam in the gas phase and a lower temperature reduce the pressure loading on containment.
However, hydrogen does not condense with the steam. Lower pressure therefore does not, by itself, show whether the atmosphere is safer. It shows that the thermodynamic state of the system has changed. The amount of hydrogen, its concentration, spatial distribution and potential for combustion are separate questions.
The central paradox: containment pressure can fall without hydrogen being brought under control.
This distinction is why containment atmosphere control cannot be assessed from a single measurement, and why spray, mixing, passive autocatalytic recombiners, igniters, inerting and monitoring do not perform the same safety function.
An important boundary: LOCA is not synonymous with a hazardous hydrogen event
A loss of coolant does not by itself mean that a hazardous hydrogen atmosphere will develop inside containment. During a design-basis LOCA, with successful operation of the emergency core cooling systems, the key objective is to prevent extensive overheating and oxidation of the fuel cladding.
This does not mean that hydrogen generation must be exactly zero in every LOCA. It does mean that a significant hydrogen problem arises primarily when core cooling is inadequate, fuel cladding temperatures rise and the zirconium-steam reaction becomes intense:
Zr + 2 H2O → ZrO2 + 2 H2 + heat
The reaction is exothermic. It not only produces hydrogen but also adds heat to an already overheated core. Hydrogen is therefore not merely a substance present in containment. Its generation rate is also an indication of the progression of core damage.
This is the point at which the article deliberately moves beyond the normal response to a design-basis LOCA and into the domain of a significant beyond-design-basis accident. If the event progresses to substantial core degradation, it is classified as a severe accident. In 2003, the US Nuclear Regulatory Commission removed the former design-basis “LOCA hydrogen release” from 10 CFR 50.44.
The conclusion was not that hydrogen was unimportant. Rather, the limited design-basis hydrogen source term was not risk-significant, while the principal combustion hazard was associated with beyond-design-basis accidents. Hydrogen monitoring therefore remained important mainly for assessing core damage and guiding the response to such events.
This was a US regulatory decision, not a universal statement about the design of every plant. It does, however, clearly demonstrate the boundary that must be preserved in any explanation: LOCA and severe accident are not synonyms.
Pressure is falling. Is the hazard falling too?
In the previous part of this series, containment spray was considered primarily as a heat removal mechanism. The droplets cool the gas mixture and surfaces, steam condenses, its partial pressure falls and the total pressure of the atmosphere falls with it.
According to Dalton’s law, total pressure is the sum of the partial pressures of the individual gases:
p(total) = p(N₂) + p(O₂) + p(H₂O) + p(H₂) + …
When some of the steam passes from the gas phase into the liquid phase, H₂O molecules are removed from the gas mixture. Under these conditions, hydrogen remains a non-condensable gas. If the amount of hydrogen remains unchanged while the total number of moles in the gas phase decreases through condensation, its mole fraction, and therefore its volume fraction, may increase:
x(H₂) = n(H₂) / n(all gases)
At the same time, cooling lowers the temperature, so the partial pressure of hydrogen may also decrease. There is no contradiction. A lower partial pressure caused by a lower temperature is not the same as hydrogen removal, and a higher volume fraction is not the same as a greater absolute quantity. A single pressure value cannot reveal these differences.
There is another important effect. Steam in a hydrogen-air-steam mixture acts as a diluent and may temporarily keep the atmosphere outside the flammability range. Its condensation can weaken that protective effect and move the mixture closer to the flammable region, even though not a single additional mole of hydrogen has been generated. The IAEA specifically identifies scenarios in which a high steam concentration initially suppresses combustion, while subsequent condensation caused by cooling or spray actuation changes the hydrogen hazard conditions.
This does not mean that spray is detrimental or that pressure reduction implies a greater hazard. Spray also cools, drives flows, affects mixing and, under certain conditions, can reduce the temperature and pressure peaks produced by combustion. Under other conditions, spray-induced turbulence can accelerate the flame. Its effect therefore depends on the current composition, the spatial distribution of gases, the direction of flame propagation and the time of actuation. This is precisely why successful spray operation cannot be equated with successful hydrogen management.
Lower pressure may be a favourable sign for containment pressure loading. It is not, by itself, evidence that the atmosphere is safe.
From hot fuel cladding to hydrogen in containment
In the scenario discussed here, the first major source of hydrogen develops in an inadequately cooled core. Containment then becomes the environment in which the consequences of its release must be managed. Hydrogen may also be generated within containment, for example through water radiolysis and corrosion reactions.
The physical sequence is clear:
1. the core remains inadequately cooled;
2. the temperature of the fuel cladding and other metal structures rises;
3. zirconium reacts more intensely with steam;
4. hydrogen and additional reaction heat are generated;
5. the resulting gas mixture is released into containment through the break, open relief paths or other connections from the primary system.
The final quantity and concentration are not the only relevant variables. The release rate also matters. The same total mass of hydrogen, released slowly into a large, well-mixed volume, does not produce the same transient conditions as a rapid release into a confined region near the source.
Zirconium-steam oxidation is not the only possible source of combustible gases during the later stages of a severe accident. Hydrogen may also be produced by oxidation of other metals. Following reactor vessel failure, molten core-concrete interaction can contribute both hydrogen and carbon monoxide. To understand the first major hydrogen source, however, it is sufficient to follow the path from an inadequately cooled core to the release of a hot steam and hydrogen mixture into containment.
The average concentration does not show where the hydrogen is
Containment is not a single, perfectly mixed vessel. It includes large open volumes, but also galleries, ceiling recesses, stairways, passages between rooms, shielded compartments and regions with limited gas exchange.
The idea that hydrogen, as a light gas, simply rises and spreads uniformly beneath the dome is also too simplistic. It is typically released from the primary system together with hot steam and with the initial momentum of the jet. Its path is then shaped by buoyancy, entrainment of surrounding gas into the jet, steam condensation on cooler surfaces, natural circulation, local obstacles and flows generated by spray, coolers or recombiners.
The result may be a well-mixed atmosphere, but it may also be transient stratification and local pockets with a higher hydrogen concentration. High, poorly ventilated or dead-end volumes and regions near the release point may be particularly vulnerable. The global average may therefore remain below a selected limit while a local mixture is already flammable.
This reveals the fundamental difference between pressure and composition. A pressure disturbance propagates very rapidly through interconnected volumes. Equalisation of composition, however, requires the actual transport and mixing of mass. Pressure can therefore be a relatively global indicator, while the hydrogen hazard may be distinctly local.
This is not merely a modelling detail. The OECD/NEA THAI programme was specifically designed to investigate the spatial distribution of hydrogen, its removal by PARs, slow combustion and, in later phases, combustion during spray operation and flame propagation between connected volumes. The need for such experiments makes the central point clear: assessing the hazard requires more than knowing how much hydrogen is present in containment as a whole. We must also know where it is and how its distribution changes over time.
Pressure may be an approximately global property of the atmosphere. The combustion hazard may be distinctly local.
Flammability is not a single number
The often repeated statement that hydrogen becomes flammable above approximately four per cent by volume is useful only as an initial reference point. It is not a universal limit for every location within containment or for every stage of an accident.
Even the lower flammability limit depends on the direction of flame propagation. For hydrogen in air at atmospheric pressure, room temperature and water-vapour saturation, the IAEA gives approximate values of 4.1% for upward propagation, 6% for horizontal propagation and 9% for downward propagation. In an actual containment atmosphere, temperature, pressure, oxygen fraction, steam fraction, turbulence and mixture homogeneity also vary. The size of the space, obstacles, and the energy and location of the ignition source are important as well.
The question is therefore not only whether the mixture is “flammable”. It is also what combustion regime can develop and what loads it will produce.
In a deflagration, the flame front propagates subsonically through the unburned mixture. A slow deflagration may produce a comparatively gradual pressure rise, but it is not necessarily insignificant for local equipment or structures. Turbulence and obstacles can wrinkle the flame and increase its surface area, thereby accelerating combustion. This flame acceleration can lead to a fast deflagration.
If the mixture, geometry and available run-up distance are sufficiently adverse, a deflagration-to-detonation transition, or DDT, may occur. In a detonation, the shock wave and reaction front are coupled, and propagation is supersonic. Direct detonation without preceding flame acceleration is generally not considered a likely initiating mode in reactor accident scenarios. The serious concern is the pathway through flame acceleration to DDT.
The difference is structural, not cinematic. The combustion regimes produce different peak pressures, rates of pressure rise, impulses and local dynamic loads. Two atmospheres with the same global hydrogen fraction therefore do not necessarily present the same challenge to containment.
One problem, different safety functions
Once pressure, hydrogen inventory, spatial distribution and combustion are considered separately, it becomes clear why several different measures exist. They are not alternative implementations of the same task. Each acts on a different part of the problem.
| Measure or system | What it actually controls | What it does not ensure on its own |
| Containment spray | Reduces temperature, condenses steam, lowers pressure, and affects flows and mixing. | Does not remove the hydrogen inventory or demonstrate that the mixture is non-flammable everywhere. |
| Passive autocatalytic recombiner (PAR) | Gradually converts hydrogen and oxygen into water vapour at the catalyst surface. The heat of reaction establishes natural flow through the unit. | Does not ensure instantaneous homogeneity or the absence of transient local peaks. Capacity depends on the local mixture and gas inflow. |
| Igniters or deliberate ignition | Initiates combustion while the mixture remains in the lower part of the flammable range, limiting the accumulation of a larger quantity of hydrogen. | Does not prevent hydrogen generation or eliminate the thermal and pressure loads caused by combustion. |
| Atmosphere mixing | Reduces local concentration peaks and makes a larger volume of air available for dilution. | Does not reduce the total amount of hydrogen. A homogeneous mixture may still be flammable, and mixing may increase the volume of the flammable region. |
| Inerting | Reduces oxygen availability and thereby prevents sustained combustion, provided the inert condition is maintained. | Does not remove hydrogen or protect a non-inerted space into which hydrogen may subsequently migrate. |
| Monitoring | Provides data for diagnosing conditions, assessing accident progression and selecting actions. | Does not itself alter the atmosphere. A single measurement point, or a delayed measurement, may not represent local conditions throughout containment. |
A PAR is a good example of the distinction between removal and distribution. It chemically reduces the quantities of hydrogen and oxygen through the reaction:
2 H2 + O2 → 2 H2O
However, it acts only where the gas mixture actually reaches it. The heat it releases establishes natural convection and therefore also affects flow patterns, but a single PAR cannot turn a complex, multi-compartment containment into an instantly homogeneous vessel. If the local gas composition is flammable and the catalyst temperature is sufficiently high, a PAR may itself become an ignition source.
Its performance must therefore be assessed together with the possibility of such ignition and its consequences.
Similarly, mixing is not chemical removal. It can dilute a local pocket, but it merely redistributes the hydrogen. Deliberate ignition consumes hydrogen, but does so through combustion, which itself produces heat and pressure. Inerting removes one of the conditions required for combustion, namely sufficient oxygen, but it does not remove the fuel. Monitoring alone does not mitigate the hazard. It allows other measures to be selected on the basis of a better understanding of the actual conditions.
Safety therefore does not derive from the “strongest” individual system. It derives from the correct integration of functions: limiting hydrogen generation by maintaining core cooling; understanding the source and release rate; controlling distribution; reducing the quantity present; limiting the potential for hazardous combustion; and monitoring the evolution of the atmosphere.
Pressure can fall. Hydrogen can remain.
Pressure is an essential parameter. It tells us a great deal about the thermal state and mechanical loading of containment, and it is one of the key indicators of containment response. The error is not in monitoring pressure. The error would be to attribute information to it that it cannot provide.
Managing the containment atmosphere is therefore not a matter of chasing a single number. Pressure, temperature, the amount of hydrogen, concentration, oxygen fraction, spatial distribution and the possible combustion regime describe different parts of the same problem. The success of one system does not automatically mean the success of another.
Pressure can fall. Hydrogen can remain. Until we know where it is, how much is present and what mixture it forms, its hazard cannot yet be considered under control.
Note on visual material
Cover image: iStock, used under an appropriate commercial licence.
The diagrams and system illustrations are original works by Elite Studio 3D / By the Protocol. They were prepared specifically for this article as simplified conceptual illustrations of physical processes and functional relationships. They do not constitute technical documentation, design drawings or the actual configuration of any specific nuclear power plant.
Sources and further reading
U.S. Nuclear Regulatory Commission, Combustible Gas Control in Containment: final rule, 68 FR 54123, 2003.
Electronic Code of Federal Regulations, 10 CFR 50.44: Combustible gas control for nuclear power reactors.
International Atomic Energy Agency, Mitigation of Hydrogen Hazards in Water Cooled Power Reactors, IAEA-TECDOC-1196, 2001.
International Atomic Energy Agency, Developments in the Analysis and Management of Combustible Gases in Severe Accidents in Water Cooled Reactors Following the Fukushima Daiichi Accident, IAEA-TECDOC-1939, 2020.
OECD Nuclear Energy Agency, Thermal-hydraulics, Hydrogen, Aerosols and Iodine (THAI) Project.
OECD Nuclear Energy Agency, Status Report on Hydrogen Management and Related Computer Codes, NEA/CSNI/R(2014)8, 2014.
U.S. Nuclear Regulatory Commission, Hydrogen Combustion Characteristics Related to Reactor Accidents, NUREG/CR-2475, 1983.
Containment containment atmosphere containment spray hydrogen hydrogen combustion LOCA nuclear safety passive autocatalytic recombiners PWR severe accident
Last modified: September 18, 2026
