Understanding of gas transport in clays for geological disposal of radioactive waste Open Access

Following international recommendations and best practices for a safe solution for the long-term management of radioactive waste, many countries have chosen to dispose of all or part of their radioactive waste in facilities constructed in stable, deep geological formations. Owing to their excellent properties for the confinement of contaminants, clays are considered as potential host formations for geological disposal in France, Belgium, and Switzerland. In Switzerland and France, these are indurated clays, and in Belgium, poorly indurated clays are considered as reference options.

Considerable amounts of gas can be generated in a geological disposal facility (GDF) for radioactive waste. The largest fraction of the gas is expected to be hydrogen produced through the anaerobic corrosion of steel and reactive metals present in the waste, in their packaging and in the engineered barrier system. Radiolysis and the degradation of organics also produce gas. Even though the gas production processes are generally slow, it is important to verify that these will not be detrimental to the good functioning of the disposal system. The low permeability of clays, which is favourable to the containment function of a repository, also limits the evacuation of the generated gas. It is possible that gas could be produced at a faster rate than it can be removed through the engineered barrier components, resulting in the development of a pressurised gas phase within the repository. If the pressure of accumulated gas in the repository is too high, gas could then escape from the repository to the host formation by creating gas-specific pathways through the engineering barrier system and/or the host formation.

To properly evaluate the impact of gas on the functioning of a deep geological repository, increasing the understanding of gas transport through low-permeability porous materials such as clayey materials was studied in the first European Joint Programme on Radioactive Waste Management (EURAD) (2019–2024) through the project EURAD-GAS. The main objectives of this project were to improve the mechanistic understanding of gas transport processes in natural and engineered clay materials, their couplings with the mechanical behaviour and their impact on the properties of these materials, but also to evaluate the gas transport regimes that can be active at the scale of a geological disposal system and their potential impact on barrier integrity and repository performance.

After some general features about the concept of geological disposal of radioactive waste in clay host formations in Section 2, this article provides a concise and integrated overview of the current state of knowledge on gas transport processes in clays as consolidated at the end of EURAD-GAS (Section 3). This common understanding, valid for three host formations considered in this article, provides the keys for transferring this knowledge gained from laboratory and in situ experiments to configurations that are commonly expected by repository designs. On the basis of shared visons on how gas is evacuated from a GDF in clay formations, assessment of the impact of gas on the functioning of a geological disposal is finally proposed in Section 4.

Geological disposal involves placing high-level and long-lived radioactive waste deep underground in stable geological formations to protect people and the environment from radioactivity. As illustrated in Figure 1, it involves the construction of a GDF, an engineered structure comprising multiple layers, whose concept depends on the type of host formation and the characteristics of the waste. The disposal system consists of the waste, the disposal facility, and the host formation, together providing multiple barriers. The multi-barrier approach ensures that the disposal system is not dependent on any single barrier, even if the main barrier remains the geological layer itself. The disposal system and its overlying layers fulfil at least (i) the containment of the radionuclides by enclosing the waste within packages and/or overpacks, (ii) the retardation of the radionuclides to limit the releases of radionuclides to a sufficient extent to prevent harm to people and the environment during the post-closure period, and (iii) the isolation of the radionuclides by providing physical separation of the wastes from people and the surface environment to a sufficient extent to protect from irradiation and the disruption by external events and environmental evolution (Levasseur et al., 2021, 2024a).

The geological disposal systems currently being developed or studied in France, Switzerland, and Belgium rely on clay host formations as geological barriers against the migration of radionuclides and chemical contaminants towards the surface environment. In Switzerland and France, indurated clays (respectively, the Opalinus Clay and the Callovo-Oxfordian claystone) have been selected to host high and intermediate level radioactive waste. In Belgium, poorly indurated clays are investigated as potential host formations, the Boom Clay is one of them. These three deep clay formations are known to present features and properties which are favourable for a radioactive waste disposal notably because of their low permeability, strong retention capacity, and self-sealing capacity. The specific behaviour and properties of the Opalinus Clay, the Callovo-Oxfordian claystone, and the Boom Clay as well as the current geological disposal concepts in Switzerland, France, and Belgium are detailed in Levasseur et al. (2021).

The transport of gas in clays has been largely studied at European level in the frame of research through various projects dealing with the geological disposal of radioactive waste. Among the projects funded by the European Commission, the following were particularly noteworthy: the MEGAS (Ortiz et al., 1994, 1997; Volckaert et al., 1994, 1995), the PROGRESS (Rodwell, 2000), and the FORGE (Norris, 2013) EC projects and more recently the EURAD-GAS project, as part of the European Joint Programme EURAD on Radioactive Waste Management (EURAD) (2019–2024) (Levasseur et al., 2024b).

The state of knowledge on gas transport processes collected over these last decades allows a fine description of each process mechanisms. As already stated by Marschall et al. (2005), gas initially present and gas generated within a geological repository will dissolve in the pore water until the concentration reaches the solubility limit under the prevailing temperature and pressure conditions. This dissolved gas will migrate by diffusion through the water and be carried along with pore water in case of water movement (advection). Diffusion is driven by a concentration gradient in the pore water. Advection of dissolved gas is limited in clays as water usually hardly moves because of the very low hydraulic conductivity. Gas can also react with or sorb onto some mineral phases of natural materials. If the gas generation rate exceeds the rate of evacuation of dissolved gas, a distinct free gas phase will form. Gas can migrate as such through the disposal system by different mechanisms of advective transport of gas phase: visco-capillary two-phase flow, dilatancy-controlled gas pathways, or gas fracturing.

Building on this knowledge, the EURAD-GAS project mainly focused on gaining insight in which transport mechanisms prevail under which range of conditions and on understanding how the coupling between pressure in the liquid and gas phases and the stresses in the solid phase control gas transport. A concise overview of the conclusions of EURAD-GAS is provided in this section for all gas transport processes. For each, particular attention is given to the shared understanding developed during EURAD-GAS, valid for the three clays considered in this article.

Where water and gas are in contact, gaseous species dissolve in the water according to Henry’s law. The diffusion of dissolved gas is driven by its concentration gradient in the pore water, following Fick’s law. The proportionality factor between the rate of diffusion (also named flux) and the gradient of dissolved gas concentration is called the diffusion coefficient. In water-saturated porous media, the effective diffusion coefficient of dissolved gas, that is, the coefficient of diffusion through the water-saturated material, is related to the morphology of the pore network (porosity, tortuosity, constrictivity) and is anisotropic (Jacops et al., 2017a, 2017b).

Today, the methodology to determine diffusion parameters of dissolved gases in water-saturated media is well developed and mastered through a broad range of clays and under various mechanical conditions (Harrington et al., 2025; Jacops, 2018; Jacops et al., 2015). Comprehensive data sets of diffusion coefficients are notably available for several gases in the three clays considered in this article (Levasseur et al., 2021). All of these coefficients were obtained from laboratory tests. The upscaling of this knowledge to larger scale remain an issue that will be soon addressed by the NEMESIS experiment (neon diffusion in MEGAS in situ). This in situ diffusion experiment developed during EURAD-GAS proposes to measure diffusion of neon on a metre scale and in different directions in Boom Clay at the level of the HADES underground research laboratory (HADES URL). This experiment was started in September 2023 and is still running at the time of writing this article. Nevertheless, the first results already suggest that the diffusion coefficient estimated in situ is around 80% of the value measured in laboratory experiments – which is very satisfactory considering the uncertainties associated to the estimation of diffusion coefficient (Jacops et al., 2025).

In the frame of EURAD-GAS, knowledge on diffusion of dissolved gas was also extended at laboratory scale to refine assessment of the anisotropic properties of diffusion process and to examine the diffusion of gas in unsaturated media. By performing diffusion tests in different directions with regards to bedding planes of Boom Clay, Harrington et al. (2025) have shown that diffusion of helium occur preferentially along bedding planes with ≈60% of the diffusional capacity of Boom Clay moving parallel to bedding. Examining the data as a whole also permitted to highlight coupling between diffusion and intrinsic permeability. A fundamental relationship was suggested between permeability and diffusivity as shown in Figure 2. If correct, such relationships could be used to predict diffusivity across a range of material and permeability scales (Harrington et al., 2025).

Gowrishankar et al. (2023) have developed a setup to measure diffusion of gases in slightly unsaturated clay samples (for a range of saturation degree between 73% and 100%) in the frame of EURAD-GAS. Results with helium and argon showed that gas diffusivity, as a combination of diffusion of dissolved gas and gaseous gas, increases only slightly, about 20%, when desaturating the sample towards 73% (Jacops and Kolditz, 2024); this would mean that the rate of gas transport is marginally affected by a partial desaturation of clay-based materials.

When a free gas phase forms, advective transport mechanisms of this gas phase may occur by visco-capillary two-phase flow, formation of dilatant pathways, or gas fracturing. The advective gas transport mechanism that will prevail in an initially fully saturated clay host rock is largely controlled by the fabric and mineralogy of the clay, as well as by in situ conditions, such as current stress state and stress history (Cuss et al., 2017; Gonzalez-Blanco et al., 2022; Harrington et al., 2017; Wiseall et al., 2015). These latter influence the capacity of gas to enter porous media (gas entry pressure), the mobility of gas (permeability), and by the susceptibility to deformation resulting from the passage of gas as described by Marschall et al. (2005).

In saturated clays, permeability is low and gas entry pressure is high. Gas can flow only after substantial pressure build-up is reached. Even in partially saturated clays, gas flow will occur with very high gas entry pressures due to the poor connectivity of the gas-filled network of macropores. When gas pressure exceeds the sum of the local stress and tensile strength of the clay, irreversible deformations can be generated and gas invasion occurs through the localised creation of new porosity, referred as dilatant pathway (Cuss et al., 2014; Harrington et al., 2012; Harrington and Horseman, 1999; Horseman et al., 1999).

In the frame of EURAD-GAS, evidence for dilatancy-controlled gas transport in Boom Clay, Opalinus Clay, and Callovo-Oxfordian claystone has been reported by Marschall et al. (2024), which builds confidence in the existing conceptual frameworks as described in Levasseur et al. (2021). As already stated by Cuss et al. (2014), gas flow rates within clay are generally low, if at all, during the initial gas injection stage of experiments conducted at sample scale during EURAD-GAS. The gas transport capacity of the clay sample increases as dilatancy-controlled gas pathways progressively develop until gas breakthrough is reached (see Figure 3). The invaded material may undergo deformation in response to gas pressure build-up, leading to a redistribution of fluid pressures and stress in the porous medium (Gonzalez-Blanco et al., 2022; Llabjani et al., 2025; Marschall et al., 2024). Dilatancy-controlled gas pathways tend to be multiple, taking advantage of local defects or plane of weakness within the materials (Figure 3). According to Gonzalez-Blanco et al. (2023), these planes of weakness could be the bedding planes in intact Boom Clay as illustrated in Figure 4.

More specifically, new insights were gained in EURAD-GAS about the volumetric behaviour of poorly indurated and indurated clays in response to water/gas injections. CIMNE conducted water/gas injection tests on Boom Clay samples and EPFL did similar tests on Opalinus Clay in a high-pressure oedometric cell, allowing for accurate measurements of the volumetric behaviour of the tested material. Both teams demonstrated independently with their experimental setups that clayey media exhibit a distinct drained/undrained volumetric response depending on the applied gas pressure build-up rates (respectively slow/high). The experiments of EPFL and CIMNE showed that the slower the rate of gas injection (i.e. gas invasion occurs in drained conditions), the less the clay expands, and thus the weaker the effect of gas-induced micro-fracturing (Gonzalez-Blanco et al., 2025).

In addition, broad consensus has been gained among the experimentalists of EURAD-GAS, confirming that gas transport through clayey materials is barely associated with any displacement of pore water from the clay matrix (Marschall et al., 2024). Corresponding evidence is not only based on the gas injection experiments carried out during EURAD-GAS, but can also be deduced indirectly from the huge existing databases, which have been compiled in recent years to characterise the water retention behaviour of clayey barrier materials (e.g. Levasseur et al., 2021). Mercury intrusion porosimetry and water retention measurements on all sorts of clays indicate that a high mass fraction of typically 80%–90% of the pore water occupies the meso- and micropores of the saturated clay matrix. This pore water can be hardly displaced by an invading gas phase.

Multiple experimental evidence has also been provided in EURAD-GAS to point out the impact of the in situ state conditions (in particular pore pressure and stress) on the gas transport mechanisms in clayey host rock. Zhang and Talandier (2023) developed empirical relationships between the gas breakthrough pressure, water permeability, and the effective confining stress for Callovo-Oxfordian claystone. Agboli et al. (2024) performed visualisation experiments on gas transport in artificially fractured Callovo-Oxfordian claystone in a triaxial cell to emphasise the fracture mechanisms and how gas permeability is affected for a wide range of stress conditions parallel and perpendicular to bedding. For different flow configurations, a systematic determination of the threshold for gas breakthrough and the formation of dilatancy-controlled gas pathways was carried out, representing the turning point from which the opening induces a significant increase in the gas permeability. They showed that the gas permeability increase is greater when the main principal stress is parallel to the bedding planes due to the development of bedding parallel microcracks (Agboli et al., 2024). This confirms in triaxial conditions the role of bedding planes observed by Gonzalez-Blanco et al. (2022) in oedometric conditions. In addition, the experimental setups developed by Agboli et al. (2024) and Zhang and Talandier (2023) in triaxial conditions as well as by Gonzalez-Blanco et al. (2023) in oedometer conditions have shown that the gas transport through the fractured clay does not limit the self-sealing ability of the clays. This ability was observed even on large fractures that could be associated with the excavation damaged zone (EDZ). Nevertheless, empirical observations suggest that gas entry pressures may still be lower in the sealed zone than in the intact clay, as already suggested by Horseman and Harrington (1994). Should gas pressure rise again, reopening of gas pathways may happen at lower pressures due to the loss of cohesion, a residual gas phase, or high gas saturation in the pore water that may modify the phase transition.

One of the outcomes of the FORGE EC project was that the modelling of gas transport by visco-capillary two-phase flow alone was not sufficient to explain the gas transport in clayey materials as observed at laboratory scale (Shaw, 2013). When the gas injection pressure increases and the material is subjected to deformations, the transport properties of the solid skeleton (permeability, relative permeability, capillary pressure relationship) can no longer be viewed as invariants but vary with deformations. To capture these phenomena, the development of relationships between transport properties and deformations was pursued in EURAD-GAS in the continuity of the work initiated in FORGE (see, for instance, Gerard et al., 2014; Pitz et al., 2023; Radeisen et al., 2024; Rodriguez-Dono et al., 2024 among others).

New insights were particularly gained during EURAD-GAS by developing advanced approaches focusing on the transport of gas in the EDZ and how gas pathways initiate, propagate, and close along planes of weakness. To model the strong interactions coupling the flow and transport properties to the mechanical behaviour, Corman et al. (2022) developed a second gradient two-phase flow hydro-mechanical (H2M) model tackling the multi-physics couplings related to gas transfers and fractures development. In this approach, the EDZ is reproduced by shear strain localisation bands using a microstructure enriched model with a second gradient approach. The gas transport is captured by a biphasic fluid transfer model. The impact of fracturing on the flow properties is addressed by relating the permeability and the water retention curve to mechanical strains. Applied to the numerical modelling of a drift in the COx claystone, they have demonstrated the non-negligible impact of the hydro-mechanical couplings inherent to the EDZ on gas migration (Figure 5).

To represent the initiation and propagation of gas pathways out of the EDZ, Corman et al. (2024) proposed another approach based on a multi-scale hydro-mechanical (HM) model capturing the influence of the microstructure features on the macroscopic gas flow, and especially on the emergence of preferential gas-filled pathways. In their approach, a constitutive model for partially saturated clay materials is developed from experimental data to perform the modelling of a representative element volume, which is then integrated into a multi-scale scheme using homogenisation and localisation techniques for the transitions to the macroscopic scale. Following similar idea, the initiation and propagation of gas pathways has been modelled by Liaudat et al. (2023) through a new pneumo-hydro-mechanical model. This modelling approach uses continuum elements to represent the mechanical and flow processes in the bulk clayey material and zero-thickness interface elements to represent existing or induced discontinuities (cracks or gas pathways). Using these tools to the modelling of a gas injection, tests have allowed to improve the mechanistic understanding of gas transport processes in natural clay barriers. The simulations of Corman et al. (2024) have particularly highlighted two essential aspects in the development of preferential pathways. On the one hand, the more continuous the connectivity between the disturbed planes, the faster the gas flow through this discrete zone. On the other hand, these connections between the planes of weakness must be repeated regularly to ensure a rapid gas propagation on a larger scale. Otherwise, the fast mechanism of gas transport by advection through the developed pathways is supplanted by the diffusion of dissolved gas in the liquid phase, which is a slower and less damaging mode of gas transfer.

To go further in process understanding, Quacquarelli et al. (2024) proposed to extend a hydro-mechanical model for an interface element to model self-sealing processes of fractures. In this approach, two low-density and relatively compressible zones around the fracture are included in the interface elements. Its application to a self-sealing test on artificially fractured COx samples shows that the injected water can penetrate the clay, first involving the EDZ and then the rest of the sample, thanks to the transmissivity of the clay. This model is able to reproduce the evolution of crack opening during wetting and drying tests as well as reproducing the influence of the confining pressure on the self-sealing process.

All these models developed during EURAD-GAS have shown to be capable of reproducing experimental results, at least qualitatively a priori, or quantitively through a posteriori calibration. The predictive capabilities of these numerical models are still inherently limited because of the small scale, local and discrete nature of gas pathways that are difficult to foresee and characterised (Corman, 2023).

Despite these limitations, the numerical modelling of gas transport at the scale of a GDF were envisaged in EURAD-GAS with the perspective to bound the evolutions of gas pressures and fluxes at repository scale (Wendling et al., 2024).

In line with the conclusions of the FORGE EC project, modelling approaches and tools that have been compared in EURAD-GAS confirm that repository-scale modelling of gas transport explicitly representing all couplings with the mechanical behaviour of the barriers and for the development of individual pathways through clayey materials is currently out of reach. It will probably remain so in the foreseeable future, both for reasons of computational cost and because of the local nature of gas pathways initiation and propagation (which cannot easily be represented by continuous numerical approaches generally used at repository scale) (Wendling et al., 2024).

Modelling at the scale of the repository is nevertheless possible using simplified, modular approaches guided by conceptual gas transport storyboards and based on dissolution, diffusion, and two-phase flow phenomena (as proposed, for instance, by Pitz et al., 2024). Smaller scale models can also be built for the different components of the system through which gas can pass through. These component models can explicitly describe the mechanical couplings and provide a better hydro-mechanical understanding of the whole system (see, for instance, Corman, 2023; Saâdi, 2024). Gas pressure evolution and fluxes estimated from these smaller scale models can then be integrated into repository-scale models as complement to the more conceptual storyboard approaches (Wendling et al., 2019, 2024).

The main specificity of the EURAD-GAS project is to have proposed a common vision on the issues related to gas transport in clays. Shared understanding of phenomenological descriptions of gas transport at repository scale was notably assembled to assess the consequences of the transport of gas on the mechanical integrity of the host rock. This shared understanding permitted to revisit the drawing of Marschall et al. (2005) with the perspective of gas flowing from a disposal gallery into a clay host rock. This conceptualisation of gas transport at repository scale and the potential consequences on the functioning of a GDF is summarised in the following and illustrated in Figures 6–9. Figure 6 presents a schematic representation of a gallery, the EDZ with its discontinuities, and the clayey host rock with bedding planes. The gas flux originates from the gallery.

On the basis of current knowledge on gas transport in clay and on what could be the architecture of a GDF in clay formations, it is obvious that the dissolution and diffusion of gaseous molecules through the pore water is an inevitable transport mechanism at repository scale as illustrated in Figure 7. Even at low gas pressures, evacuation by diffusion of dissolved gas reduces the amount of free gas in the disposal system to be dealt with by other transport processes. If gas generation is sufficiently slow (concept dependent) and transport of the dissolved gas is sufficiently rapid (clay-dependent), all the generated gas could potentially dissolve, and no free gas phase will form. It is widely accepted that this process has no impact on clay barrier integrity.

The solubility of hydrogen being low, the capacity of gas transport by diffusion of dissolved gas in water-saturated clay is limited, allowing the formation of a fee gas phase. There is a consensus that the transport of gas by visco-capillary two-phase flow is a priori possible through the porous media and/or through discontinuities as suggested in Figure 8. However, for what concerns geological disposal systems, it is expected that, if visco-capillary two-phase flow develops in clay host rocks, it will be mainly localised within the discontinuities of the EDZ (Levasseur et al., 2024b). There is no clear evidence of significant visco-capillary two-phase flow through mechanically undisturbed host rock. By definition, visco-capillary two-phase flow in a porous media involves the displacement of the liquid phase (water) by the gas phase without irreversible deformation of the media. Hence, no alteration of the properties of clay barriers materials is expected for that transport mechanism either, although displacement of water and consequent desaturation of the barrier is expected as a result.

Depending on the gas pressure and of the speed of pressure build-up, gas production rate could however be balanced steadily by a variation of pressure in the pore network and the newly created pore volume at the pathway tips. If so, the pathways will be controlled by dilatancy, and may propagate within the EDZ reopening existing fractures. In some cases, it can even extend towards natural flaws in the rock matrix such as bedding planes and tectonic features if existing as proposed in Figure 9. The formation of such dilatancy-controlled gas pathways depends on the local stress field and can be understood as the generation of new porosity and/or coalescence of pores as a result of local, gas-induced, stress redistribution (Busch and Amann-Hildenbrand, 2013; Harrington et al., 2003; Harrington and Horseman, 1999; Horseman et al., 1999; Marschall et al., 2024). This change in the microstructure of the material, through localised deformation (due to the dilatant ‘opening’ of pathways and also the localised compaction of the surrounding pores to accommodate this) increases the exchange surfaces with the clay, giving rise to significant enhancement of diffusive transport. Moreover, very little of the interstitial water present in the pores prior to the formation of dilatant pathways is expected to be displaced along the pathway (Horseman and Harrington, 1994; Jacops et al., 2014; Rodwell, 2000).

It is well recognised that clays exhibit the favourable feature of self-sealing after sustaining mechanical failure, and this ability is still evidenced in experiments showing episodic gas flow, correlated with increase and decline in gas pressure. Consequently, because of the local nature of the perturbations induced by dilatancy-controlled gas pathways and because of the observed self-sealing, it is expected that the integrity of clay barriers will not be significantly impacted by the passage of gas through this transport mechanism, even in the case that gas would be expelled in a cyclic fashion and locally perturb the stress field. The hydraulic properties of this zone will be mainly unchanged after gas passage.

Experimentally, where high gas injection rates are associated with rapid gas pressure build-up, it has been observed that gas fracturing might be occurred with propagation velocities up to the shear velocity of the material (Valkó and Economides, 1995). Nevertheless, as such high pressurisation rates are not expected in normal repository scenarios, while gas fracturing is most unlikely.

From this current state of knowledge, EURAD-GAS concluded that no significant impact on the long-term integrity of clay barriers is expected after the passage of gas provided that the repository is designed in such a way that it favours the progressive release of gas. It means that gas-induced damage is not detrimental to repository performance, so long as gas pressures can be managed with design. Possible ways to mitigate with (i.e. preserve the host rock integrity) too high gas pressures could be to design disposal facility to increase their potential gas transport capacity, and/or to limit as much as possible the quantity of metal present in the repository at closure (Levasseur et al., 2024b; Wendling et al., 2024).

This article summarises the current knowledge of gas transport through clays and how this knowledge is used in the context of the development of geological disposal systems in France, Switzerland, and Belgium. It is based on the shared understanding as developed by the partners of EURAD-GAS project of the European Joint Programme EURAD on gas transport processes and their controls in clay host formations like Callovo-Oxfordian claystone, Opalinus Clay, and Boom Clay. From these common views, it gives indications of how gas can be transported and what could be its impact at repository scale.

This article highlights how EURAD-GAS has increased confidence in the overall understanding of gas transport in clayey materials, building on the FORGE EC project and beyond and illustrates how EURAD-GAS has improved its integration into the conceptualisation process for the different components of a disposal system, supporting and justifying the use of robust evaluation approaches. This project has confirmed that the fundamental gas transport mechanisms that can develop in different clays are similar and so that knowledge gained over the last decades is of relevance for all disposal systems that include clayey barriers. In that sense, EURAD-GAS has made a major step forward on the mechanistic understanding of the transport of gas in clays.

EURAD-GAS was part of the European Joint Programme on Radioactive Waste Management (EURAD). EURAD has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement no. 847593.