| 25 Mar 2026
The role of gas injection rate on the advective movement of gas in Opalinus Clay at the field scale
Abstract. This paper presents the findings of the Gas Transport (GT) field experiment investigating how Opalinus Clay (OPA) responds to gas pressure at the Mont Terri underground research laboratory in Switzerland. The experiment aimed to determine whether advective gas movement in OPA occurs through visco‑capillary flow or through the creation of dilatant gas‑filled pathways formed by localised deformation of the clay matrix. A combined strategy was adopted: laboratory studies under controlled conditions were used to inform the design and interpretation of a large‑scale field experiment. This paper focuses on the field component, summarising results from the first gas (helium) injection test from Day 447 to 1050. The experiment used a nine‑borehole array drilled perpendicular to bedding in Gallery 08. A central borehole served as the gas injection point, while eight surrounding boreholes were instrumented to monitor porewater pressure and rock deformation using fibre‑optic sensors, extensometers, and inclinometers. Key outcomes include: 1) Gas entry occurred at 4190 kPa, close to the minimum principal stress. This was followed by a reduction in gas pressure of 960 kPa over 37 days as steady state flow was established. Deformation was mainly down-dip of the injection borehole (towards ~225°), with deformation measured predominantly as borehole-axial strain (perpendicular to bedding), suggesting that the anisotropy of the OPA controlled the propagation of gas along bedding. Overall, the observations are consistent with advective gas movement through the formation of localised dilatant pathways; 2) Increased injection rates triggered a second, more energetic event at 3720 kPa, causing rapid pressure loss and deformation perpendicular to bedding. A large amount of helium was detected at the wellhead of at least four of the monitoring boreholes, confirming gas had reached the observation boreholes and migrated upwards to the gallery. The second event is interpreted as bedding‑plane splitting and the creation of a partially open macro‑feature with limited flow capacity; 3) A clear transition from dilatant pathway growth to bedding‑plane failure demonstrates that gas transport behaviour in OPA depends strongly on injection rate under the GT field-test conditions.
This is a genuinely interesting paper that reports on a rare and well-instrumented field-scale gas injection experiment in Opalinus Clay. The combination of fibre optics, pore pressure monitoring, extensometers, and helium tracing across a nine-borehole array is impressive, and the dataset clearly has significant value for the radioactive waste disposal community. The topic matters, and the observations, particularly the apparent transition between two deformation modes, are interesting.
That said, I found the manuscript difficult to follow in places, and I have concerns about internal consistency in the conceptual framework. The distinction between "dilatancy" and "fracture" as presented here needs rethinking, and some of the fracture mechanics reasoning contains statements that are physically problematic for the system under consideration. I think the paper can be published after moderate to major revision. Below some thoughts:
1) The dilatancy/fracture distinction is not as clean as presented
The paper sets up a fairly sharp distinction between dilatancy-controlled flow and fracture-controlled flow (Section 4.1, Fig. 11). But when I read the detailed descriptions, I struggle to see where one mechanism cleanly ends and the other begins. On p.16, dilatancy is described as "not fracture mechanics driven," yet in the same paragraph we read that "bonds between clay grains may be ruptured." Bond rupture is fracturing at some scale, it's just happening at the grain scale rather than the macro scale. The processes described seem to sit on a continuum of localised deformation rather than being mechanistically distinct.
I think the authors need to be more upfront about this. It would be perfectly reasonable to frame dilatancy and macro-fracture as end-members of a deformation spectrum, with the GT experiment providing evidence for a transition between them. But the current framing implies they are categorically different things, which creates confusion when the descriptions start overlapping.
2) The fracture mechanics reasoning needs revision
The statement on p.17 that "the fracture will grow at a velocity close to the speed of sound in the rock" is problematic. This might apply to dry, brittle fracture under dynamic loading, but it does not describe fluid-driven fracture in a saturated, low-permeability claystone. In this system, crack propagation is rate-limited by fluid supply to the crack tip, leak-off into the matrix, and pore pressure dissipation ahead of the front. Propagation velocities in such settings are orders of magnitude below the speed of sound. The entire hydraulic fracturing literature in tight formations demonstrates this. Likewise, calling gas fracture a "pure fracture mechanics-controlled phenomenon" mischaracterises the process. In a saturated clay with permeability in the nD range, any fracture propagation is inherently a coupled hydro-mechanical problem, you cannot separate the mechanics from the fluid flow. I'd ask the authors to either remove the speed-of-sound claim entirely or qualify it heavily, and to reframe the fracture discussion in terms of rate-limited, fluid-coupled propagation.
3) Event 2 pressures do not satisfy the fracture criterion as stated
This is perhaps my biggest concern with internal logic. The paper presents the classical tensile fracture criterion: failure when gas pressure equals σ₃ + tensile strength. The numbers given are:
So, gas pressure didn't even reach σ₃, let alone σ₃ + T. The authors acknowledge this and invoke "sub-critical fracture" or "bedding-plane splitting," but this effectively abandons the fracture framework that was just presented two paragraphs earlier without providing a clear alternative mechanical explanation. As a reader, I'm left wondering: if it's not classical tensile failure, then what is the failure mechanism? Shear-assisted opening along a weak bedding plane? Progressive damage accumulation? Reactivation of a pre-existing discontinuity? The paper needs to address this rather than noting the discrepancy and moving past it.
Consider also noting that "sub-critical" fracture growth is typically a slow, time-dependent process, which seems at odds with the rapid, energetic character of Event 2. If the authors want to invoke subcritical mechanisms, they should explain how subcritical growth leads to such a sudden event.
4) Alternative explanations for Event 2 deserve more space
The cumulative evidence for Event 2 being qualitatively different from Event 1 is reasonably convincing, the speed of strain development, the widespread deformation, the helium detections all point to something more energetic happening. I don't dispute that. But "fracture" is not the only explanation consistent with these observations. Other possibilities include: (1) rapid coalescence of the dilatant pathway network into a connected system (a percolation-type transition), (2) connection to the borehole EDZ or grout interfaces providing a sudden low-resistance pathway, (3) a series of pathway openings that collectively look fracture-like but aren't a single discrete feature
The fibre optic data are suggestive of something different from Event 1, and fracture (or bedding-plane splitting) is a reasonable interpretation. But the paper presents it as more or less single possibility. I think a more balanced discussion acknowledging the ambiguity would actually strengthen the paper, it shows the authors have thought carefully about alternatives.
5) Stress field uncertainty and its implications
The interpretation relies on gas pressure being "close to σ₃," but the stress values come from modelling of a nearby experiment and are acknowledged as poorly constrained. There is no discussion of how the Gallery 08 excavation perturbs stresses at the test location, nor how the stress state at Mont Terri (relatively shallow, with gallery effects) compares to anticipated repository conditions at depth.
Given that the entire mechanistic interpretation rests on pressure relative to σ₃, this uncertainty deserves more than a passing mention. Consider including (1) some estimate of stress perturbation from the gallery at the injection depth, and (2) a brief discussion of how the results might transfer (or not) to repository-depth conditions.
6) Are the injection rates representative?
The transition from dilatant flow to bedding-plane splitting is attributed to increased injection rate, which is the central finding of the paper. But the injection rates used (up to 12 ml/h, with stepwise increases designed to push the system) are far higher than anticipated gas generation rates in a repository, which are typically orders of magnitude lower and sustained over millennia rather than days. This raises an obvious question: is the Event 2 transition something that could occur under repository conditions, or is it an artefact of the aggressive injection protocol? The authors touch on this in the conclusions (stating the results are "site-specific" and should be interpreted as "process-level insights"), but I think this deserves a more discussion. What are expected repository gas generation rates? At what rate might dilatancy alone be sufficient to relieve pressure?
7) The borehole disturbed zone assumption seems optimistic
The paper assumes a BDZ of ~0.5 borehole radius (~50 mm). For OPA at Mont Terri, this seems low based on what's been reported in the literature for similar boreholes. If the actual disturbed zone extends further, one to several borehole diameters, then there are implications for pathway connectivity between boreholes, interpretation of early gas arrivals, and whether gas transport through "intact" rock is really being measured.
Could the authors provide references supporting the 0.5-radius estimate, and discuss what it would mean for the interpretation if the BDZ were larger?
8) Units and terminology
A few things that tripped me up:
These may seem like small points, but I got a bit lost understanding the results because of it.
9) Readability: the event numbering
This is partly a matter of taste, but I found the constant references to "Event 1.0a, 1.0b, 1.1, 1.2, 1.3, 2.0, 2.1, 2.2, 2.3" quite hard going. I should note that these readability concerns are not minor, they slowed down the review process and will likely affect other readers similarly. Every time one appears in the text, I had to flip back to Table 1 to remind myself what it referred to. The narrative would flow much better if the authors used brief descriptive phrases inline, something like "at gas entry" or "at the secondary pressure drop" with the event codes in parentheses for cross-referencing. The precision of the numbering system is fine for a data report, but in a journal paper the readability cost is high.
10) Minor points