The pressure Gruffalo: how well do we know the basin-scale limits of geological carbon storage?

Introduction

How much carbon dioxide can the world’s saline aquifers actually hold? The question sounds like an inventory problem and is often treated as one. The current global storage catalogue is distributed across 71 countries that are almost entirely unappraised (OGCI, 2025). The catalogue now stands at 14,256 GtCO2, or roughly 14 trillion tonnes, of which about 96% lies in deep saline aquifers and about 95% is undiscovered. That is a resource number; and, with the exception of a few per cent, simply a pore volume weighted by an efficiency factor. The enormous resource also happens to be the wrong number to base expectations on when approaching the basin-scale potential of regional storage. While it faithfully answers the question of how much CO2 the pore space might accommodate in an unbounded and open system, it is hyper-concentrated into just two countries, 8 Tt in the United States and 3 Tt in China. It also does not speak to the challenge of how much a saline aquifer will accept before pressure, not space, imposes a practical limit. Which brings us to scary stories.

The Gruffalo is a children’s fable in which a mouse, walking through the wood, invents a monster to frighten off the animals that would eat it. The monster is imaginary… until, to the mouse’s alarm, it turns out to be real; but also, crucially, something the mouse can outwit. The basin-scale pressure limit of saline aquifers has been treated much the same way. We argue that the monster is real, but that it is more tractable than the worst-case scenario suggests, and that how well we know the limit is a question with a measurable answer.

Consider the two ways the question is usually answered in the absence of sufficient data, which is almost always the case. The optimist assumes the aquifer is effectively open: displaced brine migrates away indefinitely, pressure does not bind, and capacity is simply pore volume times the aforementioned storage efficiency factor. Call this the $P_0$ pole. The basin looks like a mushroom farm of neatly juxtaposed projects (Chadwick et al., 2004; Nilsen et al., 2015). The pessimist assumes the opposite: a closed-system box in which every tonne injected must be accommodated by compression, causing pressure to climb without relief, and capacity to collapse to a fraction of a per cent of the pore volume (van der Meer & van Wees, 2006; Ehlig-Economides & Economides, 2010). Compressor costs treble and red-lining operational limits pause, then ultimately cancel, the injection schedule. Call this the $P_{100}$ pole. Both are stated in good faith; both are internally consistent; and both are assumed in the absence of any direct measurement of the aquifer’s boundary. They are the Gruffalo in its two forms, dismissed or terrifying, and neither captures the middle ground.

The underlying physics is fairly straightforward (Figure 1). Injection creates two nested footprints. The near field is the CO2 plume itself: a two-phase region, perhaps 10-15 km across after decades of injection, where relative permeability, residual trapping, dissolution, and buoyancy govern how efficiently CO2 fills the accessible pore space. The far field is the pressure plume: a single-phase region of displaced brine that propagates far beyond the CO2 to around 20-200 km after several decades of injection in a representative basin (Zhou & Birkholzer, 2011), and whose governing metrics are not relative permeability but absolute permeability, total compressibility, and, decisively, the boundary conditions of the saline aquifer.

For multiple projects in large saline aquifers, it is the far field that ultimately binds. The near-field storage efficiency is worth establishing accurately, but is ultimately subordinate. What determines whether a basin stores one gigatonne or fifty is whether the pressure plume can bleed off, and that is essentially a property of the boundary, which is the dynamic time-dependent metric most entries in the catalogued resource assume away. Where a representative saline aquifer sits between the open and closed poles is not a point but a probability distribution, one we cannot easily measure directly, but we can accurately constrain for the general case.

Four conditions, one problem

Four conditions frame our analysis and each challenges the headline resource number: an undrilled saline aquifer is not assumed to be open (2.1); knowing the aquifer is not knowing the boundary (2.2); resource is not capacity (2.3); and that both poles are simplifications at the limit, leaving a semi-closed approximation as the more realistic middle-ground assessment (2.4). We show that the middle is knowable (3) and the solution lies within the modelling lineage that already brackets the space (4). We close by asking how scary the pressure Gruffalo is.

An undrilled aquifer is not necessarily open

When a saline aquifer has not been drilled and tested, which, for 95% of the catalogue, it has not, the working default is to treat it as open. This leans into reasonable, if optimistic, expectations around basin-scale prospecting; exploration teams select for good reservoir porosity and permeability, and crucially, suitable caprocks – typically mudrocks and shales. This is not often stated as an assumption, but may be folded into the storage efficiency or basin volume factor, a similar metric used in geospatial models, e.g. Gidden et al. (2025). In petrophysical terms, assuming an open system is the pressure equivalent of pasting a regional porosity-permeability transform onto an unlogged interval: a reasonable first approximation, but a placeholder nonetheless.

For a saline aquifer to behave as open in response to an injection programme, or multiple such programmes juxtaposed across space and time in a storage basin (Bump & Hovorka, 2024; Thibeau & Bump, 2026), the brine must find a path, laterally to a permeable margin or outcrop, or vertically through the seal, and quickly enough to relieve pressure within the accounting horizon. That requires either a very large saline aquifer to accommodate the pressure increase or suitably permeable boundaries that allow the pressure plume to dissipate beyond the aquifer. What would settle the matter is a far-field pressure response measured over a relevant timescale, the one observation an undrilled prospect cannot supply. So, the honest starting position is agnostic, neither open until proven closed, nor simply closed.

Knowing the reservoir is not knowing the boundary

A data well that declares ‘sandstone, 100 mD’ anchors a storage estimate, and describes the near field at the point of observation. It tells you the CO2 will go in, always a good thing; but it says little about where the displaced brine goes. Injectivity and boundary response are different subsets of the same physics, set by different rocks, in different places: the first by the reservoir at the well, the second by the seal and lateral margins.

The variable that governs the far field is the absolute permeability of the seal, and for mud-dominated systems this spans six orders of magnitude, from nanodarcy to millidarcy (Neuzil, 1994; Yang & Aplin, 2010). Within that span a saline aquifer flips from effectively closed to effectively open. Crucially, the flip is not spread evenly across the six orders: capacity is a saturating function of seal permeability. Below roughly 10 nD the system is closed and tighter seals change nothing; above roughly 100 µD the system is open and a higher permeability also changes nothing with respect to pressure; the transition is constrained to the space between (Figure 2). A wide permeability distribution therefore collapses to narrow resource estimates at both ends and bridges from one to the other somewhere in the muddy middle. Where within that range the flip happens is determined by how fast the seal and lateral boundaries can bleed off pressure versus how fast the pressure in the aquifer is being generated. Writing the leakage timescale against the accounting horizon ${\tau }_h$ gives a critical permeability or onset:

$${\kappa }_{crit}=\frac{b^2\phi \mu c_t}{{\tau }_h}\text{\hspace{1em}(1)}$$

where $b$ is seal thickness, $\phi$ its porosity, $\mu$ brine viscosity, and $c_t$ total compressibility. The dimensionless group behind the onset is the seal’s Fourier number:

$$\Lambda ={\kappa }_{seal}{\tau }_h/(b^2\phi \mu c_t)\text{\hspace{1em}(2)}$$

where $\Lambda =1$ marks the onset of the flip (Zhou & Birkholzer, 2011). Evaluated at a century of injection activity in a basin, this places the onset at around 0.1 µD, a very common permeability for mudrocks in a storage window of around 1-3 km depth. Two studies land on the same value: the derived $\Lambda =1$ onset and empirical threshold above which diffuse leakage becomes important at around 0.1 µD (Birkholzer et al., 2009); the same 0.1 µD as restated in review (Birkholzer et al., 2015); and a similar microdarcy threshold indicated by petroleum systems analysis (Cavanagh & Wildgust, 2011). The shape of the transform function is well constrained; the exact onset value is not a single number but an order-of-magnitude range, shifting with seal thickness, compressibility, and rate (Figure 2). The flip sits in the microdarcy range. And because onset carries ${\tau }_h$ (1,2), the switch is not fixed; it slides toward tighter seals the longer we account for dissipation, and away from tighter seals the faster we inject.

Resource is not capacity

The third distinction is one of definitions. The Storage Resources Management System (SRMS) reserves capacity for the most mature storable volumes under defined conditions and keeps it as a distinct set of tiers from the broader resource (SPE, 2022). The catalogue records both, but at present is 99.9% resource, a maturity distinction; a storage project delivers justified capacity once it becomes investable. The storage efficiency factor that estimates the resource from the earliest exploration phase through to effective and contingent resources is partly set by boundary condition and partly by resource maturity (Table 1). At the capacity level, the highest level of maturity, a storage efficiency heuristic is not required, as the expected tonnage is usually estimated from bespoke reservoir simulations for the storage site.

Read as a matrix, Table 1 shows storage efficiency changes by at least an order of magnitude, from a closed floor at 0.2% of aquifer pore volume, via a semi-closed middle of roughly 1-3%, to an open ceiling near 10%. Two caveats. First, the storage efficiency factor bundles every near-field variable (sweep, buoyancy, displacement) with the boundary condition layered on top (Goodman et al., 2011; Bachu, 2015), so they corroborate the structure and span of the boundary effect, but do not represent a pressure-only percentile. Second, the semi-closed columns are essentially placeholders that defer to case study simulation at the top tier of site-specific capacity, which is to say the heuristic concedes the middle is not resolvable as a single number.

Viewed as a cumulative curve, the same information reads as a distribution over the boundary – Figure 2. The closed pole coincides with the conservative $P_{90}$ and the open pole with the optimistic $P_{10}$; the boundary is a percentile axis for capacity.

The poles are simplifications; the middle is probable

It follows that the two poles are hardly rival theories, but simply the endpoints of a single equation for usable capacity: a near-field volumetric estimate and a smaller far-field pressure one:

$$C_{usable}(t)=\min [C_{vol}(k_r),C_{press}({\kappa }_{seal},\mathrm{boundary},t)]\text{\hspace{1em}(3)}$$

with the pressure ceiling in the Bump & Hovorka (2024) form:

$$C_{press}=Ah\phi \rho C_t\Delta P\text{\hspace{1em}(4)}$$

where $A$ and $h$ are the aquifer area and thickness, $\rho$ the CO2 density, and $\Delta P$ the tolerable pressure space which is capped by the fracture limit.

Now the rival positions fall out as limits. At the open pole ($P_0$), leakage through the seal and boundaries relieves pressure faster than injection builds it, the closed-system ceiling of (4) never binds, and the minimum selects $C_{vol}$: only near-field efficiency matters, the optimist’s mushroom farm. At the closed pole ($P_{100}$), nothing escapes by definition, (4) applies as written, and $C_{press}$ will usually set the minimum: the pessimist’s cancelled injection schedule. Each pole settles the minimum in advance (Figure 3), which is why each gives a firm number. Real aquifers sit in the semi-closed middle, where both terms are finite and comparable and (3) must be evaluated rather than short-circuited.

Can we say where the middle sits? We can bound it and say a little about the bounds. The physics of condition (2.2) places the bulk of the seal-permeability distribution in a prior band around $P_{30}$–$P_{70}$ a boundary condition that is useful as a default. We cannot, however, derive a single efficiency-factor, because the band’s position inside the range depends on two things we can name but not measure per basin: whether the effective seal is intact matrix, which skews toward the closed pole ($P_{100}$), or upscaled to reflect a natural population of small fractures, which skews toward the open pole ($P_0$); and the accounting horizon, which slides the whole distribution as (1) requires. The available output is a probability distribution with dependencies attached, never a point.

The poles are convenient, but they are also unlikely, collapsing to a bare minimum or absolute maximum by assuming the answer. The middle is wide because it holds uncertainty, the known unknowns, if you will. That width is the frontier of unexplored basins and undrilled saline aquifers that narrows on variables we can state in advance.

The monster really is out there

We already operate CO2 storage projects, and they populate the spectrum from closed to open (Figure 3). None is a controlled experiment, but each measures a different aspect of the far field, and together they indicate that the pressure limit is not imaginary. So far, they run the range from failed to fixed and managed success.

At the closed end sits In Salah, Algeria, with one of the richest pressure datasets we have. CO2 was injected into the flanks of a gas field with a tight reservoir and low-permeability overburden. The pressure footprint was literally visible from space: satellite radar interferometry recorded centimetres of uplift spreading kilometres from each well, an anisotropic pattern that tracked the pressurised region far beyond the CO2 itself (Vasco et al., 2010; Rinaldi & Rutqvist, 2013). This was a closed system caught in the act: a single-phase pressure plume, mapped from orbit, behaving exactly as a low-leakage boundary predicts. Concern that the rising pressure might reach the caprock fracture limit led to injection being suspended in 2011, after just six years of operation, sub-par injectivity, and many lessons learnt. What In Salah measured was unambiguous: in a tight system the pressure signal is real, large, and immediately apparent on the project’s injection timescale.

Then there is Snøhvit, in the Barents Sea: a similar lesson in a locally restricted reservoir, but this time met with a successful plan B, an engineering intervention that saved the project. Downhole pressure climbed faster than the pre-injection model had assumed, signalling far less room than expected in the original reservoir interval. Rather than shut-in, the operator moved injection to a shallower, better-connected reservoir, and pressure relaxed (Hansen et al., 2013). Snøhvit is a documented pressure-limited capacity event: the volumetric estimate was never the binding constraint; the boundary was. It also shows that a closed system is not necessarily a dead one; sometimes you can engineer around the limit, provided you have somewhere to go.

In the middle ground sits Gorgon, offshore Western Australia; a large storage operation, and a system deliberately run as semi-closed. The Dupuy Formation would over-pressure without intervention under the planned injection rate like a closed system, so the project was designed from the outset with brine-production wells that draw water off to manage pressure and maintain capacity (Flett et al., 2008). Gorgon is a closed system made operational, anticipated and actively managed at a price, in wells drilled and water produced. That operational price has proved very real in practice: so far, the ability to produce and dispose of sufficient formation water has constrained CO2 injection rates, and the pressure-management system has had to be re-engineered to increase water production and matching disposal rate. It is the clearest field demonstration that a suitable closed system can be shifted to the middle ground and made workable, if you pay for it.

At the open end sits Sleipner, in the North Sea. On 15 September 2026 the old Norwegian war horse celebrated three decades of injection into the Utsira Sand with 20 million tonnes stored, hundreds of scientific publications and dozens of PhDs, and no pressure issues at all. The Utsira is vast, laterally extensive, highly permeable, and the pressure has stayed low throughout (Eiken et al., 2011; Furre et al., 2017). Sleipner demonstrats an important fact: a saline aquifer with an effectively infinite boundary stays open, notably without an explicitly open external boundary, and operates comfortably across the full life of a project. Sleipner’s celebrated imaging records a contour-hugging plume that is close to equilibrium (Cavanagh et al., 2015). What matters for the boundary argument is that the open pole has a real-world anchor, too.

So, the axis is not a thought experiment. It is populated end to end by projects, including some we have not mentioned such as Quest and Aquistore in Canada (no active pressure management, no injectivity issues). While the pressure limit demonstrably binds at one end, hard enough to suspend one project and re-engineer another, these are not the norm but are valuable lessons learnt.

There is also a limit to what these cases can tell us. The records are short: In Salah was about six years to suspension, and Snøhvit was three years to intervention. Even Sleipner’s thirty years is short compared to the many decades over which a pressure perturbation can potentially persist or dissipate naturally (Gasda et al., 2012). Crucially, Sleipner’s record only confirms that an open saline aquifer remains stable, not that a closed or semi-closed one relaxes. No field dataset yet observes a semi-closed system bleeding its pressure down over the long horizon on which the middle-ground estimate depends. That the limit binds is measured; that it later relaxes is, for now, an extrapolation. What the field evidence shows is that the monster is real. Whether it is friendly or scary is a question the empirical data cannot yet answer, and so we turn to the physics.

The modelling lineage brackets the gap

If the field data suggests the pressure Gruffalo exists, the modelling literature shows where it lives without quite entering the forest that is the middle ground. Arranged along the boundary axis, four decades of work bracket the semi-closed solution from either side (Figure 4).

The open pole is anchored by the analytical injection solutions of Nordbotten et al. (2005), in which pressure propagates outwardly through an infinite-acting aquifer with no lateral bound – simple, elegant, and open by design. The closed pole is anchored by the fully compartmentalised case, first by van der Meer & van Wees (2006), then most starkly by Ehlig-Economides & Economides (2010), whose closed-box arithmetic returned the sub-percent efficiencies that briefly made the monster look like Godzilla (Cavanagh, 2010). That position was rebutted by two schools: by basin-scale hydrogeology, showing that real aquifers bleed pressure to their surroundings (Zhou & Birkholzer, 2011), and by petroleum systems analysis, showing that seals leak in single phase even where they hold CO2 (Cavanagh & Wildgust, 2011).

Between the poles, the interior has been mapped but not explored. Zhou et al. (2008) presented an explicit method for both closed and semi-closed systems, treating capacity as a function of compressibility and diffuse leakage. Birkholzer et al. (2009, 2015) modelled seal-permeability sensitivity to different injection rates, and Bump & Hovorka (2024) framed the closed ceiling approximation as a finite pressure-space commodity (4). Each enters the solution space; none resolves it completely, which the literature repeatedly hands off to case-by-case numerical simulation. Porosity and permeability trends (Figure 3) suggest the pressure response of a storage interval at 1-3 km depth will lie somewhere in-between as a semi-closed system.

In summary, two modelling schools, hydrogeology and petroleum systems analysis, converge on the same semi-closed solution: diffuse leakage through the seal as a single-phase flux through shale: the same physics for a transition zone in the low microdarcy range on decadal timescales, and both reject the closed extreme. The flow modelling suggests diffuse pressure-relief as the mechanism, i.e. long-horizon temporal relief of pressure in a semi-closed system.

Which brings us to the state of the art, and the gap this paper addresses. The most recent basin-scale treatment (de Jonge-Anderson et al., 2026) superimposes the open, infinite-acting Nordbotten solution to simulate pressure interference between projects, while assuming deliverability within a closed pressure-space. This is a deliberately conservative engineering choice: it efficiently resolves project interference as a first-order concern for basin-scale build-out across thousands of storage units. However, it avoids the semi-closed middle rather than entering it at some computational expense: an open solution propagated laterally, a closed ceiling imposed from above. Consider Endurance, a large storage project in the Bunter Sandstone, UK Southern North Sea. It has an excellent seal, a 200m-thick mix of shales and halites (Heinemann et al., 2012) and is underlain by the Bunter Shale and thick regional salt of the Zechstein. Given the setting and several neighbouring licences, this is the perfect setting to test such a model. How hard will pressure bind in a regional aquifer with tight top and bottom seals? De Luca et al. (2025) found that injection over 50 years created a pressure plume of 10-20 km in radius, increasing pressure at the boundary of the model by 2-3 MPa in a confined system, but only 0-1 MPa in a semi-closed system. However, the Endurance setting of regionally extensive Permian and Triassic halites is exceptional; for most regional saline aquifers, we are in the more mundane world of mudstones that dominate basin fill. Their permeability, along with the general compressibility of the system, will govern the far-field pressure response.

How scary is the pressure Gruffalo?

The monster is real, and, as we have seen, it can squeeze the life out of a project in a few short years. However, there are two things that weigh in the mouse’s favour.

The first is time. A semi-closed system does not hold pressure for ever; it bleeds it off through the seal, on a timescale set by the seal permeability that governs equation (1). The tighter the seal, the longer the bleed, but pressure drops, and the diffuse hydrogeology makes it quantifiable (Zhou & Birkholzer, 2011). The second is the carbon dioxide itself. Over decades to millennia the mobile plume is progressively pinned by residual trapping and reduced by solubility trapping. The latter contributes to pressure relaxation as free-phase CO2 is transferred into the brine, reducing the volume occupied by the separate CO2 phase. Thus, even a tight boundary sees less and less pressure with time, while a leaky boundary has less and less to leak (Gasda et al., 2011). Pressure relaxes and the cause of pressure weakens: the Gruffalo gets friendlier with time as long as it is not caged in a closed box. And the mouse gets cleverer too as the data comes in and the basin becomes a known quantity.

Both phenomena are modelled extrapolations. Few field records yet span the decadal horizon on which either process starts to operate. The fair claim is that saline aquifers are knowable, and that a friendly reading is testable with a fairly straightforward model.

The distinction is that we do not offer a global storage efficiency or basin volume factor and would not recommend that approach. What the transfer function implies is simple but not deterministic. Because capacity saturates at both ends and most mudrocks are neither pristine nor perfect seals, the bulk of the distribution sits in the semi-closed band, around $P_{30}$–$P_{70}$ on the boundary axis. Where a given aquifer falls within that band depends on the two things: whether its effective seal behaves as intact matrix, skewing toward $P_{60}$–$P_{100}$, or as an upscaled population of tiny natural fractures, skewing toward $P_{40}$– $P_0$; and the timeline, which slides the whole distribution as (1) required. The output is a band with dependencies attached. That is less than the optimist assumes and far more than the pessimist allows.

Ranked by those dependencies, real storage settings are ordered and predictable. A thick, intact mudstone seal, the Endurance case, with regional halite below and tight seal above, sits toward the closed pole: high assurance of containment, but a pressure ceiling that may bind early and hard. A shelf sandstone with a competent but ordinary marine seal, like Sleipner or Northern Lights, sits nearer the open end where the boundary rarely binds. Between them, and widest of all, sits the fluvial and sheet-sand fill that makes up most of the world’s undrilled saline aquifer volume: mixed net-to-gross, uncertain lateral continuity, an effective seal permeability that could plausibly land anywhere across the transition. The most common setting is also the least constrained which is precisely why the global resource number is so soft, and where appraisal buys the most insight. And appraisal can be booked, on a schedule. This is the part the resource caricatures miss entirely: the width of the band is not fixed. Every project that injects adds a far-field pressure history, and every year of that history clarifies the boundary estimate for a basin, a site, and by analogue for its neighbours in the same play, and similar, unexplored basins. The interference that makes basin-scale build-out necessary (de Jonge-Anderson et al., 2026) is the very activity that resolves the boundary: overlapping pressure plumes are a basin-scale pump test we are running whether we design it or not. So, the uncertainty does not merely sit there to be feared; it can be explored and progressively reduced as a staggered build-out converts decades of injection into decades of pressure record.

None of this tames the Gruffalo. It makes it approachable. The sensible reading is extrapolation today and measurement tomorrow, and the path between the two is knowable now. The right output for a saline aquifer is therefore not a number but a range with a horizon and a stated way to close it: conservative where we are ignorant, and demonstrably less so each year we are not.

The mouse and the Gruffalo

We began with an inventory question and end with a slightly different one. How much CO2 can the world’s saline aquifers hold is, at basin scale, not a question about pore volume but about boundaries, and boundaries are the one property the global resource numbers assume rather than measure (Cavanagh et al., 2026). While the 14,256 GtCO2 in the catalogue is an answer to a question no storage project asks, it genuinely matters when the allocation to this technology is weighted in mitigation pathways.

The pressure limit is real, suspending In Salah and re-engineering Snøhvit, both within a few years of first injection; it is designed for, at cost, at Gorgon; and it is entirely absent where the boundary is genuinely open, as at Sleipner. That spectrum, populated by real projects and governed by seal permeability, is a transfer function that saturates at both ends and flips at a narrow microdarcy transition. The semi-closed middle ground is where we expect most unexplored aquifers to sit based on the physics of boundary flow, compression, and dissolution.

Two things follow, and they pull like a tug-of-war only if a dogmatic position is taken. The monster is more tractable than the closed-box worst case scenario: over time, pressure bleeds off and CO2 immobilises, making a semi-closed system friendlier with time. The fair reading is therefore a capacity that is a range – conservative where we are ignorant, and narrowing on a schedule as staggered build-out turns injection into pressure history.

That is a shift in mindset, neither $P_0$ complacency, nor $P_{100}$ alarmism, but a calibrated middle that states its uncertainty and the timetable for closing it. The global resource-scale reconciliation and the deployment build-out model (de Jonge-Anderson et al., 2026) both rest on this middle, the physics and probability distribution it yields. The OGCI catalogue resources average 100-400 Mt per entry. Conservatively assuming an average 200 Mt per site, that requires thousands of projects by the end of the century to meet a forecast demand of around 0.7-1.4 trillion tonnes of geological carbon storage (IPCC, 2022). We already have tens of projects scheduled for Europe by the early 2030s. The hundreds of projects by mid-century will provide the insight we need to accurately forecast the boundary and limit long before it is reached.

The mouse invented a monster; the monster turned out to be real and is almost certainly not as scary as feared. The pressure Gruffalo is real enough to plan for, and, uniquely for a monster, one we can ask questions of. The mouse and the Gruffalo share the wood. With data.

References

• Bachu, S. [2015]. Review of CO2 storage efficiency in deep saline aquifers. International Journal of Greenhouse Gas Control, 40, 188-202.
• Birkholzer, J.T., Zhou, Q. and Tsang, C.-F. [2009]. Large-scale impact of CO2 storage in deep saline aquifers: a sensitivity study on pressure response in stratified systems. International Journal of Greenhouse Gas Control, 3(2), 181-194.
• Birkholzer, J.T., Oldenburg, C.M. and Zhou, Q. [2015]. CO2 migration and pressure evolution in deep saline aquifers. International Journal of Greenhouse Gas Control, 40, 203-220.
• Bump, A.P. and Hovorka, S.D. [2024]. Pressure space: the key subsurface commodity for CCS. International Journal of Greenhouse Gas Control, 136, 104174.
• Cavanagh, A.J. [2010]. Pressurisation and brine displacement issues for deep saline formation CO2 storage. IEAGHG Technical Report 2010/15, November 2010. IEA Greenhouse Gas R&D Programme, Cheltenham, UK.
• Cavanagh, A.J., Haszeldine, R.S., Alcalde, J., Gasda, S., Roberts, J.J., de Jonge-Anderson, I., Ringrose, P.R., Dixon, T., Johnson, G., Blunt, M., Maitland, G. and Vilarrasa, V. [2026]. A prudent planetary limit for geological carbon storage: correcting a crucial underestimate. Abstract, 18th International Conference on Greenhouse Gas Control Technologies (GHGT-18).
• Cavanagh, A.J. and Wildgust, N. [2011]. Pressurization and brine displacement issues for deep saline formation CO2 storage. Energy Procedia, 4, 4814-4821.
• Cavanagh, A.J., Haszeldine, R.S. and Nazarian, B. [2015]. The Sleipner CO2 storage site: using a basin model to understand reservoir simulations of plume dynamics. First Break, 33(6), 61-68.
• Cavanagh, A.J., Wilkinson, M. and Haszeldine, R.S. [2020]. Bridging the gap: storage resource assessment methodologies. EU H2020 STRATEGY CCUS Project 837754, Report, 67 pp.
• Chadwick, R.A., Zweigel, P., Gregersen, U., Kirby, G.A., Holloway, S. and Johannessen, P.N. [2004]. Geological reservoir characterization of a CO2 storage site: the Utsira Sand, Sleipner, northern North Sea. Energy, 29(9-10), 1371-1381.
• de Jonge-Anderson, I., Johnson, G., Alcalde, J. and Roberts, J.J. [2026]. Global limits on CO2 geological storage deployment using a coupled industry growth-pressure constrained model. Preprint submitted to International Journal of Greenhouse Gas Control. EarthArXiv.
• De Luca, M., Mackay, E.J., Good, T.R., Scisciani, V., MacBeth, C., Patruno, S. and Sutcliffe, J. [2025]. Regional pressure response to large-scale carbon storage in the Bunter Sandstone Formation, Silverpit Basin (Southern North Sea, UK). International Journal of Greenhouse Gas Control, 148, 104530.
• Ehlig-Economides, C. and Economides, M.J. [2010]. Sequestering carbon dioxide in a closed underground volume. Journal of Petroleum Science and Engineering, 70(1-2), 123-130.
• Eiken, O., Ringrose, P., Hermanrud, C., Nazarian, B., Torp, T.A. and Høier, L. [2011]. Lessons learned from 14 years of CCS operations: Sleipner, In Salah and Snøhvit. Energy Procedia, 4, 5541-5548.
• Flett, M., Beacher, G., Brantjes, J., Burt, A., Dauth, C., Koelmeyer, F., Lawrence, R., Leigh, S., McKenna, J., Gurton, R., Robinson IV, W.F. and Tankersley, T. [2008]. Gorgon Project: subsurface evaluation of carbon dioxide disposal under Barrow Island. SPE Asia Pacific Oil and Gas Conference and Exhibition, Perth, Australia, 20-22 October 2008, SPE-116372-MS.
• Furre, A.-K., Eiken, O., Alnes, H., Vevatne, J.N. and Kiær, A.F. [2017]. 20 years of monitoring CO2-injection at Sleipner. Energy Procedia, 114, 3916-3926.
• Gasda, S.E., Nordbotten, J.M. and Celia, M.A. [2009]. Vertical equilibrium with sub-scale analytical methods for geological CO2 sequestration. Computational Geosciences, 13(4), 469-481.
• Gasda, S.E., Nordbotten, J.M. and Celia, M.A. [2011]. Vertically averaged approaches for CO2 migration with solubility trapping. Water Resources Research, 47, W05528.
• Gasda, S.E., Nordbotten, J.M. and Celia, M.A. [2012]. Application of simplified models to CO2 migration and immobilization in large-scale geological systems. International Journal of Greenhouse Gas Control, 9, 72-84. 
• Gidden, M.J., Joshi, S., Armitage, J.J., Christ, A.-B., Boettcher, M., Brutschin, E., Köberle, A.C., Riahi, K., Schellnhuber, H.J., Schleussner, C.-F. and Rogelj, J. [2025]. A prudent planetary limit for geologic carbon storage. Nature, 645(8079), 124-132.
• Goodman, A., Hakala, A., Bromhal, G., Deel, D., Rodosta, T., Frailey, S., Small, M., Allen, D., Romanov, V., Fazio, J., Huerta, N., McIntyre, D., Kutchko, B. and Guthrie, G. [2011]. U.S. DOE methodology for the development of geologic storage potential for carbon dioxide at the national and regional scale. International Journal of Greenhouse Gas Control, 5(4), 952-965.
• Hansen, O., Gilding, D., Nazarian, B., Osdal, B., Ringrose, P., Kristoffersen, J.B., Eiken, O. and Hansen, H. [2013]. Snøhvit: the history of injecting and storing 1 Mt CO2 in the fluvial Tubåen Fm. Energy Procedia, 37, 3565-3573.
• Heinemann, N., Wilkinson, M., Pickup, G.E., Haszeldine, R.S. and Cutler, N.A. [2012]. CO2 storage in the offshore UK Bunter Sandstone Formation. International Journal of Greenhouse Gas Control, 6, 210-219.
• IPCC [2022]. Climate Change 2022: Mitigation of Climate Change. Contribution of Working Group III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK and New York, NY, USA.
• Krevor, S., Blunt, M.J., Benson, S.M., Pentland, C.H., Reynolds, C., Al-Menhali, A. and Niu, B. [2015]. Capillary trapping for geologic carbon dioxide storage – from pore scale physics to field scale implications. International Journal of Greenhouse Gas Control, 40, 221-237.
• Neuzil, C.E. [1994]. How permeable are clays and shales? Water Resources Research, 30(2), 145-150.
• Nilsen, H.M., Lie, K.-A. and Andersen, O. [2015]. Analysis of CO2 Trapping Capacities and Long-Term Migration for Geological Formations in the Norwegian North Sea Using MRST-co2lab. Computers & Geosciences (pre-print).
• Nordbotten, J.M., Celia, M.A. and Bachu, S. [2005]. Injection and storage of CO2 in deep saline aquifers: analytical solution for CO2 plume evolution during injection. Transport in Porous Media, 58(3), 339-360.
• OGCI [2025]. CO2 Storage Resource Catalogue – Cycle 5 Report: Global Summary. Oil and Gas Climate Initiative, August 2025. Prepared by Halliburton and the Global CCS Institute.
• Rinaldi, A.P. and Rutqvist, J. [2013]. Modeling of deep fracture zone opening and transient ground surface uplift at KB-502 CO2 injection well, In Salah, Algeria. International Journal of Greenhouse Gas Control, 12, 155-167.
• SPE [2022]. CO2 Storage Resources Management System (SRMS). Society of Petroleum Engineers.
• Thibeau, S. and Bump, A. [2026]. Under pressure: the role of regional geological boundaries in estimating storage resources. International Journal of Greenhouse Gas Control, 155, 104755.
• Thibeau, S., Bachu, S., Birkholzer, J., Holloway, S., Neele, F. and Zhou, Q. [2014]. Using pressure and volumetric approaches to estimate CO2 storage capacity in deep saline aquifers. Energy Procedia, 63, 5294-5304.
• van der Meer, L.G.H. and van Wees, J.-D. [2006]. Limitations to storage pressure in finite saline aquifers and the effect of CO2 solubility on storage pressure. SPE Annual Technical Conference and Exhibition, San Antonio, Texas, USA, 24-27 September 2006, SPE-103342-MS.
• Vasco, D.W., Rucci, A., Ferretti, A., Novali, F., Bissell, R.C., Ringrose, P.S., Mathieson, A.S. and Wright, I.W. [2010]. Satellite-based measurements of surface deformation reveal fluid flow associated with the geological storage of carbon dioxide. Geophysical Research Letters, 37(3), L03303.
• Yang, Y. and Aplin, A.C. [2010]. A permeability-porosity relationship for mudstones. Marine and Petroleum Geology, 27, 1692-1697.
• Zhou, Q. and Birkholzer, J.T. [2011]. On scale and magnitude of pressure build-up induced by large-scale geologic storage of CO2. Greenhouse Gases: Science and Technology, 1(1), 11-20.
• Zhou, Q., Birkholzer, J.T., Tsang, C.-F. and Rutqvist, J. [2008]. A method for quick assessment of CO2 storage capacity in closed and semi-closed saline formations. International Journal of Greenhouse Gas Control, 2, 626-639.

Nomenclature 

aquifer area (m²)

seal thickness (m)

total compressibility (Pa⁻¹)

far-field, pressure-limited capacity ceiling (GtCO2)

usable storage capacity, = min capacity (GtCO2)

near-field, volumetric capacity ceiling (GtCO2)

aquifer thickness (m)

relative permeability (–)

open (optimistic) boundary pole (P0)

closed (pessimistic) boundary pole (P100)

boundary percentile band of the semi-closed middle (≈ P30–P70)

far-field pressure-front radius (km)

tolerable pressure rise to the fracture limit (MPa)

critical (onset) seal permeability (m²)

seal/boundary absolute permeability (m²)

closed / open saturation thresholds (m²)

seal Fourier number (–)

brine viscosity (Pa·s)

CO2 density (kg m⁻³)

accounting (injection) timeline (yr)

aquifer porosity (%)

seal porosity (%)

Abbreviations 

CO2 – carbon dioxide; InSAR – interferometric synthetic-aperture radar; N:G – net-to-gross; OGCI – Oil and Gas Climate Initiative; SEF – storage efficiency factor; SRMS – Storage Resources Management System; VE – vertical equilibrium.

Units Mt, Gt, Tt: million / giga / tera tonnes; GtCO2, TtCO2, mass of CO2.