Lost in Translation: Why CCS Projects Misread the Underground

Executive Summary

Carbon capture and storage projects consistently underperform their design targets—not because the technology is immature, but because knowledge fails to translate across the professional boundaries that separate those who understand the subsurface from those who finance it.

This analysis identifies three distinct mechanisms of knowledge failure. Transmission failure occurs when relevant expertise never reaches decision-makers. Translation failure occurs when identical terminology—most dangerously, the P90 metric—carries fundamentally different meanings for geologists and bankers. Misapplication occurs when extraction logic developed for hydrocarbon production is deployed for CO₂ injection without recognizing that the physics reverses.

These mechanisms converge at Final Investment Decision, where geological uncertainty must be converted into financial parameters. The Independent Engineers retained to bridge technical and financial communities are structurally compromised: paid by sponsors, competing for future mandates, they produce "constructive ambiguity" that softens warnings without eliminating risks. The result is systematic optimism bias that sets projects up for operational disappointment.

Examination of five major CCS projects—Sleipner, Snøhvit, Gorgon, In Salah, and Quest—confirms this pattern. Projects that succeeded adopted conservative geological assumptions, integrated petroleum expertise directly into CCS design, and operated within regulatory frameworks that rewarded honest uncertainty assessment. Projects that struggled treated petroleum knowledge as automatically transferable.

Successful knowledge transfer in adjacent domains—geothermal drilling, energy meteorology, electricity capacity markets—demonstrates that translation failure is not inevitable. It requires explicit translation agents, deliberate adaptation rather than copying, and incentive-compatible institutional structures. CCS can scale successfully, but only if the industry acknowledges the translation problem and designs mechanisms to address it.

The knowledge exists. The translation does not.

In the spring of 2009, a conference room in Perth hosted a meeting that would commit $54 billion to one of the most ambitious energy projects ever attempted. Around the table sat geologists who had spent careers learning to read the earth's secrets, reservoir engineers who could model fluid flow through rock formations invisible to the human eye, and bankers whose expertise lay in converting uncertainty into spreadsheets. They were there to approve the Final Investment Decision for the Gorgon project—a liquefied natural gas development that would include the world's largest dedicated carbon capture and storage facility.

The geologists had done their work. Seismic surveys had mapped the Dupuy formation, a Cretaceous sandstone lying beneath Barrow Island. Core samples had been analyzed. Simulation models had been run. The formation, they concluded, had adequate porosity to store the project's CO₂ emissions for millennia. The P90 estimate—the conservative case that the geological team believed had a 90% probability of being exceeded—indicated sufficient capacity.

The bankers had done their work too. They had stress-tested the financial models, calculated debt service coverage ratios, negotiated risk allocation among the joint venture partners. The project, they concluded, was bankable. The numbers worked.

What neither group fully grasped was that they had been speaking different languages while believing they understood each other. When the geologists said "P90 capacity," they meant a conservative estimate of how much CO₂ the rock could physically hold. When the bankers heard "P90," they interpreted it as a reliable floor for annual injection rates—the throughput that would keep the project in compliance with its environmental permits. Capacity and injectivity are not the same thing. A formation can have vast storage potential while being unable to accept CO₂ at the rates a project requires. The Dupuy formation, it would turn out, was exactly such a case.

Fifteen years later, Gorgon's carbon capture system has never achieved its design performance. Capture rates have fluctuated between 30% and 68% of the promised 80%+, making it one of the most expensive underperformers in CCS history.[1] Sand production has clogged wells. Pressure management systems have failed. Equipment corrosion has forced repeated shutdowns. The project has faced regulatory proceedings for excess emissions. Chevron, the operator, has spent hundreds of millions on remediation while the atmosphere has received millions of tonnes of CO₂ that were supposed to go underground.

The Gorgon story is not a tale of incompetence. Chevron employs some of the world's most experienced subsurface engineers. It is not a tale of deception—there is no evidence that anyone deliberately misrepresented the geological risks. It is, rather, a tale of translation failure: the systematic loss of meaning that occurs when knowledge moves between professional communities that use the same words to denote different things.

This is a problem that extends far beyond a single project in Western Australia. As the preceding analysis of knowledge stranding established, the fossil fuel industry has accumulated a century of hard-won expertise in working with the deep subsurface—expertise that net-zero technologies urgently require. But the existence of knowledge is not the same as its successful application. Between the engineers who possess subsurface expertise and the projects that need it stands a series of institutional interfaces where meaning can be garbled, warnings can be softened, and epistemic humility can be stripped away. The most consequential of these interfaces is the Final Investment Decision.

This analysis examines how translation fails. It identifies three distinct mechanisms through which knowledge goes astray: transmission failure, where information never reaches those who need it; translation failure, where identical words carry different meanings across professional boundaries; and misapplication, where knowledge is deployed without the contextual adaptation its original domain requires. It then traces these mechanisms through the FID process, paying particular attention to the structural position of Independent Engineers—the consultants retained to bridge technical and financial communities—and to the semantic slippage that transforms geological uncertainty into bankable parameters. Five major CCS projects provide the empirical foundation: Sleipner, Snøhvit, Gorgon, In Salah, and Quest. Three successful cases of cross-domain knowledge transfer—geothermal drilling, energy meteorology, and electricity capacity markets—offer counterpoint, revealing conditions under which translation succeeds rather than fails.

The stakes are considerable. If CCS is to contribute meaningfully to climate mitigation, it must scale from a handful of demonstration projects to hundreds of commercial facilities. That scaling depends on mobilizing private capital, which in turn depends on FID processes that accurately characterize risk. If those processes systematically mistranslate geological uncertainty into false financial confidence, projects will continue to disappoint, investors will demand prohibitive risk premiums, and a technology that climate models deem essential will fail to deploy at the pace the atmosphere requires.

Three Ways Knowledge Goes Wrong

The conventional story about knowledge transfer focuses on barriers to transmission—the obstacles that prevent information from flowing between organizations, industries, or generations. Such barriers certainly exist. But a framework limited to transmission cannot explain why projects fail even when staffed by experienced professionals from the relevant domain. Chevron's Gorgon team included reservoir engineers with decades of subsurface experience. Equinor's Snøhvit team drew on Norway's deep tradition of offshore expertise. If the knowledge was present, why did the projects struggle?

The answer requires distinguishing three mechanisms that are often conflated.

Transmission failure occurs when relevant knowledge simply does not reach those who need it. Information exists somewhere in the system but fails to flow across organizational boundaries, disciplinary divides, or generational gaps. The oil and gas industry's professional networks—SPE conferences, technical journals, informal communities of practice—circulate knowledge efficiently within the petroleum world but have limited connections to the renewable energy networks organized around different conferences, different journals, different career paths. A reservoir engineer's insight about injection-induced pressure compartmentalization may be well-known within her professional community while remaining invisible to the project finance specialists structuring the deal.

Research on workforce transitions confirms that transmission failure is pervasive. Studies by Robert Gordon University's Energy Transition Institute found that while approximately 90% of oil and gas technical skills are theoretically transferable to adjacent sectors, in practice 42% of workers who left the petroleum industry during the 2015-2020 downturn exited the energy sector entirely rather than moving to renewables or CCS.[2] The knowledge they carried did not transfer to where it was needed. It simply left.

Translation failure is more insidious because it occurs even when communication appears successful. The same terminology is used by different professional communities to denote different concepts, creating an illusion of mutual understanding that masks fundamental disagreement. Both oil and gas engineers and renewable energy developers use "P90" to describe conservative estimates. Both employ "independent technical review" to validate project parameters. Both speak of "risk" and "uncertainty" and "contingency." But these surface similarities conceal deep differences in what the terms actually mean.

Consider "well integrity"—a concept central to any project involving injection into geological formations. In the oil and gas context, well integrity encompasses the comprehensive system of barriers (casing, cement, wellhead equipment) that prevent hydrocarbon migration to the surface or to unintended formations. Standards like API RP 90 and NORSOK D-010 codify decades of hard-won knowledge about how wells fail and how to prevent failure.[3] The core concern is preventing what is underground from escaping upward.

When this terminology migrates to CCS, its meaning subtly shifts. CO₂ injection involves different chemistry—supercritical CO₂ reacts with cement and steel in ways that hydrocarbons typically do not. The pressure regime differs: injection builds pressure over time rather than depleting it. And the liability horizon extends from decades to centuries or millennia. The phrase "well integrity" travels across contexts, but the engineering content it denotes does not translate without adaptation. A completions engineer who has mastered well integrity for hydrocarbon production has relevant expertise for CO₂ injection—but not identical expertise. The gap between relevant and identical is where projects get into trouble.

Misapplication occurs when knowledge is correctly transmitted and accurately understood at an abstract level but deployed without the contextual modifications its original domain requires. The paradigmatic case in CCS involves the transfer of extraction logic to injection problems.

The oil and gas industry accumulated its expertise through a century of extracting fluids from underground reservoirs. This experience generated sophisticated understanding of how to characterize formations, model fluid flow, design wells, and manage production decline. The relevant physics involves flow from reservoir to wellbore, pressure declining over time, geological structures acting as passive containers from which value is withdrawn.

Injection reverses these dynamics. Flow moves from wellbore to formation. Pressure accumulates rather than depletes. The geological structure must actively accommodate increasing volumes while maintaining seal integrity. Mental models that serve petroleum engineers well for extraction can mislead when applied to injection. Formation characteristics that represent assets in extraction contexts—high permeability, extensive connectivity, minimal compartmentalization—may become liabilities in injection contexts, where pressure dissipation rather than pressure maintenance is the operational challenge.

Gorgon exemplifies this logic error. The project design assumed that water production wells could manage reservoir pressure by withdrawing formation water and reinjecting it into overlying formations—essentially applying production logic (withdraw fluid to create space) to an injection problem. The approach worked in theory. In operation, sand production clogged the water handling system, formation heterogeneity exceeded model predictions, and the pressure management concept that seemed sound for extraction failed to translate to injection.[4]

These three mechanisms—transmission, translation, and misapplication—are analytically distinct but frequently co-occur. A single project can suffer from all three simultaneously. What unites them is that none is reducible to individual error or incompetence. Each reflects structural features of how knowledge is organized across professional communities. And each concentrates at a particular institutional chokepoint: the Final Investment Decision.

The FID as Translation Zone

Final Investment Decision is the moment when a project transitions from concept to commitment. Capital is allocated, contracts are signed, construction begins. In the project finance structures typical of large energy investments, FID is preceded by elaborate due diligence designed to give investors confidence that the project will deliver its promised returns. Technical consultants prepare independent assessments of engineering feasibility. Financial advisors structure debt and equity components. Legal teams draft the contractual architecture that allocates risk among parties.

The conventional narrative presents this as a rational, linear process. Front-End Engineering Design establishes the technical solution. Commercial negotiation locks in market risk through offtake agreements. The investment committee reviews projections and approves the project if returns exceed hurdle rates. This narrative, while not entirely false, obscures the epistemic complexity of what actually occurs. FID is not simply a decision point. It is a translation zone—a space where fundamentally different ways of understanding uncertainty must be reconciled into a single go/no-go judgment.

The geological and engineering teams approach FID with an epistemological stance shaped by decades of subsurface work. They understand that their models are approximations, that the subsurface is invisible and therefore imperfectly known, that adaptive management will be required as operations reveal formation characteristics that pre-drilling data could not capture. Uncertainty, in their framework, is not merely a reflection of data limitations that additional surveys could resolve. It is an inherent feature of working with geological systems whose complexity exceeds any model's capacity to represent.

The financial teams approach FID with a different epistemological stance. Project finance requires all uncertainties to be either quantified as risks (assigned probability distributions and incorporated into sensitivity analyses) or allocated to parties contractually (transferred to insurers, guaranteed by sponsors, absorbed by off-takers). Uncertainty that can be neither quantified nor allocated is, by definition, unbankable. The imperative of bankability creates pressure to translate every geological concern into a numerical parameter that financial models can process.

Something is necessarily lost in this translation. A geologist's statement that "formation heterogeneity may be greater than our model captures" must become a percentage adder to the CAPEX contingency or a Monte Carlo probability distribution over injection rates. The qualitative texture of geological uncertainty—the sense, cultivated through years of experience, for what kinds of surprises the subsurface typically delivers—cannot be directly input to a spreadsheet. It must be converted, and the conversion is lossy.

The Independent Engineer's Dilemma

Between the technical and financial communities stands a critical intermediary: the Independent Engineer. IE firms—DNV, RISC Advisory, Mott MacDonald, Wood Mackenzie—are retained to provide objective technical assessments that lenders can rely upon. Their mandate is to review the sponsor's engineering work, identify risks, and opine on whether the project is technically feasible and likely to achieve its design performance. In principle, the IE serves as a bridge between epistemologies, translating geological uncertainty into terms that financial decision-makers can interpret.

In practice, the IE's structural position compromises this bridging function. Although Independent Engineers nominally serve lender interests—protecting the bank's capital from poorly conceived projects—their fees are typically paid by project sponsors. More fundamentally, IEs operate within a competitive market for due diligence services. A consultancy that develops a reputation for raising too many concerns, for being too conservative in its assessments, may find itself excluded from future mandates. Commercial pressures push IEs toward what might be called constructive ambiguity—language that technically acknowledges risks while presenting them in forms that do not threaten project viability.[5]

This dynamic manifests in characteristic linguistic patterns. Rather than stating "the geological data are insufficient to characterize this formation with confidence," an IE report might note that "data density is consistent with industry norms at this project stage, with additional characterization recommended during execution." Both statements may be technically accurate. They carry different implications for decision-makers. The first suggests a problem that should give pause. The second suggests normal project development practice.

The "red flag" mechanism illustrates this softening process. Due diligence protocols distinguish between red flags (fatal flaws that should prevent financing), amber flags (significant concerns requiring mitigation), and green lights (acceptable risks). A common dynamic in FID processes involves the negotiated downgrading of red flags to amber status. A geological concern is raised. The sponsor proposes a mitigation measure—additional monitoring, a contingency budget, an insurance policy. The IE accepts the mitigation as adequate and downgrades the flag. The underlying physical risk has not changed. The formation remains as uncertain as before. But the risk has been translated from a dealbreaker into a manageable concern.

Consider the logic embedded in this translation. If a geologist identifies "caprock seal integrity uncertainty" as a potential red flag, the sponsor might propose an extensive Measurement, Monitoring, and Verification plan including time-lapse seismic, microseismic arrays, and groundwater sampling. The IE can then record that "caprock risk will be actively managed through the MMV program" and downgrade to amber.

But monitoring can only detect leakage after it occurs. It cannot prevent leakage. The MMV plan converts geological risk into operational expense—something that financial models can accommodate—while leaving the physical probability of failure unchanged. The translation achieves bankability at the cost of epistemic accuracy.

The Boundary Object Problem

As the preceding analysis of stock and flux epistemologies established, P90 functions as a dangerous boundary object—a term that different professional communities use together without agreeing on what it means.[6] This semantic slippage operates with particular force within the FID process.

When geologists present a P90 storage capacity estimate, they intend it as a conservative bound on how much CO₂ the formation can physically accommodate over its lifetime, given current understanding of its geometry, porosity, and connectivity. This is a statement about volume—a stock. It carries an implicit assumption that actual performance will require adaptive management as injection reveals formation characteristics that models could not predict.

When bankers encounter the same P90 figure, they interpret it through project finance conventions. In renewable energy project finance, P90 denotes the annual energy production level that the project has a 90% probability of exceeding—a statement about flow rather than stock, used to size debt service obligations. Bankers accustomed to this meaning may interpret a geological P90 as a reliable floor for annual injection rates rather than a lifetime volume estimate.

The gap between these interpretations creates systematic optimism bias. Parameters that geologists consider conservative become planning cases that financiers treat as near-certainties. No one is lying. No one is even necessarily making an error within their own professional framework. The error emerges in the translation between frameworks—and no one is specifically responsible for ensuring that translation is accurate.

The Responsibility Vacuum

Examine the disclaimers that accompany IE reports and technical due diligence documents. Standard language includes formulations like "this report is provided on an as-is basis" and "we make no representations regarding the accuracy of information provided by the sponsor."[7] Such disclaimers serve legitimate legal purposes. They also create what might be called a responsibility vacuum.

The project sponsor can claim: "We relied on expert advisors." The IE can claim: "We assessed only what we were asked to assess, based on data the sponsor provided." The lenders can claim: "We relied on the IE's expert opinion." No party bears ultimate responsibility for the geological assumptions that underpin the investment case. The physical risks are not so much allocated as diffused—passed around the table like a cognitive hot potato until the music stops. And when the music does stop, when injection performance falls short of projections, each party can point to others in the chain.

This diffusion of responsibility has epistemic consequences. When no party bears clear accountability for geological assumptions, no party has adequate incentive to subject those assumptions to rigorous scrutiny. The IE's incentive is to produce a report that satisfies formal requirements while not antagonizing the sponsor. The lender's incentive is to close the deal and earn arrangement fees while documenting sufficient due diligence for internal compliance. The sponsor's incentive is to present the most favorable plausible case. At no point does any party have strong incentive to adopt the pessimistic stance that geological uncertainty warrants.

Five Projects Under Audit

To move from mechanism to evidence, we now examine five major CCS projects that span the range of geological settings, operator types, and outcomes the industry has produced. For each, we reconstruct what was known at the moment of Final Investment Decision and compare it against what operations subsequently revealed.

Sleipner: Success That Teaches Humility

Sleipner, operated by Equinor in the Norwegian North Sea, commenced injection in 1996 and represents the longest-running offshore CCS operation. The target formation, the Utsira Sand, is a massive, highly porous and permeable saline aquifer. Pre-injection models predicted that supercritical CO₂ would rise buoyantly through the formation and spread laterally beneath the caprock, forming a coherent plume whose migration could be tracked over time.

The first time-lapse seismic survey in 1999 revealed something different. Rather than a single rising column, the CO₂ had formed a layered structure—what geophysicists came to call a "wedding cake"—with separate accumulations trapped beneath thin intra-formational mudstones. These mudstones, typically 1-2 meters thick, had been below the resolution of pre-injection seismic data. They were invisible to the characterization program. Yet they fundamentally controlled plume geometry.[8]

Sleipner is generally counted as a success, and it is one. CO₂ has been injected continuously for nearly three decades without evidence of leakage beyond the storage complex. But the epistemological lesson is more nuanced than "CCS works." The lesson is that even in a geologically favorable setting, with a technically sophisticated operator, the pre-FID model was wrong in important respects. Features that control fluid migration were invisible to available characterization technologies. The project succeeded not because predictions were accurate but because the Utsira formation is forgiving—its high permeability and open boundaries accommodated prediction errors without generating the pressure problems that plagued other projects.

Sleipner's success is real but not automatically replicable. It depended on geological fortune as much as engineering skill.

Snøhvit: When Compartments Close In

Snøhvit, also operated by Equinor, began injection in 2008 into the Tubåen formation beneath the Barents Sea. The geological model characterized Tubåen as a high-permeability deltaic sandstone with regional aquifer connectivity. FID projections indicated that injection could continue for over eighteen years without approaching pressure limits, as pressure dissipation through connected aquifer volumes would accommodate CO₂ displacement.

Within eighteen months, bottom-hole pressure had spiked to within striking distance of the formation's fracture gradient. The reservoir that models showed as well-connected proved to be compartmentalized—a complex of fluvial channel sands separated by low-permeability floodplain deposits and cut by sealing faults. The injection point sat within an isolated compartment whose boundaries the pre-injection seismic could not resolve. Additionally, salt precipitation near the wellbore—dry supercritical CO₂ evaporating formation water, causing dissolved salts to crystallize and clog pore throats—created near-well damage that further restricted injectivity.[9]

Equinor was forced to suspend injection, conduct an expensive workover, and redirect operations to the shallower Stø formation, which had not been the primary target at FID. The Stø formation proved more accommodating—a piece of geological luck rather than foresight—and injection has continued successfully in the alternate zone.

Financial literature sometimes cites Snøhvit as a successful example of "real options" thinking—the project had the option to switch formations when the primary target failed, demonstrating flexibility and adaptive management. This interpretation requires scrutiny. The Stø formation was not developed as a planned contingency with allocated budget and pre-characterized parameters. It was a fallback discovered under duress after the primary plan collapsed. If Stø had also proved unsuitable—a geological possibility that cannot be ruled out for uncharacterized formations—the project would have faced catastrophic failure rather than expensive recovery. Calling this "real options" mistakes luck for strategy.[10]

Gorgon: Cathedral of Mistranslation

Gorgon merits extended examination because it concentrates every variety of translation failure into a single, exceptionally expensive case.

The project was approved with a CCS component because of regulatory requirements attached to the exploitation of offshore gas fields. Western Australia conditioned development approval on Chevron's commitment to reinject reservoir CO₂ rather than venting it to atmosphere. The Dupuy formation, a low-permeability Cretaceous sandstone underlying Barrow Island, was designated as the injection target. FID documents projected injection of 3.4 to 4 million tonnes annually, representing approximately 80% capture of CO₂ separated from produced gas.[11]

The geological challenges were acknowledged in technical documentation but translated into financial terms that understated their severity. The Dupuy's low permeability meant that injection at design rates required active pressure management—water would be produced from the formation via dedicated wells, treated, and reinjected into overlying formations to create space for CO₂. This pressure management system represented a major engineering departure from conventional CCS designs, which typically rely on passive pressure dissipation. The system's complexity created multiple potential failure points, but FID analysis treated these as manageable operational risks rather than fundamental design vulnerabilities.

Operations have vindicated the geologists' concerns and exceeded their worst projections. Sand production from the Dupuy formation clogged filters and damaged pumps in the water handling system. Formation heterogeneity exceeded model predictions, creating differential pressure responses across the injection area. The pressure management approach that worked conceptually in simulation proved operationally unreliable in practice. Capture rates have ranged from 30% to 68% of design capacity, with periods of complete system shutdown. Fiscal year 2024-25 saw the lowest performance yet: 1.33 million tonnes injected, barely a quarter of the design target, at an implied cost exceeding AU$265 per tonne of CO₂ stored.[12]

Multiple translation failures converged at Gorgon. The assumption that extraction experience (producing water from formations) would transfer to injection contexts (maintaining pressure equilibrium during CO₂ storage) exemplified misapplication. The treatment of system complexity as manageable operational cost rather than fundamental reliability risk illustrated how financial models struggle to capture non-linear failure modes. And the P90 boundary object problem manifested directly: massive storage capacity in the geological sense did not translate to adequate injection performance in the operational sense.

In Salah: The Geomechanical Blind Spot

In Salah, operated by a BP-Sonatrach joint venture in Algeria, injected CO₂ into a tight sandstone formation from 2004 to 2011. The project contributed what should have been an early warning about geomechanical risks—a warning imperfectly absorbed by subsequent projects.

By 2007, satellite interferometry detected surface uplift of approximately 5 millimeters per year above the injection zone. The pattern of deformation indicated that injection was creating pressure changes extending beyond the immediate vicinity of the wells, altering the regional stress field. Subsequent analysis suggested that pressure-induced fracturing might be propagating toward the caprock sequence.[13] No evidence of CO₂ leakage to the surface was detected, but the geomechanical response raised concerns about long-term containment integrity. Injection was suspended in 2011, before the originally planned endpoint.

The failure at In Salah was disciplinary rather than organizational. FID-era analysis focused primarily on fluid mechanics—would the rock accept CO₂ at planned rates, and would buoyancy-driven migration remain within the storage complex? Geomechanics received less attention. The working assumption was that injection pressures would remain below levels that could cause rock failure, and that any elastic deformation would be too small to matter.

Both assumptions proved wrong. The lesson is that CCS site characterization cannot be decomposed into independent assessments of geology, fluid mechanics, and geomechanics. These processes are coupled: injection changes pressure fields, pressure changes stress states, stress changes can reactivate fractures or create new ones, and fractures can alter permeability and containment. Models that treat these processes in isolation will fail to capture system behavior.

Quest: What Conservative Looks Like

Quest, operated by Shell in Alberta, began injection in 2015 and has consistently met or exceeded its targets, storing over nine million tonnes of CO₂ by 2024.[14] Quest represents the clearest success case in the CCS portfolio, offering a comparative benchmark for identifying what distinguishes effective practice from the failures that characterize other projects.

Three factors stand out.

First, Quest adopted deliberately conservative geological assumptions. Where other projects' FID analyses used mid-range or optimistic parameter estimates, Quest's models employed pessimistic values for permeability, porosity, and aquifer connectivity. The project was designed to succeed even if geological conditions proved worse than expected. This conservatism had costs—it likely led to over-investment in injection capacity relative to minimum requirements—but it created margin for error that cushioned the project against surprises.

Second, Shell integrated its internal petroleum expertise directly into the CCS design process. Rather than treating CCS as a separate technology requiring different expertise, the project drew on Shell's organizational knowledge of Alberta Basin geology accumulated through decades of oil and gas operations in the region. The translation from petroleum to CCS occurred within a single organizational context, reducing the risk that critical tacit knowledge would be lost crossing institutional boundaries.

Third, Quest benefited from supportive institutional structures. The Alberta government's liability transfer mechanism, under which the province assumes long-term stewardship responsibility after operators demonstrate storage stability over a defined monitoring period, removed a source of uncertainty that complicates CCS investment elsewhere.[15] The existence of a clear pathway from operational injection to transferred liability created incentive alignment that facilitated honest geological assessment. Operators had reason to be conservative because they would need to demonstrate performance to achieve liability transfer.

Quest does not prove that translation problems are easily solved. The project benefited from favorable geology, an experienced operator with directly relevant regional knowledge, and a regulatory framework other jurisdictions have not replicated. But it demonstrates that translation failure is not inevitable. When operators adopt epistemic humility, when organizational structures facilitate knowledge integration, and when institutional incentives reward honesty about uncertainty, CCS can achieve its objectives.

When Translation Succeeds

An account of failure would be incomplete without examining success. If cross-domain knowledge transfer always failed, there would be little hope for improving CCS outcomes. The cases below demonstrate that effective translation is achievable—under appropriate conditions.

Petroleum Well Control to Geothermal Drilling

The transfer of well control technology from petroleum to geothermal applications represents perhaps the clearest success in subsurface knowledge transfer. Both domains involve drilling into high-pressure formations where uncontrolled fluid release poses significant safety and environmental risks. The physics of well control—maintaining hydrostatic pressure balance, detecting kicks, executing kill procedures—translates directly between contexts.

Yet the translation was not automatic. Geothermal formations present conditions that petroleum experience does not directly address. Temperatures are typically higher, sometimes exceeding equipment limits. Formation fluids often contain aggressive chemical species. Rock mechanical properties differ. Successful translation required explicit recognition of what could transfer directly and what required adaptation.

The key translation agents were individual practitioners with experience in both industries. Iceland's geothermal development relied heavily on drilling engineers who had worked in North Sea petroleum operations and brought their well control expertise to a new geological context.[16] These individuals functioned as living bridges between communities, carrying tacit knowledge that written standards could not fully capture. Their cross-over experience enabled them to recognize both the applicability and the limits of petroleum approaches.

Marine Meteorology to Energy Meteorology

The emergence of energy meteorology illustrates a different pattern—the adaptation of scientific knowledge to commercial applications with entirely different evaluation criteria.

Traditional marine meteorology developed to serve navigational safety. Energy meteorology repurposes the same fundamental atmospheric science—numerical weather prediction, boundary layer physics, statistical post-processing—for power system operation, where forecast accuracy determines grid balancing costs and market revenues.[17]

The translation required multiple adaptations. Temporal resolution had to be refined from six-hour windows to sub-hourly predictions. Evaluation metrics changed from identifying hazardous conditions (where false alarms are acceptable) to minimizing root mean square error (where over- and under-prediction are equally penalized). Workflows had to integrate forecast information with operational decisions through automated systems rather than human judgment.

Translation agents included standards organizations like the IEC, which developed technical specifications bridging meteorological practice and project finance requirements. Regulatory bodies played a role, with grid operators imposing accuracy requirements that created demand for improved services. Academic institutions developed training programs combining atmospheric science with power system engineering, producing graduates capable of operating at the disciplinary interface.

Strategic Reserves to Capacity Markets

The conceptual transfer from strategic petroleum reserves to electricity capacity markets illustrates translation at the policy level rather than the technical level.

The Strategic Petroleum Reserve was established following the 1970s oil embargoes to provide a buffer against supply disruptions. The underlying logic—that stockpiling a critical commodity enhances system resilience against low-probability, high-consequence events—is not specific to petroleum. It applies wherever supply adequacy concerns justify holding reserves above immediate consumption needs.

This reserve logic was translated to electricity markets through FERC Order 841, which mandated that wholesale electricity markets establish mechanisms for energy storage resources to participate in capacity, energy, and ancillary services markets.[18] The order's underlying logic reflects the petroleum reserve principle: electricity systems benefit from resources that can provide capacity when needed, even if those resources sit idle most of the time.

The translation agent was FERC itself, acting as an authoritative regulatory body with power to impose new market structures. The petroleum reserve concept provided a template for thinking about resource adequacy that would not have emerged organically from electricity industry traditions. The result has been explosive growth in battery storage investment, as resources that previously had no clear revenue pathway gained access to capacity payments and ancillary service markets.

Conditions for Success

Comparing successful translation to CCS failures reveals patterns.

First, successful translation involves explicit recognition that translation is necessary—that knowledge cannot simply be lifted from one context and applied without modification. The petroleum engineers who moved to geothermal understood that their expertise was a starting point, not a finished product. The meteorologists who developed energy forecasting recognized that their models needed coupling with power system models. This awareness was often absent in CCS, where the assumption seemed to be that petroleum expertise automatically applied.

Second, successful translation requires identifiable agents who take explicit responsibility for the bridging function—individuals with cross-over experience, standards organizations developing consensus documents, or regulatory bodies imposing translation through authoritative rule-making. In CCS, the Independent Engineer occupies this structural position but is compromised by commercial pressures and legal disclaimers.

Third, successful translation benefits from incentive-compatible institutional structures. When regulatory frameworks reward honest uncertainty assessment, when liability structures ensure that optimistic projections impose costs on those who make them, actors have reason to invest in effective translation. Quest's relative success owed something to Alberta's liability transfer mechanism; Gorgon's struggles occurred in a regulatory context with weaker incentive alignment.

What Would Have to Change

The analysis presented here has implications at multiple levels—for individual projects, for industry practice, and for regulatory policy.

At the project level, FID processes should explicitly budget for translation as a distinct activity. Rather than assuming that petroleum expertise automatically applies to CCS, project developers should commission translation assessments that identify what knowledge transfers directly, what requires adaptation, and what does not apply. These assessments should be conducted by parties with appropriate incentives—ideally agents who bear consequences for translation failure.

At the industry level, CCS needs translation infrastructure analogous to what exists in other domains. Standards bodies should invest in documents explicitly addressing the adaptation of petroleum practices to injection contexts. Professional societies should develop training and certification for practitioners who specialize in boundary-spanning roles. Knowledge repositories should capture both successful and unsuccessful translation attempts, enabling organizational learning.

At the regulatory level, current liability structures are inadequate. The diffusion of responsibility that characterizes FID processes allows optimistic projections to pass through multiple checkpoints without accountability. Regulatory reforms should concentrate responsibility—requiring specific parties to certify geological assumptions and bear consequences if those assumptions prove wrong. Additionally, regulators should develop mechanisms for public assumption of tail risks that markets cannot price. Quest's success owes something to Alberta's willingness to accept transferred liability; similar mechanisms elsewhere could create incentives for honest uncertainty assessment.

The Translation Imperative

CCS occupies an uncomfortable position in the climate policy landscape. It is presented as essential infrastructure for net-zero emissions—a judgment supported by most integrated assessment models—while its operational track record inspires limited confidence. This gap between ambition and achievement reflects not technological immaturity but institutional failure. The knowledge required to execute CCS projects successfully largely exists, distributed across petroleum engineering, reservoir characterization, financial structuring, and regulatory design. What fails is the translation of that knowledge across the professional boundaries that separate its holders.

The mechanisms of failure—transmission, translation, misapplication—are not mysterious forces. They are predictable consequences of how expertise is organized in complex technical domains. They concentrate at the FID because that is where different ways of knowing must be reconciled into a single decision. And they produce systematic bias because the institutional pressures at FID favor optimism over accuracy.

The direction of energy transition is not in question. Climate physics and international commitments point toward decarbonization pathways that include substantial CCS deployment. The question is whether that deployment can occur without repeating the costly disappointments of the first generation of projects. The analysis here suggests that the answer depends on whether the industry and its regulators are willing to acknowledge the translation problem and design explicit mechanisms to address it.

The knowledge exists. The question is whether we can learn to translate it.

References

[1]: Institute for Energy Economics and Financial Analysis (IEEFA), Gorgon Carbon Capture Facility Hits New Lows in 2023-24, IEEFA Report (Sydney: IEEFA, September 2024)

[2]: Robert Gordon University Energy Transition Institute, Powering Up the Workforce: The Future of the UK Offshore Energy Workforce (Aberdeen: RGU, 2023), 34-47.

[3]: American Petroleum Institute, API RP 90: Annular Casing Pressure Management for Offshore Wells (Washington, DC: API, 2020); Norwegian Oil and Gas Association, NORSOK D-010: Well Integrity in Drilling and Well Operations, Rev. 5 (Lysaker: Standards Norway, 2021).

[4]: RISC Advisory, "Asia-Pacific CCS Overview & Gorgon—'Not a CCS Problem'" (presentation, SPE CCUS Symposium, Perth, October 2023)

[5]: Benjamin C. Esty, Modern Project Finance: A Casebook (New York: John Wiley & Sons, 2004), 78-112; Clean Air Task Force, Risk Allocation and Regulation for CO₂ Infrastructure (Boston: CATF, 2024), 23-31

[6]: Susan Leigh Star and James R. Griesemer, "Institutional Ecology, 'Translations' and Boundary Objects: Amateurs and Professionals in Berkeley's Museum of Vertebrate Zoology, 1907-39," Social Studies of Science 19, no. 3 (1989): 387-420

[7]: DNV, Independent Project Review of Storage Component of the CCS Project (report prepared for Government of Alberta, 2012), Alberta Open Government Portal

[8]: R. Andy Chadwick et al., "Geophysical Monitoring of the CO₂ Plume at Sleipner, North Sea: An Outline Review," in Proceedings of the 9th International Conference on Greenhouse Gas Control Technologies (Oxford: Elsevier, 2009), 303-311; Equinor, "CO₂ Injection Operations: Insights from Sleipner and Snøhvit" (presentation, SPE Aberdeen CCUS Technical Section, October 26, 2020)

[9]: IEEFA, Norway's Sleipner and Snøhvit CCS: Industry Models or Cautionary Tales?, IEEFA Report (Sydney: IEEFA, February 2023); Equinor, Snøhvit LNG: Lessons Learned from CO₂ Storage Operations, Technical Report (Stavanger: Equinor, 2020); Joshua A. White et al., "Snøhvit CO₂ Storage Project: Understanding the Role of Injection-Induced Mechanical Deformation" (presentation, NETL Carbon Storage Program Annual Review, Pittsburgh, August 2014)

[10]: Graham Davis and Brandon Owens, "Optimizing the Level of Renewable Electric R&D Expenditures Using Real Options Analysis," Energy Policy 31, no. 15 (2003): 1589-1608

[11]: Government of Western Australia, Gorgon Project: Carbon Dioxide Disposal Management Plan (Perth: Department of Mines and Petroleum, 2015), Tabled Paper 2730, Parliament of Western Australia

[12]: IEEFA, Gorgon Carbon Capture Facility Hits New Lows in 2023-24, September 2024; Boiling Cold, "Time's Up on Gorgon's Five Years of Carbon Storage Failure," Boiling Cold, October 15, 2024；Boiling Cold, "Gorgon LNG Emissions to Rise as Sand Clogs $3.1B CO₂ System," Boiling Cold, November 8, 2024

[13]: Donald W. Vasco et al., "Satellite-Based Measurements of Surface Deformation Reveal Fluid Flow Associated with the Geological Storage of Carbon Dioxide," Geophysical Research Letters 37, no. 3 (2010): L03303; Joshua P. Morris et al., "A Coupled Reservoir Simulation-Geomechanical Modelling Study of the CO₂ Injection-Induced Ground Surface Uplift Observed at Krechba, In Salah," Energy Procedia 4 (2011): 3495-3502; Jonny W. Johnson et al., "Geomechanical Behavior of the Reservoir and Caprock System at the In Salah CO₂ Storage Project," Proceedings of the National Academy of Sciences 111, no. 24 (2014): 8747-8752

[14]: Shell Canada, Quest Carbon Capture and Storage Facility: Annual Summary Report 2023 (Calgary: Shell Canada, 2024); Global CCS Institute, Global Status of CCS 2024 (Melbourne: Global CCS Institute, 2024), 45-48; Clean Air Task Force, "The Shell Quest Carbon Capture and Storage Project," CATF Technical Review (Boston: CATF, 2024)

[15]: Government of Alberta, Carbon Capture and Storage Statutes Amendment Act, Statutes of Alberta, 2010, Chapter C-2.5 (Edmonton: Alberta Queen's Printer, 2010); Alberta Energy Regulator, Quest Carbon Capture and Storage Project: Closure Plan (Calgary: AER, 2023)

[16]: International Energy Agency, The Future of Geothermal Energy (Paris: IEA, 2024), 87-102; Guðmundur Þórhallsson and Sverrir Sveinbjörnsson, "Geothermal Drilling Technology," in Geothermal Power Generation: Developments and Innovation, ed. Ronald DiPippo (Amsterdam: Elsevier, 2016), 197-231

[17]: Pierre Pinson, "Wind Energy: Forecasting Challenges for Its Operational Management," Statistical Science 28, no. 4 (2013): 564-585; ANEMOS Project Consortium, Advanced Short-Term Forecasting of Wind Generation, Final Report, European Commission Framework Programme 6, Contract No. ENK5-CT-2002-00665 (Brussels: European Commission, 2008)

[18]: Federal Energy Regulatory Commission, Electric Storage Participation in Markets Operated by Regional Transmission Organizations and Independent System Operators, Order No. 841, 162 FERC ¶ 61,127 (Washington, DC: FERC, February 15, 2018); Resources for the Future, "FERC Order 841 and the Transformation of Energy Storage Markets," RFF Working Paper 24-08 (Washington, DC: RFF, 2024)

Copyright

© 2026 Alex Yang Liu. All rights reserved.
Publisher: Terawatt Times Institute | ISSN: 3070-0108
Version: 1.0 | Date: January 2026
Citation: Liu, A.Y. "Lost in Translation: Why CCS Projects Misread the Underground." Terawatt Times, January 2026. ISSN 3070-0108.

License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0).
You are free to share and distribute this work with proper attribution, for non-commercial purposes, without modification.
For permissions beyond this license, contact the author.

Disclaimer

This paper represents the author's independent analysis and does not necessarily represent the views of Terawatt Times Institute or any other organization. While every effort has been made to ensure accuracy, the author and publisher assume no responsibility for errors or omissions. Data and projections are based on publicly available sources as of January 2026.