Why you need quantitative risk analysis to store carbon effectively

Today, subsurface risk assessment methods typically rely on deterministic models and best-estimate scenarios. These often fail to capture the full range of uncertainties associated with subsurface storage, leaving developers exposed to unexpected capex and opex costs. By leveraging subsurface modeling capabilities, however, it’s possible to shift toward quantitative, model-based risk analysis instead.

The aim is to provide a more detailed and objective assessment of potential risks by incorporating probabilistic methods, sensitivity analysis, and scenario modeling. In other words, quantitative risk analysis reduces uncertainty, which then improves project predictability, strengthens financial planning, and builds investor and regulatory confidence—all of which are crucial for successful carbon storage projects.

Managing carbon storage uncertainty with model ensembles

Decades of exploration and production activities in the oil and gas industry have resulted in extensive datasets for many reservoirs. This is part of what allows geologists and engineers to now build highly calibrated models with relatively low uncertainty. CCS projects, on the other hand, tend to lack enough subsurface data.

Many potential CO₂ storage sites are deep saline aquifers that haven’t been characterized as thoroughly as their oil and gas counterparts. Injected CO₂ doesn’t generate a consistent signal like oil and gas production rates either, making it much harder to validate and refine a single model. And because CO₂ injection occurs over decades, some key risks—such as plume migration or caprock integrity issues—may only manifest after many years.

Rather than seeking that single best-fit model, I recommend leveraging a collection of multiple subsurface models, each representing a different plausible scenario given how your project’s uncertain parameters might vary. These model ensembles allow you to explore a range of possible outcomes that, when analyzed using a probabilistic approach, help quantify risks and improve your decision making.

A proposed methodology for quantitative risk analysis

When assessing your CO₂ injection risks, it’s crucial to understand how fluid flow and geomechanical responses interact. Traditional reservoir modeling focuses primarily on the former, simulating how CO₂ moves through the storage formation. But this approach is insufficient on its own. Injection-induced pressure changes can alter the stress state of the subsurface, which potentially means undesirable consequences to containment.

Coupled flow and geomechanical modeling solves this by simulating both pressure-driven fluid movement and the resulting mechanical alterations in your formations—as well as any changes in the porosity and permeability of intact rocks and discontinuities. It's an integrated approach that allows you to simulate

• fault reactivation
• surface heave
• induced seismicity
• leakages through caprock and faults
• lateral migration of the CO₂ plume outside permit limits.

Subsurface environments are inherently uncertain though, and many key parameters (e.g., permeability, fault properties, and rock strength) aren't known with absolute precision. That’s where a model ensemble comes in: It accounts for the uncertainty by varying the parameters across a range of plausible values and running multiple simulations.

Using these different parameter combinations, the model ensemble captures the full range of possible subsurface behaviors, providing developers with an opportunity to numerically quantify the probability of those events occurring. How? By dividing how many of the ensemble's models exhibit one of the events by the total number of simulations.

This brings us to risk, which is commonly defined as the probability of an event occurring (the formula we just discussed) multiplied by its severity. Severity can be quantified directly from the subsurface model by evaluating the magnitude of the hazard. Here are some examples:

• CO₂ leakage—Severity can be assessed by calculating the total amount of CO₂ released and considering the zone where the leakage occurs. For instance, a minor leak in deep geological layers may be less critical than one reaching shallow aquifers or the atmosphere.
• Fault reactivation—Severity can be quantified by evaluating deformation along the fault plane, which indicates the potential for significant structural movement.
• Surface heave—Severity can be assessed by calculating the vertical displacement at the surface, as excessive uplift may affect infrastructure or land stability.
• Induced seismicity—Severity is determined by the earthquake magnitude generated by fault slip, with larger magnitudes posing greater risks to nearby structures and public safety.

All the above are outputs of coupled subsurface models that can be used to numerically quantify and classify severities. Multiplying numerical probability by severity results in a geospatial distribution of each risk that can be obtained at different stages of CO₂ injection.

Why is numerical risk quantification essential?

A modeling-based risk quantification not only enhances technical decision making but also facilitates clear and efficient communication with regulators and stakeholders. Assessing the risk of a CCS project involves multiple parties, including operators, regulatory agencies, policymakers, and the public, all of whom have different levels of expertise and varying perceptions of what constitutes an acceptable risk. By quantifying risk, you ground discussions in objective data rather than subjective interpretations, thereby improving transparency and reducing the odds of parties misunderstanding.

The way you classify risk must also be meaningful and actionable. What constitutes high, medium, or low risk? Without a structured framework, risk perception can vary significantly based on personal experience or bias. Try involving regulators and stakeholders early in the process to help define that classification and avoid inconsistent evaluations. If you collaboratively establish thresholds for acceptable risk levels from the outset, then all parties work within a shared reference system.

And let’s not forget that numerical quantification also helps communicate complex subsurface risks in a way that nonexperts can understand. Instead of presenting highly technical geological and engineering details, you can update your stakeholders with quantified risk values that indicate the probability and impact of potential failures. This both fosters trust and enables proactive engagement; stakeholders can see that risks are systematically assessed and managed rather than being subject to individual judgment.

The ever-present and ever-necessary role of digital

Significant progress has been made in developing digital solutions for quantitative risk analysis in CCS projects, and ongoing developments continue to advance our subsurface understanding. Computational power has improved dramatically, allowing for the efficient execution of large model ensembles that allow us to better predict the severity of an event.

Regulatory acceptance of digital risk assessment is also evolving, with many agencies recognizing the value of probabilistic, data-driven approaches over traditional qualitative methods. The transparency and explainability of digital models are also improving, which adds to the trust among regulators and stakeholders.

This is important because incorporating advanced subsurface characterization and modeling into CCS can help reduce uncertainties, quantify risks, and optimize CO₂ injection performance. As a result, operators can significantly improve investment confidence and accelerate regulatory approvals. And through ongoing advances in computational modeling and machine learning, digital solutions enable a more comprehensive understanding of complex subsurface behavior—along with the potential risks associated with CO₂ storage—for everyone.