How CMG Reduced CO₂ Storage Simulation Times by Up to 80% While Preserving Area of Review Accuracy

For many CCS projects, the biggest challenge is no longer building a model. It is running enough simulations to support uncertainty analysis, optimization, and regulatory approval.

The regulatory approval depends on demonstrating:

• CO₂ plume containment 
• Pressure propagation 
• Area of Review (AoR) 
• Long-term storage integrity 

However, the most accurate representation of the aquifer is achieved through explicit high-resolution modelling, an approach that is often required to support regulatory confidence, but that also results in very large models that make optimization and uncertainty analysis prohibitively expensive.

In this study, CMG evaluated multiple approaches for reducing computational complexity in a large-scale saline aquifer storage model containing more than 52 million grid cells, while preserving the accuracy required for regulatory assessments.

Key Insight

For CO₂ storage projects, simulation speed is only valuable if pressure propagation remains accurate. The fastest model is not always the most reliable model.

The Challenge: Regulatory Confidence Requires Large Models

Accurately representing large saline aquifers requires very large simulation models.

In this study, the reference model contained:

• 52 million active cells 
• 5 CO₂ injectors 
• 25 years of injection 
• heterogeneous porosity and permeability distributions

Why This Matters

One simulation may be manageable.

Hundreds or thousands of simulations required for:

• uncertainty analysis 
• optimization studies 
• risk assessment 
• regulatory workflows 

are not.

The question became: How can simulation time be reduced without compromising regulatory accuracy?

Defining What Really Matters: Area of Review

The AoR represents the region where pressure changes or CO₂ migration could affect storage integrity and therefore plays a critical role in permitting and regulatory approval.

For this study, the Area of Review was defined using:

• CO₂ saturation thresholds 
• pressure increase thresholds 

These parameters became the benchmark for evaluating every reduced model. 

Testing Multiple Reduction Strategies

Several methodologies were evaluated:

• Analytical aquifers 
• Flux boundaries 
• Volume and transmissibility modifiers 
• Tartan grids 
• Static amalgamation 
• Dynamic gridding 

Each was assessed using two criteria:

1. Accuracy of AoR prediction 
2. Simulation time reduction

Insight 1: The Fastest Model Was Not the Most Reliable

Analytical aquifer models produced the largest computational gains.

• Simulation time improved by approximately 7.5x faster

However, pressure propagation accuracy suffered significantly.

• Critical pressure area error reached −54.9%

while plume predictions remained relatively accurate.

Insight 2: Pressure Is Harder to Preserve Than the CO₂ Plume

Plume area errors remained below approximately 3% across all approaches. 

Pressure propagation, however, proved far more sensitive.

Several techniques that appeared acceptable based on plume matching alone produced significant AoR errors.

Insight 3: Preserving Model Structure Produced Better Results

Methods that retained the original grid architecture consistently delivered more accurate results.

These included:

• Tartan grids 
• Static amalgamation 
• Dynamic gridding 

All maintained low pressure errors while still reducing computational effort.

Insight 4: Static Amalgamation Delivered the Best Balance of Speed and Accuracy

Among all methods tested, static amalgamation delivered the best results for both accuracy, speed and ease of implementation.

Using CMOST optimization, the team evaluated 110 candidate models and identified an optimal amalgamation strategy. 

The resulting model:

• reduced simulation time by approximately 80% 
• maintained AoR error near 3% 
• preserved pressure behavior 
• retained plume accuracy

A faster model (higher amalgamation) could be chosen if the error is acceptable.

Why It Matters

The goal is not simply to create faster models.

The goal is to create models that are:

• fast enough for uncertainty analysis 
• accurate enough for regulatory approval 

Static amalgamation achieved both.

Business Impact

Reducing simulation time by up to 80% fundamentally changes what is possible.

Instead of evaluating:

• a handful of scenarios 

Operators can evaluate:

• hundreds of uncertainty realizations 
• optimization studies 
• injection strategies 
• regulatory sensitivities 

within practical project timelines. 

Why CMG Matters

This study highlights an increasingly important challenge in CCUS:

Regulatory models must be both accurate and computationally practical to ensure that regulatory requirements are met while maximizing storage capacity.

CMG's workflow combined:

• GEM for compositional CO₂ storage simulation 
• CMOST for optimization and calibration 
• advanced gridding and model-reduction techniques 

to preserve the physics that matter most for regulatory decision-making.

Conclusion

For carbon storage projects, the challenge is no longer just building accurate models.

The challenge is building models that are accurate enough for regulatory approval and efficient enough for large-scale uncertainty analysis.

This study demonstrated that not all model-reduction techniques are equal. While some approaches delivered impressive speed gains, they compromised the pressure predictions required for reliable AoR assessments.

By combining GEM, CMOST, and optimized grid-reduction workflows, CMG showed that it is possible to dramatically reduce simulation time while preserving the regulatory metrics that matter most.

Ultimately, for large-scale CO₂ storage projects, the ability to run more, not just larger, simulations enables better decisions.