An update on our dynamic modeling framework

Reactive impurities represent one of the least understood, yet potentially most critical challenges for large-scale deployment of carbon capture and storage (CCS). While CO₂ purity specifications are becoming increasingly conservative, they are still not grounded in a mechanistic understanding of when, where, and why highly corrosive acids form in real CCS infrastructure.

Since 2024, supported by Gassnova, we have been developing a physics-based, mechanistic modelling engine to predict chemical reactions between trace impurities in CO₂-rich mixtures. Over the past six months, we have systematically compared our model against 9394 hours of experimental data published in the last decade by the Institute of Energy Technology (IFE).

Reactive impurities in CCS must be understood before widespread deployment

CO₂ captured from different industrial sources inevitably contains trace amounts of impurities at the parts per million (ppm) level. Some of these impurities are chemically reactive and can interact in ways that are not captured by conventional purity specifications.

Nitrogen dioxide (NO₂) is a particularly problematic impurity that is prevalent in CO₂ streams from a wide range of capture technologies. In the presence of water, NO₂ forms nitric acid (HNO₃), which is highly corrosive to carbon steel. If sulfur-containing species such as SO₂ or H₂S are also present, reactions between impurities can lead to the formation of sulfuric acid (H₂SO₄). If sufficient sulfuric acid forms in the gas-phase from chemical reactions, it will eventually precipitate out as an aqueous phase, which catalyzes further reactions and accelerates acid production.

For a decade, experimental work on reactive impurities has been carried out at IFE in Norway. As early as 2014, experiments conducted at Kjeller in Norway demonstrated that highly corrosive acids precipitate from CO₂ mixtures containing trace impurities. Mixtures of nitric and sulfuric acid were shown to compromise the integrity of carbon steel over surprisingly short time scales of months, or even weeks. These findings underline the importance of understanding not only if acids form, but where and under what conditions they precipitate in CCS infrastructure.

Concern over acid formation is the basis for today’s extremely conservative CO₂ purity specifications, such as those adopted in the Northern Lights project. In these specifications, the allowable NO₂ concentration has lately been reduced to 1 ppm, which is even lower than the limit for food-grade CO₂. Removing impurities to this degree is costly, and it remains unclear whether such limits are sufficient, or even necessary, to prevent acid formation under all operating conditions. For example, acids may still precipitate at even lower impurity levels if the temperature and pressure drop.

At present, uncertainty around acid formation is not just a corrosion issue. It represents one of the remaining systemic risks that can halt cost-effective CCS deployment if efficient mitigation strategies are not developed.

Why a dynamic modelling framework is needed

Acid formation in CO₂ pipelines and infrastructure is governed by a complex interplay of physical and chemical mechanisms. Impurities may adsorb onto pipeline walls, where surface reactions become important. Other reactions occur in the gas phase, while solids such as sulfur or salts may precipitate under certain conditions. If an aqueous acidic phase forms, it will act as a catalyst for additional reactions, rapidly increasing acid production.

These processes are dynamic and strongly dependent on local conditions such as temperature, pressure, flow regime, mixing of streams, and transient water availability. Static impurity specifications or equilibrium-type models cannot capture these effects.

A dynamic mechanistic modelling engine is therefore essential to:

• Develop effective strategies for acid mitigation and handling

• Interpret and understand experimental observations

• Design targeted experiments that reduce uncertainty

• Predict if and where acids will precipitate in real CCS infrastructure

The Status on our modeling framework

Development of our dynamic modelling framework for reactive impurities in CO₂ streams started late 2024, following discussions with the Institute of Energy Technology (IFE), which highlighted the lack of predictive tools. Since then, we have made substantial progress.

Our current framework is based on a physics-based thermodynamic description of reactive electrolyte systems, combined with state-of-the-art models for adsorption onto carbon steel surfaces. A comprehensive reaction network has been implemented, covering gas-phase, liquid-phase, and surface reactions, together with thermodynamic prediction of precipitation of relevant solid phases. Where experimental data are unavailable, reaction kinetics have been estimated using first-principles molecular methods.

The modelling engine has been extensively validated against published experimental data from IFE. Using the exact same initial and boundary conditions as in the experimental rigs, and without tuning parameters to individual experiments, the model reproduces measured outlet concentrations and dynamic behavior. The figure below illustrates this by comparing measured water concentrations at the outlet of the experimental rig (markers) with model predictions (solid line), demonstrating an excellent agreement.

In total, similar comparisons have been performed for 29 independent experiments, corresponding to 9,394 hours of operation of the experimental rig over the past ten years. For all published experiments conducted above 15 °C, the model reproduces the observed dynamic concentration profiles and predicts the onset of acid precipitation in agreement with experimental observations at IFE.

This level of agreement indicates that the modelling engine captures the dominant physical and chemical mechanisms governing reactive impurities in CO₂-rich systems, at least at higher temperatures, and that it has the potential to become a predictive basis for engineering analysis rather than as a case-specific fitting tool.

From dynamic modeling framework to practical engineering tool

Experimental rigs at IFE typically use pipes with diameters on the order of a few centimeters, while real CO₂ transport pipelines may be tens of times larger. In addition, experimental setups often involve continuous injection and removal of impurities, which differs fundamentally from pipeline transport or CO₂ storage in tanks.

To bridge this gap, we have translated our validated mechanistic model to representations of large pipelines, tanks, and other CCS infrastructure components. A beta version of this functionality has been implemented in our software platform, allowing reactive CO₂ mixtures to be propagated through pipeline networks under realistic operating conditions.

This enables identification of locations where acids may precipitate due to cooling, mixing of streams, or changes in flow conditions, effects that cannot be assessed using static specifications alone.

The way forward – reducing uncertainty

While the modelling framework has reached a level where it can reproduce existing experimental data, further work is required to reduce uncertainty in key reactions before the model can fully support high-confidence engineering decisions. The largest uncertainties are now associated with low-temperature conditions relevant for ship transport and certain storage scenarios.

Addressing these gaps requires targeted experimental campaigns combined with continued model development. This is not incremental refinement, but work that directly determines whether CCS systems must rely on conservative over-design, or can be engineered for safety and cost-efficiency based on mechanistic understanding.

Reactive impurities and acid formation remain one of the last unresolved challenges for large-scale CCS deployment. Mechanistic modelling, anchored in experimental data, is a necessary step toward overcoming this challenge and enabling CCS at the scale and cost required for meaningful climate impact.

An update on our dynamic modeling framework

Reactive impurities represent one of the least understood, yet potentially most critical challenges for large-scale deployment of carbon capture and storage (CCS). While CO₂ purity specifications are becoming increasingly conservative, they are still not grounded in a mechanistic understanding of when, where, and why highly corrosive acids form in real CCS infrastructure.

Since 2024, supported by Gassnova, we have been developing a physics-based, mechanistic modelling engine to predict chemical reactions between trace impurities in CO₂-rich mixtures. Over the past six months, we have systematically compared our model against 9394 hours of experimental data published in the last decade by the Institute of Energy Technology (IFE).

Reactive impurities in CCS must be understood before widespread deployment

CO₂ captured from different industrial sources inevitably contains trace amounts of impurities at the parts per million (ppm) level. Some of these impurities are chemically reactive and can interact in ways that are not captured by conventional purity specifications.

Nitrogen dioxide (NO₂) is a particularly problematic impurity that is prevalent in CO₂ streams from a wide range of capture technologies. In the presence of water, NO₂ forms nitric acid (HNO₃), which is highly corrosive to carbon steel. If sulfur-containing species such as SO₂ or H₂S are also present, reactions between impurities can lead to the formation of sulfuric acid (H₂SO₄). If sufficient sulfuric acid forms in the gas-phase from chemical reactions, it will eventually precipitate out as an aqueous phase, which catalyzes further reactions and accelerates acid production.

For a decade, experimental work on reactive impurities has been carried out at IFE in Norway. As early as 2014, experiments conducted at Kjeller in Norway demonstrated that highly corrosive acids precipitate from CO₂ mixtures containing trace impurities. Mixtures of nitric and sulfuric acid were shown to compromise the integrity of carbon steel over surprisingly short time scales of months, or even weeks. These findings underline the importance of understanding not only if acids form, but where and under what conditions they precipitate in CCS infrastructure.

Concern over acid formation is the basis for today’s extremely conservative CO₂ purity specifications, such as those adopted in the Northern Lights project. In these specifications, the allowable NO₂ concentration has lately been reduced to 1 ppm, which is even lower than the limit for food-grade CO₂. Removing impurities to this degree is costly, and it remains unclear whether such limits are sufficient, or even necessary, to prevent acid formation under all operating conditions. For example, acids may still precipitate at even lower impurity levels if the temperature and pressure drop.

At present, uncertainty around acid formation is not just a corrosion issue. It represents one of the remaining systemic risks that can halt cost-effective CCS deployment if efficient mitigation strategies are not developed.

Why a dynamic modelling framework is needed

Acid formation in CO₂ pipelines and infrastructure is governed by a complex interplay of physical and chemical mechanisms. Impurities may adsorb onto pipeline walls, where surface reactions become important. Other reactions occur in the gas phase, while solids such as sulfur or salts may precipitate under certain conditions. If an aqueous acidic phase forms, it will act as a catalyst for additional reactions, rapidly increasing acid production.

These processes are dynamic and strongly dependent on local conditions such as temperature, pressure, flow regime, mixing of streams, and transient water availability. Static impurity specifications or equilibrium-type models cannot capture these effects.

A dynamic mechanistic modelling engine is therefore essential to:

• Develop effective strategies for acid mitigation and handling

• Interpret and understand experimental observations

• Design targeted experiments that reduce uncertainty

• Predict if and where acids will precipitate in real CCS infrastructure

The Status on our modeling framework

Development of our dynamic modelling framework for reactive impurities in CO₂ streams started late 2024, following discussions with the Institute of Energy Technology (IFE), which highlighted the lack of predictive tools. Since then, we have made substantial progress.

Our current framework is based on a physics-based thermodynamic description of reactive electrolyte systems, combined with state-of-the-art models for adsorption onto carbon steel surfaces. A comprehensive reaction network has been implemented, covering gas-phase, liquid-phase, and surface reactions, together with thermodynamic prediction of precipitation of relevant solid phases. Where experimental data are unavailable, reaction kinetics have been estimated using first-principles molecular methods.

The modelling engine has been extensively validated against published experimental data from IFE. Using the exact same initial and boundary conditions as in the experimental rigs, and without tuning parameters to individual experiments, the model reproduces measured outlet concentrations and dynamic behavior. The figure below illustrates this by comparing measured water concentrations at the outlet of the experimental rig (markers) with model predictions (solid line), demonstrating an excellent agreement.

In total, similar comparisons have been performed for 29 independent experiments, corresponding to 9,394 hours of operation of the experimental rig over the past ten years. For all published experiments conducted above 15 °C, the model reproduces the observed dynamic concentration profiles and predicts the onset of acid precipitation in agreement with experimental observations at IFE.

This level of agreement indicates that the modelling engine captures the dominant physical and chemical mechanisms governing reactive impurities in CO₂-rich systems, at least at higher temperatures, and that it has the potential to become a predictive basis for engineering analysis rather than as a case-specific fitting tool.

From dynamic modeling framework to practical engineering tool

Experimental rigs at IFE typically use pipes with diameters on the order of a few centimeters, while real CO₂ transport pipelines may be tens of times larger. In addition, experimental setups often involve continuous injection and removal of impurities, which differs fundamentally from pipeline transport or CO₂ storage in tanks.

To bridge this gap, we have translated our validated mechanistic model to representations of large pipelines, tanks, and other CCS infrastructure components. A beta version of this functionality has been implemented in our software platform, allowing reactive CO₂ mixtures to be propagated through pipeline networks under realistic operating conditions.

This enables identification of locations where acids may precipitate due to cooling, mixing of streams, or changes in flow conditions, effects that cannot be assessed using static specifications alone.

The way forward – reducing uncertainty

While the modelling framework has reached a level where it can reproduce existing experimental data, further work is required to reduce uncertainty in key reactions before the model can fully support high-confidence engineering decisions. The largest uncertainties are now associated with low-temperature conditions relevant for ship transport and certain storage scenarios.

Addressing these gaps requires targeted experimental campaigns combined with continued model development. This is not incremental refinement, but work that directly determines whether CCS systems must rely on conservative over-design, or can be engineered for safety and cost-efficiency based on mechanistic understanding.

Reactive impurities and acid formation remain one of the last unresolved challenges for large-scale CCS deployment. Mechanistic modelling, anchored in experimental data, is a necessary step toward overcoming this challenge and enabling CCS at the scale and cost required for meaningful climate impact.