3D simulations predict the underground journey of CO₂

Deep beneath the North Sea, engineers have been quietly carrying out one of the most ambitious climate experiments in history. Since the 1990s, millions of tons of carbon dioxide (CO₂) have been pumped into layers of porous rock more than a kilometer underground, locked away in salty aquifers where it can’t reach the atmosphere. To the casual observer, this might sound like stacking boxes in a basement: once you put them there, they stay there forever. But once the CO₂ is pumped below the Earth’s surface, what happens to it? Will it stay safely locked away like those basement boxes, or might it leak back out over centuries? Answering this question isn’t easy, since the process unfolds out of sight, deep underground, over timescales far longer than any human experiment can run.

That’s where a team of researchers from the University of Twente (Netherlands), TU Wien (Austria), Newcastle University (UK), and Sapienza University of Rome (Italy) stepped in. In their paper published in Geophisical Research Letters, Marco De Paoli and colleagues tackled this puzzle using some of the world’s most powerful supercomputers. Their work offers the most detailed look yet at how CO₂ mixes with salty groundwater, that is, brine, inside porous rocks, and what that means for the future of carbon storage.

When CO₂ is injected underground, it doesn’t just sit there as a giant bubble. At first, it forms a layer beneath the rock that caps the storage formation. This might sound worrying, since if that cap rock cracked, CO₂ could seep upward. Fortunately, physics lends a helping hand. Some of the gas dissolves into the brine. However, CO₂-saturated brine is heavier than normal brine. That density difference triggers convection. Think of cream swirling into coffee, but in slow motion, over the course of decades or centuries. Dense plumes of CO₂-rich brine sink downward, pulling in more CO₂ from above. Over time, this mixing spreads the gas throughout the formation, keeping it safely trapped and less likely to leak.

The problem is that this process is fiendishly complex. Underground rocks are irregular, flows are chaotic, and the timeframes are immense. Direct experiments are nearly impossible. That’s why computer simulations are so powerful: they let scientists fast-forward through thousands of years of underground mixing.

Previous models of CO₂ mixing mostly relied on simplified two-dimensional (2D) simulations. These are easier to run but miss some of the nuances of real-world, three-dimensional (3D) flows.

The new study by De Paoli’s team breaks through that limitation. Using the LUMI supercomputer in Finland (one of the fastest in the world) they ran large-scale 3D simulations of CO₂ convection in porous rock at unprecedented detail. They found that there are three stages of mixing. First, CO₂ spreads slowly by diffusion (like sugar dissolving in tea). Then, gravity takes over, forming “fingers” of CO₂-rich brine that plunge downward. Finally, as the rock becomes saturated, the mixing slows and eventually shuts down. Also, hey realized thata 3D mixing is stronger than 2D, but less so than expected. Earlier studies suggested that 3D flows might boost mixing efficiency by about 25%. This team found it’s closer to 13%. That might sound like splitting hairs, but in the world of climate modeling, where estimates guide billion-dollar storage projects, precision matters. Despite the complex simulations, the researchers distilled their results into a straightforward physical model. This model can predict how much CO₂ dissolves over time, making it a valuable tool for designing and monitoring real storage sites.

Climate change is a long game, and so is carbon storage. Policymakers, energy companies, and scientists need reliable predictions: How fast will injected CO₂ mix? How much will dissolve over decades? What’s the risk of it escaping? This new research strengthens confidence in those predictions. The study narrows the uncertainty in long-term storage models, showing that 3D effects are smaller than previously thought. It also provides a ready-to-use formula that engineers can plug into their large-scale simulations, saving time and improving accuracy.

The team even applied their model to the Sleipner site in the North Sea, one of the world’s first commercial CO₂ storage projects. They found that about half the injected CO₂ there could dissolve into brine in less than 20 years, while reaching 90% saturation would take more than a century. This kind of insight is crucial for evaluating the safety and performance of storage operations.

Of course, the Earth isn’t made of perfectly uniform, sponge-like rocks. Real storage formations are riddled with fractures, layers, and chemical quirks. Future studies will need to take these complexities into account. But back to the basement boxes, you can think of CO₂ sequestration as tucking away our industrial emissions in the planet’s basement. Unlike boxes, the air in that basement circulates, but as the paper concludes, over the long haul, like the boxes, the air is supposed to stay in, as long as the door remains shut.

If you want to learn more, the original article titled "Simulation and Modeling of Convective Mixing of Carbon Dioxide in Geological Formations" on Geophisical Research Letters at https://doi.org/10.1029/2025GL114804.