Digital Twins Make CO₂ Storage Safer

As greenhouse gases accumulate in the Earth’s atmosphere, scientists are developing technologies to pull billions of tons of carbon dioxide (CO2) from the air and inject it deep underground. 

The idea isn’t new. In the 1970s, Italian physicist Cesare Marchetti suggested that the carbon dioxide polluting the air and warming the planet could be stored underground. The reality of how to do it cost-effectively and safely has challenged scientists for decades. 

Geologic carbon storage — the subterranean storage of CO2 — comes with significant challenges, most importantly, how to avoid fracturing underground rock layers and letting gas escape into the atmosphere. Carbon, a gas, can behave erratically or leak whenever it’s stored in a compressed space, making areas geologically unstable and potentially causing legal headaches for corporations that invest in it. This uncertainty, coupled with the expense of the carbon capture process and its infrastructure, means the industry needs reliable predictions to justify it. 

Georgia Tech researcher and Georgia Research Alliance Eminent Scholar Felix J. Herrmann has an answer. His lab, Seismic Laboratory for Imaging and Modeling (SLIM), uses advanced artificial intelligence (AI) techniques to create algorithms that monitor and optimize carbon storage. The algorithms work as “digital twins,” or digital replicas of underground systems, facilitating the safe, efficient storage of CO2 underground.

“The trick is you want the carbon to stay put — to avoid the risk of, say, triggering an earthquake or the carbon leaking out,” said Herrmann, professor in the School of Earth and Atmospheric Sciences and the School of Electrical and Computer Engineering. “We’re developing a digital twin that allows us to monitor and control what is happening underground.”

Predicting the Best Place

Waveform Variational Inference via Subsurface Extensions with Refinements (WISER) is an algorithm that uses sound waves to analyze underground structures. WISER runs on AI, enabling it to work more efficiently than most algorithms while remaining computationally feasible. To improve accuracy, WISER makes small adjustments using sound wave physics to show how fast sound travels through different materials and where there’s variation in underground layers. This helps to create detailed, reliable images of underground areas for better predictions of carbon storage. 

WISER allows researchers to work with uncertainties, which is vital for understanding the risk of these underground storage projects.

Scaling the Algorithm

While Herrmann’s lab has been working to apply neural networks to seismic imaging for years now, WISER required them to increase the networks’ scale. Making multiple predictions is a much larger problem that requires a bigger, more potent network, but these types of neural networks only run on graphics processing units (GPUs), which are known for speed but are limited in memory.

To optimize the GPU, Rafael Orozco, a computational science and engineering Ph.D. student, created a new type of neural network that can train with very little memory. This open-source package, InvertibleNetworks, enables the network to train on very large inputs and create multiple output images conditioned on the observed seismic data.

WISER’s fundamental innovation is for the lab’s next concept: creating digital twins for carbon storage. These twins can act as monitoring systems to optimize and mitigate risks of carbon storage projects. 

Devising the Digital Twin

Digital twins are dynamic virtual models of objects in the real world, capable of replicating their behavior and performance. They rely on real-time data to evolve and have been used to replicate factories, cities, spacecraft, and bodies, to make informed decisions about healthcare, maintenance, production, supply chains, and — in Herrmann’s case — geologic carbon storage.

Herrmann and his team have developed an “uncertainty-aware” digital twin. That means the tool can manage risks and make decisions in an uncertain, unseen environment — because it’s been designed to recognize, quantify, and incorporate uncertainties in CO2 storage.

Probing the Unseen

Subsurface conditions are diverse and complex, making the management of greenhouse gas storage a delicate process. Without careful monitoring, the injection of CO2 can increase pressure in rock formations, potentially fracturing the cap rock that is supposed to keep the gas underground.

“The digital twin addresses this through simulations in tandem with observations,” said Herrmann, whose team linked two different scientific fields — geophysics and reservoir engineering — for a more comprehensive understanding of the subsurface environment. Specifically, they combined geophysical well observations with seismic imaging.

Geophysical well observation involves drilling a hole in the subsurface in a geological area of interest and collecting data by lowering a probe into the borehole to take measurements. Seismic imaging, on the other hand,  uses acoustic waves to create images based on the analysis of wave vibrations.

“Bridging the gap between different fields of research and combining various data sources allows our digital twin to provide a more accurate and detailed picture of what’s happening underground,” Herrmann said. 

To integrate and leverage these diverse datasets built from observations and simulations, the team used advanced AI techniques like simulation-based inference and sequential Bayesian inference, a method of updating information as more data becomes available. The ongoing learning allows researchers to quantify uncertainties in the subsurface environment and predict how that system will respond to CO2 injection. The digital twin updates its understanding as new data becomes available.

Making Informed Decisions

Herrmann’s team tested the digital twin, simulating different states of an underground reservoir, including permeability, which is the measure of how easily fluids flow through rock. The goal was to find the maximum injection rate of CO2 without causing fractures in the cap rock.

“The work highlights how dynamic digital twins can play a key role in mitigating the risks associated with geologic carbon storage,” said Herrmann, whose research group is supported in part by large oil and gas companies, including Chevron and  ExxonMobil. “Companies are now in the process of starting large offshore projects for which the digital twin is being developed.”

But there is still plenty of work to be done, he added. For instance, the digital twin can monitor the subsurface and provide critical information about that uncertain environment. It can inform. But it still needs adjustments by humans for each new CO2 injection site, and Herrmann and his team are working on further developing the technology — giving the digital twins the ability to quickly replicate themselves so they can be deployed massively and quickly to meet the demands of mitigating climate change.

“Our aim is to make them smarter,” Herrmann said. “To make them more adaptable, so they can control CO2 injections, become more responsive to risks, and adapt to a wide range of complex situations in real time.”