Heriot-Watt University, Edinburgh – 24-28 March 2025 —
The Digital Lab for Porous Media and Flow Processes (DigiPorFlow) put its signature pore-scale expertise in the spotlight during last week’s ECO-AI Workshop and Hackathon, a UK-wide gathering focused on AI for carbon-capture and storage (CCS). Based in Heriot-Watt’s Institute of GeoEnergy Engineering, DigiPorFlow pairs high-resolution imaging with advanced numerics and machine-learning models to unravel how fluids move through complex rocks — knowledge that is critical for safe, cost-effective CO₂ storage.
Workshop highlights
At the two-day workshop (24–25 March), Marcos Cirne and Zhenkai “Josh” Bo showcased the group’s newest tools. Cirne demonstrated an AI-accelerated reactive-transport pipeline that shrinks simulation times from days to minutes, while Bo outlined a multi-scale pore-network framework that links micro-CT images to reservoir-scale predictions. Their talks illustrated how deep neural networks can de-risk injection strategies long before a well is drilled.
Impact in the hackathon arena
DigiPorFlow talent was equally visible in the 72-hour ECO-AI Hackathon (26–28 March). Group members joined the “Sub-surface Modelling” track, fusing their open-source GeoChemFoam solvers with convolutional–transformer architectures to forecast two-phase CO₂ flow through heterogeneous sandstones in under a second. The prototype earned an honourable mention from judges and will now be incubated in the ECO-AI project’s industry extension.
Building skills for the Net-Zero workforce
Beyond code and competition, the hackathon proved a hands-on training ground. Mentors ran clinics on reproducible workflows, responsible AI and open-source licensing, while mixed teams of geoscientists, data engineers and policy analysts swapped techniques in real time. DigiPorFlow participants highlighted the experience as “three days of accelerated learning,” crediting the event with sharpening their software-engineering habits, broadening their AI toolbox and forging cross-disciplinary networks they can draw on long after the laptops close.
