Technology giant Nvidia has revealed new artificial intelligence (AI) powered software to accelerate carbon capture and storage modeling. Using the Fourier Neural Operators (FNO) architecture, Nvidia Modulus, Nvidia Omniverse and AI software, the company can speed up CCS simulations by up to 700,000 times to aid engineers in selecting the best location and optimal depth and spacing of wells, in addition to the best injection rates and pressures for CO2.
By accelerating simulations, CCS decision making tasks which were previously time consuming and expensive can now be conducted more efficiently and cost effectively. A probabilistic assessment for maximum pressure buildup and CO2 plume footprint, for example, can now be completed in 2.8 seconds using Nested FNO, compared to a time period of two years when using numerical simulators.
Video link: https://youtu.be/u-M5LQvx1cQ
The above video shows the process of storing CO2 in an underground reservoir consisting of saline aquifer. Once the CO2 arrives at the surface facility, it is then distributed to four wells and pumped to an underground reservoir. Inside the reservoir, pressure buildup and the evolution of CO2 saturation is shown over 30 years.
Through the use of the Nvidia platform, engineers are able to predict, visualize and interact with their respective models to enhance the safety and reliability of proposed CCS solutions.