Nvidia Corporation, a giant tech company that’s a leading supplier of AI software, displayed its new approach to carbon capture and storage (CCS) that scientists can utilize to accelerate carbon storage.
CCS is a way to alleviate climate change by redirecting carbon deep underground. During the process, scientists must prevent fracturing geological formations where carbon is injected, leaking CO2 into aquifers or back into the atmosphere.
That can happen if there’s too much pressure buildup due to the process of injecting carbon into rock formations. This is what Nvidia’s Ai-powered technology addresses to help improve carbon storage.
There are more than a hundred CCS facilities are under construction around the world. Traditional simulators for carbon storage are costly and require a lot of time to complete. Machine learning and AI models deliver the same level of accuracy but lower cost and time.
Nvidia introduces its AI approach to carbon storage that CCS scientists can readily use in real-world applications through Nvidia Modulus and Nvidia Omniverse. It’s AI-powered technology accelerates CCS modeling 700,000X using Fourier Neural Operators (FNO) architecture.
FNO architecture offers more accurate predictions of pressure buildup and CO2 saturation. It’s 2X as accurate while needing only a third of the training data compared to other computer models.
The software helps engineers to choose the best injection sites fast, identify the optimal spacing and depth of wells, determine the best injection pressure and rate for the captured carbon and make sure that rock formations don’t get fractured. Also, engineers can visualize and optimize the entire inspection process through Nvidia Omniverse.
A strong assessment for CO2 plume and pressure buildup usually takes about 2 years using numerical simulators. But with Nvidia’s FNO, it may only need 2.8 seconds. It allows scientists to simulate how pressure levels build up and where CO2 spreads throughout the 30 years of injection.