Transforming an Oil and Gas Giant with AI
Forbes recently published an article highlighting the ways Royal Dutch Shell (Shell Oil) is leveraging AI, "The Incredible Ways Shell Uses Artificial Intelligence To Help Transform The Oil And Gas Giant". It is an interesting read about how a company in oil and gas is taking advantage of cutting edge technologies.
Lets take a look at what they are developing and protecting. Delving into the patent filings will give us some additional insight into this story.
The interesting thing about the AI field is its relative agnostic application across industries. When analyzing the players developing AI technology and filing patent applications, we see players from industries ranging from electronic devices to farming equipment to fintech to media to automotive to oil and gas and beyond. Shell is in the top 50 global patent filers in the AI space, with over 320 applications. This puts them in the company of organizations like SAP, Nuance and Huawei. In the oil and gas industry, Shell is leading the pack. Competitor oil companies are working in the field, however:
Exxon-Mobil and BP each have holdings a little over half the size of Shell.
Saudi Aramco, Chevron and Italian firm ENI come in with just under a third of what Shell owns.
On the lower end, we see French Total with a dozen or so filings and Chinese firms Sinopec and China National Petroleum with just a handful between the pair.
If you bring services companies into the mix, Halliburton is about on par with Shell and Schlumberger falls into the ranks of the Saudi Aramco, Chevron and ENI.
The vast majority of Shell's IP come out of their Netherlands-based Shell Internationale Research company. Many of the applications focus on prediction of oil properties and enhanced downhole operations.
For an interesting patent read, take a look at this application filed in 2016 on technology which Shell developed in collaboration with MIT:
"Structural health monitoring (SHM) instruments a structure with sensors to collect data and derives some information from the collected data to determine whether the structure has changed. A detected change in the structure can then be attributed to damage that can be more closely investigated. In general, collected data is processed into features that may indicate these changes in the structure, and, in some cases, statistical or probabilistic discrimination of these features can separate data collected from intact and changed structures."