Royal Dutch Shell is heavily investing in research and development of artificial intelligence (AI), which it hopes will provide solutions to some of its most pressing challenges.
From meeting the demands of a transitioning energy market, urgently in need of cleaner and more efficient power, to improving safety on the forecourts of its service stations, AI is at the top of the agenda. I have been working with Shell over the past months to help create a data strategy, which gave me a thorough insight into Shell’s AI priorities and initiatives.
Current initiatives include deploying reinforcement learning in its exploration and drilling programme, to reduce the cost of extracting the gas that still drives a significant proportion of its revenues.
Elsewhere across its global business, Shell is rolling out AI at its public electric car charging stations, to manage the shifting demand for power throughout a day. It has also installed computer vision-enabled cameras at service stations, which are capable of detecting customers lighting cigarettes – a severe hazard.
During the data strategy development, I worked with Daniel Jeavons, Shell’s general manager for data science. Jeavons talked to me about Shell’s AI-first strategy and said “What it means in practise is that we as a data science team are in a great position because we can make our current business more effective, more efficient, more reliable, safer – by applying AI into those settings.
“But we can also play a role in creating some of the new business models that we want to create, and that’s really exciting because we’re playing our part in taking Shell into the next generation of energy sources, new fuels, and new sources of revenue.”
Shell is involved in the entire oil and gas supply chain – from mining raw hydrocarbons from the earth to refining them into fuel and various other products, to retailing them to businesses and individuals. AI is being rolled out or trialled at each step of this process. Recent developments include the adoption of reinforcement learning – a form of “semi-supervised” machine learning, to control its drilling equipment.
While machine learning can work with either labelled data (supervised learning) or unlabelled data (unsupervised learning), reinforcement learning takes a middle-ground approach by incorporating a reward system, dependent on the outcome of the AI’s “choices.”
As Jeavons says, “The key thing is you’re giving the [AI] agent the autonomy to make the decision. But you’re providing input into the model, so you’re providing reward or penalty functions on the basis of what’s happening in the model, and how the model responds to the set of conditions that you give it.”
Algorithms designed to guide the drills as they move through a subsurface are trained on historical data from Shell’s drilling records, as well as information gathered from simulated exploration. It covers mechanical information from the drill bit, such as temperature and pressures, as well as data on the subsurface from seismic surveys.
The result is that a Shell geosteerer – the human operator of the drilling machine – is able to understand the environment more accurately they are operating in, leading to faster results and less wear, tear and damage to machinery.
In many ways the challenge was similar to those faced by developers working on self-driving cars – only instead of navigating hazards a vehicle might encounter on the road, the drilling machinery must autonomously adapt to changing conditions under the ground.
Jeavons says “We talk a lot about augmented intelligence, and the reason is that this isn’t about removing people from the operation … what we’re trying to do is help the people who make the decisions to make those decisions with additional support from the intelligence that we’ve created.
“What we expect is that this will probably never fully replace geosteering as a discipline, but it will allow a single geosteerer to support many more wells.”
Encouraging motorists to switch to an electric vehicle is seen as key to reducing the Co2 emissions caused by humanity, and limiting their effect on climate change. But it involves something of a chicken-and-egg problem. Motorists are put off making the switch due to a lack of public charging terminals, and forecourt operators may be slow to adopt them due to a lack of demand.
Shell’s answer to this problem involves deploying AI to monitor and predict the demand for terminals throughout the day, enabling power to be supplied more efficiently.
“If you think about it, ” says Jeavons, “as a grid operator you’re operating many, many electric charging posts … if all the cars plug in at the same time and automatically start charging, you create a big load on the grid t, by the way, can’t be filled by solar, because it’s 7 am or 8 am in the morning.”
“So, what we can do by understanding people’s charge profiles is we can spread the load during the day, which basically means we can save the consumer money.
“But also, more renewables are used – because if you can charge more people at lunchtime, there’s going to be more solar on the grid at that point.
“It’s an example of where we see the role of artificial intelligence playing a key part – thinking about not just how we can make things more efficient, but also how we can change energy consumption patterns to take more advantage of renewable sources.”
The programme, known as RechargePlus, is currently being rolled out in California.
Another initiative being trialled in Singapore and Thailand involves the use of computer vision at service station forecourts. Computer vision – cameras which can “think” and understand what they are filming – are trained to watch out for the potential hazard of customers lighting cigarettes in the vicinity of pumps and refuelling vehicles.
Camera data is processed by what is essentially the same technology powering Google’s reverse image search, which allows the content of the picture to be labelled and categorised.
When an image is detected that matches what the algorithms “know” (through training) is a person lighting a cigarette, alerts can be issued allowing the forecourt staff to close down nearby pumps and reduce the risk of fires or explosions.
This relies on “edge processing, ” with camera data being analysed locally to avoid the delay that would be inevitably caused by sending it to the cloud and back before action could be taken. While it currently focuses on spotting smokers, in the future the technology could also be trained to detect other hazards such as reckless driving, criminal damage or theft.
Shell has certainly progressed a long way on the path towards becoming a truly AI-first organisation, and the key to this has been identifying use cases where AI can drive real and immediate value. However, it is likely to face even more significant challenges in the future, if it is to meet its responsibilities around energy transition.
Jeavons says “Artificial intelligence has a huge role to play in energy transformation – we are trying to paint a picture of a way forward for a society that would meet the Paris targets … I think what’s important about that, is that oil and gas are going to be a part of that future because it’s very hard to move away from it altogether. AI is essential if we’re going to make our existing carbon sources of energy more efficient – optimization is going to be massive.
“But there’s also a whole bunch of other energy sources in there as well – and many of the emerging technologies are going to need AI in order to be effective – smart charging is just one.”