Enterprise AI

Title: The Digital Transformation of Shell

Duration : {15:11} minutes

{Opening and introductions}

Helen Forde

I do remember my first time stepping on a refinery going, wow, this is huge.

Dan Jeavons

The dimensions of it are just mind blowing.

Helen Forde

It has thousands of miles of pipeline.

Sander Buhling

Up to 10,000 valves and a million instruments.

Dan Jeavons

All of those things have sensors embedded in them. These vast data stores create a huge opportunity for digital transformation.

Ulrika Wising

And that can really lead to some significant new solutions.

Dan Jeavons

Digital transformation is really about two things. It's about recognizing that the amount of data we have around our existing business gives us a big opportunity to transform the way in which we operate. And digital transformation is also about finding ways in which we can operate in the world that's coming.

Dan Brennan

Companies in the energy industry today are going through a massive transition. If these companies can't manage the energy transition, which is really about the world demanding more energy, but more energy from sustainable sources, they're gonna have challenges operating both from a social license to operate, but also from just the basic economics of doing business.

Dan Jeavons

Shell has a really rich history in digital technologies and data science in particular. We developed a statistics group in the 1970s. And so when I started working in this space in 2013, I was building on a history that we already had. We saw that things were changing. And in particular, what was changing was the ability to take large datasets, to store them in a cloud based environment, and to use cloud based computation to accelerate the development of new software.

Alisa Choong

In 2015, I was really blessed to be given a chance to lead the digital transformation team.

Dan Jeavons

Alisa was a phenomenal visionary. She understood where the technology was going. And she understood the transformation that that could lead to for the company.

Alisa Choong

I think I've been working with Dan for about 10 years. Yeah. When he came in, he was this young punk with his hair standing up. Even though today, he still look like a punk. Yeah. What I really treasure about Dan is his passion, his passion in using technology to solve problems.

Arnold Hes

In 2014, we had a big incident where we were lucky that no people got injured. A control valve failed.

Sander Buhling

The plug came loose from the stem, and it starts oscillating in the process. And at one point, it actually broke the body of the valve.

Arnold Hes

We had to repair the plant, it cost us three weeks of unplanned downtime.

Sander Buhling

We started thinking about how we could use the data that we have already collected to prevent future failures.

Arnold Hes

We contacted the Shell data science department, they had much more experience in this.

Dan Jeavons

What was fortuitous about the work with Pernis was that it came at a time where we'd been working in this space for quite some time in Shearwater in the UK. And what we'd been looking at was the opportunity to effectively take all of the telemetry data that we had, and to use that to train data driven machine learning models to detect anomalies that humans otherwise wouldn't detect. And so we very quickly saw the opportunity to leverage this approach in Pernis.

Arnold Hes

We have years and years of data. We started with 15 valves.

Dan Jeavons

We're looking at things like the temperature of that valve, the flow rate of that valve.

Arnold Hes

Pressures - once every minute or once every second.

Dan Jeavons

And we're using that to forecast, if you will, the normal behavior of that valve, and then understand where the behavior of that valve deviates from normal conditions. And when that happens significantly enough to generate an alert.

Arnold Hes

The challenge was how to automate it without a lot of human effort. As I said, 20,000 valves in Pernis, but we have, like 100,000 valves in Shell.

Dan Jeavons

How do you scale that, to run that globally? We needed a machine learning operation solution, we also needed a solution that would scale on the vast quantities of data that's generated from these assets. We were looking to say, well do we build this ourselves? And I think what we realized quite quickly, is we don’t want to be a software platform company. Our focus is around our understanding of the problems that our industry faces.

Alisa Choong

We spotted a company that is called C3. They were at the cutting edge of developing predictive maintenance modeling, a lot of algorithms that we saw was appropriate for us.

Tom Siebel

We were approached by Shell some years ago. And we demonstrated that we can identify device failure with very, very high levels of precision. This is being accomplished through the application of an entirely new generation of technology called predictive analytics or AI.

Adi Bhashyam

Shell was interested in what C3.ai was building, and whether it would be relevant to Shell operations. The challenges they were running into were really around scaling and deploying their software and their machine learning algorithms to the entire footprint of Shell. While the mechanics of the valve were simple, the context of the valve required that each valve had to be treated as its own independent piece of equipment. Which meant that you really needed at least one independent machine learning model per valve.

Tom Siebel

Shell has an unusually gifted team of data scientists. They are professionals, they understand the industry cold. The C3.ai platform enables the Shell data scientists using the tools they've always used before to immediately put these applications to work at full enterprise, Shell scale, which is as large an enterprise as there is in the world.

Dan Jeavons

What C3 demonstrated was that their platform was capable of managing 2 million of these models. That was what excited us because it gave us confidence that they understood the problem that we had, and that their platform was scalable enough to be able to deal with it as we put these models into production.

Adi Bhashyam

The types of these successes tend to be somewhat under the radar because they're all avoided failures. Right, so nothing happened. And that's a good thing.

Arnold Hes

We have repaired 83 valves basically, that we would not have found without the analytics. And we have millions of savings on unplanned downtime.

Dan Jeavons

We've now managed to deploy multiple applications into production at scale, in really critical areas. We've developed together mechanisms to monitor over 8,000 pieces of equipment every day.

Dan Brennan

We've been able to expand the scope from just valves. There's literally hundreds of use cases.

Alisa Choong

We also now collect quite a lot of data. We collect data using robotics.

Arnold Hes

They can monitor the position of hand valves, they can detect leaks, gas leaks, but also liquid leaks. They are really helping the operators doing their daily routines.

Alisa Choong

We then looked at how we can turn this data into patterns.

Dan Jeavons

A lot of the opportunities are in optimization.

Alisa Choong

Optimization is important. We again look at the data, we can see where is best place to put the warehouse, and how's the best way to get it up to our offshore location.

Helen Forde

There will be hydrocarbons for certain industries, so the question is how do you produce it in a way that is responsible, how do you produce it in a way that gets net zero emissions, how do you do it in a way that helps support the planet.

Dan Jeavons

We've also developed technology to optimize performance of things like liquefied natural gas trains. A recent algorithm we deployed to one of these trains in Nigeria resulted in the equivalent of taking about 28,000 vehicles off the road. It's a great example of how AI is starting to have a big impact.

Ulrika Wising

Shell does a lot to try to understand our customers. And that's where AI comes in as well, and they can use multiple sources of data to actually give a much, much more holistic picture of what our customer needs.

Dan Jeavons

When you look at the energy system, everybody has a piece of the puzzle. The original equipment manufacturer has a piece of the puzzle. Baker Hughes stepped in with their deep understanding of how the equipment operates. The operator also has a piece of the puzzle because they understand how that equipment operates in context. Of course when you talk about aggregating large volumes of data you need to do that in a cloud based environment. Microsoft stepped in there to help us to accelerate the speed of development of our solutions collectively.

Alisa Choong

We collaborate with C3, Baker Hughes, as well as Microsoft, to really look at how we can accelerate our AI journey.

Dan Brennan

Shell is a long standing customer of Baker Hughes as well. So naturally, as our relationship, Baker Hughes relationship with C3 was formed, Shell became a natural party for us to collaborate on some, some new innovations on a C3 platform.

Adi Bhashyam

We have at C3.ai a deep and committed relationship with Baker Hughes. Baker Hughes has been our partner and customer for the last several years in our bid to take our solution and our capabilities to the energy field altogether.

Dan Brennan

The open AI Energy Initiative is a really exciting partnership between four founding members: Baker Hughes, C3.ai, Shell, and Microsoft. And really what it aims to do is to start to create a set of industry standards and references around how we bring AI technology to the industry.

Adi Bhashyam

The whole point of this effort was to take the learnings and innovation that we have been doing so far with Shell and bring the entire industry along.

Dan Jeavons

If we wanted to be really successful transforming the energy industry, we had to work with a broader ecosystem of partners. What the Open AI Energy Initiative tries to do, is it tries to say let’s work together on a common digital infrastructure from which we can all benefit through fair value exchange, with a common set of open data standards underpinning it, which can accelerate the digital transformation of an asset or a site in the energy sector.

Tom Siebel

This was really driven by the visionaries at Shell that are by far the leaders in applying artificial intelligence to deliver cleaner, safer, more reliable energy with less environmental impact.

Ulrika Wising

Nobody can go at this alone. It's a societal challenge that we're faced with. And with that we need to collaborate. It takes all of us to, to drive this decarbonization agenda.

Helen Forde

I think this is an increasingly important part of how we look at digital is how we help it enable our greenhouse gas ambitions.

Tom Siebel

Let's think about what Shell is doing. These guys have set the goal to reinvent themselves as a zero net carbon footprint company by 2050. This is ambitious.

Ulrika Wising

We’ve made a commitment to net zero by 2050, we have an aim of reaching 50 million customers by 2030. Which is very ambitious, but the demand for green electricity is just growing across the globe. We couldn’t do this without digital solutions.

Dan Jeavons

If you look at, for example, electric vehicles, we can look at ways in which we charge those vehicles to make sure that we maximize the amount of renewables that are on the grid at any given time. We have the opportunity to optimize the grid behavior using AI, and to balance all of those new loads that are coming onto the grid.

Ulrika Wising

If you take the example of the home, where solar panels are generating solar electricity during the day, but we're storing them in a battery and using them when people are home at night, that's a digital solution. So you wouldn't be able to do that without the right software solutions.

Dan Jeavons

At the end of the day, energy transition is urgent, we're going to have to transform the energy system faster than we've ever thought possible. And digital is one key lever to do that.

Tom Siebel

Companies that do not digitally transform themselves with this new generation of technologies will cease to be competitive.

Dan Brennan

My hope is in ten years we have an energy industry that’s producing, certainly an abundance of clean, safe, efficient energy and is able to distribute it to parts of the world that frankly don’t have access to it today.

Alisa Choong

I am very optimistic about the future. I believe technology is there to help humans solve many problems. And Shell is here to produce energy, cleaner energy for the people and for the planet.