Rolls-Royce’s Caroline Gorski believes AI analytics and open collaboration will be essential to the future of manufacturing innovation

 

“Being able to understand and work with data in a transformational way is one of the key factors that will drive success in our industry.”

 Caroline Gorski, Group Director,
R2 Data Labs, Rolls-Royce

Launched a year ago this month, Rolls-Royce’s ambitious new R2 (pronounced, R-squared) Data Labs division is tasked with harnessing the company’s vast data assets by using advanced analytics, artificial intelligence, and new machine learning technologies to unlock design, manufacturing, and operational efficiencies within its factories, and to help create new data-driven services for its customers around the world.

Established over a hundred years ago, Rolls-Royce is now one of the world’s leading aerospace and power systems companies with a global footprint including multi-billion dollar facilities and over 6,000 employees in the U.S. The company first pioneered the concept of selling aircraft engine power as a service, rather than as a traditional physical product, over 50 years ago, and today, according to some estimates, generates over 70 trillion data points every year from its commercial airline engines.
In our latest Dialogue with a manufacturing industry thought leader, Caroline Gorski, Group Director at Rolls-Royce’s R2 Data Labs division, talks to Manufacturing Leadership Executive Editor Paul Tate about the need for more agile and collaborative approaches to digital innovation in manufacturing, the importance of breaking down data siloes and applying new AI technologies to release operational value, and why an inquisitive mind may be one of the most important qualities that future manufacturing leaders need to possess.

Q: What was Rolls-Royce’s motivation in forming its new R2 Data Labs division?
A:
Rolls-Royce has been tracking IoT data from within its civil aerospace engines for over 30 years. That activity enabled the fundamental shift that allowed us to move to the “Power-by-the-Hour” business model some years ago. By offering power as a service, where we focus on the ability to power a plane rather than charging for the engine up front, we needed to understand how that engine was operating and performing, so we needed to have the right data to be able to meet the service level agreements with our customers.

Over the last two or three years we recognised that we could develop that data-driven approach in a number of different ways. For example, we can make that kind of data capability much more broadly available to other business units in the company, like land-based power generation or marine power systems. There’s also growing demand from our customers for data that not just supports power as a service, but also helps them to run their own businesses more effectively. That was another real trigger for the company as it could see enormous potential sitting its data innovation capacity.

Today, we believe that being able to understand and work with data in a transformational way is one of the key factors that will drive success in our industry. It’s become a fundamental part of our group strategy. It’s about perfecting the digital twin, about getting the most value possible from data innovation, and about becoming a truly digital-first company. Those three strands of our digital strategy are really critical to what we see driving success both in our own space, and in the broader industrial context.

Q: What’s the goal of the new R2 Data Labs?
A:
Essentially, R² is about extracting value from data innovation. Our responsibility is to look at how we can use the sizeable data assets the we have at Rolls-Royce, that we hold in trust for our customers, and that we can draw from contextual data sets in the outside world. We’re focused on how we can take those deep data sets and apply advanced data innovation techniques to them in order to generate value – both for our own business, in terms of operational cost savings and efficiency gains, and for our customers, in terms of how we support new services that help them operate their businesses and make better decisions. That’s really what R² is about.

Q: What excites you most about your role at R2?
A:
I feel enormously privileged to be working inside an organization that has the sort of intellectual and innovation firepower that Rolls-Royce has – in its engineering teams, in its manufacturing teams, and in its service and operations teams. When you combine that with the best of the external marketplace, with outside creativity and disruptive thinking, with the ability to challenge assumed ways of doing things, and an opportunity to really break down barriers and share ideas – across vertical domain specialisms, across geographies, and across different and adjacent industrial players – those are all incredibly exciting things. It’s an amazing agenda to be working on.

“We’re focused on how we can take deep data sets and apply advanced data innovation techniques to them in order to generate value.”

Q: What makes R2 different to traditional innovation approaches in the industry?
A:
Traditional engineering mindsets are often focused on long-term innovation cycles that are quite closed and strongly defensive around intellectual property. This can lead to the development of very deep internal expertise that then becomes a barrier to entry for competitors. But that model doesn’t really work in the digital world. The digital world has much faster innovation and development cycles. Much more of what is done, is done in a collaborative or partnership way.

Indeed, if you think about the nature of data itself, data wants to be horizontal. Data wants to be shared. Data wants to be augmenting other data in order to create the sorts of interesting patterns, insights, and golden nuggets that can really release value.

Taking that on board, we recognized that we needed to think actively about how we can build a more collaborative, innovative ecosystem with both internal and external partners – we have over 100 external partners now – who can really work together to unlock new opportunities in the digital universe. That partner ecosystem is now fundamental to the way that we think about delivering innovation at Rolls-Royce and delivering digital value to our customers.

Q: How does this level of internal and external collaboration work in practice?
A:
We work in what we call Innovation Cells – groups of individuals from different backgrounds and domains focused on a priority challenge or specific business opportunity. It may include a domain expert or a customer stakeholder, and a technical project manager who effectively runs that cell and makes sure it’s doing what it should be doing. It may also include a design thinking expert who can help to articulate the problem that they’re seeking to address. It would likely include a data architect who can help to structure the data so that it can be worked on.  It also could include a number of data scientists, like a data engineer, whose role might be to connect the data to other data, or clean up the data that can’t be used because it’s not been stored in the right way. And it may involve a service representative, someone who can help us to productionize the product and make sure that it actually is deployable in the business.

The participants in the cell will change over the course of a sprint, because different skills are needed at different stages in the sprint.

But we also use other approaches. For example, earlier this year we held a “Manufacturing Hackathon” in partnership with one of our manufacturing plants that produces blades for our jet engines. We put a real set of data around one of our machining processes out into the field for AI innovators and new businesses to look at, to work on how they could improve the outputs of that process, and how they could help us to look at the process to be able to make adaptations to it in a much more automated, predictive, and adaptive fashion. It was very successful and produced lots of diverse ideas.

Q: What challenges still keep you awake at night?
A:
One of the biggest challenges is simply the state that data is in today. It would be disingenuous of me to suggest that it’s not hard to wrangle all of the sources of data that we try to work with. It’s hard because they’re in existing enterprise siloes and legacy systems, and in many instances, the data isn’t coded, or managed, or architected in a way that makes it easy to work with.

Plus, we’re doing new things with it. We’re applying artificial intelligence. We’re doing deep learning and machine intelligence with those data sets. So we’re really challenging the way that we think about how we can innovate with data. In some instances, that means that we are taking databases that may have been put together decades ago, and using them to meet new opportunities and new challenges. That piece of the puzzle is difficult.

Q: How do you overcome these kinds of legacy data issues?
A:
We don’t try to boil the ocean. We don’t try to solve the problem across the entire piece in one go. We attack it by taking a very agile, atypical digital development model. We work on 90-day sprints and a significant chunk of that will be about calling all of the data together. We then prototype ideas and iterate around those prototypes until they get to a state where they start to demonstrate their potential value to our internal and external customers. It’s very different from a traditional waterfall development process. We’re working to create minimum viable products that we can then launch into the business. If we didn’t do it that way, it would end up being an enormous challenge to try to solve that entire problem in one go.

Q: How important are advanced technologies like Artificial Intelligence to this process?
A:
We use a series of technologies that are all linked to how the digital world is evolving the way we use data. One of the biggest recent technology shifts is between traditional big data analytics and new artificial intelligence data analytics technologies that provide the ability to predict. Big data analytics, by its very nature, is retrospective. It looks for patterns in big collections of data, but it stops at the point that its reports end. It can tell you what has happened, but it can’t tell you what’s going to happen. That’s the biggest difference between big data analytics and AI data analytics.

Those AI techniques are the underlying capability that fit into our smart factory work with predictive maintenance. For example, giving indications and advice when parts of our fleet might need to be called off into our maintenance or repair shops. And we’re increasingly using that AI capability to look at examples where parts of our product may not be performing as we would expect and we challenge ourselves to understand, using AI, how we can predict that, how we can take action to prevent that reoccurring, or how we can help our customers to take action to mitigate any associated risks.  

Q: What other key technologies excite you for the future?
A:
One is the difference in the way the data is processed and handled. One of the big challenges with a big data analytics position is that, to all intents and purposes, you end up with an enormous silo of data expensively sitting on databases and servers in your organization. Then you have to keep that fresh and maintain all the services that you need to actually keep that data working and keep access to it possible. We’re really interested in the notions of moving toward distributed data management models. That might mean, in the first instance, thinking about how you securely work with data in the cloud. That’s clearly an issue for industrial players like us who work in highly-secured environments, but it’s fundamental to the possibility of making access to data available across our business in a much more democratized way, so that we can turn everybody in our business into a data scientist, not just a small pool of data scientists in the middle.

In the longer term, we’re also looking at how you think about technologies like blockchain, which allow for the distributed processing and analysis of data without there necessarily being a central authority. We’re looking at how you distribute trust tokens so you can enable your supply chain to get involved much more intimately in working with the same data that we’re working with.

We are also looking at the next step on from IoT with things like edge processing and analysis of data. It’s all very well putting sensors into a host of fixed or mobile devices, as in our case with engines, in order to collect information. The next step is how you put the processing capability that looks at that data and makes decisions about it, out at the edge so that you greatly reduce the latency of any decisions that you make. If you can put your processing capability and some artificial intelligence and autonomy at the edge, suddenly you remove the latency.

Those are all really interesting technological advances that we are either piloting, or investigating, or actively deploying across the business environment.  

  “One of the biggest recent technology shifts is between traditional big data analytics and new artificial
intelligence analytics
that provide the ability to predict.”

Q: What do you see as the biggest challenges for the manufacturing industry over the next few years?
A:
There are significant challenges for manufacturing in being able to put enough investment into the kinds of instrumentation and the refreshing of manufacturing plants to generate the sorts of data needed to deliver real transformation. The world of manufacturing operation technology has a significant number of proprietary protocols and updating these can be very expensive. Given that the prime output of the manufacturing process is the product, not the data that goes around it, it’s a really difficult call to then spend a significant proportion of your operating budget on instrumenting your space to generate data that may not, at first, appear directly relevant to the output of the product itself.

The challenge is, though, that competitors who do make that investment decision are seeing, or will be seeing, significant benefits in their ability to meet changing customer demand, which is trending evermore toward shorter production runs and much greater customization and personalization. That kind of adaptive manufacturing approach is something that is fundamentally made possible by being able to collect data on manufacturing processes and manage, much more adaptively, how you go about producing your product to meet the new demands.

Plus, the world is becoming evermore connected. So as new economies come to the forefront, whether it’s the Middle East, the Far East, or Eastern Europe, those new manufacturing economies have the potential to establish green field capabilities, where data and connectivity come out of the box. So they can hit the road running with more efficient production rates than existing legacy manufacturing estates that haven’t been able to make the transition.

Q: What new skills will leaders need to succeed in this fast-changing, digitally-driven future?
A:
I think the strongest differentiating mindset is going to be around collaboration. The digital world, the world of data, is not a world that happily accommodates siloes. It’s a world where the more data you have access to, and the more adaptively you can respond to that data, the more likely you are to be successful. So you have to be more collaborative and work in partnerships across silos, whether that’s inside or outside any single corporate entity. That means working more closely with your supply chain, but it may also mean working more closely with your competitors. Those sorts of behaviors, the idea that you need to partner broadly and widely to be successful, an openness and willingness to take an outside-in approach to innovation, and understanding that the marketplace may have new ways of thinking about what you are doing, and that those ideas might come from completely adjacent markets, not from your own industry at all, but from somewhere completely left-field. Those are all skills that I think the manufacturing and industrial sector is going to need to get better at in order to continue to see the benefits from data innovation as they progress and mature.  

Q: Finally, if you had to focus on one thing as a watchword for the future of manufacturing, what would that be?
A:
It would be “inquisitiveness”. I think that’s a natural instinct for most engineers, most industrial designers, and most manufacturers. The notion of asking questions, of being inquisitive, of looking for answers in unusual and exciting places, that has to be the thing that makes the difference between an organization that can be successful at generating value from data innovation, and an organization that can’t. So for me, “inquisitiveness” is the watchword for the future of manufacturing because that’s the thing that we need to possess, and to reward, and to hold dear.  M

Roll-Royce Headquarters, London

FACT FILE: Rolls-Royce Holdings plc
– Location: 
London, UK
– Business Sector:
Aerospace, Defense, Energy, Marine
– Revenues: $21.45 billion (£16.31 billion, 2017)
– Net Income: $
5.54 billion (£4.21 billion, 2017)
– People: 50,000 Employees
– Market Presence: 150 Countries

 

EXECUTIVE PROFILE: Caroline Gorski
–  Group Director, R2 Data Labs, Rolls-Royce
– Nationality: British
– Education: Master’s degree, Modern History and English Literature, Oxford University
– Languages: English, French
– Previous Roles Include: Director, Global Ecosystem and Partnerships, R2 Data Labs, Rolls-Royce
-Head of IoT & Digital Manufacturing, Digital Catapult, UK Government Innovation Network
-Head of IoT Business Development, Telefonica
-Managing Partner, Retail & Leisure, O2 Telefonica
-Head of Strategy & Operations, SMB, O2 Telefonica
-Proprietor, Better Business Thinking
-Project Manager, B&I Restructuring, Sodexo
-Consultant, eBenchmarkers
-Analyst, Forrester Research