COMPANY CONFIDENTIAL. CAN BE SHARED WITHIN JMP AND ROLLS ROYCE.
The Rolls-Royce of factories
A company long synonymous with quality takes cyber-physical systems integration to the next level with inbuilt quality and reliability analytics
_____________________________________________________________________________________
Challenge: Reduce costs and increase yield across Rolls-Royce manufacturing facilities around the world by adopting an international standard for improvement. Remove single-point failures from the supply chain and generate intellectual property.
Solution: After learning that Rolls-Royce’s Engine Health Monitoring group had used JMP® to develop an improvement that saved the company £1 million a year, Chief of Manufacturing Engineering Systems Dan Middleton made the decision to license the software in manufacturing. A direct comparison of available tools – in the form of an internal hack-a-thon – was proof of concept: JMP could do in 8 hours what took engineers using Python 3 days and engineers using Minitab 6 days to complete.
Results: Rolls-Royce’s capacity for making more data-driven decisions faster with JMP has resulted in widespread value-added services and cost efficiencies. In just 9 months, a single engineer used JMP to develop solutions for 15 unique manufacturing challenges at a variety of Rolls-Royce factories in different geographies. The company is now working aggressively to scale up these improvements.
_____________________________________________________________________________________
Industry 4.0 offers a transformative vision of the future of manufacturing: cyber-physical systems integration will generate continuous, free-flowing data streams. Cloud computing will offer new ways to access and analyze that data. And machine learning will put data to more creative, accurate and efficient uses. A very appealing proposition, no doubt. But while most companies see smart manufacturing as an exciting possibility, few have successfully put theory into practice.
Few, that is, except Rolls-Royce.
In 2018, Rolls-Royce invested over £60 million in one manufacturing facility alone; £7 million of which was earmarked explicitly for IT integration. It was a major investment that the company’s Chief of Manufacturing Engineering Systems Dan Middleton calls a ‘no-brainer’ because, he says, “the business case is that [integration] pays for itself.”
Rolls-Royce’s capacity for making more data-driven decisions faster with JMP has resulted in widespread value-added services and cost efficiencies.
That estimation isn’t far off. Rolls-Royce is already seeing the returns on its investment, Middleton says. In fact, the renovated site is now achieving between 35 and 45 percent more output. A portion of that benefit can be attributed to upgraded machinery of course, but there’s no doubt analytics integration is making an outsize impact. And having the right software, Middleton says, is part of the reason that ROI has been so fast in coming.
Learning from another division’s £1 million-a-year success
Just as the upgraded facility was coming online in 2018, Middleton was beginning an audit of the company’s tooling landscape. He sought to evaluate Rolls-Royce’s already extensive portfolio of software to identify a one-stop-shop tool that could help engineering teams take advantage of new data streams. In the end, the solution lay just across campus at Rolls-Royce’s Engine Health Monitoring (EHM) division.
When in early 2018 the EHM announced an informal analytics seminar, Middleton and colleague Russell Bayes, a Manufacturing Systems Engineer from Rolls-Royce’s Advanced Blade Casting Facility (ABCF) in Derby, UK, decided to attend. What they saw was exciting: an EHM team had used a statistical software called JMP® to develop a new monitoring method that was likely to save the company more than £1 million a year.
It was the first anyone in manufacturing at Rolls-Royce had given much thought to JMP, Middleton explains. “We were blown away with it actually – the capabilities of the tool. We thought okay, let’s play around with JMP and see what it can really do,” he says. “We were a bit skeptical at first. Within around 4 months, though, we’d established that there was something about this tool. We weren’t sure what it was just yet, but it was allowing us to do some very clever maths.”
Middleton brought on a new engineer, Aman Datta, to focus exclusively on learning JMP and determining what role it could play in Rolls-Royce’s manufacturing operations. “That’s the importance we see in [JMP],” says Middleton, “that we’ve effectively freed up an engineer to go play on a piece of software for a year.” New tool in hand, Datta, Middleton and Bayes worked together to triage a number of issues that had plagued the ABCF for years. It was a test case for JMP, and they were not disappointed.
Executive leadership takes notice: “How can we do more of this?”
Like Rolls-Royce’s other smart factory sites, the ABCF can create a detailed data map of the life cycle of every single engine component passing through the facility’s value streams. With a sufficiently robust software, engineers can then use these data signatures to construct dynamic statistical models that help to identify which components are most likely to fail or nonconform. This information in turn equips them to make strategic decisions about which parts should advance to the next step and which should be scrapped or deprioritized.
Eliminating out-of-specification parts earlier in the production cycle enables manufacturing engineers like Middleton and his colleagues to reduce the time and resources they invest in non-quality components that require rework. Cumulatively, these savings can have an enormous impact on bottom-line costs.
In their initial trials with JMP, Middleton explains, the team solved four key problems that yielded significant benefits. “We [achieved] so much that our executive vice president asked, ‘How can we do more of this?’ So we purchased 41 more JMP licenses – we went from one [copy of JMP] to 41 in the space of 9 months.” The group also now has 30 manufacturing engineers using JMP Pro.
Since realizing the scalability of these improvements, Rolls-Royce has repositioned Middleton and Datta so as to utilize their expertise at its facilities around the world. Datta spends time on the factory floor to understand any site-specific problems his colleagues face – and then to use JMP to find a solution. The proposition, Middleton explains, is this: “I’ve got this wicked piece of software, a super computer and a cloud network I can link to. Let me have a go at [your problem] for you.”
In just 9 months, Datta developed solutions for 15 unique manufacturing challenges at Rolls-Royce factories in different locations around the world. And, says Middleton, every new challenge is a data point in the bigger picture of Rolls-Royce’s operations.
The problems they’re seeing, he adds, are more often similar than they are different. “This is part of the cultural shift: rolling out benefits from one facility to the other.” And as more Rolls-Royce facilities come online with digitally-integrated analytics systems, it will become easier to be more proactive – and more predictive – about manufacturing engineering; “You have to change the mindset and say, let’s not waste our time or resources.”
Embracing the fail-fast philosophy, scaling up with cloud computing
One of the ways that culture shift has manifested is in an embrace of the fail-fast philosophy. It may seem strange that an organization as highly regarded as Rolls-Royce views failure as acceptable – even encouraged. But the company has learned that it’s better to abandon projects early when they lack promise. (Middleton says for example, that any exploratory analysis with an R value less than 0.5 might be shelved as the team moves on to investigate more promising trends).
To help prioritize, Rolls-Royce operates weekly governance sessions. Engineers who are not seeing promise in their projects will take them off the agenda and move on. “It’s a very mature way of doing things,” Middleton says. Previously, engineers may have become very protective of a project, but the fail-fast ethos empowers them to discard improvement ideas unlikely to yield results. And that means increased capacity and more improvement projects.
“The problem [with smart factory systems] is that we’re finding things too fast now,” says Middleton. “And this is where we now use cloud technology. Human beings can only work for so long. But machines – we can set them to task creating models.
“JMP speeds all of that up. Effectively, we’ve gone from analyzing data down over the course of a week to doing it in two days.” Upload reams of data to the cloud and run scores of models simultaneously. See something interesting, and tweak or rerun any analysis in a short span of time.
“To some extent, we no longer have to do it ourselves,” Middleton explains. “We create the model ourselves, but modeling has become so much easier. You can leave it running over the weekend and come back in on Monday and it’s ingested all this data and gives you an answer or a new direction. That direction might be to go and do a different investigation or do a DOE (design of experiments).”
Cloud-based technology, coupled with modeling in JMP, promotes problem-solving and support; engineers can access data in way that, Middleton says, “makes Rolls-Royce flow seamlessly as an organization.”
“Our entire culture changed overnight”
Industry 4.0 transformations don’t usually flow so quickly. But at Rolls-Royce, the growing body of proof motivated a sea change in statistical thinking across all echelons of the organization, Middleton explains. One of the most direct examples came from a company hack-a-thon in 2018 where teams from different divisions across Rolls-Royce competed to solve a previously unsolvable manufacturing challenge. And in so doing, they ended up with a direct comparison of the software options available.
The case was quite complex and involved an array of processes and interactions. One team used Minitab, one used Python, and another R. Middleton’s team used JMP. “We decided to tackle this is a different way because it was sensitive from an output perspective, so our team used JMP,” he says. “And within 8 hours we fixed the problem.”
By the time Middleton’s team had arrived at a solution, he says, “The people using Python had only just started to create a model. The people using Minitab still hadn’t even structured the data yet.”
In the end, the teams using open-source software arrived at a solution in 3 days. The team using Minitab took 6 days. And the team using JMP only needed 8 hours. It was a telling result and Rolls-Royce’s leadership took notice, says Middleton; “our entire culture changed overnight.”
Rolls-Royce is now scaling up the hack-a-thon model so that teams of subject matter experts and data scientists around the world can work together to solve some of the company’s most challenging problems – and come up with big ideas. “[The hack-a-thon model] allows us to better understand our engineering challenges and push the envelope of physics even further,” Middleton says, adding that a global approach will help Rolls-Royce to adopt an international standard and remove single-point failures from the supply chain.
Creating intellectual property
Beyond standardizing operations, the company has come to a critical realization: by using analytics so effectively, Rolls-Royce is actually creating intellectual property (IP). That IP can then be leveraged to generate new funding streams on top of traditional continuous improvement, Middleton contends. “We never even imagined that. When we started, we just thought about using analytics to solve the problems of today, to make our products and processes better and give our customers a better performing engine.”
Take, for example, the case presented in the hack-a-thon. “Through that one problem, we effectively generated two pieces of IP. One [is highly protected] and will guide how Rolls-Royce sets its factories up. The other we will share with our suppliers and customers. And supply chain,” Middleton explains.
“JMP has allowed us to ask different questions and challenge the thinking of our subject matter experts…. For example, we've been trying to solve a defect on [a] high-angle boundary for about 20 years now. And what we showed [with JMP] is that there was something fundamentally wrong with the design of the equipment. The subject matter experts had never thought about it before. It’s only because you can bring so much data to them, model it and bring a modeling tool with you.
“That's now allowed us to go back to the equipment provider and tell them there is something fundamentally wrong with their design. Of course we didn’t tell them exactly what it is we found. Because we've effectively developed IP.”
In the one year since the company’s newly integrated facilities have come online, Middleton reports, everyone involved is now making smarter decisions and learning more from the experience. Those decisions have resulted in both value-added services and cost efficiencies, he adds. “If we can run an engine hotter, then we’ll get a more efficient engine. If we can make components faster, then we can supply our market. If we make engines cheaper, then we will be more competitive.”
Ultimately, these improvements not only benefit Rolls-Royce and its supply chain, but also its customers. And that’s the ultimate goal, Middleton explains; “Rolls-Royce’s customers are at the heart of everything we do. And that’s why we have leveraged technology: to give our customers a better product.”
