BAE Systems produces components for military aircraft including the Lockheed Martin F-35 and the Eurofighter Typhoon fighter aircraft. Turning toward the future, with Rolls Royce, Leonardo and MBDA, it is developing what is hoped to be the successor of the latter plane after 2035, called Tempest.
For the last few years, a team within the company has been imagining how the airframe might best be made, by drawing on emerging technologies that aim to reduce sunk costs, improve production flexibility and process integrity. The ambitious R&D project, called Factory of the Future,may end up changing not only its production, but also the wider UK aerospace supply chain.
The story starts six years ago, when the aerospace supplier developed a way to speed up the job of countersinking holes on the Lockheed Martin F-35 fighter plane body. The project found a way to automate the process by using robots. Off the shelf, their positioning accuracy is too poor for this application, so the engineers developed a way to improve their accuracy by more than an order of magnitude by bolting on extra measuring technology.
Technical lead for the project is principal technologist in emerging technologies & systems Austin Cook. He recalls: “We don’t use lots of robots, but we do use lots of machine tools. The installation was very advanced and very successful, and that changed our appetite to looking at more automation for tasks that were repetitive and very accurate.”
That same year, it began collaborating with an engineering team at the University of Nottingham that was looking to apply robotic assembly methods common in the automotive industry into aerospace production. Doing so requires grappling with the complexity and tight tolerances of aerospace parts. And in a way, that’s the least of it, adds Cook.
“In 2018, we started to work out that if we’re going to look at this type of technology for future aircraft, we need to not just look at transforming the building, but the engineering, and how the toolset is integrated, how to manage the data, and how to monitor the system accurately and quickly to get efficiencies. And the connectivity to a new type of factory network, principally the Internet of Things.”
Since then, these drivers culminated in a physical building that opened in mid-2020, delayed a mere three months by COVID-19.
The project, based in Warton, Lancashire, continues to be developed alongside some 40 supplier partners, including DXC, Siemens, TWI and Renishaw alongside tech firms such as True Position Robotics and ElectroImpact, as well as the Universities of Sheffield and Nottingham.
Demonstration projects began in November and continue until the end of the year, and then with a focus toward production readiness up until 2023, to inform the potential Tempest build programme (but in the meantime, other projects have already been exported to other areas; see box below). Although the company has built a number of other R&D facilities, including for composites and additive manufacturing, this one is completely new. Cook points out: “The white space idea is helpful in not being burdened by those things that we do today. We’ve always done that; it cultivates transformational thinking. If we start from scratch, how would we do something? How would others?”
FIXTURELESS ASSEMBLY
One of the principal projects is to do away with assembly jigs or fixtures, physical structures that hold components in place as they are joined to each other using a number of manual processes such as drilling, attaching shims, adding sealant and connecting other parts. Fixtures are expensive, complex and difficult to make; in addition, they have to be constantly checked to confirm tolerances. Although they do the job of producing the part at the required precision, they are inflexible: any part changes require their alteration. This is old-fashioned; the trend today is mass customisation and product upgrades.
Explains the technologist: “We are approaching building in more flexibility; lower-cost and quicker, less bespoke, more flexible.” Instead of using a static jig, this project imagines using a robot as a jig, and then reprogramming it to position in different places, and reconfigure itself.
The researchers already knew, through the previous collaboration with researchers at Nottingham, that it was possible to raise the robots’ accuracy by integrating additional devices, so it wanted to explore how extending the concept might provide more affordable, agile manufacturing that was ultimately product-agnostic. That means that it could be reconfigured to make different products from one day to the next, in contrast to traditional manufacturing robotic cells, which are customised for a single process.
Initially, just a single robotic cell was installed in the factory; next a second was added. Then the two started to work in sequence: the first would finish a process, and then the second would start. And in the last few months, engineers there have brought the two robots together to trial a more efficient approach: parallel working, in which one component is held in position by one robot and worked on by another.
If that were not complex enough, the next step is to put people in the mix as well, to carry out what Cook calls ‘particularly dextrous’ physical manipulations alongside and within the robots’ working envelope. That requires the development of new safety provisions, and to that end one of the engineers on the project, Jane Byford, is liaising with standards committees to extend collaboration standards.
One issue that the trials have thrown up is distortion. Many of these large metal airframe parts contain huge voids; this minimises their weight while meeting stiffness and strength requirements. Sometimes when they are unbolted from the machine bed workholding, they might twist a bit. While that deviation may be small, and ultimately be cancelled out once the piece is bolted into its final strain state, it’s still enough to surprise the robot, whose expectations of the dimensions of the unstressed part come from the digital model. The result could be an unexpected clash or collision, wasting a very expensive part. As a result, the team has shifted to study modelling and distortion control in part machining.
This process also reimagines the part manufacturing process. Historically, complete components that had been finish-machined and assembled had to be sent to separate drilling machines to be drilled, and that process inevitably put holes through the skin. As the robotic production system is reconfigurable, individual robots can automatically change the processing head used, to include drilling. In the car industry, robots welding carbody shells can automatically change welding heads to offer different reach or access. Factory of the Future uses the same principle, but with greater accuracy. The cell will be able to perform assembly, drilling and inspection automatically without a separate machine. In future, its capability will be further augmented to include fastener installation, too.
PRIMACY OF DATA
The discovery of the problems of distortion is just one example of the importance of managing component data in the manufacturing process. Several other strands of research cover this topic in more detail. First, the team is exploring the benefits of an integrated engineering toolset. At the moment, BAE Systems buys best-of-breed tools for every stage of the product development. Doing so inevitably brings in multiple brands of software for product lifecycle management, computer aided design and manufacturing execution systems. In the real world, those systems’ interfaces don’t quite match up, according to Cook, which necessitates manual intervention to support the enterprise.
Instead, an integrated toolset would allow model-based definition: product manufacturing information. Instead of relying on 2D drawings to make parts, designers add data into a digital model when it is designed, and then the machine could possibly read the data from the model, rather than having to program it manually.
The other big data project explores the benefits of linking operational technology systems. States Cook: “Once we abstract the data, we can turn it into useful information: predictive maintenance, using analytics to do that.” That would require developing software routines that turn production data into intelligence. He offers a contemporary application: automatic defect recognition. There are certain defects that can be picked up using today’s vision systems and data extraction technologies. The team aims to build on that. He adds: “As we move forward, accessing data from deployed connected infrastructure, our understanding of what types of data we need and don’t need continues to improve.”
These ideas are also being explored in another large project, chiming with Factory of the Future, but grant-funded by government: the 5G test bed. This joint project with AMRC is installing IOT connectivity using a 5G wireless private network. It began in September and runs through April 2022, and will connect BAE Systems’ robot assembly cell, the AMRC main campus in Sheffield and the AMRC North West site in Preston.
As data transmission rates are expected to be so much faster, production control systems might be able to collect and analyse data in real time, Cook speculates, so that engineers might be able to intervene very quickly in a process to prevent it going out of tolerance, or becoming dangerous.
Another application being explored in the 5G test bed is the possibility of real-time mixed-reality livestreaming, for production or maintenance. An engineer might be able to project an animation or schematic of a system on to the live physical assets themselves, to understand how they are working or how to carry out some particular work.
Finally, there is also a 5G pilot that involves tracking and tracing digital assets on a connected network. Is it possible to develop real-time data transfers within the supply chain? The company is developing a use case of this technology with the supplier Miralis.
Concludes Cook: “Although there’s a big focus on manufacturing technology at Factory of the Future, we’re looking wider than that. How do we engineer products in the first place? How do we engineer them to accommodate new manufacturing process? We are going to the left in the production engineering process into design and early structural layouts when thinking about how to make a product.”
BOX: DIGITALISING THE SUPPLY CHAIN
In addition to its work within the Factory of the Future, BAE Systems is also working in collaborative teams to develop digital supply chain technology. The project is aimed at digital transformation of the supply chain. In particular, the work aims to study the effects of using advanced digital manufacturing processes on the supply chain.
Summarising the project, Austin Cook says: “What are our requirements; how will we help them respond. How can we use data in a smart way. It’s not just what’s inside the building.”
BOX: INTELLIGENT WORKSTATIONS
One Factory of the Future project that has already been integrated into existing production is an intelligent workbench that guides an operator through tasks such as part assembly by using lights and a video screen, and which monitors progress using camera systems (see also www.is.gd/awesac). This technology was developed in collaboration with the Advanced Manufacturing Research Centre and Fairfield Control Systems. The system is in use within BAE Systems’ Typhoon Major Units facility in Samlesbury, Lancashire and involves the use of light-assisted assembly, integrated work instructions and increased ergonomics, as the bench automatically adjusts for height and light intensity for each operator upon log-in. Given a new task on the unit, an apprentice was able to build the part safely and to the required quality within three hours. Future developments include a sensor-enabled cobotic arm to work safely alongside others.