Pulling the digital thread with AI and systems thinking

Industries across the board are starting to use and seriously benefit from artificial intelligence, and manufacturing is no exception. So, how are today’s product-based companies applying intelligent technology to solve the many challenges of getting high-quality products to market faster?

During our speaker series, “Pulling the Digital Thread,” I spoke with three of the industry’s leading thinkers to share perspectives and ideas on how AI is augmenting systems thinking and model-based systems engineering (MBSE) to streamline and accelerate design and engineering processes leading to faster time to market.

Rob McAveney
Rik Rasor
Researcher at Fraunhofer IEM & Co-CEO of AI Marketplace
Jochen-Thomas Morr
Partner at Pricewaterhouse Coopers
Rob McAveney
Chief Technology Officer at Aras

 

Read on to learn how AI is reshaping systems engineering and model-based systems engineering (MBSE), with tips on how to focus your AI investments. Hear about AI-assisted asynchronous simulations, getting your experts on board, creating the perfect coffee machine line, and how the Red Bull Formula 1 race team can put a newly designed car on the track weekly.

How AI is helping companies get to market fast

Whether in the automotive, aerospace, medical technology, food and beverage, or other manufacturing sectors, all companies face the challenge of getting their products to market before competitors without sacrificing quality.

Jochen-Thomas Morr shared that in PwC’s most recent CEO survey, 75% of companies want to reach customers directly through digital platforms. “They want to get fast feedback and continually reinvent their products to respond to customer needs, creating an even faster race to market as companies move to shorten the time from idea to launch. That’s the game now.”

But this race is strewn with challenges, from knowing compliance requirements to simulating, iterating, and validating to shaping the right product portfolio. Here’s how AI is helping companies overcome obstacles and accelerate processes.

Ensure compliance

It’s fruitless to design a new component or model without understanding the full spectrum of ever-changing laws, regulations, safety specifications, and compliance requirements your products must adhere to in your target markets. AI can now carry out this time-consuming scan, letting companies ensure that battery systems and crash tests remain up-to-date.

“The rise of large language models (LLMs) lets us look at every text-based engineering artifact,” according to Rik Rasor. “There’s been a lot of hype around how AI can help with this, but we’re starting to see the first real-world applications of handling specs in different ways.”

Optimize design parameters

For decades, experienced engineers and product architects have used MBSE in the early stages of design to respond to high-level system constraints like weight, size, and power. These constraints are then spread out into many different departments to develop, simulate, build, and validate. Now, we can integrate AI to make this process easier, not only in the early stages but later in the process, too, stretching MBSE into the entire design process.

“AI is letting us find patents, do multi-domain simulations, generate models efficiently, and explore every possible solution asynchronously to truly discover the simplest, most innovative, and most optimized design parameters and architecture to lead to the best product,” said Rik Rasor. “This kind of innovation would not be possible without AI-supported MBSE.”

Accelerate simulation

Generating models and automating simulations using AI is helping industries with large and complex products, such as cars, planes, spacecraft, and medical equipment, gain more accurate models to test their current designs’ function and safety performance. Typically, simulation results are not immediate. AI can help you quickly change input parameters and see the results faster while doing multi-fidelity simulations and real-time simulation monitoring. Rik Rasor looks forward to seeing simultaneous simulations used to train the models and generate synthetic data that can be used later in product development.

“AI is helping the automotive industry rapidly innovate,” says Jochen-Thomas Morr. “One example is Formula One, where teams perpetually redesign their cars, but typically on a delay of 4-6 weeks. The RedBull team is using AI to put a new car on the track every week. They’re also designing cars for different drivers and tracks, whether Singapore or Monaco.”

Automate validations

Validating products against specifications and requirements is time-consuming and can extend the time to market, especially with complex products. AI is changing the game for organizations like NASA and automotive companies.

“One coffee machine maker had three product lines, from basic to professional,” said Jochen-Thomas Morr. “The company let AI calculate a more optimal product profile based on customer needs, parts costs and features like how many types of beans the machine could handle. AI suggested that having two product lines was more optimal.”

How to make AI work

You may wonder how to start using AI if your company, like many others, is still trying to achieve Industry 4.0, update and integrate digital systems, and move from a hardware-driven to a software-driven company. How do you ensure AI recognizes and leverages your engineering methodologies and models? And, while respecting the deep expertise of your professional staff, how can you ensure machines and people play nice? These elements and many more can become barriers to successfully using AI. Address them, and you will make them enablers instead.

Establish a digital thread first

While it’s true that AI can do some amazing things, it’s not the silver bullet to all of our problems. Without a digital thread to link parameters across disciplines, AI won’t work as you need it to.

“I could set AI loose on all my separate 2D drawings, requirements, specification PDFs, 3D CAD models, and costing models, then hope for the best,” says Rob McAveney, Aras CTO. “But your company has a lot of interrelated disciplines, processes, and data. You don’t have room for guesswork when you’re building a product that has to be safe and keep risk low. You need the checkpoints and validations allowed by a digital thread to be able to trust the results.”

Engage your people

When it comes down to it, it’s your people who do the work at your company. If you want to be faster and more efficient, you must change how you work. If your people don’t embrace AI as a tool, they may feel their work is devalued. If they don’t understand the strengths and limitations of AI, mistakes will happen.

“Formally adopting AI into your work processes requires a lot of change management,” said Jochen-Thomas Morr. “You’re not just changing the IT system. You need to make sure the people using AI are trained to use the new tools and can see the benefits the technology offers. It’s not replacing their work but enhancing it and freeing them up for more interesting work.”

Scale your solution

The other question on AI is: can you scale it? “There are so many ideas out there, but we’ve seen that 90% of companies fail to use Industry 4.0 processes end-to-end,” said Jochen-Thomas Morr. “If you can integrate the solution into the tool suite and ecosystem rather than just here and there, you’ll see benefits.”

Secure your data

AI uses large data banks to learn and deliver results. However, it’s important to ensure that data security and federation are respected.

“What you don’t want to do is drop all product-related data in a data lake and let AI loose on it,” said Rob McAveney. “Some of your data may be classified through government agencies, export controlled, or a trade secret. Data must be federated according to a project’s phases and disciplines in an enterprise-grade access control model that does not leak data but reflects the serious nature of data security.”

Teach AI your methodologies

Training AI on your company data, engineering systems, and methodologies holds a great deal of promise. “But you’ll need to talk to your engineers who have been working in your company for years to bring the internal knowledge of how all the methods are applied to engineering processes and data,” said Rik Rasor. “For example, AI may know how you’ve created a CAD, but it may not know how your company is doing the containment tree.”

The future of AI in manufacturing

Some manufacturing sectors are ahead of others in adopting AI. The automotive industry jumped right in, with North American and European car makers keen to keep pace with rapid innovation in China. AI is helping despite existing brownfield technologies.

The food and beverage and fast-moving consumer goods sectors are seeing the benefits of AI in processes like managing SKU numbers and developing recipes; their less complex infrastructure and products help them more easily adopt these new technologies. The aerospace industry has been a slower adopter, given that its new product lifecycle is in the 30-year range. The medical technology sector has also lagged since its margins remain high.

In five to ten years, all engineering software tools like CAD, PLM, and MPC will integrate AI, and all sectors will move naturally into using AI. Said Rik Rasor: “When that happens, your LLMs will be reaching conclusions based not on random data scraped from the internet but on your own digital thread, your PLM, your formalized MPC models. That’s when we’ll see some unexpected innovation from the tooling side.”

Is your company using AI? Have you seen any upside, major or minor, from using AI? Feel free to share any learnings with our readers in the form of comments. And stay tuned for our next webinar and blog post in the “Pulling the Digital Thread” series: Harnessing the Power of Simulation and AI for the Future of Engineering.