Skip to main content

According to research by TechTarget’s Enterprise Strategy Group (ESG), as artificial intelligence (AI) continues its “meteoric” rise into business and IT environments, organisations are rapidly assembling or accelerating strategies to support AI technologies across every applicable area.

Unlike niche technologies that impact only certain processes or personnel, AI has wide-ranging potential to transform entire businesses, IT environments and associated teams. The underlying infrastructure and other supportive elements must be fully capable of supporting that strategy.

This means, say the report’s authors, AI strategies must be multi-pronged efforts that properly align business objectives with AI initiatives and expectations, which requires thorough participation from stakeholders across an organisation.

There are not many industries where teams, stakeholders and tandem work strategies are more fundamental than in engineering and manufacturing, a world of constant collaboration and communication. And this is a world where AI, in particular generative AI (GenAI), is having a profoundly increased effect on the connected networks that the industry depends on and the skills required to address its needs over the course of the next decade.

Aligning manufacturing and AI

The ESG study forecast as many as 92% of organisations will have GenAI in production by the end of this year. But before looking at where, how and by whom it will likely be applied in manufacturing, it is worth looking at the modern manufacturing process and where business objectives could be properly aligned with AI initiatives and expectations.

Fundamentally, the modern manufacturing process is based on many interconnected disciplines taking place, often at the same time. Each member of design and manufacturing teams needs to have access to the right product information at the right time, meaning information availability is mission critical.

These team members are involved in tasks encompassing conceptual design; initial engineering design; prototyping, including the likes of wind tunnel analysis and increasingly computational fluid dynamics; design simulation and crash tests; procurement, supply and ordering; marketing; service and maintenance; and production.

The complexity of the systems we have and the complexity of the context in which we operate requires an incredible transformation of the tools we are using, such as the data model and data flows
Olivier Ribet, Dassault Systèmes

Project success depends on constant collaboration and communication between the various people engaged in carrying out their individual activities, who may be located virtually anywhere in the world.

As distributed processes become more established, a knock-on effect has been a huge increase in data, making product lifecycle management (PLM) more complicated. All of the individual process elements have to be completely interlinked, with team members, manufacturers, suppliers and partners authorised to have access the appropriate data at the appropriate stage of the manufacturing process. And all of the individual processes can and are seeing the influence of AI, most interestingly in creative design.

Improving design and manufacturing

As it looks to what will be the future of the industries it has served since 1981, engineering software company Dassault Systèmes is looking to see where GenAI can affect and improve upon the design and manufacturing process and paradigms that it believes its markets will be based on.

The company says that it aims to combine art, science and technology for a more sustainable world. It employs 22,500 people in 136 countries, has one global R&D centre and 76 labs. The company’s products are used in industries such aerospace and defence; architecture, engineering and construction; business services; cities and public services; consumer packaged goods in retail; high-tech; home and lifestyle; industrial equipment; infrastructure, energy and materials; life sciences and healthcare; marine and offshore; transportation and mobility. Projects on which the firm’s system have been used range from a shampoo bottle to a satellite.

As an indication of how it has evolved over the last 43 years, the company has grown from being a design and engineering software provider to now having three core pillars of operation – products and services; business models and operations; workforce of the future and organisation.

These are presented as part of a virtual twin strategy that proposes to reshape standard methods used in engineering design. The aim is to realise the power of technology not only in modelling but also in the processes that are going to make products become real, and in the sustainable lifecycle of their production and operation.

That is, to have the ability to perform tests, simulations and what-if scenarios, not just with changing sizes and dimension. For example: what if you were to do virtual testing on the manufacturing? What if you change the manufacturer? What if design requirements need to change? How would manufacturers need to adapt? What would be the sustainability impact of changing materials? The carbon footprint of the device? The energy consumption? What if you wanted to make a product lighter? What are the costs and other trade-offs?

Essentially, the result is to take the offer beyond just mechanical engineering and software simulation and bring about a totally different way of thinking about engineering. Including, says Dassault executive vice-president for EMEA Olivier Ribet, “a projection” of what should not happen.

“We are coming out of a new era of digitalisation where we’ve pretty much digitised paper and workflows. We believe that once we are capable of modelling product in context, once you are capable of bringing experiences that you can have with this product, then suddenly you can do something extremely fundamental – putting the human being back in the centre of the innovation process,” he says.

“So everybody can project [an idea] in context of innovation in complex production, in context of cooperation, [in a] cycle of human interaction that doesn’t already exist. That’s what we call a disassembly loop, moving from virtual to real and back and learning from the real world and equally again the model with the intent to make the [process] get better and better. Virtual technologies have a very powerful dimension and [we have] a strong interest to enable sustainable innovation.”

Sustainability and innovation

By sustainability, Dassault essentially means processes that use fewer materials and less energy at less cost with less product waste, tapping the existing knowledge, know-how and expertise of people who have previously generated design and manufacturing intellectual insight. But how will the likes of Dassault pass on knowledge that has been created? How is this brought to a new generation of designers to evolve and transform what was done before?

Innovation, says Ribet, is not just creating something new from scratch and implementing something that was much more difficult to design in the past. This, he says, questions the fundamental assumption of how it was initially designed. Innovation must be seen as a sustainable continuous process that supports not only the designers and engineers but also suppliers and partners, to ensure the product designed is as good or better than before.

“How do we categorise insights, intellectual property expertise and pass back to the next generation to make better? With a business model, how do you make sure that you can continue to invest beyond manufacturing capacity or R&D capacity and project what you need in terms of infrastructure process for the future. That’s the way we think about sustainability for people and profitability of the business. So once you are set up, another question is, how do you ensure not just the success story of people building software, but the transformation of models, approaches, ways of talking to each other? That’s a totally different collaboration,” he says.

“The complexity of the systems we have and the complexity of the context in which we operate requires an incredible transformation of the tools we are using, such as the data model and data flows that we are operating on a daily basis. But more importantly, the profound transformation of skills and competencies of people in a way of representing the future in a totally different way than in the past, not throwing away 50, 100, 200 years of innovation. [Making sure this] is capitalised and we learn from that moving to the next evolution of innovation.”

At the vanguard of this next evolution is the use of GenAI in the design phase of the process. And in this domain, digital twins will act as a representation of an asset or a product facility, fully capturing what the real-life product or part includes. That is the physics and science which goes around a product.

Connected infrastructure

A year ago at Mobile World Congress (MWC) in Barcelona, Dassault demonstrated for the first time the way in which 5G networks could assist this process, connecting with the internet of things (IoT) and building up a connected digital infrastructure. For example, modelling a factory robot showing every asset contained within the working environment, with all of the required communications access points to plan for the coverage zones of the localised communications infrastructure. These could take the form of understanding the connected environment of, for example, a private 5G network and converting connectivity points to a 3D model containing all of the building enclosure and equipment contained within, but also using this model for asset monitoring and simulation.

At MWC 2024, this paradigm entered the virtual realm with Dassault talking to users about how to put in practice the modelling and communications infrastructure to bring value.

Dassault Systèmes vice-president for high-tech industry, Stéphane Sireau, explains: “Say you’ve got a massive open air facility, and you’ve got a lot of moving parts, or an automated guided vehicle, for example. If it needs to connect in an area where there’s a pretty wide area with a lot of critical assets where you need the industrial grade, then the business case for a 5G network makes sense.

“Equally, there might be cases where you’re not able to find a business case for 5G so that’s where the virtual twin comes in, because it allows you to basically model 5G [coverage] within the wider context of your manufacturing facility by sight. So the conversation, despite all the excitement, is around 5G. The discussion has to be about value and needs to be done by organisations that deeply understand the manufacturing process. Should [the site] need a 5G network, the question has got to be about value in the context of the industrial process. So 5G is not an end in itself, but as an enabler to this process.”

For Sireau, the process starts with having experts who deeply understand the process of design, engineering and manufacturing. Then, he says, by use of digital twins engineers can play around with value by being able to orchestrate what you want to do in a virtual world before actually designing. What this implies is that product design isn’t the be all and end all. Projects are a lot more use-case driven, and the modus operandi is that you take product design, create a virtual twin of your product, and then the opportunities are limitless.

Engineering LLMs

The projects now include GenAI. The Dassault CAD models allow virtual data to be mapped to that of the real world and create a feedback loop using work instructions, which can then be reused, as all design intent is captured, meaning large language models (LLMs) of engineering data now have massive value.

“The large language model can process different use cases. Now what we focus on is the actual physical model that today is something where we use generative design. And in [our] particular kind of generative design system, you start from your name and then you ask for the AI to generate what is the model that best suits your mindset, and you can see a piece being shaped to meet your requirements,” Sireau says.

The current Dassault GenAI design models show how to build a digital facsimile, including clustering, to interrogate the modelling system to analyse parts that may have the required shape and materials, and so on. Generative design is seen as a completely new way to design instead of sketching information. The user explains the part in general and its key attributes, not only including dimensions but where it is attached or what is its function; physical attributes such as the strength of torque it may experience or minimum displacement as you start to put pressure on it. And when the system offers an optimal shape, simulation software then checks mechanical constraints.

This way designers are given access to more meaningful assembly information based on dedicated key details to construct a map of data based on the 3D digital image. This map can support a number of use cases. For example, check to see a design path already exists, comparing the model database for paths that may have the same shape, the same usage or the same criteria. Dassault compared the process to Amazon Marketplace giving out several possible options to use.

While the generative design space is moving quickly, Sireau points out that the company has been working with generative design for many years and stresses the recent work around large language models is part of a continuing shift. The key to it all, he says, is the value such functionality brings.

“The adoption of virtual twins often depends on the value it brings to the customer. So often, it’s a question of customer perception and how the customer is able to articulate the value from a virtual environment,” he says.

“It’s important to separate between the technology and the value it brings to customers and their ability to be able to articulate and really have something meaningful. What we’re seeing is customers already understanding the value of virtual and increasingly going to virtualise their environment, whether it’s product design or manufacturing design, And AI, effectively, is leveraging more value from your data – it’s not an end in itself. So it’s important to recentre the conversation all the time.”

Educating in engineering

Cranfield University in the UK, a leading education establishment in aerospace engineering, is one of the places where such conversations are taking place. In September 2023, Dassault Systèmes and Cranfield launched the 3DExperience Edu Centre of Excellence, a joint programme to push forward digital experiences and present the knowledge and know-how that will enable the next generation of students, technicians, engineers and innovators to thrive in the aerospace industry’s transformation toward digitisation and sustainability.

The partnership marked the first Edu Centre of Excellence in the UK, joining a network of 16 centres in the worldwide programme. The centres are designed to accelerate experiential lifelong learning locally, by providing businesses and governments with a base to develop the 3DExperience platform expertise needed for what is described as a virtual and collaborative leap of the industry, and the reduction of the skills gap. They also rely on Dassault Systèmes’ certification programme to validate skills and knowledge.

Dassault and Cranfield have guaranteed to offer students a unique virtual/real learning experience based on the Dassault Systèmes 3DExperience engineering platform. In January 2024 the two parties expanded their programme to see how postgraduate engineers can push forward the realms of connected design. Cranfield has already started delivering courses based on the Dassault software platform, with customisation of some of the aspects of the standalone desktop software platform.

Cranfield is regarded as a world-leading hub for aerospace and vehicle design, even having its own airport and real (grounded) Boeing jet on which to test design and manufacturing projects. The university attracts around 100 postgraduate students each year who undertake what is essentially a very non-conventional course, simulating what it is to be part of a large aerospace design team. The college works on the assumption that if it is to solve the forthcoming challenges of the aerospace industry, it has to find good and effective ways of communication and collaboration.

Ian Grace, director of aerospace at Cranfield, says: “We as a sector face one of the biggest challenges we’ve ever faced. And so for me, this isn’t just about manufacturing in the supply chain. This is about how we can provide shared data across the whole ecosystem. If we are going to resolve the aviation challenges of 2050, we need to find good effective ways of sharing information across the ecosystem. And that for me is what this is about.”

For Jafar Jamshidi, senior lecturer in integrated product development in the school of aerospace, transport and manufacturing at Cranfield, the partnership is about a need to push on in digital product development and digitisation of manufacturing.

“We have seen via discussions from our industrial partners that there is a need to align our teaching and training activities more in the direction of industrial needs. We would like to utilise the latest technology when it comes to product design and development experience. We will have some in-house software that we will use for our training activities, some open source and other commercial software,” he says.

“In terms of research projects, again MSc and PhD students have already started using [the Dassault 3Dsoftware]. We will define new projects and seek new opportunities to utilise the platform mode. I like the idea of having [an engineering company] to be able to come and see the journey from conceptual design, initial design, prototyping, testing and eventually having the product.”

And reflecting how Sireau predicts the future of engineering, Cranfield stresses how engineering data and insight will be reused. A single cohort will be tasked with all different components of the aircraft, such as key structural parts and systems. But it is very likely that the team will be building on the work of previous projects delivered by another cohort of students, meaning it is essential to have an infrastructure with seamless data integration.

The first step for students will be bringing legacy data into the platform to build a new educational experience. The bottom line is that designing every single system and component from scratch is just not efficient – and with the new GenAI tools increasingly meshed into the Dassault system, that approach is no longer the future.

Leave a Reply