“Computer simulation gives us a competitive edge”
Essential today in our product design, computer simulation is also necessary for improving manufacturing processes. The complete picture with André Colom, Simulation and Data Science Director at Ladoux (MICHELIN Research Center, France).
What does computer simulation mean?
Computer simulation uses a series of scientific calculations made by the computer and reproduces physical phenomena. The idea is to virtually assess and test product performances using scientific calculation software. In the tire world, that can be wear, grip, noise, cracks in the tread, etc. In parallel, simulation increasingly uses data from the field, what makes our simulation models perform better.
How is it used at Michelin?
We use simulation in two specific areas: the development of our tires, from a few dozen grams for bikes to several tons for mining and for everything that involves the manufacturing process. We can therefore simulate industrialization, different manufacturing procedures and product processing.
What are the challenges for the Group?
Computer simulation is at the very heart of innovation. Simulation tools are present at every stage of our design process. Moreover, simulation is continuing to develop. It is moving from the expert design office to be even more integrated; it is a source of creativity and a catalyst for acceleration. It is becoming a key element to make the customer experience more fluid. By combining physics and data processing, we will be able to offer our customers high-performing predictions and advice. This will help us remain a key stakeholder on a market where newcomers are very aggressive.
Let’s start with product simulation...
Our simulation strategy includes everything that varies performance, from molecules to the finished tire and vehicle wear. We study all elements that contribute to performance. To do this, we use extremely varied expertise in chemistry, materials and every aspect of physics. Our tools incorporate models at several levels to recreate real performances and generate a virtual twin of the tire. Our teams of mathematicians, mechanics and chemists determine the right math and physics laws to incorporate into our models. We then have several levels of simulation that progressively lead us to models of tires for vehicles: bikes, cars, trucks, planes, etc. The simulation will help us understand how a new and a worn tire behaves and how accidental damage can affect performance. It is very useful in the logic of sustainability and it helps us increase the lifetime of our products safely and securely.
How exactly do you work?
We use our internal expertise at the Ladoux site and in the Group’s other research centers. However, we also use an entire ecosystem of academics, mathematicians, physicists and chemists from France and around the world. We contribute to the Applied Mathematics Chair at Laval University in Quebec, which is helping us develop some of our simulation tools.
Do you have other partners?
Dassault Systèmes is a key stakeholder. It provides the basis for our CAD tool (computer-aided design). Using geometry, we add our own meshing, calculation and design tools. For calculations, we use partner software like Abaqus by Dassault Systèmes and products by Siemens and others. A tire can be used in a multitude of configurations and environments so we need the different corresponding skills and software. One piece of software would not be enough to simulate a tire and its uses. Nevertheless, we develop the core of our design tool in-house.
What about the manufacturing process?
It is less simple to understand the product itself but it is just as important for us. A tire has around two hundred component parts: rubber, chemicals, fabrics, textiles. It is an ensemble of materials that we assemble in a progressive process to achieve the finished product: the tire. This includes phases during which we work the rubber in mixers adding chemicals, with physical properties that will depend on how we mix them; extrusion phases, assembly and heating, etc. The rubber shaping process is also modeled, and the physics become complex when you move from raw to heated rubber. The quality of the finished product and productivity will therefore depend on the simulation of the different processes, with modeling tools at every stage of manufacture. Here again, there are no global tools. Which is why we use several modeling resources depending on our needs, some of which are developed in-house, others as partnerships.
Are there any particular difficulties?
On a technical level, one of the particularities is that we are working with rubber. Modeling mechanical procedures is pretty commonplace, like in the steel industry, but it is far more complicated with rubber because of its very particular and very demanding physics. Here are the complexity and charm of the work we do. We benefit from decades of expertise and today that gives us a competitive edge over the competition.
Internally, how do you work with the production teams?
The collaborative side, working with people, is fundamental. The Ladoux campus was designed and built with this in mind, to encourage collaboration between teams. This is the beauty and difficulty of associating all the activities. Simulation is an excellent vector and a great medium for getting teams to work closely together: industrialization, design and performance analysis.
How many people are working on these projects at Michelin today?
In total, in the Department of Applied Mathematics “Simulation and Data Science” there are around a hundred people in R&D working on these questions, mainly at our Ladoux site.
Can we evaluate the benefit?
Computer simulation offers the possibility of virtually testing hundreds of possible configurations. It helps validate design intentions and eliminates two-thirds of prototypes. Over time, we want there to be only one prototype required. This is why we continue to invest in improving the performance and design of our simulation tools for our products and our manufacturing process. If we did not have computer simulation today, it would be like working in the dark; it would be very difficult.
What about security?
For security reasons we do not use the Cloud for our computer simulations. When you have secret data, the only way to keep it safe is to remove it from the Internet. That is why our main process uses a calculation basis installed internally, with calculation power of nearly a petaFLOP (1 million billion operations per second).
Let’s talk about data and data science which is also your responsibility...
Thanks to the huge volume of collected data and artificial intelligence, we are going through a period of tremendous change. Data and AI are evolving our manufacturing processes, our products and our relations with customers. Today, we have the possibility of using increasing amounts of data collected from the field, from actual use. This data feeds our calculation models and means we can predict performances better and offer personalized services to our customers. This is a major evolution with a real impact on our processes. With smart tires, the data provided by customers will feed the analysis tools and help better understand how our products are used. You will understand why a tire did 50,000 or 100,000 miles. You will then be able to reinterpret the design of the tire and feed the models, not with hypotheses but with data from actual use. You will also be able to talk to the customer to predict product maintenance. It is particularly useful for B2B relations (Business to Business).
What are the applications?
On the B2B market, we already offer predictive maintenance services. It helps fleets better manage their vehicles. There are fewer unexpected problems and profitability is improved. Big machines for mining and the heavy truck sector are both pioneers in the development of predictive maintenance services.
André Colom, Simulation and Data Science Director at Ladoux (MICHELIN Research Center, France)