How AI and machine learning can be used to assess manufacturability
Can a component really be made with the selected geometry? This question can be answered by precise, but time consuming, simulations. The process can be speeded up by a new program which is currently being developed at the Institute of Vehicle System Technology (FAST) at the Karlsruhe Institute of Technology (KIT). The scientists are combining process simulation with machine learning and AI to enable the algorithm to recognise patterns and quickly assess whether the part can be manufactured or not. This saves development and manufacturing costs. Clemens Zimmerling, who is developing the program, will be presenting the enormous opportunities which AI and co. offer for lightweight technology at a fascinating talk to be given at the 6th Technology Day Hybrid Lightweighting (20th and 21st May in Stuttgart).
The Development Agency for Lightweighting Baden-Württemberg presents this innovation in the March 2019 edition of the ThinKing. Leichtbau BW GmbH uses this label to showcase excellent lightweight technology products or services from Baden-Württemberg every month.
The scientists refer to the new development in Karlsruhe as a “virtual AI process expert”. Machine learning and AI enable the algorithm to learn to recognise patterns and by doing so to assess whether a component can be built using the selected geometry or not. “It's a bit like packing the know-how and intuition of many different engineers into a single program and making it usable at the touch of a button,” explains Clemens Zimmerling from the Institute for Vehicle System Technology (FAST) at the Karlsruhe Institute of Technology. The plan is for this virtual process expert to be used in the design of lightweight structures made of fiber composites to quickly optimise both the part itself and the process of manufacturing it.
It’s better to have good information early on rather than precise information when it's already too late
The “Research Bridge project” funded by Baden-Württemberg’s Ministry of Science, Research and Arts focuses primarily on understanding and speeding up design processes. The problem at present is that although the production process can be simulated to assess the feasibility of manufacturing components, they are very cumbersome and time consuming. Successive optimisation loops multiply computing time. “The feedback we have received from developers is that it is better to have a good assessment early on in the process than to have a very precise evaluation later on,” says Zimmerling. This is where the “virtual AI process expert” comes in. “If it is used in the early phase of designing a component, it is possible to obtain feedback about how easy or difficult it would be to manufacture the part as well as an assessment of potential alternative manufacturing processes. This means that AI methods can be used to eliminate unsuitable process variants from the optimisation calculation and to concentrate resource-intensive simulations on the most promising ones,” explains Zimmerling.
“Digitalisation is having an electrifying impact on lightweighting in particular. It is now important to jump on the bandwagon and to discover and make profitable use of AI or machine learning – because others most certainly will,” affirms Dr. Wolfgang Seeliger, Managing Director of Leichtbau BW. Reducing time-to-market is an enormously effective value lever. “The complex simulations and optimisation loops in lightweight construction depend on digitalisation. The full potential of lightweight technology can be leveraged if this is taken into account from the very start of product development,” continues Seeliger.
“Our program can be used to save tool and manufacturing costs. Trial and error testing can also be minimised in development to arrive at error-free parts production much faster,” Zimmerling is convinced. The program also enables small batches in which alternating components are manufactured on the same production line to be produced more cost effectively. This is especially useful in industries with a high level of derivitization, such as automotive, mechanical engineering or aerospace.
Stay one step ahead and be a game changer: 6th Technology Day Hybrid Lightweighting
Digitalisation, machine learning, AI, digital development chains – at the 6th Technology Day Hybrid Lightweighting this is anything but an empty phrase. On 20th and 21st May, trend setters and experts will be demonstrating how these technologies can be put to effective use in lightweighting to reduce time to market, to optimise processes or to design better products and stay one step ahead of competitors. One of the speakers is Clemens Zimmerling. The attendance fee is only 190 euros. Go to www.lightweight-techday.com for more information about the programme and to register.