Grant: Discovery Grants Program - Individual

Manufacturing industry faces a new challenge: producing in low batch sizes highly diversified products. Reduced product life cycle, on demand manufacturing and increasing demand for customized and personalized products are among the main drivers. If customisation is known since several decades, personalisation is a relatively new trend. In a personalized product, the customer no longer chooses among predefined options, but gets actively involved in the design. Industry aims to develop the new concept of mass personalisation where one aims to produce massively personalized products at similar costs than in mass production. As a result, production shifts towards the fabrication of parts in low volume but high diversity. If technologically the production of personalized parts is possible, an economically viable production is, except few exceptions, not yet feasible. Mass personalisation requires flexible manufacturing technologies of a new kind which are able to adapt to specific customer inputs. They have to avoid specialized tooling, operations requiring long setup times, recalibration and must handle complex geometries. A major cost driver are the recalibration and optimisation operations triggered by each new customer requirement. The long-term objective of the present program is to develop novel manufacturing technologies able to address the challenges resulting from the production of parts in low volume and high diversity. On the short-term, the program will develop tools addressing the challenge of reducing costs resulting from frequent recalibration and optimisation of the production processes triggered by design changes. For this, we develop a digital framework within which production processes learn from past experience in order to adapt their process parameters to the customer requirements, taking into account the present state of the production line. The driving idea is that, as a junior operator becomes a senior operator over time, a production line progressively improves its operation as parts are produced. As demonstration, electropolishing, electroplating, electroforming, and glass micro-machining by spark assisted chemical engraving are selected. For such electrochemical manufacturing processes, a specific challenge is the electrolyte which over the time, as it gets used, changes its properties resulting in different manufacturing outcomes under identical process parameters. The frequent customer design changes avoid that electrochemical baths can be kept within a stable state by optimising counter-measures (e.g. adding periodically defined quantities of fresh electrolyte). In the frame of our program, we develop models learning from historical production data in order to predict optimal parameters to be used considering the current state of the electrochemical bath. By further combining them with first principle models of the process, the change on part geometry can be predicted and consequently the production adapted to it.