Within the automotive industry, there is an increasing demand for a paradigm shift in terms of which
materials are used for the manufacturing of the automotive body. Global climate goals and customer
demands are forcing rapid adoption of new, advanced, sustainable material grades. Here, one
opportunity is metals created through secondary production, i.e. recycled materials [1-2]. However,
using materials with increased scarp contents does not come without consequences. Increasing the
content of recycled material can lead to reduction in formability [3] or increase in the in-coil variation
of material parameters [4], potentially leading to a situation as outlined in Figure 1.
Figure 1. Illustration of current situation (left) where a single set of material parameters can be used for a coil, and potential
future situation (right) where multiple sets of material parameters must be determined for a coil.
The following study will present how Artificial Neural Networks (ANN) can be used to adopt materials
with high in-coil variation into the manufacturing of automotive body components. The study will
present an industrial case as a demonstrator - a Volvo XC90 Front Door Inner component. Initially, the
study will focus on training an ANN with data generated from high-fidelity stochastic Finite Element
simulations, to predict the state of 15 quality points around the industrial demonstrator. Secondly, the
study will compare the predictions made from the ANN to experiments carried out at the shop floor at
the Volvo Cars stamping plant in Olofström, Sweden. Finally, the study will present initial findings of
re-training the ANN to act as a support tool for the operator to proactively adjust for varying material
properties, and thereby reducing material scrap in production.