Coupled process chain modelling to minimise the earing formation in industrially rolled aluminium strips for beverage can production using data-based methods

Abstract The production of aluminium strip involves a long sequence of thermomechanical processing steps that significantly influence the material’s mechanical properties and may induce anisotropy. This anisotropy can manifest as earing during deep drawing operations - such as those used in beverage can manufacturing - resulting in increased trimming scrap, process downtimes, and reduced economic viability. To assess formability and quantify earing, the cup drawing test is employed as a standard evaluation method. Understanding and minimizing earing formation requires comprehensive modelling of the entire process chain, which is traditionally performed manually by domain experts - a time-consuming, error-prone, and costly effort. This study presents a novel, scalable, and flexible approach to model a process chain by integrating production data with process models on the Microsoft Azure Databricks platform. The proposed method is validated on an industrial aluminium strip production line, demonstrating its capability to automate data processing, extract actionable insights, and support process optimisation. The approach successfully identifies an optimum processing route that minimises the earing integral, as determined by a dedicated evaluation function.