An Integrated Offline and Online Optimization Framework for Large Scale Additive Manufacturing

The large-scale additive manufacturing system using thermoplastic materials has been widely applied in the aerospace and automotive industries. The improper surface temperature of layers can cause quality issues. Therefore, an accurate prediction of layer deposition time, or layer time, can significantly improve product quality and production efficiency. Due to varying temperature requirements for different experimental designs and differences in temperature cooling curves among various geometries, layer times need to be estimated accounting for multiple conditions such as geometry, material, and ambient temperatures. However, since conducting repetitive experiments using real production process data is expensive and inflexible, temperature data generated from a physics-based FEA model, which can simulate the printing process, needs to be used to find the optimal layer time before printing (offline design). The optimal layer time provided by this method will be highly consistent on each layer for a homogeneous geometry due to the ideal temperature ignoring uncertain environmental influences and sacrificing fidelity. Therefore, during printing based on the optimal layer time provided by offline control, it is necessary to use real-time information captured by the IR camera to further optimize layer time and make corresponding adjustments. We propose this approach as an integrated optimization framework, which is further verified using a real case study.