Model-based optimization of process parameters in high energy impact additive manufacturing processes

Cold gas spraying is an emerging technology in additive manufacturing, known for its versatility and broad range of applications. This process enables the deposition of various materials, such as metals, ceramics and polymers, onto substrates by accelerating particles to high velocities within a Laval nozzle. To achieve optimal manufacturing quality at low cost, continuous and precise adjustment of process parameters is essential. However, due to the complex behavior of the gas dynamics, assessing quality during manufacturing is challenging. To address this issue, two data-driven modeling approaches are described and compared that connect process parameters to particle descriptors within the spray jet: a fast and easy-to-implement radial basis function network (RBFN) method and a low-parametric copula-based method in order to probabilistically model the high-dimensional dependencies among particle descriptors. These two modeling approaches are illustrated through an example of data-based optimization of process parameters for a cold gas spray in free jet, but are directly applicable to non-free jet data, if available. Additionally, both methods have low computational cost, making them suitable for applications even in autonomous process control. • Two data-driven modeling approaches for characterizing cold gas spraying. • Modeling of dependencies between particle descriptors using particle tracking velocimetry. • Fast and easy-to-implement radial basis function network approach. • Low-parametric copula-based approach using R-Vine copulas. • Model-based optimization of process parameters in free jet setting.