Inconsistencies in build quality part-to-part and build-to-build continue to be a problem in additive manufacturing (AM). The flexibility of AM often enables low-volume and custom production, making conventional methods of machine qualification and health monitoring challenging to implement. Machine health has been difficult to separate from the effects of design and process decisions, and therefore inferring machine health through part quality has been similarly complicated.
This conceptual paper proposes a framework for monitoring machine health by monitoring two types of witness parts, in the form of witness builds and witness artifacts, to provide sources of data for potential indicators of machine health.
The proposed conceptual framework with witness builds and witness artifacts permits the implementation into AM techniques to monitor machine health according to part quality. Subsequently, probabilistic models can be used to optimize machine costs and repairs, as opposed to statistical approaches that are less ideal for AM. Bayesian networks, hidden Markov models and Markov decision processes may be well-suited to accomplishing this task.
Though variations of witness builds have been created for use in AM to measure build quality and machine capabilities, the literature contains no previously proposed framework that permits the evaluation of machine health and its influence on quality through a combination of witness builds and witness artifacts, both of which can be easily added into AM production.
Hale, J. and Jin, M. (2021), "A conceptual framework for witness builds and witness artifacts in additive manufacturing", Rapid Prototyping Journal, Vol. 27 No. 6, pp. 1133-1137. https://doi.org/10.1108/RPJ-10-2020-0253
Emerald Publishing Limited
Copyright © 2021, Emerald Publishing Limited