Investigating the influence of fidelity on the capability of a digital twin to detect material extrusion failures

This paper investigates the influence of fidelity on the capability of digital twins (DT) to detect two common material extrusion (MEX) failures: extrusion error and layer shift. Monitoring of the failures is first explored using an image-based DT to provide a benchmark. The fidelity-performance analysis of DTs with significantly different levels of fidelity ranging from low to high fidelity is then studied and presented. In this paper, digital twin fidelity is defined as the expression of the necessary levels of accuracy in different attributes of the twin. The basis for the DTs employed in this study is image comparison. Consequently, six attributes are selected to describe the fidelity, including image resolution, mesh number, shadow method, illumination model, light source and state threshold. The effects of varying the fidelity of the six attributes on the capability to detect printing failures are then analysed and discussed. It is shown that, for the two failure cases considered, lower-fidelity DTs can deliver comparable capability to what are considered DTs with high or ultra-high fidelity. In addition, benefits in terms of configuration cost, data storage, simulation time, and detection time are demonstrated.