Defect prediction model for wrapping machines assembly

Purpose: Development of a defect prediction model for the assembly of wrapping machines.
Design/methodology/approach: The assembly process of wrapping machines is firstly decomposed into several steps, called workstations, each one potentially critical in generating defects. According
to previous studies, two assembly complexity factors related to the process and the design are evaluated. Experimental defect rates in each workstation are collected and a bivariate prediction
model is developed.
Findings: Defects occurring in low-volume production, such as those of wrapping machines, may be predicted by exploiting the complexity based on the process and the design of the assembly.
Research limitations/implications Although the defect prediction model is designed for the assembly of wrapping machines, the research approach can provide a framework for future
investigation on other low-volume productions of similar electromechanical and mechanical products.
Practical implications: The defect prediction model is a powerful tool for quantitatively estimating defects of newly developed wrapping machines and supporting decisions for assembly qualityoriented
design and optimisation.
Originality/value: The proposed model is one of the first attempts to predict defects in low-volume production, where the limited historical data available and the inadequacy of traditional statistical
approaches make the quality control extremely challenging.