A data-driven reconfigurability index for manufacturing systems: a PCA-based approach

Abstract Reconfigurable manufacturing systems (RMS) have emerged as a strategy for enhancing production responsiveness in volatile markets. While reconfigurability is essential for adjusting production capacity and functionality, existing literature often relies on subjective multi-criteria decision methods to assess its implementation, which may overlook the complex interdependencies between core characteristics. This study develops a quantitative and transferable reconfigurability index (RI) by applying principal component analysis (PCA) to empirical data from 112 manufacturing companies located in Portugal. The PCA model demonstrated merit for factor analysis (KMO = 0.73) and accounted for 66.05% of the total variance explained. By leveraging the intrinsic statistical variance of the data, the proposed method derives objective weights for five key characteristics: modularity, integrability, customization, adaptability, and diagnosability. The results reveal statically significant associations between adaptability and manufacturing cost, and diagnosability and delivery performance. Specifically, adaptability and diagnosability emerged as the primary drivers of reconfigurability, each contributing 25% to the overall index. The application of the RI demonstrates its utility as a benchmarking tool, although its current findings are bounded by the geographical context of Portuguese manufacturing sector. The study suggests that soft reconfigurability (diagnostics and logical adjustments) shows a more dominant association with system maturity than hard mechanical modularity in this specific context. This research provides a data-driven framework that enables managers to evaluate system gaps objectively and optimize investments in reconfigurable capabilities.