"background": "The evaluation of industrial machinery fleet efficiency in developing economies is hindered by a lack of controlled field data and the logistical impossibility of randomised trials in operational industrial settings. Existing methods often rely on theoretical models or self-reported data, which lack robustness for causal inference. ", "purpose and objectives": "This article presents a novel quasi-experimental framework designed to measure causal efficiency gains from interventions, such as maintenance protocol changes or telematics system adoption, within operational industrial machinery fleets. The primary objective is to provide a rigorous, field-applicable methodology for structural and mechanical engineers. ", "methodology": "The framework employs a difference-in-differences design, comparing treated and control machine groups before and after an intervention. Key efficiency metrics, including fuel consumption per output unit and operational availability, are tracked via sensor data. The core statistical model is a fixed-effects panel regression: Y{it = \ + \ (Treati \ Postt) + \ Xit + \ + \ + \₈ₓ, where robust standard errors are clustered at the fleet level to account for serial correlation. ", "findings": "As a methodology article, this paper presents no empirical results from a specific application. However, the framework's validation using simulated data demonstrates its capacity to isolate intervention effects from confounding factors; for instance, it reliably detected a simulated 8. 5% improvement in fuel efficiency with a 95% confidence interval of 7. 1%, 9. 9% under typical field conditions. ", "conclusion": "The proposed quasi-experimental framework provides a technically sound and practical methodological tool for engineers and managers seeking to empirically validate efficiency improvements in capital-intensive industrial operations, moving beyond anecdotal evidence. ", "recommendations": "Practitioners applying this method should ensure a minimum of six months of baseline data collection, carefully select control assets to satisfy the parallel trends assumption, and integrate direct sensor logging to minimise measurement error in key performance indicators. ", "key words": "qu
Sarr et al. (Tue,) studied this question.