A Quasi-Experimental Design for Reliability Diagnostics in Senegalese Manufacturing Systems: A Case Study

{ "background": "Manufacturing systems in developing economies often operate under unique constraints, including intermittent power supply and variable maintenance regimes, which challenge conventional reliability engineering models. Existing diagnostic frameworks, largely developed for stable industrial contexts, lack validation in such environments, creating a knowledge gap in predictive maintenance strategies for these regions.", "purpose and objectives": "This case study aimed to develop and evaluate a quasi-experimental design for conducting reliability diagnostics within the specific operational context of Senegalese manufacturing. The primary objective was to measure the causal impact of scheduled preventive interventions on system mean time between failures (MTBF).", "methodology": "A longitudinal, non-equivalent control group design was implemented across two comparable production lines in a Dakar-based plant. The treatment group received a structured preventive maintenance protocol, while the control group continued with reactive maintenance. System failure data were collected over an operational period. Reliability was modelled using a Weibull proportional hazards model: $h(t|X) = \\frac{\\beta}{\\eta} \\left(\\frac{t}{\\eta}\\right)^{\\beta-1} \\exp(\\theta X)$, where $X$ indicates treatment status. Inference was based on robust standard errors clustered by machine type.", "findings": "The intervention produced a statistically significant increase in MTBF. The treatment group exhibited a 34% improvement in MTBF compared to the control group (95% CI: 22% to 46%). The shape parameter ($\\beta$) estimate indicated a shift from a wear-out to a more random failure pattern post-intervention, suggesting improved system stabilisation.", "conclusion": "The quasi-experimental design proved effective for isolating the effect of maintenance strategies on system reliability in this context. It provides a robust methodological framework for reliability diagnostics where fully randomised trials are impractical.", "recommendations": "Manufacturing plants in similar settings should adopt structured preventive maintenance protocols informed by locally collected reliability data. Further research should apply this quasi-experimental design to other critical infrastructure sectors to validate its transfer