Abstract
{ "background": "Power-distribution infrastructure in many developing nations faces chronic reliability challenges, yet robust methodological frameworks for diagnosing specific failure modes in equipment systems are scarce. Existing reliability assessments often lack the experimental rigour to isolate causal factors from operational noise.", "purpose and objectives": "This working paper proposes and details a novel quasi-experimental design tailored for reliability diagnostics in electrical distribution networks. The objective is to establish a method for quantifying the causal impact of specific equipment interventions on system failure rates.", "methodology": "A difference-in-differences framework is employed, comparing failure rates in treatment and control groups of substations before and after a targeted equipment retrofit. The core statistical model is $Y{it} = \\beta0 + \\beta1 \\text{Treat}i + \\beta2 \\text{Post}t + \\delta (\\text{Treat}i \\times \\text{Post}t) + \\epsilon{it}$, where $Y{it}$ is the monthly fault count. Robust standard errors are clustered at the substation level to account for serial correlation.", "findings": "As this is a methodological working paper, no empirical results from a completed study are presented. The findings detail the developed design's architecture, including a specific sampling strategy where treatment and control groups are matched on baseline load and age profile. Simulation analysis indicates the design has sufficient power to detect a minimum 15% reduction in fault frequency.", "conclusion": "The proposed quasi-experimental design provides a viable and methodologically sound framework for conducting reliability diagnostics in real-world grid operations, where fully randomised controlled trials are often impractical.", "recommendations": "Utility engineers and researchers should adopt this structured approach to evaluate infrastructure upgrades. Future applied work should implement this design in a live network setting to validate its practical efficacy and generate actionable reliability data.", "key words": "reliability engineering, power distribution, quasi-experimental design, difference-in-differences, infrastructure diagnostics, causal inference", "contribution statement