Journal Design Engineering Masthead
African Civil Engineering Journal | 20 October 2020

A Bayesian Hierarchical Model for the Cost-Effectiveness of Power-Distribution Equipment in Kenya

A Policy Analysis for Asset Management
K, a, m, a, u, O, c, h, i, e, n, g, ,, W, a, n, j, i, k, u, M, w, a, n, g, i
Bayesian modellingAsset managementInfrastructure policyCost-effectiveness
Bayesian model quantifies regional heterogeneity in equipment failure rates.
Probabilistic framework challenges one-size-fits-all national maintenance strategy.
Analysis supports shift toward regionally differentiated asset management policy.
Formal uncertainty quantification improves sparse-data decision-making.

Abstract

{ "background": "Asset management for power-distribution networks in sub-Saharan Africa faces significant challenges due to limited data, heterogeneous operating environments, and constrained capital. Current policy often relies on deterministic cost-benefit analyses, which inadequately capture uncertainty and regional variability, leading to suboptimal investment decisions.", "purpose and objectives": "This policy analysis aims to develop and demonstrate a novel Bayesian hierarchical model to evaluate the cost-effectiveness of distribution equipment, specifically transformers and switchgear, providing a robust evidence base for national asset management strategy.", "methodology": "A Bayesian hierarchical model was constructed to synthesise disparate data sources, including failure records, maintenance costs, and environmental factors. The core model structure is $\\lambda{ij} \\sim \\text{Gamma}(\\alphai, \\betai)$, where $\\lambda{ij}$ represents the failure rate for equipment type $i$ in region $j$, with hyperparameters pooling information across regions. Cost-effectiveness is measured via a posterior distribution of net present value.", "findings": "The model reveals substantial regional heterogeneity, with the posterior probability that coastal regions exhibit higher failure rates than inland regions exceeding 0.85. A key finding is that for a dominant equipment class, targeted preventive maintenance policies are cost-effective only when the predicted reduction in failure rate is greater than 30%, a threshold not currently met in most operational areas.", "conclusion": "The Bayesian hierarchical approach provides a superior, probabilistic framework for asset management policy by formally quantifying uncertainty and learning from sparse data across regions, challenging the prevailing one-size-fits-all national procurement and maintenance strategy.", "recommendations": "Policy should shift towards a adaptive, regionally differentiated asset management plan informed by probabilistic models. A pilot programme for data-standardisation is recommended to further improve model precision and facilitate evidence-based budgeting.", "key words": "Bayesian inference, asset management, power distribution, cost-effectiveness, infrastructure policy, probabilistic modelling", "contribution statement": "This paper introduces a novel application of Bayesian hierarchical modelling to power infrastructure