"background": "Evaluating the cost-effectiveness of power-distribution equipment is critical for infrastructure investment in regions with constrained resources. Traditional life-cycle cost analyses often fail to adequately incorporate operational variability and epistemic uncertainty inherent in network performance data. ", "purpose and objectives": "This case study develops and applies a novel Bayesian hierarchical model to assess the comparative cost-effectiveness of different medium-voltage cable systems within a national utility context. The objective is to provide a robust, probabilistic framework for decision-making that quantifies uncertainty in total ownership cost. ", "methodology": "A case study methodology was employed, analysing historical procurement, failure, and maintenance data for cross-linked polyethylene and paper-insulated lead-covered cable networks. The core statistical model is a hierarchical linear model: C{ij = \ + \ Xij +, where Cij is the total cost for installation i in region j, \ \ N (\\, \²\) are region-specific random effects, and Xij denotes covariates. Parameters were estimated using Hamiltonian Monte Carlo sampling. ", "findings": "The model quantified substantial regional heterogeneity in cost drivers, with the credible interval for the regional effect variance parameter \²\ being 0. 14, 0. 31. One cable type demonstrated a median 17% lower predicted total life-cycle cost, but with an 85% posterior probability of being more cost-effective, not a deterministic certainty. ", "conclusion": "The Bayesian hierarchical approach provides a superior framework for cost-effectiveness evaluation by formally integrating geographical and operational uncertainty, moving beyond point estimates to probabilistic rankings of infrastructure options. ", "recommendations": "Utilities should adopt probabilistic, hierarchical modelling for asset investment decisions. Future work should integrate this model with reliability-centred maintenance schedules and expand the covariate set to include environmental stressors. ", "key words": "Bayesian inference, hierarchical modelling, life-cycle cost, power
Thandiwe van der Merwe (Mon,) studied this question.