Portfolio-Level Reliability Analysis of a Multi-Asset System: A Frailty Proportional Intensity Modeling Framework

Artikel i vetenskaplig tidskrift, 2026

Understanding and modeling the equipment failure behavior is crucial for reducing downtime, optimizing maintenance strategies, and improving operational safety. This challenge becomes particularly critical in industrial contexts involving multi-asset portfolios, where heterogeneous conditions influence the reliability of individual assets and, consequently, the system as a whole. This study proposes a structured, data-driven framework for reliability analysis of multi-asset portfolios in large-scale operations. Building on this framework, a frailty-based proportional intensity model is applied to a real-world case study of a mining transportation system consisting of 81 trucks operating in different sites with distinct operational and environmental conditions. Model parameters are estimated using penalized partial likelihood, and model adequacy is evaluated using likelihood ratio tests (LRT), Akaike Information Criterion (AIC), and Schoenfeld residual diagnostics. The LRT results indicate that the frailty term is significant (θ = 0.50; p < 0.001), confirming the presence of unobserved heterogeneity and supporting the use of the frailty-based model. The analysis identifies key risk factors, including operator skill, working shift, haulage distance, slope direction, and road condition, as major drivers of failure in the case study. The proposed approach provides an improved model fit and a more realistic representation of failure processes in heterogeneous multi-asset systems. These findings also provide practical guidance for managers by enabling targeted interventions, such as operator training, road condition improvement, and haulage route optimization, to reduce failure intensity and enhance system reliability. This study contributes to the reliability literature by extending frailty-based modeling to portfolio-level reliability assessment under heterogeneous conditions. This integration promotes a system-wide perspective on reliability analysis and decision-making beyond traditional single-asset approaches.