A Data-Informed Physical Asset Management Model for Maintenance Strategy Selection Under Reliability, Risk, and Uncertainty: A Scenario-Based Approach

Abstract

Ensuring high levels of reliability and availability of industrial assets requires effective integration of maintenance decision-making within asset management frameworks. However, most existing approaches treat asset criticality, reliability assessment, and predictive maintenance separately, limiting their applicability in complex and uncertain decision environments. This paper proposes an integrated data-informed decision-support framework for maintenance strategy selection that combines asset criticality, reliability modeling, and predictive information within a unified structure. The proposed method integrates a fuzzy analytic hierarchy process (Fuzzy AHP) for asset criticality assessment, a condition-adjusted Weibull reliability model, and a hybrid remaining useful life (RUL) indicator derived from condition monitoring data, degradation trends and predictive confidence. These elements are aggregated into an Integrated Maintenance Decision Score (IMDS), enabling consistent classification among corrective, preventive, and predictive maintenance strategies. The framework is evaluated using a scenario-based case study representing different operational conditions, degradation levels, and data availability contexts. Results show that IMDS provides stable and interpretable maintenance classification across scenarios, clearly differentiating between corrective and predictive regimes depending on asset condition and failure risk. Monte Carlo-based uncertainty propagation further confirms the robustness of the decision mechanism, with narrow confidence intervals and high Decision Robustness Index (DRI) values. Sensitivity analysis identifies asset criticality and probability of failure as the dominant drivers of decision outcomes, while predictive confidence acts as a stabilizing factor under uncertainty. The principal contribution of this work is the development of a transparent and modular decision-support architecture that explicitly integrates asset criticality, reliability modeling, predictive condition assessment, and uncertainty propagation into a single maintenance strategy selection framework. Unlike existing approaches that typically address these dimensions separately, the proposed IPAM framework provides an explainable and condition-aware mechanism for selecting corrective, preventive, or predictive maintenance strategies.