Preventive maintenance (PM) policies for repairable systems are commonly modeled using static restoration assumptions, despite growing empirical evidence that maintenance effectiveness degrades with system age. This study develops a non-parametric, age-dependent preventive maintenance effectiveness framework for repairable systems operating under imperfect maintenance. Building upon a virtual-age non-homogeneous Poisson process (NHPP) model with multiple PM types, the proposed approach estimates age-specific restoration factors directly from field data without imposing restrictive functional forms. The model is applied to a fleet of underground load-haul-dump (LHD) trucks using real-world failure and maintenance records from the mobile subsystem. Preventive maintenance intervals are subsequently optimized with the objective of maximizing system availability under operational feasibility constraints. Results demonstrate that maintenance effectiveness varies substantially across age bands and PM types, with light PM actions exhibiting greater sensitivity to aging effects than moderate or major interventions. The optimized age-dependent PM schedules differ systematically from static OEM policies, particularly in mid- and late-life operating regimes. These findings highlight the limitations of constant-effectiveness maintenance models and illustrate the practical value of incorporating age-dependent restoration behavior into preventive maintenance planning for complex industrial systems.
Uthman Said (Fri,) studied this question.