Group for Research in Decision Analysis


Long-Term Fleet Maintenance of Hydroelectric Generators Using Proportional Hazard Model and Nonlinear Programming

, , and

This paper presents a framework to determine optimal maintenance planning of a fleet of complex and independent systems. They are made up of several major components which operate in different environment conditions, and are built with different technologies. This framework uses proportional hazard model (PHM) to characterise the failure rates of components and the effects of the environment conditions and the load levels. A nonlinear programme is developed to minimise the fleet maintenance cost under age replacement policy of its components and a set of organisational and technical constraints. Lindo API and NOMAD are used to solve the nonlinear model. The framework is applied to set a preliminary plan to overhaul a fleet of 90 hydroelectric generators in 6 power plants over 50 years. Sensitivity and performance indexes are built to interpret optimisation results in two settings: normal and 50% increase in load.

, 20 pages