Markov-Based Optimization Model for Building Facilities Management

Abstract

An effective and efficient building facilities management system calls for a systematic performance-based methodology. This paper proposes an integer-programming model based on the Markov decision process: Markov chains are used to model the change of condition index, a consistent scale to measure building performance, and integer programming employed to optimize annual management actions and annual budget allocation, subject to various types of constraints. A wide scope of useful outputs can be obtained from this model. This includes: (1) annual management actions to take; (2) annual budget allocation; (3) the expected condition index (CI) values at the beginning of each year before and after management actions are taken; (4) the expected annual performance levels; and (5) sensitivity analysis and corresponding outputs to different budget scenarios, for all the elements, components, systems and buildings of a building network and for each year over a long time horizon. As a general model, it can be applied in any type of building network.