Existing studies paid most attentions to the strategic
planning of building retrofitting project. During the actual
operation, the overall energy performance of the retrofitting
project deteriorates over time owing to the usage and failures
of the retrofitted facilities. For the sustainability of energy
efficiency, the maintenance plan optimization for the retrofitted
facilities must be considered, which suggests the possibility
to sustain the building energy efficiency during operating.
However, the overall energy performance reveals strong time
dynamics subject to the failures and maintenance actions,
which make the maintenance plan optimization a complicated
problem. This paper characterizes the dynamics of the totality
of the retrofitted facilities by the multi-state system models
governed by a discrete-time Markov process. The impacts of
both preventive and corrective maintenance actions on the
over energy efficiency of the plant are then quantified by the
present model. The optimization problem is thus casted into a
stochastic optimal control problem which aims at maximizing
the long-term energy efficiency and financial payback. An
MPC based approach is employed to solve the present control
problem. Finally, a case study is conducted to demonstrate the
effectiveness of the present approach.