This paper presents the application of an offset-free model predictive controller (OF-MPC) for an energy efficient operation of the central chiller plant in a case study hotel on a tropical island. The OF-MPC performance is simulated under the mismatch between the real system's dynamics and the controller's internal model, the incidence of stochastic unmeasured disturbances, and different weather conditions and building occupancies over weekly periods throughout seven months. This controller is compared with two frequently used open-loop control methods (i.e., Baseline and Conventional). The simulations suggested that the proposed OF-MPC is able to track the set-point temperature of the worst-case reference room with lower thermal comfort violations than the Conventional. Furthermore, there is a significant energy savings potential for the OF-MPC compared to both strategies, which is deteriorated under extreme weather conditions. The promising OF-MPC ensures that the cooling capacity delivered by the central chiller plant can match the actual building cooling load, making the chilling system operate energy efficiently. (C) 2016 Elsevier B.V. All rights reserved.