In the refinery industry, especially in large-scale modern refineries, production scheduling modeling is an essential tool in the operation and management. By scheduling systems, better economic performance can be achieved. In order to implement the scheduling schemes, the influence of control systems has to be considered in scheduling models. This paper presents a novel refinery scheduling model where several controllable and realizable operation modes for production units are established based on unitwide predictive control. Furthermore, operational transitions resulting from switching operation modes of production units are taken into account to reflect the dynamic nature of production. A novel mixed integer linear programming (MILP) discrete-time model is developed, and computational tests show the efficiency of the proposed model.