A novel technique for solving a dynamic optimal chiller loading problem is presented. This method reduces the complexity of the dynamic problem by considering all chillers to be a single, optimal chiller, which significantly reduces the number of decision variables. A static optimal chiller loading problem is solved as a sub-problem to give the optimal total power required at each time interval. The static optimization problem is solved by choosing the best solution from a series of convex quadratic programming problems, thus ensuring global optimality at each time interval. This hierarchical structure takes advantage of the extra degrees of freedom provided by thermal energy storage, while effectively breaking the problem down into a much simpler problem. This methodology is applied to a district cooling system in Austin, TX. Results are compared for three operating strategies: equal ratio chiller loading, static optimal chiller loading with no storage, and dynamic optimal chiller loading with storage. The problem is solved with two different objective functions: to minimize total energy consumption and to minimize total cost over a 24-h time horizon. The results show energy savings as high as 9.4% and cost savings as high as 17.4% when applying a conservative time-of-use pricing structure. (C) 2012 Elsevier Ltd. All rights reserved.