The cost of electricity used for pumping in water-distribution systems typically represents the largest part of the total operational costs. Therefore, optimization of pump operations is a major concern for water utilities around the world, especially in recent years with significantly increasing energy prices. Recently, to maximize cost and energy savings, pump scheduling is frequently done in real time by integrating the relevant optimization software into the water company’s supervisory control and data acquisition (SCADA) system. This, however, requires solving a complex, large, nonlinear optimization problem in a computationally efficient manner, typically in less than 1 h. To achieve this, the pump-scheduling problem is solved in this paper by a novel hybrid optimization method that uses linear programming (LP) and a greedy algorithm: LPG. The new methodology is applied to two case studies: the artificial, benchmark case study of Anytown network and the real-life pump-scheduling problem of the Richmond water-distribution network (WDN) in the UK. The results obtained clearly demonstrate that the LPG hybrid method is capable of solving real-life pump-scheduling problems in an extremely computationally efficient manner while preserving the accuracy (i.e., the near optimality) of the obtained solution. This makes the method particularly appealing for use in real-time pump-scheduling applications.
Fast hybrid optmization method for effective pump scheduling
NICOLINI, Matteo
2013-01-01
Abstract
The cost of electricity used for pumping in water-distribution systems typically represents the largest part of the total operational costs. Therefore, optimization of pump operations is a major concern for water utilities around the world, especially in recent years with significantly increasing energy prices. Recently, to maximize cost and energy savings, pump scheduling is frequently done in real time by integrating the relevant optimization software into the water company’s supervisory control and data acquisition (SCADA) system. This, however, requires solving a complex, large, nonlinear optimization problem in a computationally efficient manner, typically in less than 1 h. To achieve this, the pump-scheduling problem is solved in this paper by a novel hybrid optimization method that uses linear programming (LP) and a greedy algorithm: LPG. The new methodology is applied to two case studies: the artificial, benchmark case study of Anytown network and the real-life pump-scheduling problem of the Richmond water-distribution network (WDN) in the UK. The results obtained clearly demonstrate that the LPG hybrid method is capable of solving real-life pump-scheduling problems in an extremely computationally efficient manner while preserving the accuracy (i.e., the near optimality) of the obtained solution. This makes the method particularly appealing for use in real-time pump-scheduling applications.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
