Real-time optimization of water supply system operation / Otimização energética em tempo real da operação de sistemas de abastecimento de água

AUTOR(ES)
DATA DE PUBLICAÇÃO

2009

RESUMO

This work presents a computational model for real-time optimization of water-distribution networks operation. An integrated software tool has been developed which is composed of three main modules: (i) a hydraulic simulator that performs the extended period simulation of the system (EPANET); (ii) a short-term demand-forecasting model, based on the moving Fourier series; and (iii) an optimization module, using genetic algorithm, to minimize pumping costs. The optimization process is divided in two routines, the off-line optimization and the on-line optimization. The aim of the first routine is to find the ideal operation levels for the reservoirs which are used as guidance for the on-line optimization. The schedules to be implemented in the network are determined, however, by an on-line optimization which runs a new optimization processes at each SCADA (Supervisory Control and Data Acquisition) update. In the off-line optimization, near-optimal pump and valve settings for a selected operating horizon are found on the basis of the typical 24-hour water demand cycle in which each reservoir must also start and end with the same water level. The levels of the reservoirs at the end of each hour in this optimization process, called ideal levels, are considered in the on-line optimization as a constraint that represents a penalty added to the objective function whenever a reservoir level is below the ideal level at the end of each hourly operation. The on-line optimization starts updating the state of the network transmitted from the SCADA system and the demand forecast for the next 24-hour period. Near-optimal pump and valve settings for a selected operating horizon are then determined but only the control settings for the next hourly time step are used to implement the operation. At the next SCADA update (next-time step), the whole process is repeated on a continuous basis and a new operating strategy is computed. In this way, the control process moves forward in time, correcting any discrepancies as it progresses. This integrated software tool has been applied to a small example network. The results obtained from this application have shown that the proposed model offers a practical tool for finding feasible operation strategies for a water pipeline system, obtaining reductions in pumping energy costs.

ASSUNTO(S)

otimização em tempo real operação de sistemas de abastecimento de água eficiência energética genetic algorithm real time optimization water-distribution networks operation algoritmo genético energy costs

ACESSO AO ARTIGO

Documentos Relacionados