|Ph.D Student||Price Eyal|
|Subject||Inclusion of Reliability and Leakage in Multi-Objective|
Operation of Multi-Quality Water Distribution
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Avi Ostfeld|
|Full Thesis text|
Pump scheduling is a major consideration to the operational costs of a water distribution system. Optimal pump scheduling to minimize system operational costs, must take into account many considerations such as changing electrical tariff rates over time, reducing water leakage from the system, considering the cost of water from different sources, changes in pump energy consumption according to changing hydraulic conditions, reducing pump wear be minimizing pump operation switching, minimum or maximum service pressure constraints along the network and water quality constraints. The pump-scheduling problem is typically solved by using techniques such as evolutionary algorithms (EA), linear programing (LP), non-linear programing (NLP) or mixed integer programing (MIP). The main drawbacks of the above techniques are long solution times (EA, NLP, MIP), complex hydraulic modeling (NLP, MIP), and the returning of a non-discreet pump scheduling solution to discretely operated pumps (LP, NLP). In most cases the existing techniques deal with only part of the supply considerations. In this research two primary optimization algorithm were developed. The first, is a successive linearization program algorithm (SLP) based on linear programing (LP) which has short solution times and global minimization. The non-linear, convex, equations for headloss, water leakage, dynamic pump total head (TH) are linearized and solved using successive linear programing. The complete SLP algorithm developed can find the optimal pump scheduling for a water system, using water balance constraints, hydraulic constraints (min/max water head constraints throughout the network), minimizing leakage, dynamic pump energy calculation (according to resulting TH in the system) and source cost fines. The algorithm demonstrates short solution time, and converges to global minimum. The second, is an operational graph optimization (OGO) algorithm combining graph theory, to optimally find pump scheduling, and the hydraulic solver EPANET2.0, to deal with the hydraulic and water quality analyses. A skeletonized representation of the key control elements is held in the form of a directed graph. The operational graph is used to locate the optimal pumping units to operate to satisfy service constraints in conjunction with the hydraulic solver. The water quality is controlled by reducing operational storage volumes to reduce water age. The algorithm demonstrates short solution times and includes water balance, headloss, water quality, discrete pump operation and dynamic pump energy.