|M.Sc Student||Ohar Ziv|
|Subject||Optimal Booster Chlorination Stations Layout in Water|
Systems for Minimizing Disinfection by Products
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Avi Ostfeld|
|Full Thesis text|
This work describes a new solution methodology for the well known disinfection booster design problem (location and operation) in a water distribution system. In this thesis the above problem is extended and includes in addition to the customary chlorine residual concentrations constraints, disinfection by-products regulations constraints.
The methodology is based on a water quality multi-species model combined with hydraulic solver and genetic algorithm optimization procedure. Optimizing the boosters design will lead to minimize the operational cost of disinfectant injections and can improve water quality. The cost objective is calculated based on the set of boosters to be installed, their magnitude and the chlorine injected mass.
The multi-species quality model in this work solves the propagation and proposed reaction kinetics mechanism of chlorine and fictitious reactant in order to estimate chlorine decay, total trihalometane formation and their concentrations in a water distribution system.
The essence of a network can be captured by describing a periodic demand cycle. In this work we address the need to perform a close quality cycle, meaning that the concentrations of all species throughout the WDS at the end of a simulation period will be as the concentrations at the beginning of the period. Two approaches to this issue are formulated in this work. The first, Repetitive cyclical simulation (RCS), achieve cyclical closure by performing numerous demand cycles until water quality stabilization is found without the need to address the initial concentration of the species in the water system. The second, Cyclical constrained species (CCS), achieve the above in a minimal number of demand cycles by addressing the species initial concentration as a decision variables and by adding a set of constraints and penalties.
The methodology was coded and applied in MATLAB using EPANET-MSX as the water quality solver, EPANET as the hydraulic simulator and MATLAB genetic algorithm toolbox as the base for the optimization process. Both approaches are demonstrated and examined by using well known benchmark networks reported in the water distribution literature and several sensitivity analyses.
The results had shown that optimal booster configuration and exposure to TTHM are competitive objectives and when comparing the above two approaches, the CCS method was found to be less computationally expensive, to have better closure solutions, but higher operational costs.