Date of Award

2019

Publication Type

Master Thesis

Degree Name

M.A.Sc.

Department

Civil and Environmental Engineering

Keywords

ANN, Artificial Neural Networks, Demand Forecasting, Energy Conservation, Smart Water Grid, Water Distribution networks

Supervisor

Rupp Carriveau

Supervisor

David Ting

Rights

info:eu-repo/semantics/openAccess

Abstract

The unyielding interconnection between water and energy has made demand forecasting a necessity for water utilities. Electricity prices driven by the time of use has impelled water utilities towards short-term water demand forecasting. The progressive new Smart Water Grid platform has helped water utilities in utilizing their Water Distribution Networks. This two-way platform has provided developers and decision makers with robust models that rely on consumer feedback. Among these models is the water demand forecasting models. Multitudinous demand forecasting methods have been developed but none have utilized model implementation practicality. Utilities differ in size, capacity, and interest. While small size utilities focus on model simplicity, larger utilities prioritize model accuracy. This work focuses on a water utility located in Essex County, Ontario, Canada. This study presents three papers that focus on investigation and evaluation of short-term water demand forecasting techniques. The first paper compares water usage between two crops (tomatoes and bell peppers) in an effort to evaluate a crop to crop forecast technique that relies on one crops watering data in order to produce forecasts for another crop, The second paper examines the effect of model type, input type, and data size on model performance and computational load. The third paper proposes a new methodology where model performance is not sacrificed for model simplification.

Download

Share

COinS