Agent-based modelling as a decision support tool for water resources planning and managemen

The research presented in this thesis focuses on developing a water-use forecasting method using an agent-based model. The novelty of this method is that it allows for a given population to be represented heterogeneously and that the household are connected amongst themselves and can transmit information and modify their behaviours. The model is showcased using a case study in Kingston, Ontario where households are modeled at the individual level as agents that separate water use into 6 household fixtures. Agents are given a set of attributes that enable them to make decisions and adapt behaviour based on social-networks and communication. Available data from Statistics Canada is used to characterize 40 neighbourhoods within Kingston, Ontario. The modelling framework is utilized to test population responses and potential water savings achieved through conservation campaigns. The research consists of evaluating the change in water demand using the model and running the results into a pipe-network hydraulic solver (EPANET 2.0) to calculate the change in energy use from the distribution system associated with conservation programs. The model and the case study are used to answer a series of research questions concerning the sensitivity of the ABM to social communication parameters, the potential water and energy savings that are achievable in the Kingston distribution system and finally, how the spatial distribution of water savings affects energy savings in the Kingston system.