Water Journal : Water Journal May 2011
technical features 82 MAY 2011 water demand management refereed paper and the emergent behaviour can be observed (Perugini et al., 2008). Agent-based modelling is the underlying technology behind the SimulAIt platform. Each "agent" represents a particular type of consumer. Agents are based on the human cognitive model and thus can mimic consumer behaviours and decision making. Using prescribed rules, each agent can mimic the complex behaviour of a broad range of individual consumers. Rules are constructed by integrating different types of (social, economic, environmental and political) data to describe how different people make decisions under different circumstances. The bottom-up approach is used where many agents are used to mimic the large numbers of people in the real world to simulate mass-consumers (i.e., a population). The simulation can be used to observe the emergent behaviour of the system (i.e., the forecasts), as well as the specific behaviours of individual consumers (i.e., the "why" behind the forecasts). Agent-based modelling has been successfully applied to environmental problems (Perugini et al., 2008, Rixon et al., 2007, Athanasiadis et al., 2004, Berger, 2001, Nuttal et al., 2009, Dietz et al., 2009), including the modelling of water pricing and water trading structures, and modelling household carbon emissions. Some of these models are still in their infancy. The CHW Forecasting Model Creation In 2009, ISD was appointed by the Board of Central Highlands Water (Victoria) (CHW) to use SimulAIt to generate a behavioural micro-simulation of thousands of individual households in Ballarat (and, later, in Bendigo). The initial project (of two-month duration) was a retrospective analysis of efficacy of past water conservation programs. A second project (three-month duration) prospectively modelled demand management scenarios into the future. SimulAIt's (cognitive) agents modelled individual consumers within households, how they made decisions in the house and garden, and how their decisions were influenced by different policies and communication signals, such as restrictions, prices, marketing and media communications. The model was created by integrating a wide range of complex qualitative and quantitative social, economic, environmental (engineering) and political data based on detailed demographic data (ABS 2006) and population dynamics. This included market research data, end-use studies (Willis et al., 2009), weather data, and economic and statistical data. The data was used to identify how different consumers made decisions under different situations or influences over time. The CHW forecasting model was created using a bottom-up approach and a detailed model of individual households and their occupants was constructed. The model details each area within individual households (for example, toilet, laundry), the composition of water-related products and appliances and water use behaviours, outdoor water usage and behaviours, and the change in product composition and community behaviours over time in response to demand management programs. The behavioural model was run over time (past and future) to assess the effects of various influences (scenarios) on decisions for change and decisions for usage for water consumption. Influence-Behaviour Change Model The influence-behavioural change model describes in detail how social (including economic) influences may alter behaviours, decisions for change and, ultimately, water consumption. Influences are defined by four categories in the model: • Constraints (restrictions): Behavioural change forced upon residents (no legal choice). • Barriers (financial and other measures): Financial measures including prices, rebates and taxes (if applicable). The financial measures provide a barrier (when a choice is available) for water use and product acquisition behaviour. • Obligation (severity of the situation and/or communications): Behavioural change resulting from (personal and social) obligation to reduce water consumption in response to the severity of the environmental situation (for example, current water levels or restriction levels). • Feedback (monitoring): Enables residents to obtain feedback from their behaviours, such as monthly statements and smart meters, to assist in controlling their behaviours. How influences ultimately result in behaviour change depends on messages received, as well as motivation and actions of individuals. The influence- behaviour change model describes the exposure of individuals to influences communicated (messages) through various media including radio, television, newspapers, phone, internet, billboards, or other people. Different individuals have different levels of exposure to particular media. The motivation of an individual to change behaviour is dependent on both the message and the individual, which is defined by the influence type, trust in the source of the message, message content, schedule of communication (e.g. frequency of communication), relevance of the message to the individual (e.g. garden restrictions have less relevance to those without a garden), and the level of interest (concern) that an individual may vest in the message. Individuals that are motivated to change their behaviour may translate this into action. The action (behaviour change) of an individual is determined by the causal relationship between possible behaviours and motivation. The extent to which behaviour changes (amount and persistence) is largely dependent on the level of motivation, the level of Figure 1: Historical demand forecasts (red) versus actual demand (grey) in Ballarat.
Water Journal April 2011
Water Journal July 2011