Appendix 2: Forecasting the demand for supplying drinking water

What demand forecasting involves

Demand forecasting involves identifying the main factors and trends that influence demand for an asset or service and then preparing projections for demand over time. Once the factors are understood, mathematical modelling is often used to assess the effect of these factors on future demand. Demand forecasts can vary from simple linear trends (for example, assuming a direct linear relationship with population growth) to regression models based on historical trends in a number of factors that influence demand. Demand forecasts are considered more reliable when prepared for separate classifications of use and sectors.

There are several commonly used methods of projecting water demand. These include projecting:

• from the historical (bulk) rate of increase in water consumption;
• based on historical consumption per capita data and the projected population growth rate; and
• based on historical consumption per user category (for example, domestic, industrial/commercial) and expected changes (increases or decreases) in user category over the forecasting period.

Demand forecast methodologies can include quantitative and qualitative techniques. If possible, more than one approach to demand forecasting should be used, and the different methodologies compared.

Quantitative demand forecasting techniques often require mathematical modelling, such as regression analysis. This approach establishes a relationship between various drivers of demand (for example, demographic factors such as family size, expectations of large users, and temperature or seasonal variability), and the relationship is used to project future demand. Historical data is used to verify and adjust the model.

Qualitative forecasting can include consultation with experts to reach a consensus on the forecast demand, or a market analysis of the main drivers of demand.

Demand forecasts should be treated with some caution because factors that influence demand cannot be predicted with certainty over long periods of time. Risks and sensitivity of forecasts should be considered, including considering:

• the consequence of partial or total loss of supply;
• the effect that conservation strategies may have; and
• the effect of variations in rates income.