Making better predictions of peak water demand

The summer of 2018 was an extremely dry one and 2019 is on course to follow suit. A lack of precipitation has major consequences for our water sources: groundwater levels fall and low river water levels allow salt-water wedges to advance a long way inland. At the same time the need for water is greater, for watering the garden, for example, for filling the swimming pool or for taking more frequent refreshing showers. Working closely with water companies WML and Brabant Water, KWR investigated the factors that determine the demand for water on a given day. The water companies now have greater insight into the connection with garden area, number of residents and number of swimming pools, for example, and can respond better to future peaks in the demand for water.

Climate change and the water company

The Dutch national weather service KNMI expects dry periods such as those in 2018 to become more commonplace and maybe also more intense because of climate change. How much water will we want to use in such a future summer? This is something the water companies would like to know because some of the infrastructure needed to provide us with water in the future is already under construction. The required capacity of the infrastructure depends mainly on the peaks in demand for water. But: how much more capacity do we need exactly? And also: what causes peaks in the demand for water? Can knowing these things help limit water consumption on peak days?


Brabant Water, WML and KWR have carried out a joint research project into this in two supply areas, one in Brabant (de Peelkant) and one in Limburg (Plasmolen). By entering weather data and recorded water consumption for the last 20 years into a machine learning model, correlations can be found between temperature or precipitation and water consumption, for example. What did we find? More than temperature, it is evaporation that is a predictor of our water consumption. Weather data, such as temperature, precipitation and evaporation, can therefore be used to predict peaks in water consumption. Taking the KNMI’14 climate scenarios, in which the consequences for temperature, precipitation and evaporation were outlined for four climate scenarios, as a basis, the computer model then gives predictions of future water consumption. We can use this as a basis for calculating the level of the peak factor that occurs once every ten years, the so-called day peak factor.


The calculations show that peaks in water consumption will be higher in the future, especially in the summer. Thanks to this research we now have a better idea of how much higher they will be. Water companies are incorporating this knowledge into their planning process: for example, WML is provisionally working on the basis of an increase in the day peak factor of 10% by 2050 compared with now. The expected increase in peakdemand is placed alongside other trends, such as population growth or contraction, water conservation measures and economic growth. Water companies can use the overall sum to determine what increases in water infrastructure are needed. If these increases are to be ready in 15 years’ time, the preparations for a water company start now, so that customers never find themselves without water.