Exploration of the relationship between water main breaks and temperature covariates

Water utilities (especially in colder climates) often experience an increase in water main breaks in colder seasons. Some observers argue that this increase largely occurs during the period when there are sudden and prolonged changes in water and air temperatures, which typically occur during the late fall to early winter (temperature drop) and late winter to early spring periods (temperature rise). This paper examines the impact of temperature changes on observed pipe breakage rate for three pipe materials, namely, cast iron, ductile iron and galvanised steel. Several water and air temperature-based covariates were tested in conjunction with a non-homogeneous Poisson pipe break model to assess their impact on water main breaks, using data sets from three different water utilities in the USA and Canada. Temperature-based covariates, such as average mean air temperature, maximum air temperature increase and decrease, and how fast the air temperature increase and decrease over a specific period of days, were found to be consistently significant. While the availability of water temperature data (which most utilities do not have) can enhanced the prediction of water main breaks, it appears that air temperature data alone (which most utilities can access) are usually sufficient.