Statistical sampling of water uses for leakage assessment via water budget

The paper describes a promising statistical sampling method of water consumptions for drafting water balances at network district scale in order to quantify the extent of leakages. It was studied the option of providing only few selected customers with perfectly working and updated metering devices, which could also be read continuously from remote, extending the results obtained to similar users so as to assess the whole district water use. Statistical sampling enables to obtain information on a given characteristic of a population through a small subset of its, called sample. Information is conveyed through a synthetic value, such as the mean of the characteristic or its total, adopting as an estimator the same function of sampled values. Sampling techniques vary according to how sample units are selected out of the population. In particular, simple random sampling and stratified sampling were compared with reference to the sampling fraction necessary to obtain estimates of the mean or total daily use with given precision and confidence level. An early theoretical application, based on real consumption data from a small homogeneous group of connections located in a district of Palermo (Italy), shows how stratified sampling entails considerably smaller sampling fractions. A later full field application of the statistical sampling methodologies to the water uses of a small municipality in Sicily (Italy) shows that they can be actually applied in technical practice, providing results with the accuracy level predictable by the theory. The method is also applicable to other purposes, different from leakages assessment, such as more general studies on water uses and consumption models.