Analysis of Flow and Ambient Sound Data to Identify the Microcomponents of Domestic Water Consumption for Large Households

Conventional approaches to microcomponent identification are typically confined to analysing limited periods of flow data (e.g. selected one to two week periods) for houses with up to 5–6 occupants where clear repeating patterns in the flow signal can be identified and associated with particular water use devices. However, this approach is not feasible when there are large numbers of occupants in single households (e.g. 15–20 occupants for extended families) due to the complex nature of the flow signal. In order to address these limitations, two innovative developments were undertaken. Firstly, sound frequency monitoring of water use devices in multiple rooms in a single house in Kuwait was undertaken, and this additional information was used to enhance microcomponent identification. Secondly, new data processing and microcomponent identification software was developed which can utilize both flow and sound frequency data for microcomponent identification, and which also allows the automated processing of large data records for extended periods. A probabilistic approach to identification was developed which quantifies the uncertainty in attributing a sound signal to a water use device. A procedure was also developed whereby the results obtained from detailed flow and sound data from one house were extrapolated to a set of 18 houses for which flow data only were available. The main result from this study is that the mean consumption for the set of 19 Kuwaiti extended family houses in the study area is 246 ± 27 (95% CI) litres per capita per day (lcd) and the largest use is for showering (31%).