M.Sc Thesis

M.Sc StudentShteynberg Denis
SubjectMeasurement and Simulation of the Diurnal Pattern of
Domestic Water Demand on Micro-Component Scale
DepartmentDepartment of Civil and Environmental Engineering
Supervisor PROF. Eran Friedler
Full Thesis textFull thesis text - English Version


Detailed knowledge of the water end-point usage (WEPU) such as shower, dishwasher, etc., can be important for water demand management, prediction of endpoint water-use and evaluation of onsite water reuse potential (e.g. greywater).

Organizations have come to the realization that it is almost impossible to evaluate the effectiveness of residential water demand management schemes without accurate and appropriate measurement of actual water consumption at an end-use or micro-component level. Measuring the flow of individual water-using appliances, although looks simple, is not straight-forward task as it involves interfering with the in-house plumbing. Due to this complexity, for years hardly any study was performed on a micro-component level. In recent years, methodologies were developed to enable derive individual appliances water-use from a single measurement point and advanced data analysis techniques. Several techniques such as SIMDEUM, HydroSense, Identiflow, Flow Trace Analysis and "classic" Diary Surveys/Questionnaires, have been used in the past for that purpose, each with its own limitations, mainly fair accuracy, expensive and inaccessible for public use.

In this study, new pattern recognition system was designed, based on smart flow meter and programmable logger controller (PLC), which were able to measure and record domestic water flowrates using only one sampling location at the water inlet of each house. WEPU is classified out of the acquisition system’s data using an adaptive decision tree (DT) developed as part of this research. The DT developed is able to identify specific properties of the in-house water-using appliances such as flow pattern fitting (recognition) for toilet flushing identification, and recursive event recognition for automated appliances (dishwashers & washing machine) identification. The overall accuracy of the DT-based algorithm was 91%, best for faucet identification (94% accuracy) and worst for washing machine identification (58% accuracy).  In order to train the system, real water usage data from 16 domestic houses in the north of Israel was obtained and compered to diary surveys filled during a one week period by families who volunteered to participate.

The study reveals that the average in-house water usage in the participating homes was 104 liters/capita/day (L/c/d) and 30 events/capita/day (E/c/d). The most water consuming appliance was the shower with 42 and 66 L/c/d in weekdays and weekend days respectively. Faucets were found to be the most frequently used with 17 and 23 E/c/d in weekdays and weekend days respectively. Nevertheless, faucets use consume only 8 L/c/d, comprising 7.7% of the daily in-house water volume used. The data analyses all aspects of the in-house WEPU and revel diurnal usage distribution as well, by week days/weekend day's differences.