Evaluation of In-Service Residential Water Meters: Analysis of Registration Error and Metering Infrastructure Upgrades

The American Water Works Association (AWWA) and the International Water Association (IWA) have designated the volume of water not registered by water meters as a form of "apparent loss" in a distribution system. The term apparent is given because this volume is not technically a water loss, as is the case of wasted water from real leaks in the distribution system. Large volumes of apparent losses hurt the revenue of utilities that rely on water metering to bill their customers. This is critical to utilities given that billed consumption is often the main source of income to provide adequate service. This form of apparent losses is a challenge to water management, particularly, in the case of significant drought because of the uncertainty about the real volume of water consumed. Although the impact of apparent losses from a single residential service connection is not as significant compared to an industrial meter with low accuracy, the cumulative effect of apparent losses across residential users can be very significant.

Until the early 2000's water utilities in the U.S. relied on mechanical water meters to measure residential water use. Since then, electronic meters with higher accuracy at low flow rates have been developed. Data collection from meters has also evolved as well, from the manual reading by an operator, to drive-by systems and most recently to remote readings using a network of transmitters/receivers (i.e., advanced metering infrastructure or AMI).

An expectation of this dissertation is that it will help water utilities to have a better idea of the volume of apparent losses due to metering inaccuracy (i.e., registration error) and provide insights into the effects of installing AMI systems to residential metered water (MW). To achieve this goal, two main objectives are fulfilled 1) to expand on the knowledge of registration error (RE) in mechanical nutating-disc (ND) meters used to monitor residential consumption, and 2) to evaluate the impact of metering infrastructure upgrades on the volume of metered water (MW) from residential service connections. This dissertation follows the manuscript format with three journal articles constituting the main chapters after a general introduction characterizing the issues in Chapter 1.

Chapter 2 is an experimental study that evaluates the influence of service time (ST) and volumetric throughput (TP) on the accuracy of ND meters within the recommended flow rates set by the U.S. water industry for meters with an internal diameter of ⅝-in. (15-mm). Over 300 meters removed from service were tested for accuracy. Key findings of this study are 1) ND meters that have been in service over 25 years have a greater likelihood of poor accuracy at the minimum recommended flow rate (Q^min) of 0.25 gallons per minute (gpm) (57 liters per hour (L/h)) and 0.5 gpm (114 L/h) independent of TP, and 2) comparison with data from accelerated laboratory testing showed that simulated use may not necessarily reflect the actual performance of ND meters in service, particularly, at 0.25 and 0.5 gpm.

Chapter 3 is an experimental study that investigates REs of ND meters below the minimum recommended flow rate (Q^min = 0.25 gpm), particularly, at ½, ¼ and ⅛ of Q^min. Over 100 meters removed from service were tested in this study. Key findings of this study are 1) confirmed how performance decreases with reducing flow rate below Q^min, 2) of the variables considered, TP was found to be a better indicator of RE at Q_(1/8)^min up to an approximate meter reading of 0.66 MG (2.5 ML) compared to ST for 10 ≤ ST ≤ 24 years, with minimal influence at Q_(1/4)^min and none at Q_(1/2)^min, and 3) a strong linear relationship was found between RE at Q_(1/2)^min and RE at Q^min independent of TP or ST.

Chapter 4 is a study that evaluates the extent to which the implementation of a new AMI system combined with a system-wide installation of new ND meters impacted the volume of MW from residential service connections of a 22,000-person municipality in southwest Virginia. Time-series analysis techniques were employed to evaluate changes in the trend of bimonthly MW and median daily MW over a six-year period. Key findings of this study are 1) the AMI system improved the accountability of MW for the utility, 2) despite an ongoing downward annual trend in MW, average bimonthly MW mildly increased after the AMI system was fully operational, and 3) annual MW increased by 2.2% in the 12-month period immediately following the metering infrastructure upgrade.