How a Large Texas Utility Leveraged Data Analytics to Optimize their Flow Metering Maintenance Approach

After a deficiency was experienced in one of their lift stations, the North Texas Municipal Water District requested for their wastewater system meters to be evaluated for accuracy and functionality for 141 of their billing meters, of which 90 were for billing purposes. The goal of the evaluation was to evaluate the accuracy and functionality of the existing meters. Using data recorded at five-minute increments from January 2022 to March 2024, meter downtime analysis, flow balances, and comparisons with billing data were conducted. Wastewater systems use meters to record flow rates of wastewater flowing throughout to determine how much to bill customers. If meters are calibrated accurately and placed in strategic locations, a clear picture on how the system cascades can be painted. If not, unmetered flows from undeterminable sources occur leading to potentially inaccurate billing of customers. A combination of statistical methods was employed to assess the accuracy of each meter. First, flow data was compiled using Python scripts to organize data for easier processing in the future and graph raw data and calculate daily averages. Daily averages were calculated using a rolling 24-hour sum to eliminate noisy data. Next, the meter data and monthly billing data were graphically compared and triaged into layers of severity based on discrepancies noticed. The severities ranged from a few months of no readings or spikes to obvious magnitude differences, with highest severity noticed in meters that were not metering over the two years of data. By triaging the meters, the municipality would know which meters to target first for maintenance and which to ignore in an efficient manner. Then, a meter downtime analysis was conducted to determine which meter would need to be properly calibrated or replaced. The percentage of how often the meter was down, which was determined by a threshold decided upon after manually inspecting the data, was calculated and used to highlight meters that needed maintenance or replacement. Finally, a flow balance of the system was conducted. Three general cases were identified. The ideal case was when the basin is under metered by around a 10% to 20% margin due to inflow and infiltration in the gravity mains. The other two cases were over metering and under metering. Over metering was noticeably caused by possible double counting by meters near a treatment plant or inaccurate metering and warranted a recalibration if the problems occurred more recently or replacement if it had been a constant issue. Under metering was noticeably caused by unmetered residential flows in between meters, significant down time, which was verified by the meter downtime analysis, or inaccurate metering which warranted a recalibration, replacement, or relocation of a meter. The presentation is intended to inform the methods on processing and assessing data from existing flow meters to assist a municipality in managing their assets.