Located northeast of Atlanta, Georgia, the Gwinnett County Department of Water Resources (GCDWR) manages a service area with a population of 950,000 (Figure 1). GCDWR has a non-revenue water (NRW) rate of approximately 10%, which is considerably lower than 30%, which is what Pinney et al., (Pinney, 2018) reported was the average NRW for a sample of US utilities. Non-revenue water is water lost in the distribution system before reaching a customer meter, either through real losses (primarily leaks) or apparent losses (primarily metering inaccuracies where some amount of water passes through the meter without being registered). Leaks on the customer side of the meter are not considered non-revenue water because they have been measured by the meter, and in fact, these leaks add revenue at the expense of wasted water, missed conservation goals, and potential property damage.
FIGURE 1.
AMI Pilot System Architecture
Even though its NRW losses were lower than similar utilities, GCDWR implemented a pilot project to further reduce its water losses using Advanced Metering Infrastructure (AMI). The specific technologies implemented during the study were for flow monitoring, hydrant pressure sensors, cellular communications, and advanced data analytics. Data was collected from August 2017 through July 2018. In the end, the AMI pilot study heled identify leaks on the utility and customer sides of the meter and improved security by generating backflow and tamper alerts. The AMI system provided insights into the distribution system that allowed GCDWR to further reduce water loss and improve its response during periods of insufficient rainfall, thus increasing the utility’s resilience to drought conditions. The pilot study also demonstrated the reliability of using a cellular network for AMI applications.
OBJECTIVES
GCDWR’s primary objective with its pilot project was to understand how best to integrate proven technologies with its internal practices to further reduce water loss. NRW is attributed to multiple causes including leaking or broken pipelines, water theft, and poorly operating meters, but typical NRW projects focus on or at least begin with addressing leaks and broken pipelines. On the other side of the meter, customers can experience water loss through leaks in irrigation systems, water heaters, toilets, and other parts of the premise plumbing. Minimizing distribution system and customer water losses are important parts of improving a utility’s overall resiliency because it can allow a utility to build and conserve its reserves during water shortages caused by drought without requiring additional sources of supply.
Another study objective was to demonstrate the effectiveness of cellular technology for communication between the AMI sensors and the data management system. When the pilot project was originally conceived a few years prior, cellular technology for AMI was not considered mainstream. Since then, cellular communication for AMI has been tested in multiple pilot applications (including the GCDWR’s 1-year evaluation described herein) and has now been implemented by many utilities across North America. This is not surprising because cellular communication protocols have been used for other types of Internet of Things (IoT) field sensors for well over a decade.
Over the last 3 years, USEPA’s Water Security Division has evaluated the use of AMI as a supplement to the Physical Security Monitoring component of a Surveillance and Response System (SRS). Thus, the third objective of the AMI study was to improve system security by generating alerts for backflow and tampering incidents. Meter tampering includes any actions that prevent the meter from communicating with the utility or reading the flow properly. EPA Water Security Division members were invited to join the pilot project to evaluate the backflow and meter tampering data streams, and EPA experts conducted a resiliency and security workshop with GCDWR. EPA also funded the development of advanced analytics for evaluating backflow and tampering data.
COMMUNITY OUTREACH
Prior to installing the AMI meters, GCDWR reached out to customers by placing door hangers on each home in the test area and hosting a community meeting. The door hangers explained the purpose of the project and that all residents in the pilot neighborhood would have a new water meter installed. At the community meeting, GCDWR personnel displayed an AMI meter as a hands-on exhibit and described the project in detail by discussing the sensors being installed (residential meters, district meter, and pressure sensors) and the data they would collect. GCDWR responded to questions and explained that the project would improve water loss control efforts while helping the utility to learn more about the latest AMI technologies. GCDWR personnel also explained that the pilot is a district-metered study, meaning that all homeowner’s meters were required to be connected to the AMI system to account for all flows and identify losses. Approximately 12 residents attended the meeting and they were generally supportive of the project after the outreach.
IMPLEMENTATION
The pilot AMI system was tested in a neighborhood that could be served by water routed through a single pipeline to form a District Metered Area (DMA); a DMA is a discrete area within a distribution system where flow and pressure can be strictly segregated and controlled from the rest of the system. Other pipeline connections to the neighborhood were valved off for this study after fire flow tests were performed. A flowmeter was installed on the incoming pipeline to serve as the DMA meter, and residential meters were provided for all service connections in the pilot neighborhood including 502 residential homes and two meters serving neighborhood amenities (e.g., community center). Pressure sensors were provided on four fire hydrants in the pilot area to capture pressure profiles.
SENSORS
Each residential and DMA meter measured flow rate and direction (i.e., forward or reverse). Residential meters also had the ability to generate alerts for backflow and continuous and intermittent leaks. A single check valve was installed upstream of each residential meter to prevent reverse flow.
The DMA and residential meters used a Cellular Meter Interface Unit (CMIU), which used a CAT3 chip on a 4G cellular network to wirelessly transmit data from the meter to the data management system. The CMIU also generated tampering alerts by detecting if the cable to the meter had been disconnected or if the CMIU had been moved based on the GPS function built into the CAT3 chip. The DMA and residential meters recorded flow every 15 minutes, stored the data internally, and transmitted the accumulated data to the data management system every 6 hours via the CMIU. If the meter is disconnected from the CMIU, the meter continues collecting and storing data and transmits the accumulated values to the data management system after the connection is restored so that there are no data gaps.
The data transmission frequency for CMIUs is typically once every 24 hours, but a six-hour transmission frequency was used during the study to test the durability of the CMIU batteries. (Note that some newer CMIUs have a default transmission frequency of six hours.) Alerts generated by the meter and CMIU were sent immediately to the data management system rather than waiting for the next scheduled data transmission. The DMA meter was a full-bore electromagnetic flow sensor. Residential meters consisted of 448 ultrasonic meters and 56 positive displacement meters.
Four of the 51 hydrants in the pilot area were equipped with a pressure sensor. Each pressure sensor performed a measurement every second, stored the data internally, and transmitted the accumulated data to the data management system every hour. The relatively frequent pressure measurements were required for identifying pressure transients.
DATA MANAGEMENT
All flow and pressure sensors used cellular technology to wirelessly transmit measured values to the data management system. The flow sensors transmitted their data to a server hosted by the CMIU manufacturer, while the pressure sensors transmitted their data to a server hosted by their manufacturer. Using secure socket layer connections over the Internet, the flow and pressure servers communicated with the data management system, which consisted of a geographic information system (GIS), analytic, web, and operational data storage servers. The system architecture is shown in Figure 1.
The data management system stored and applied advanced analytics on the incoming data and provided information to the dashboard including:
Flow values for the entire pilot area
Individual flow information for user-specified meters
Pressure values from each of the four sensors in the pilot area
Alert information
The pilot evaluation team used the dashboard to analyze data and investigate alerts and unusual flow patterns. Installation of the flow and pressure sensors began in June 2017, and data was collected from August 2017 to July 2018.
DATA ANALYSIS
Each day after checking for alerts and pressure incidents, the field team reviewed the dashboard for the highest residential flow values as well as trends in the DMA incoming flow and total metered flow. The dashboard GIS overlay allowed staff to determine whether alerts were geographically clustered, which would indicate a system-level issue rather than an isolated incident. Trends were used to investigate pressure spikes, unusually high flows, and leak alarms from the meters, which were useful when determining if a water loss or backflow incident was occurring.
DASHBOARD DESIGN
The dashboard included a map view that displayed flows and pressures as well as alerts and their locations. The dashboard included a “Usage Details” tile that showed the 25 highest flows in the system in descending order of water use. Clicking on any of these accounts provides user information and usage trends at that location. An orange icon on the area map showed the location of any alerts, and alert thresholds were adjusted over the course of the pilot to minimize false positives. The alerts and their thresholds were as follows:
Pressure alert. Detected a >10 psi pressure drop or spike
Tamper alert. The cable between the meter and CMIU was disconnected resulting in loss of signal
Continuous leak alert. Detected a non-zero baseline for a period longer than 24- hours.
Intermittent leak alert. Detected unusual flow characteristics over the past 24- hours.
Backflow alert. Detected > 1 gpm of reverse flow, which is higher than the normal expansion of water from a water heater back through a meter.
Non-reporting meter alert. The data management system has not received data from a meter for >30 hours, which indicates a communications outage or battery failure.
The dashboard is shown in Figure 2.
FIGURE 2.
Dashboard Map View
RESULTS – WATER LOSS REDUCTION
Water loss on the utility side of the meter (NRW) was calculated by comparing DMA flow to the total of residential flows. During the study, the pilot area’s NRW was approximately 4%, which is far below the United States average. Each type of water loss can be correlated with pressure to identify the root cause of the incident based on a unique set of patterns and relationships in the data as described in the following:
Pipeline Leaks and Breaks: Leaks can be identified by slight, continuous pressure drops and a constant or slightly growing amount of water loss in the DMA. Pipeline breaks in the DMA will have a sustained pressure loss and a larger, continuous amount of water loss.
Water Theft: When water theft in the distribution system occurs (i.e., connecting a tanker truck to a fire hydrant), system pressure drops sharply and returns to normal after the activity is complete. Additionally, there is an increase in NRW during this period. Water theft can also occur at the meter if a homeowner bypasses or disables the meter. This type of theft is detected when a customer’s flow drops to zero and remains flatlined for a prolonged period.
Meter Issues: Conventional positive displacement meters wear out over time, resulting in lower-than-actual readings. DMAs that show a constant pressure with minimal fluctuation and a continuous magnitude of water loss can indicate that some of the meters are reporting low or not reporting at all.
Key Performance Indicators (KPIs) were developed for the project to assess NRW. Each KPI and its corresponding results are described in Table 1.
TABLE 1.
AMI Pilot Results for NRW Key Performance Indicators
| KPI | Description | Study Results |
| Faulty Meters | Number of meters that failed and could no longer accurately measure water flow |
|
| Failed CMIU | Number of CMIUs that stopped transmitting data |
|
| Water Theft Occurrences | Number of identified occurrences of water theft |
|
| Continuous Leaks – Utility | Number of leaks that were discovered on the utility’s side of the meter. The DMA flow was compared against the totalized flow from all residential meters to measure this KPI. |
|
Water losses on the customer side of the meter were identified by reviewing the dashboard for residential meters that indicated unusually high usage. On a typical day, there were approximately 70 to 75 continuous leaks on the customer side of the meter, which ranged from extremely small (<100 gpd) to very large (9,000 gpd), although the vast majority were less than 500 gpd).
When a sudden increase in usage was detected, the GCDWR customer service department was notified to reach out to the homeowner. During the initial call, GCDWR customer service personnel explained that high water use had been detected and scheduled an on-site visit. During the site visit, GCDWR customer service personnel assisted the resident with finding the source of high usage by providing dye tablets and an in-home walkthrough. In several cases, dye tablet testing indicated that a toilet flapper valve needed to be replaced. For other incidents, a leaking water heater or irrigation system was found to be the cause. Incidents with very large flows are described subsequently.
Case studies 1 through 7 are examples of unusual flow and pressure patterns that GCDWR personnel observed on the AMI dashboard. The underlying cause and remedy for each incident are discussed. To reduce water consumption and loss, the GCDWR conducted a system pressure-reduction study, and Case Study 8 describes the study’s objectives, methods, and results.
CASE STUDY – DMA REVERSE FLOW CONDITION
Case Study 1 centers around a contractor hired by GCDWR to exercise large valves in the distribution system. The contractor was exercising one of the valves that had been closed to isolate the DMA and failed to close it after working on it. The result was that water flowed backwards through the DMA meter as shown in Figure 3. Detecting large flow reversals in the system can provide important information when responding to customer complaints about suspended particles in pipelines (i.e., discolored water).
FIGURE 3. Backflow through the DMA Meter.
Note: The bottom trend is an 8-month view of DMA incoming flow data with the current view highlighted.
By viewing the DMA and total incoming flow trends on the dashboard, it was quickly determined that the DMA was experiencing reverse flow. Following an investigation, staff found the valve had been exercised earlier in the day. Because the pilot evaluation team identified this incident in a timely manner, the response time to address this reverse flow condition was minimized, with GCDWR personnel identifying the issue on a Thursday and correcting the issue by the next Monday. The team conducted an evaluation of the incident and determined that there were no significant customer impacts, such as pressure loss or turbid flow, related to the reverse flow conditions.
CASE STUDY – CONTINUOUS LEAK FROM RENTER
This incident was a large and prolonged 4,000 gpd leak on the customer’s side of a residential meter, as shown in Figure 4. GCDWR personnel went to the customer’s home and left door hangers and also attempted to contact the resident by phone. After several days with multiple attempts to make contact and no response, the GCDWR identified the customer as a renter and notified the landlord about the potential problem. The landlord visited the home, found the faucets running, and turned them off. This resulted in the GCDWR revising its investigation procedures to first evaluate if the home is a rental property, which would require GCDWR personnel to contact both the renter and landlord about any potential problem.
FIGURE 4.
Increase in Flow from a Continuous Leak
CASE STUDY – PERIODIC LEAK FROM IN AN IRRIGATION SYSTEM
In Gwinnett County, some residents turn on their irrigation system as early as April. A typical irrigation system uses a few thousand gallons per day, but, in the case of one customer, the system was using approximately 24,000 gpd due to a large leak. The long-range graph at the bottom of Figure 5 shows that the leak started the previous fall. The system was shut down over the winter but resumed again in the Spring.
FIGURE 5.
Periodic Leaks from an Irrigation System
The regular intervals of this event and time of day distinguishes it from other large, short-term usages in the spring such as filling swimming pools and pressure washing, which would typically be a singular, non-recurring event. These results demonstrate the importance of communication between utilities and customers about thoroughly checking their irrigation systems during the initial spring start-up and winter shut down.
CASE STUDY – MASSIVE WATER SYSTEM LEAK
When the AMI system detected a rapid increase in a customer’s water usage, GCDWR personnel notified the customer. The customer stopped the leak without assistance from the GCDWR, and the cause of the incident was not disclosed. Helping customers reduce their losses and preserve the County’s water resources was a very important goal for GCDWR. Figure 6 provides one example where a customer had a very small continuous leak that suddenly grew to approximately 8,400 gpd.
FIGURE 6.
Evidence of a Massive Leak
CASE STUDY – SYSTEM PRESSURE DROP
A sudden pressure drop was detected on May 27, 2018, and the incident was reported to the GCDWR and the cause was correlated to a pump station experiencing a failure and temporarily shutting down as shown in the pump station’s pressure profile in Figure 7. GCDWR personnel were surprised to learn that the pressure wave caused by the sudden drop propagated as far as the pilot area. The pump station is located in the Central Zone adjacent to the pilot DMA. The pilot evaluation team is continuing its evaluation of potential transient events associated with this sudden shutdown.
FIGURE 7.
Pressure Profile Showing Pump Station Failure
SYSTEM PRESSURE REDUCTION STUDY
The pressure at the beginning of the pilot test period ranged from approximately 110 to 150 psi (Figure 8), which is much greater than the upper limit of approximately 80 psi recommended for residential customers. System pressures exceeding 80 psi typically require a pressure regulator; however, these often go unchecked by their owners. Poorly operating home pressure regulators can lead to elevated pressure in the home and cause leaks in appliances and household fixtures. It has also been demonstrated by the European Commission that reducing pressure in the distribution system can often reduce water loss (European Commission, 2015).
FIGURE 8. Phase I and Phase II Pressure Reduction Steps.
(Note: The pressure dip on May 27 described is shown for all four pressure sensors because the pressure sensor malfunction was repaired, and the data recorded by the sensor during the outage was backfilled into the data management system.)
With a better understanding of the pilot area’s water consumption and losses on the utility side of the meter, GCDWR’s team evaluated the effectiveness of pressure reduction to reduce water loss by installing a pressure-reducing device downstream of the DMA meter to allow pressure reduction and stabilization in the pilot neighborhood.
After the pressure reduction valve was installed, the pressure study was conducted in two phases.
Phase I.
Pressure was reduced on May 23, 2018, by approximately 10 psi for a two-week period. For example, pressure in the low area of the system, which was at 140 psi, was reduced to 130 psi. Figure 8 shows the pressures measured by the four sensors in the pilot area.
Phase II.
Pressure was reduced on June 6, 2018, by an additional 8–10 psi and was evaluated over a two-week period.
Table 2 summarizes the DMA and total metered flows (i.e., sum of the residential flows) and loss values for the period before and during the phase I and II pressure-reduction steps.
TABLE 2.
DMA, Total Metered Flows, and Loss During the Pressure Reduction Study
| Phase | Date Range | Average DMA Flow1 (gal/day) | Average Total Metered Flows2 (gal/day) | Average Loss3 (gal/day) | Percent Loss4 |
| Pre-Phase I | 5/1 – 5/22/2018 | 93,600 | 90,100 | 3,503 | 3.77% |
| Phase I | 5/23 – 6/5/2018 | 86,100 | 83,200 | 2,840 | 3.30% |
| Phase II | 6/6 – 6/20/2018 | 91,000 | 88,200 | 2,840 | 3.12% |
Notes:
Average DMA Flow is the total flow entering the pilot area averaged over the date range.
Average Total Metered Flows is the sum of flow from all residential meters averaged over the date range. This is the amount of water used by customers (i.e., consumption).
Average Loss = (Average DMA Flow) – (Average Total Metered Flows). This is the amount of water lost in the system between the DMA meter and residential meters averaged over the date range.
Percent Loss = Loss / (Average DMA Flow) × 100
The stabilization of pressure in the pilot area was an important benefit of installing the pressure reducing valve. Consistent pressure conditions allowed the pilot evaluation team to detect smaller pressure deviations because of less noisy data. Furthermore, steadier pressures are less stressful to distribution system pipes and customer plumbing.
The DMA and total metered flow closely tracked one another. Table 2 shows that the average total metered flow (i.e., consumption) was 90,100 gpd during pre-phase I, which decreased to 83,200 gpd during phase I and increased to 88,200 gpd during phase II. The decrease in consumption during phase I could be attributed to less water loss on the customer side of the meter. This was likely due to lower pressures and less lawn irrigation because the weather during this period was relatively cool and wet. The increase in consumption during phase II was attributed to dry conditions and rising temperatures during this period, which led to more outdoor irrigation. However, consumption did not return to the pre-phase I amount, which could be caused by less water loss from operating at the lower pressure as well as lower customer use at the lower pressure (e.g., same shower length but less water used because of lower pressure). Regardless, the overall reduction in consumption between pre-phase I and phase II was 2.1%.
Table 2 also shows that the percent loss on the utility side of the meters was 3.77% during pre-phase I, decreased to 3.30% during phase I, and decreased further to 3.12% during phase II. Thus, even with an increase in consumption during phase II, water loss on the utility side of the meters decreased as a percentage of DMA flow, which correlates to the reduction in system pressure.
BACKFLOW AND TAMPERING
Other than the DMA reverse flow condition described in the first case study, no backflow incidents were detected by the residential meters in the pilot neighborhood during the evaluation period. However, there were two tampering incidents during the evaluation period. Both were at the same residence and caused by cutting the cable between the meter and the CMIU, which the pilot evaluation team suspected were attempts at water theft. The meter continued to collect and store flow data during each incident and backfilled the data management system after the cable was replaced. Thus, there were no gaps in the data.
The EPA and GCDWR recognized that responding to such incidents, although infrequent, was an important part of securing the utility’s distribution system. Thus, a security and response workshop was conducted. During the workshop, participants discussed AMI alert response procedures and system-level topics such as customer notification and incorporating AMI response procedures into standard operating procedures and the emergency response plan. The workshop participants also discussed general steps that could be part of an AMI investigation, including the following:
Review the customer’s water usage, payment history, and history of calls or complaints.
Determine whether the customer is a renter or owner.
Call the customer.
Dispatch a technician to read the meter, visually inspect the area around the meter for potential issues.
Discuss with the customer ways to identify and fix leaks; provide dye tablets if a leaking toilet tank is suspected.
If the resident is not home, leave a doorhanger with information about identifying and fixing leaks.
LESSONS LEARNED
WATER LOSS REDUCTION
The usage and pressure data provided by the AMI system allowed the pilot evaluation team to effectively detect and remedy leaks on both sides of the meter. With this data in hand, the utility was able to alert homeowners of potential problems with premise plumbing or irrigation systems. The homeowners were grateful for the quick diagnoses of high usage because it prevented an unexpectedly large water bill, especially when the owner did not live at the residence such as a landlord with a rental property. Also, some homeowners asked if they could receive high usage alerts on their smartphones.
Water loss also decreased on both sides of the meter as a result of reducing the system pressure, with water loss on the utility side of the meter decreasing from 3.77% to 3.12% of average DMA flow and overall consumption decreasing from 93,600 to 91, 000 gal/day. For utilities that have areas with relatively high pressure, system pressure reductions should be considered.
CELLULAR COMMUNICATION
The wireless transmission of data from the CMIUs to the data management system via cellular communications was very reliable. Although there were four defective CMIU batteries that failed soon after installation, the remaining CMIU batteries functioned throughout the pilot. Even when a CMIU battery failed, the meter continued to collect and store data every 15 minutes, and it transmitted the accumulated values to the data management system after the CMIU was replaced. The pilot evaluation team recommended continuing to collect data every 15 minutes but reducing data transmissions to every twelve hours. The meter manufacturer indicated that this frequency should provide a battery life in excess of 10 years.
SYSTEM SECURITY
Including backflow and tamper alerts on the dashboard improved GCDWR’s situational awareness in the pilot area. The following important points came out of the security GCDWR’s workshop with EPA:
Tamper: Workshop participants agreed that 24-hours was an acceptable time frame for tamper alerts because such alerts would typically be caused by water theft (e.g., cutting the cable between the CMIU and meter), which has minimal public health risk.
Backflow: Workshop participants agreed that an immediate response was required for backflow alerts because it could indicate a main break, cross connection, or intentional contamination—all of which are acute public health risks.
ENHANCED RESILIENCY
By reducing water loss and improving system security, the overall resiliency of the AMI area was enhanced. GCDWR is in a drought-prone region, so minimizing water losses system-wide helps maintain service, especially during periods of insufficient rainfall. Also, the AMI system provided the GCDWR the information they needed to quickly respond to potential water security incidents. The granular, time-series flow data generated by the AMI system can also improve a utility’s resiliency by allowing for more refined calibration of hydraulic models and more precise descriptions of current usage patterns.
CONCLUSION
GCDWR’s AMI pilot achieved its objectives of reducing water loss and improving security. The AMI system reduced water loss on the utility side of the meters (NRW) by identifying leaks and reducing water loss on the customer side of the meter by generating high usage alerts. Also, data from the exploration of pressure reduction showed that installing a pressure-reducing valve can reduce water consumption and loss while stabilizing system pressures. The cellular communications system used by the AMI system performed well and did not have any significant outages during the pilot period. The AMI system also improved security by generating backflow and tamper alerts.
KEY TAKEAWAYS.
A pilot study was conducted to understand the use of advanced metering infrastructure in a neighborhood northeast of Atlanta, Ga.
The dashboard system displayed locations of flows and pressures as well as alerts that the team flagged for further investigation.
The usage and pressure data from the pilot project allowed detection and remediation of leaks on both sides of the meter.
ACKNOWLEDGEMENTS
The authors would like to recognize the following organizations for their assistance and contributions to the AMI pilot project. Their engagement is greatly appreciated:
AT&T
Qualcomm
Neptune
Trimble
The authors would also like to recognize the following individuals from the GCDWR for their active involvement in the AMI pilot project. Their contributions are much appreciated.
Eric Swett
Tommy Gunter
Glyn Fowler
Contributor Information
Nelson Mix, US Environmental Protection Agency Office of Ground Water and Drinking Water (www.epa.gov/ground-water-and-drinking-water), Washington, D.C..
Alan Lai, Jacobs (www.jacobs.com), Cincinnati, Ohio..
Kenneth Thompson, Internet of Things and smart sensors at Jacobs (www.jacobs.com), Denver, Colo..
Steven C. Seachrist, Gwinnett County Department of Water Resources (gwinnettcounty.com), Lawrenceville, Ga..
REFERENCES
- Pinney Dan. (2018, February 20). Stem the Tide of Non-Revenue Water. Water Finance & Management. Retrieved on 4/22/21 from https://waterfm.com/stem-tide-non-revenue-water/
- European Commission. (2015). EU Reference document Good Practices on Leakage Management WFD CIS WG PoM. Retrieved on 4/22/21 from https://circabc.europa.eu/sd/a/1ddfba34-e1ce-4888-b031-6c559cb28e47/Good%20Practices%20on%20Leakage%20Management%20-%20Main%20Report_Final.pdf
AWWA RESOURCES
Advanced Metering Infrastructure: Lifeblood for Water Utilities. Moore, S. & Hughes, D.M., 2008. Journal AWWA, 100:4:64. https://doi.org/10.1002/j.1551-8833.2008.tb09605.x.
Consider Best Practices for Deploying Advanced Metering Infrastructure. Ball, J., 2014. Opflow, 40:12:14. https://doi.org/10.5991/OPF.2014.40.0077.
Improving Water System Resiliency and Security: Advanced Metering Infrastructure. Mix, N. & Thompson, K.A., 2016. Journal AWWA, 108:6:E310. https://doi.org/10.5942/jawwa.2016.108.0048.
Water and Electric AMI Differences: What Water Utility Leaders Need to Know. Brueck, T.M.; Williams, C.; Varner, J.; & Tirakian, E., 2018. Journal AWWA, 110:6:36. https://doi.org/10.1002/awwa.1096.
These resources have been supplied by Journal AWWA staff. For information on these and other AWWA resources, visit www.awwa.org.