September-October 2006

Until recently, an underground water line in Manville, NJ, has been leaking at a rate of 225 gallons each minute. How long it had been leaking is not known because the line was situated in a shale-type soil so the leaking drinking water did not surface but instead drained through the porous shale. In a year’s time at that 225-gallon-per-minute rate, 118,260,000 gallons of water would have been lost from that single leak. At a replacement cost of $0.65 per thousand gallons of treated drinking water, the annual costs of that leak would have been $77,000.

Could it have gone on leaking undetected for a year or more? Theoretically, yes, says Wayne Morgan, vice president, service delivery, for American Water’s Northeast Region. “Since this particular leak was caused by a circumferential break, it generated very little noise, so we probably never would have found it surveying with geophones or other conventional listening devices that traditionally have been used to try and find underground leaks. It would seem to corroborate the American Society of Civil Engineers’ [ASCE’s] mind-boggling estimate of this country’s loss from drinking-water leakage.”

In its 2005 Report Card for America’s Infrastructure, the ASCE states, “Each day, six billion gallons of clean, treated drinking water disappears, mostly due to old, leaky pipes and mains. That’s enough water to serve the population of a state the size of California.” Water leaks are a problem plaguing water systems nationwide, and one of the biggest challenges facing the United States is how to improve and maintain infrastructure for generations to come.

The case of the Manville leak is instructional as to the loss potential of underground pipe leaks, but after an unknown leak duration, it was detected this April during one of the pilot programs American Water is conducting in several northeastern states. The core technology that enabled this difficult leak to be discovered and located is acoustic leakage monitoring. This technology listens in part to the moving water itself and hence is effective with plastic pipe, ductile iron pipe, or copper service lines. This technology, placed in the field for continuous monitoring, has come a long way, and now it shows great promise to allow utilities to monitor leaks before they surface, to pinpoint the location of these leaks, and to repair them before there is serious water loss and/or expensive collateral damage to landscaping, paving, or sidewalks.

American Water and at least one individual water district (see sidebar) have been working with several leak detection companies in search of a low-cost, permanent leak detection system or systems that could be distributed throughout a water system and provide a leak analysis “on demand,” Morgan recalls. “We have been using Flow Metrix’s MLOG device for most of our pilot programs. “Strapped to active service pipes, the MLOG is designed to monitor sound on a nightly basis. Its range extends about 500 feet for metal pipes, reaching into the distribution system and detecting ‘noise’ from other nearby mains or services. Because of the sensitivity of the sensor, only about 10% of services must be equipped with the unit in order to monitor the entire distribution system. The percentage depends on the density of services, the pipe materials in use, and the distribution system. Given the layout and the relatively low cost of the MLOG unit, the effective price of the equipment is less than $15 per connection.

“Pipe noise monitoring is programmed to occur frequently during a quiet time [12:30 a.m. to 4:30 a.m.] The monitor catalogs the data in 10-minute segments and selects the least noisy time period as a single data point for comparison with other MLOGs in the field. Designed for a walk-by meter reader to collect, the original MLOG had its own radio transmitter that could transmit stored files of information. The data would be effectively communicated to the meter reader as he collected meter readings.”

Automatic Meter Reading Linkage
This walk-by collection of acoustic leak data so closely parallels meter-reading activity that it inevitably led to an examination of the feasibility of applying fixed network automatic meter-reading (AMR) technologies for both functions. AMR has become an increasingly reliable cost-effective method to convey customer meter readings with minimal error and personnel to perform utility billing and meter operations. The meter-reading factors that are normally considered in deciding whether to proceed with an investment in an AMR program include the following:

• The increasing difficulty in accessing meters set inside of residential buildings
• Concerns about personal security
• Hard-to-read and sometimes hard-to-access meters that are set in pits
• Significant operational costs for walk-by drive-by readings
• Inclement weather conditions

What is often overlooked is the value of the communication network established by AMR that might be used to relay other information about the water system. If systems that supply that information generate sufficient value, they can increase the cost-effectiveness of the combined system and justify the capital expenditure. With its innovative acoustical monitoring technology, Morgan asserts, American Water has not only found a way to add value to AMR but may have developed a revolutionary approach to maintaining minimal leak levels in a water system.

In a fixed-network AMR system, collectors stationed on poles, rooftops, or tanks throughout the meter routes pick up meter data at regular intervals from two to 12 times a day. Wireless devices transmit the data to the AMR control center that organizes the data, transmits data to other locations, and prepares reports. Alternatively, drive-by AMR is also used in some communities, and an MLOG collector could be used as a separate recovery device in the common drive-by vehicle. However, higher operational drive-by costs, a possible slowed rate of MLOG collection, and longer intervals between leak data collection indicate that the fixed network AMR system is operationally preferable if a utility can afford its higher capital costs and frequent water leaks are suspected. For the proposed MLOG link, Morgan points out, the fixed-network AMR provides the opportunity to collect the acoustic information daily. This translates into near real-time evaluation of potential leaks as they “grow.” Conversely, the drive-by AMR would store a month of data unless supplemental trips were made. The vehicle and the MLOG receiver system must pass by slowly enough to collect the data stream. Part of the goal of the pilot program is to evaluate the value of assessing data on a daily or weekly basis as opposed to examining data on a monthly basis.

“We had an AMR company [Hexagram] develop a pilot system combing MLOG with a fixed-network AMR system,” Morgan recalls. “Hexagram obtained an MLOG prototype and successfully transferred the data stream through an AMR system in its Ohio factory. Since the MLOG generates its daily reports in four packets, it was necessary to increase the daily transmissions from the originally proposed twice-daily transmission to four times per day. The dual-port Hexagram AMR transmitter unit, designed for compound meters, used one port for the meter [encoder] signal and one port for the acoustic MLOG device. For installation purposes, both units had waterproof connections installed to their devices with an agreed waterproof connector to be linked in the field.”

Pilot Testing
American Water has embarked on several pilot programs to test (and improve) the equipment and the various configurations. In Uniontown, PA, and Manville, NJ, the MLOG sensors were tested using the drive-by method to collect data. Another pilot is just starting up in Short Hills, NJ. It, too, will test the acoustic leak detection system without integration with AMR.

The first-of-its-kind pilot program that uses acoustic leak detection linked to an AMR system for data retrieval began in the spring of 2005 in the Pittsburgh suburb of Connellsville. The pilot was designed to specifically link the two technologies to alleviate non-revenue water (NRW) loss, which in Connellsville was estimated to be at least 25% of the city’s estimated 250,000 to 300,000 gallons per day of undetected leakage. The goals of the pilot project were to capture 50% to 60% of the undetected leakage and reduce the NRW loss by a comparable percentage.

Connellsville’s water system has an aging infrastructure of small-diameter galvanized steel and cast-iron water mains that has affected its high NRW losses, Morgan explains. “The city’s unique geography contributes to the problem. Since the city is built on a hillside along the Youghiogheny River, water from leaks in the distribution system have been flowing undetected underground into the river for many years. On average, the city’s water system experiences 35 surfacing breaks each year. The estimated annual cost of NRW loss is about $200,000—a number that provided a strong financial incentive to target leak reduction as a top priority.”

In the Connellsville pilot study, Hexagram connected 486 pre-existing AMR-compatible meters with its send-out radio-frequency multiple-tenant units and nine collector units placed throughout a system to roughly simulate the AMR system throughout town (5,000 customers). Then 486 MLOG units were added at the same locations to fully simulate coverage of the system by acoustic monitoring. The MLOG units recorded sound vibrations over a period of four hours each night. Specially designed software analyzed these acoustic patterns and assigned them into one of four leak categories:

1. No leak
2. Possible leak
3. Probable leak
4. No data available

Once a leak was detected, the AMR system provided notification of the failure so that the leak could be located and repaired. Within the first nine months of the program, the system had located 27 leaks with an estimated loss rate of 180,000 gallons per day. Coupled with a significant reduction in flushing water, the city’s NRW loss level has dropped from over 25% to under 10%, which, at a water purchase cost of $1.90 per thousand gallons, resulted in estimated annual savings of $170,000. “Using predictive tools,” Morgan concludes, “we estimated that the investment [primarily the equipment itself] will be repaid within two to three years.”

With the better-than-anticipated results from Connellsville and the steady improvement being made in the equipment, American Water is initiating additional pilots with different fixed-network AMR systems in Irvington, NJ, and with Sensus in Cape May County, NJ. And the company is performing a study on behalf of the American Water Works Association (AWWA) Research Foundation to determine whether the leaks it is detecting and repairing would have otherwise become breaks. The results of this study might indicate whether it is more cost-effective for a utility to find a leak early or to accept a delay of 30 days or more. And this might help determine a utility’s choice of drive-by versus fixed-network monitoring with or without AMR. In either case, the utility will have a much clearer picture of the actual condition of its pipe system.

Although Morgan cautions that the pilot projects are just entering Phase B status, he is clearly encouraged and excited by the results to date. And with AWWA’s decision to accept the International Water Association’s (IWA’s) standard definition of NWR and its components, it should reduce regulatory delays for utilities to implement systems when they have been perfected and fully tested. American Water’s leadership in this development program should further accelerate its acceptance throughout the country where the company has a presence in 29 states. It is not a case of “one system fits all”; the conditions vary too much from utility to utility and region to region. However, the acoustic leak detection technology holds out such promise that utilities in some states may be anxious to implement this technology with the least possible delay. For example, New Jersey American faces a combination of high purchased water costs and NRW levels of 15% to 16%.

Both Wayne Morgan and Marcellus Jones, distribution systems manager at the Las Vegas Valley Water District (see sidebar), are excited about the prospects of this technology for the water industry. Morgan’s confidence and enthusiasm, tempered only by his engineering caution, can best be illustrated by quoting the conclusions in the paper he, Dave Hughes, and Russell Titus delivered at the IWA Water Loss Task Force Conference in Halifax, Nova Scotia, last September:

It appears that the combination of AMR and acoustic monitoring can provide an effective network for system leakage monitoring. Although the value of the daily monitoring for system leakage is still being assessed, preliminary data strongly suggests that NRW will be significantly reduced. As the pilot program in Connellsville progresses to bring leaks to minimal levels, the full capability of the acoustic monitoring and companion pinpointing tools will be evaluated. As a result, American Water will continue to study the pilot program with a broader research effort being contemplated. Part of the research will look at other cost savings that might be realized. By identifying leaks in the early stages, repairs can be conveniently scheduled. Cost savings per repair should be reduced as the extent of damage to the pipe and the environment around the pipe will be less severe. If most leaks are arrested before the pipe fully fails, the extent of damage to other infrastructure and private property will be reduced. We are looking forward to learning and reporting further on the evolution of a truly innovative leak monitoring effort.