The Financial and Policy Implications of Water Loss

Transcription

1 E77 The Financial and Policy Implications of Water Loss ERIN RESS 1 AND J. ALAN ROBERSON 2 1 Tetra Tech Inc., Fairfax, Va. 2 AWWA, Washington, D.C. This research analyzed the financial and policy implications of potentially using a regulatory requirement as a mechanism to reduce water loss. This research compared validated water audits from North America, voluntarily submitted in calendar years 2010 to 2013, with audits from Georgia, required to be submitted to the state in calendar years 2011 and The analysis did not show any significant differences between the two data sets. This article also compares performance indicators on the basis of the size of the utilities and their geographic distributions and found no trends. Two methodologies are presented for calculating the benefits of reducing losses, which could be used as part of an analysis to assess the return on investment when prioritizing among competing infrastructure needs and when evaluating a water loss control program. Keywords: nonrevenue water, performance indicators, return on investment, water loss Public drinking water systems are treating and distributing billions of gallons of safe drinking water every day. In a perfect world where pipes never leak, the total volume of water that enters the distribution system should be equal to the volume of water that reaches, and is accurately billed to, its consumers; however, this is never the case. Distribution systems are losing 10 to 30% of their water supply every day with some reporting close to 50% losses (Fiut et al. 2013). The nation s population is expected to increase by 100 million by the middle of the 21st century, which will increase the demand for water and increase stress on water resources for many drinking water systems (Cohen 2012). Water system managers need to take a holistic risk-management approach to their enterprise when planning to meet future demands. A range of tools need to be considered for this holistic approach, including drought management, water conservation, energy efficiency, and the focus of this paper, water loss reductions. The need to appropriately manage water losses is becoming especially important in some areas during the summer months when temperature and water demand typically increase and water supplies typically decrease. Droughts can exacerbate water supply problems for systems that are near capacity during peak summer demands or are facing increasing demands as a result of population and job growth. Drought management plans are an important part of the overall risk management for water systems (AWWA 2011). Water conservation is another tool that can be used in tandem with the appropriate management of water losses to stretch water supplies. Traditionally, some states have established regulations promoting demand-side water conservation, which includes limiting outdoor water use and providing incentives for purchasing water-efficient appliances such as faucets, toilets, and showerheads. Beyond traditional conservation measures, efforts to stretch water supplies also need to take place on the supply side (e.g., implementing water loss control programs at utilities). Leaking pipes across the nation lose an estimated 2.6 tril gal of total water every year, which results in an annual loss of $4.1 billion in electricity costs (Cohen 2012). The water energy nexus is the connection and dependence of water and energy between and on one another; water is needed by electricitygenerating stations to produce energy, while energy is essential to treat, pump, and transport water. Reducing water losses not only increases the amount of available water, but through the water energy nexus, operational costs can be reduced by requiring less energy input use (Baird 2011). Through the water energy nexus, increasing the water efficiency at a water utility will result in more available water and a reduction in energy input costs, which in turn reduces greenhouse gas emissions from power-generating stations (Aubuchon & Roberson 2014). MANAGING DISTRIBUTION SYSTEMS AND LOSS As discussed previously, individual water systems are losing millions of gallons of drinking water every day. This is water that could be generating revenues for the system if not lost in the distribution system. In order to adequately support their customers needs and be good stewards of their financial resources, water systems need to take the necessary steps to better manage their distribution system and their water loss. Nonrevenue water (NRW). To improve water and energy efficiency at drinking water systems and decrease water losses, systems need to know exactly how much water is being lost in the distribution system, which is just one component of NRW. The water balance in Table 1 shows that water loss, either real or apparent, is the difference between the total system input volume

2 E78 and the volume of authorized consumption (AWWA 2009). Real losses are physical losses that result from leaking service connections or storage tank overflows and leaks, while apparent losses are commercial losses that result from theft, customer metering inaccuracies, or systematic data handling errors (AWWA 2009). Typically, real losses are valued using the production cost, which includes the cost of treating and pumping the water, while apparent losses are valued using the retail cost because the water has been consumed but not paid for (Mathis et al. 2012). The use of production and retail costs will be discussed later in this article as part of the return on investment (ROI) analysis for reducing water losses. The other component of NRW, unbilled authorized consumption, is water that is used for flushing of the distribution system, street cleaning, firefighting, and other acknowledged or permitted uses for which no payment is received or measured. Managing NRW is important for utilities to benchmark and track their progress toward target goals to conserve water and energy and to become better stewards of their financial resources. Cost implications. Generally, decreasing NRW is cost-effective, but several factors need to be considered when evaluating the cost benefit analysis for a specific water system. Financial resources must be allocated for the upfront cost to pay for equipment and hiring a crew to find leaks, inaccurate or faulty meters, and to make repairs, which are then followed by a payback period. Fixing every leak or faulty meter is not economically feasible or even possible in some cases. The water saved from repairing an individual leak can vary significantly, from saving 53.4 gpm for repairing a 0.5-in.-diameter hole to saving gpm for a 1.0-in.-diameter hole when operating at 80 psi (AWWA 2009). Therefore, water system managers must decide the level of financial investment that will result in a positive ROI. ROI is the benefit that results from an investment, and in this case the benefits are the value of the water and the potential energy savings. To produce a positive ROI, systems will repair only the leaks that provide more benefits than the repair costs. In other words, the money saved by reducing losses needs to be greater than the cost of repairing the cause of those losses. However, a water system with scarce water resources could make the policy decision to have an aggressive water loss reduction program (with potentially a smaller ROI). With a fixed budget, critical decisions must be made beforehand regarding which leaks and meters to repair in order to maximize the amount of water and energy saved. This article presents two water system case studies using two methodologies that could be used to determine the maximum amount of money to spend on water loss control programs to produce a positive ROI. Policy implications. Currently no national standard regulates the amount of water loss permissible at a drinking water system. Frequent droughts and increasing energy costs are two of the reasons some states have imposed regulations with the intention of reducing water losses and operating costs while increasing water supplies. The need for a national regulation can be debated on both sides one side could say that a single standard is not feasible because of all of the site-specific factors inherent in a water loss program, while the other side might say a national standard would provide uniformity and be comparable to energyefficiency standards, automobile fuel-efficiency standards, and standards for low-flow plumbing fixtures. OBJECTIVES This water research investigated some of the policy and financial implications of water loss programs through three analyses: Determining whether a regulatory requirement for water audits would decrease water loss by comparing validated water audits from North America, which were voluntarily submitted in calendar years 2010 to 2013, with audits from Georgia, which were required to be submitted to the state in calendar years 2011 and Comparing performance indicators on the basis of the size of the utilities and their geographic distributions. TABLE 1 AWWA/IWA water balance Water exported (corrected for known errors) Billed water exported Revenue water Volume from own sources (corrected for known errors) System input volume Water supplied Authorized consumption Water losses Billed authorized consumption Unbilled authorized consumption Apparent losses Billed metered consumption Billed unmetered consumption Unbilled metered consumption Unbilled unmetered consumption Customer metering inaccuracies Unauthorized consumption Systematic data handling errors Leakage on transmission and distribution mains Revenue water Nonrevenue water Water imported (corrected for known errors) Real losses Leakage and overflows at utility s storage tanks Leakage on service connections up to the point of customer metering Source: AWWA 2009 IWA International Water Association

3 E79 Using two methodologies for calculating the benefits of reducing losses that could be used to calculate the ROI when prioritizing among competing infrastructure needs and when evaluating a water loss control program. BACKGROUND Water loss has been an issue for water systems for many years, but the terminology and calculations for the concept have evolved over time. When assessing water flow into and out of the distribution system, utilities traditionally referred to water losses as unaccounted-for water. In the past, the definition of unaccounted-for water varied between utilities across North America, as many utilities would account for some known amount of losses as accounted-for and other losses as unaccounted-for (USEPA 2010). The lack of standardized terms and definitions led to communication challenges between utilities and inconsistent measurements that made benchmarking difficult. New and consistent terminology was needed, as well as a standard methodology for calculating water losses. This led to the International Water Association (IWA) and AWWA partnering to develop a water audit methodology currently recommended by the AWWA Water Loss Control Committee (WLCC) as the best management practice for accounting and controlling losses in drinking water systems (AWWA 2009). In 2009, AWWA published the third edition of Manual of Water Supply Practices M36, Water Audits and Loss Control Programs, a manual that is used as guidance for drinking water utilities on how to use water audit methodology to economically control water and revenue losses by tracking how effectively water is moved from its source to its consumers. It uses consistent and sound definitions and terms that should be adopted by every drinking water utility. This methodology discourages the use of the term unaccounted-for water to assess water loss and instead recommends the use of the term nonrevenue water. It also discusses the inaccuracies inherent in using water loss on a percent volume basis as an operational performance indicator and instead recommends the use of several other performance indicators. Performance indicators. Water loss expressed as a percentage of total volume is not the optimal performance indicator because of the influence of variable consumer demand. For example, a water utility with an average consumption of 200 gpcd with a total system input of 3,000 mil gal/year and a total metered consumption of 2,700 mil gal/year would have total losses of 300 mil gal/ year (10%). Assume in this example that this utility decreases consumption to 150 gpcd through demand-side conservation, which results in total metered consumption of 2,025 mil gal/year and a total system input of 2,325 mil gal/year. With no reduction in losses, the system is still losing 300 mil gal/year, which results in a 12.9% loss compared with the initial 10%. The percentage of water loss increased without any real change in losses. This example shows why the percentage of system input volume is a poor operational performance indicator for assessing water losses (Thornton et al. 2008). Two operational performance indicators used to assess water losses as suggested in M36 are apparent losses and real losses expressed as gallons per service connection per day. These performance indicators can be used for benchmarking within the utility to show improvements in efficiency. Using these performance indicators for comparison with other utilities is challenging at best because of the differences in the gallons per service connection per day. Therefore, the infrastructure leakage index (ILI) was designed specifically as an operational performance indicator for comparison among utilities and benchmarking within a utility. As shown in Eq 1, ILI is the ratio of the current annual real losses (CARL) to the unavoidable annual real losses (UARL) and quantifies how well a distribution system is managed for controlling real losses at the current operating pressure (AWWA 2009). ILI = CARL UARL (1) Ultimately, utilities aim to achieve a low ILI by reducing the CARL as close as possible to the UARL. Financial performance indicators are also used to view the overall financial impact of the water losses at the utility. This research analyzed NRW percent by cost and NRW percent by volume, which are defined by M36 in Eqs 2 and 3, respectively: NRW % cost = NRW % volume = Cost of NRW Total cost of operating the system Volume of NRW Total system input volume As previously discussed, NRW percent by volume cannot be used to demonstrate operational efficiency because of changes in demand and the variability between systems (AWWA 2009). Therefore, it is used only as a reference point and not for any analyses. Other financial indicators used include the annual cost of real losses and the annual cost of apparent losses. Water audits. Another tool used for managing water losses is the AWWA/IWA Free Water Audit Software, which is used to conduct a water audit and serves as an initial assessment of the losses and the associated costs (Chastain-Howley 2007). The software provides specific results for operational and financial performance indicators after entering site-specific factors. The water audit methodology and the software are two tools that serve as best management practices for utilities to easily track their progress toward better managing NRW (AWWA 2009). An important feature of this software is a validity score, which is the confidence behind the values that are submitted in the audit. That is, with each entered value, a validity score is submitted on a scale of zero to 100 as an estimate of the confidence in the accuracy of that specific number, leading to an overall average validity score. Validating the data is an important process because of the many calculations and potential for errors when assessing water losses (Jernigan 2014). Developing a water audit is the first step in assessing water loss. Once the water audit is completed, utilities have some insight into where they are losing the most water, whether it is from faulty meters, leaking pipes, or elsewhere. Once the problem is found, the associated cost must be assessed to produce a positive ROI (2) (3)

4 E80 (i.e., the net benefits of an investment). In this case, the net benefits are the value of water saved and that the cost covers all of the necessary interventions, including hiring workers and crew, training the crew, purchasing equipment, allotting time, and many other site-specific factors. If the total cost is greater than the net benefit of repairs, the result is a negative ROI, meaning more money is being spent than recovered. Therefore, in most cases, it is not economically sensible to make the investment. According to M36, in most cases controlling losses will follow the law of diminishing returns in that when losses are excessive, a relatively large reduction in losses can occur at a relatively low cost. As utilities employ additional resources to manage water losses, the effort requires greater costs and efforts to recover everdiminishing returns. Repairing every leak is not economically feasible; therefore, an important part of ensuring a positive ROI is balancing the costs and benefits by operating at the economic leakage level (ELL) (Thornton 2008). The ELL is when the value of the water lost real and apparent plus the cost to reduce losses, known as the cost of intervention, is at a minimum (Thornton 2008). Above the ELL, the effort to control the losses costs more than the value of the recoveries, and it is not economically feasible to pursue. When examining a marginal cost curve, many utilities lack sufficient resources or the necessary expertise in water loss management to decide what leaks to repair to produce a positive ROI (WRF 2014). This article examines the potential savings and benefits from reducing water loss by a small percentage and reducing the ILI. National and state regulations. The Energy Policy Act of 1992 (EPACT92, P.L ) mandated the development of regulations for low-flow plumbing fixtures such as showerheads, faucets, and toilets. In response to frequent droughts, consumers are sometimes required to reduce water use by using less water for outdoor purposes and to promote additional reductions in indoor water use. While both of these efforts decrease water demand, this type of water conservation can only decrease water use to a certain point because of baseline indoor demand. Diminishing returns on demand-side conservation means that water systems must turn to supply-side efficiency but, unlike demand-side use (e.g., low-flow plumbing fixtures), there is no national regulation. Currently no national policy or regulation mandates the specific amount of water loss permissible from a public water distribution system. This lack of a national policy or regulation has prompted some states to pass their own legislation that requires utilities to submit water audits. Most state requirements focus on submitting audits as opposed to setting a maximum amount of water loss, but by requiring audits the goal is for utilities to benchmark themselves, compare among other similar utilities, and appropriately manage their water loss. Georgia has adopted the most stringent regulations for managing water loss. While Georgia has abundant water resources, they are not evenly distributed across the state, and the rainfall is unable to replenish the sources evenly (CNT 2014). A task force found three key goals that led to the passing of the Georgia Water Stewardship Act, signed into law in June 2010: to conserve water by reducing demand, to capture and identify new sources, and to control the water supply by changing policies and procedures (Ashley & Kirkpatrick 2011). The act requires all drinking water utilities serving more than 3,300 people to conduct an annual water audit using the AWWA/ IWA methodology, which includes validation and the software to implement a water loss control program (CNT 2014). The regulated systems provide 80% of the potable water in the state (CNT 2014). The audit requirement in Georgia and other states could be the start of a regulatory trend toward better management of water losses across the nation. This article discusses the results and trends from the water audits submitted by the Georgia utilities from calendar years (CYs) 2011 and 2012, as well as a North American data set collected by the AWWA WLCC from CYs 2010 to The goal of this research was to analyze whether the regulatory requirement for water utilities in Georgia to submit audits results in decreased water losses or overall improved efficiency when compared with the North American data that were voluntarily submitted. DATA AND METHODS Two data sets, one from North America and the other from Georgia, were used for this research. The North American data were compared by their respective US Environmental Protection Agency (USEPA) region and by the system size. The Georgia data were compared by the system size and then compared with the North American data using a hypothesis test. These comparisons were made using the following performance indicators: apparent losses in gallons per service connection per day, real losses in gallons per service connection per day, and ILI and NRW percent by cost. NRW percent by volume was not used for comparisons, but only as a reference point. The North American data used in these comparisons were validated water audits for four CYs, 2010 through These data were collected by the AWWA WLCC using the AWWA Free Water Audit Software ( Table 2 shows the summary statistics for the four performance indicators used in the analysis and for NRW percent by volume. Each of the four years were subdivided into two system sizes large systems serving a population greater than 10,000 and small systems serving a population less than 10,000 using an online tool provided by USEPA to determine system size (USEPA 2015). Over the four years, audits were submitted from 42 utilities from across North America (Figure 1). Only 11 utilities submitted audits for all four years. The four performance indicators used for comparisons in this study are real losses in gallons per service connection per day, apparent losses in gallons per service connection per day, ILI, and NRW percent by cost. These indicators were first analyzed across all four years for any trends that could result from conservation efforts, decreases in demands, changes in population, or any combination of these. It was hypothesized that the 11 utilities that reported across all four years would show a decrease in performance indicators, as these systems may have been improving system efficiency. Second, given the geographical distribution as shown in Figure 1, an analysis was conducted to determine whether any regional differences could be made from these data, based on the USEPA regions. Because only 42 utilities reported, not every region

5 E81 TABLE 2 North American summary statistics, CY Apparent losses gal/service connection/day Range n = n = n = n = 26 Mean Median th percentile Real losses gal/service connection/day Range Mean Median th percentile Infrastructure leakage index CARL/UARL Range Mean Median th percentile Nonrevenue water % by cost Range Mean Median th percentile Nonrevenue water % by volume Range CARL current annual real losses, CY calendar year, UARL unavoidable annual real losses Mean Median th percentile was accounted for. All four years had a majority of the utilities reporting only from USEPA Regions 3, 4, and 6, which cover a large portion of the East Coast from Pennsylvania down to the southeast and stretching across the southwest to New Mexico. Lastly, the utilities were compared by system size small versus large to determine any differences in performance indicators. The Georgia data were validated water audits for CYs 2011 and 2012 for large and small utilities, respectively, as shown in Table 3. FIGURE 1 North American utilities by location (calendar years ) These audits were submitted to the Georgia Environmental Protection Division as a requirement under the Stewardship Act previously discussed. Large systems (those serving more than 10,000 in population) were required to submit audits by March 2012 covering CY 2011, while small systems (those serving 3,300 10,000 in population) were required to submit audits by March 2013 for CY This data set was a much larger sample size that included 107 large systems reporting in 2011 and 100 small systems reporting in The performance indicators from the small and large systems in Georgia were compared to determine whether utility size made any difference in water loss. Lastly, the Georgia data and the North American data were compared to test the hypothesis of this research: does having a state requirement reduce the water loss from a system? To test this, a hypothesis test specifically a t-test was performed to compare the four performance indicators among the large systems from the North American data set with the large systems from the Georgia data set using a p value of To lessen the complexity in assessing the potential financial implications of water losses, this research calculates the potential savings of reducing water losses using two methods. The first method calculated the value of water saved by reducing water losses by 0.5, 1.0, 1.5, and 2.0%. Using the given amount of water loss in millions of gallons per year and the percent reduction, two potential savings were calculated: one using the

6 E82 production cost and one using the retail cost. The two costs provide the lower bound (production) and upper bound (retail) costs that can be used by a water system in determining a system-specific ROI. The actual benefit (water and energy saved) from a water loss program is likely somewhere in between these two numbers, depending on several system-specific factors. The second method assesses the savings of reducing only real losses using a target ILI approach. This methodology values the TABLE 3 Georgia summary statistics, CYs Apparent losses gal/service connection/day Real losses gal/service connection/day Infrastructure leakage index CARL/UARL Nonrevenue water % by cost Nonrevenue water % by volume 2011 n = 107 (Large) 2012 n = 100 (Small) Range Mean Median th percentile Range Mean Median th percentile Range Mean Median th percentile Range Mean Median th percentile Range Mean Median th percentile CARL current annual real losses, CY calendar year, UARL unavoidable annual real losses FIGURE 2 Water gal/service connection/day Real losses versus validity score for Orange County, Fla. Real losses Validity score Year Validity Score saved water at the production cost, which includes the cost for treating and pumping because the water never reaches the consumers, compared with apparent losses, which are valued at the retail cost because the water has been consumed but not paid for. The necessary reduction in real losses is calculated by multiplying the target ILI by the given UARL specific for each system. The new level of water losses are then multiplied by the production cost to give the annual cost of the real losses, which is compared with the current annual cost of real losses to determine the potential savings. The potential savings helps determine the maximum amount of money the utility should spend in order to keep the benefits greater than the costs. Utilities can use either or both of these methods to decide on an appropriate investment for a water loss control program that will result in a positive ROI. RESULTS North America. The North American data set is summarized across all four years reporting the range, mean, median and 90th percentile for each of the performance indicators and NRW percent by volume in Table 2. The mean apparent losses decreased from 14.9 gal/service connection/day in 2010 to 12.0 in However, the mean real losses increased from 63.3 gal/service connection/day to 69.7 over the four years. Also, the NRW percent by volume for CY 2013 shows that 90% of the systems that reported water audits have NRW percent volume less than 40.8%, with 45.0% being the highest reported in This means this system is not recovering revenue for almost half of its total water supply. Although the means of the performance indicators did not decrease for all of the systems, of the 11 utilities that reported all four years, a few had one or two performance indicators decrease. For example, Orange County, Fla., had a reduction in apparent losses from 7.4 to 5.8 gal/service connection/day. On the other hand, Orange County experienced a significant increase in real losses starting at 17.1 gal/service connection/day in 2010 and increasing to 55.8 in This dramatic increase in real losses is most likely a result of the simultaneous increase in the validity score, which increased from 75 in 2010 to 87 in Figure 2 shows the comparison of real losses with validity score for Orange County, both of which increased significantly over the four years. This figure shows that with increasing validity, the parameters in the water balance are likely to shift. As more annual audits are submitted, more trends may be visible and more familiarity with the process will likely lead to an increase in validity, which may or may not lead to increases in performance indicators. An examination of all four performance indicators showed that five utilities were consistently higher than the mean for the performance indicators across all four years: Birmingham Water Works Board (Ala.), DC Water and Sewer Authority (Washington, D.C.), City of Griffin (Ga.), Philadelphia Water Department (Pa.), and the City of Wilmington (Del.). Figure 3 shows the real losses for CY 2013 of these five utilities against the mean real losses for all utilities that year; however, this study examined only 26 utilities for that year out of the approximately 53,000 community water systems across the United States (USEPA 2012). Future research is necessary to explain

7 E83 FIGURE 3 Water gal/service connection/day Birmingham, Ala. Select real losses versus average of all utilities in calendar year 2013 Real losses Average all utilities a Washington, D.C. Griffin, Ga. a Average of all utilities for calendar year 2013 Philadelphia, Pa. Wilmington, Del. why these utilities have above-average values in these performance indicators over the four years. Geographically, USEPA Region 3 had much higher means for all performance indicators than Regions 4 and 6 as seen in Figure 4. Region 3 had the highest mean real losses of gal/service connection/day in 2011, which is more than double the 48.0 and 37.9 mean real losses for Regions 4 and 6, respectively. Philadelphia Water and DC Water are in Region 3, which are both large systems and two of the oldest systems involved in the data collection. These systems have older pipes and service lines, which leads to a greater chance of leaking pipes, main breaks, and meter inaccuracies. Both of these cities encompass a large population that will continue to expand as the nation s population increases. Therefore, these systems will need to maintain their water loss control program to meet expected future needs. Analyzing the size of the utilities for the North American data across the four years showed the large utilities typically had higher mean performance indicators than the small utilities except on two occasions: ILI in 2012 and real losses in Table 4 shows the means of the performance indicators for CY 2013 in which the small systems had mean real losses of 47.5 gal/service connection/day compared with mean real losses of FIGURE 4 North American performance indicators by region Region 3 Region 4 Region 6 40 A Apparent Losses 140 B Real Losses Apparent Losses gal/ service connection/day Real Losses gal/ service connection/day Calendar Year Calendar Year ILI CARL/UARL C Infrastructure Leakage Index Calendar Year Nonrevenue Water % by cost D Nonrevenue Water % cost CARL current annual real losses, ILI infrastructure leakage index, UARL unavoidable annual real losses Calendar Year

8 E gal/service connection/day for the large systems. Table 4 also shows the apparent losses in gallons per service connection per day between the small and large systems at 4.2 and 13.5, respectively. The higher volume of water at large systems requires more meters to maintain and more chances for error in data handling. The large systems also have a higher audit score of 78 compared with the small systems, which have a mean audit score of 68. This indicates that the large systems were more confident with the information submitted into the water audit. Because the audit scores are not that high and are still increasing, the authors cannot conclude any apparent trends between the small and large systems. Once the validity of the data is higher and consistent over multiple years, comparisons and observation of trends could likely be made. A review of priority areas for potential improvement for CY 2013 showed that 18 of the 26 utilities had volume from own sources listed as the first-priority area out of nine other options. Volume from own sources is the amount of water that leaves the treatment plant and is recorded by the production master meter(s) and includes raw water and sources such as wells, rivers, lake reservoirs, or aqueduct turnouts (AWWA 2009). The AWWA Free Water Audit Software uses a meter error adjustment for this value because no water meter is 100% accurate, which makes it a common first-priority area. TABLE 4 North American small and large system averages, CY 2013 Small System Large System Apparent losses gal/service connection/day Real losses gal/service connection/day ILI Nonrevenue water % by cost Audit score CY calendar year, ILI infrastructure leakage index TABLE 5 Decrease in Water Loss % CY calendar year Reduction in water losses and the potential savings for Birmingham Water Works Board, a CY 2013 Production Cost b ($339.00/mil gal) $ Potential Savings Retail Cost c ($3.90/1,000 gal) $ , , , , , , , ,198 a Birmingham Water Works Board with a water loss of 11, mil gal/year b Lower bound c Upper bound Georgia. The Georgia data set is summarized for 2011 and 2012 reporting the range, mean, median, and 90th percentile for each of the performance indicators and NRW percent by volume (Table 3). The range for NRW percent by volume is notable. Both years show a maximum above 50%, meaning there is at least one system that is not producing revenue for over half the amount of water that goes into the system. While the North American data set showed the large utilities with higher performance indicator means than the small utilities, the Georgia data set had different results. The Georgia data showed that the small systems had a higher mean ILI and real losses than the large systems. As seen in Table 3, the mean ILI for large systems is 2.5 compared with 3.3 for small systems. Also, the mean real losses for the large systems were 49.8 gal/ service connection/day compared with the 52.1 mean for the small systems; however, the large systems had a higher mean of apparent losses of 9.8 gal/service connection/day compared with 8.9 for small systems. Small systems may have higher ILI and real losses because they typically lack the technical, financial, and managerial capacity to repair leaks and overflows as quickly and effectively as large systems. Although there is a difference between the size comparisons within the North American and Georgia data sets, the large sample size for Georgia increased the reliability of the data. Comparison. No significant difference was found between any of the four performance indicators for the large Georgia and the large North American utilities for CY For example, there is no significant difference between the North American ILI (mean 3.5 ± 3.2) and the Georgia ILI (mean 2.5 ± 2.0) (p = 0.20). The p value was greater than 0.05 for all four performance indicators for the large systems; so far, these results suggest that requiring utilities to submit water audits does not lower the amount of water loss. This result could largely be a result of the Georgia data being from the first year of the required audits; thus, utilities were not able to benchmark or compare against other systems to potentially lower their water losses. A comparison of performance indicators between systems may add political or social pressure to increase the efficiency of the system and reduce losses. Financial. The first method to view the potential savings by reducing water losses is shown in Table 5. Birmingham Water Works Board lost 11,887 mil gal of water at a production cost of $339.00/mil gal in CY If it was to reduce its losses by 0.5%, it could potentially save $20,149 on the basis of production cost. However, it can be argued whether to value the water losses at the production cost or the retail cost, so using the same methodology, the retail cost of $3.90/1,000 gal was converted into cost per million gallons for consistency with units. This resulted in potential savings of $231,800 at a 0.5% reduction. These two numbers are the lower and upper bounds, respectively, of the range of potential savings for Birmingham. This potential investment could be economically feasible if Birmingham spends between $20,149 and $231,800, depending on which cost of water is used. The second method to view the potential savings by reducing losses is the target ILI approach, as shown in Table 6. In 2013,

9 E85 DC Water lost 5,622 mil gal with an ILI of To reach an ILI of 7.0, DC Water would have to reduce its real losses by 270 mil gal to 5,352 mil gal/year. This results in a potential savings of $214,109 based on retail cost, and the investment would be economically feasible if DC Water did not spend more than this amount when reducing real losses to reach an ILI of 7.0. Considerations. Several considerations arose from these analyses. Most importantly, higher confidence is needed for the parameters submitted in the audits. Systems that submit audits in sequential calendar years are typically increasing the validity in their data, which results in more accurate water loss calculations. Of the 27 systems that reported two or more years, 12 increased their validity while 15 stayed about the same or decreased a small amount. Once the validity is high and consistent, systems will have a more accurate assessment of the amount they are losing and have better knowledge of the cause of the water losses. Additionally, the North American utilities self-reported their audits; that is, they may be above or below the national averages for the performance indicator, but there is no way to know how they compare with the balance of the 53,000 community water systems in the United States. Lastly, the dollar values for reported potential savings represent the value of water that will be saved for that year. It does not take into account the other benefits from reducing water losses such as deferred treatment facility upgrades, recovery of lost revenue, reduced pressure on water resources, and many other benefits not quantified in this paper. CONCLUSIONS AND RECOMMENDATIONS Two water loss data sets from North America (four years) and the state of Georgia (two years) were analyzed for trends by location and system size. For the North American data set, USEPA Region 3 had higher means than Regions 4 and 6 for all of the performance indicators because of two older systems (Philadelphia and Washington) located in that region. By system TABLE 6 DC Water target ILI reduction strategy, CY 2013 ILI Real Losses mil gal/year Annual Real Loss Cost $ Potential Savings $ ,622 4,685,840 Current level 7.0 5,352 4,471, , ,588 3,828, , ,832 3,186,446 1,499, ,058 2,549,158 2,136, ,294 1,911,868 2,773, ,529 1,274,579 3,411, ,289 4,048,550 CY calendar year, ILI infrastructure leakage index size, the large utilities typically had higher mean performance indicators than small systems except for one year: ILI and real losses in The limited number of systems in the North American data set did not allow for a national comparison between regions, and future research should focus on collecting validated data from a larger number of systems. The Georgia data showed that smaller systems had a higher mean ILI and real losses than large systems. Future research should focus on analyzing more than two years of Georgia data. No significant difference was found between any of the four performance indicators for the large Georgia and the large North American utilities for The small number of systems and the limited number of calendar years of data used in this study lead to the conclusion that a national regulation mandating the maximum amount of water loss allowed from a system may not be necessary at this time. More years of validated water loss data and more research are needed to evaluate what an appropriate number might be and which performance indicator might be used. States that have recurring water scarcity issues are already developing their own regulations, and this is likely the most appropriate national policy direction at this time. More research is needed with a larger number of systems and more years of validated water loss data before any national water loss policy can be considered. Given water resource constraints, some states may move forward with their own water loss regulatory requirements, and that may be appropriate at the state level on the basis of state-specific issues. Additionally, water losses should not be ignored by any water system, as tracking water losses by submitting water audits is potentially beneficial for all systems. Once water loss is minimized, the benefits go beyond reducing production costs through using less energy. Not only will it result in more available water, but also in increased efficiency, deferred treatment facility upgrades, recovery of lost revenue, and reduced pressure on water resources, and may lead to long-term water sustainability. An active water loss control program may also lead to the replacement and repair of pipes before the occurrence of larger leaks that could result in more damages and potential liability costs to the utility. Appropriately managing water loss reduces energy use, which then indirectly reduces greenhouse gas emissions. While managing water loss has many benefits, the economics behind managing water losses must be addressed using systemspecific factors before any investment is made, because repairing every leak is not economically feasible. Using a percent reduction in total water losses or a target ILI reduction strategy are two methods to view the potential savings for reducing water loss. The resulting savings from the production or retail cost can be used as the lower and upper bounds, respectively, in determining the appropriate investment for controlling or minimizing these losses. Analyzing and understanding the necessary investment needed to produce a positive ROI is an essential part of managing water losses to improve the overall efficiency of the system. Water system managers need to take a holistic risk management approach when planning for future supplies to meet future demands, and understanding and minimizing water loss should be a part of that approach.

10 E86 ACKNOWLEDGMENT The authors acknowledge the AWWA Water Loss Control Committee; Craig Aubuchon, manager at The Analysis Group Inc.; and Samantha Rucinski of GE (both former AWWA interns), as well as Steve Cavanaugh of Cavanaugh and Associates P.A. for their input and advice on this paper. Erin Ress internship at AWWA was funded by the Water Industry Technical Action Fund (WITAF). WITAF is administered by AWWA and is funded through AWWA organizational members dues. WITAF funds information collection and analysis and other activities in support of sound and effective legislation and regulations. ABOUT THE AUTHORS Erin Ress (to whom correspondence may be addressed) is an environmental scientist at Tetra Tech Inc., Eaton Pl., Ste. 340, Fairfax, VA USA; com. Ress received her BS degree in environmental science from Virginia Polytechnic Institute and State University, Blacksburg, Va. J. Alan Roberson is the director of federal regulations at AWWA, Washington, D.C. PEER REVIEW Date of submission: 03/27/2015 Date of acceptance: 10/06/2015 REFERENCES Ashley, D.M. & Kirkpatrick, K., The Governor s Water Task Force and the Georgia Water Stewardship Act. Georgia Institute of Technology, Atlanta. Aubuchon, C. & Roberson, J.A., Evaluating the Embedded Energy in Real Water Loss. Journal AWWA, 106:3:E jawwa AWWA, 2011 (1st ed.). AWWA Manual of Water Supply Practices M60, Drought Preparedness and Response. AWWA, Denver. AWWA, 2009 (3rd ed.). AWWA Manual of Water Supply Practices M36, Water Audits and Loss Control Programs. AWWA, Denver. Baird, G.M., Money Matters Who Stole My Water? The Case for Water Loss Control and Annual Water Audits. Journal AWWA, 103:10:22. CNT (Center for Neighborhood Technology), Stepping Up Water Loss Control Lessons From the State of Georgia. Chastain-Howley, A., Water Audits Got a Little Easier in Journal AWWA, 99:2:36. Cohen, B.R., Fixing America s Crumbling Underground Water Infrastructure. Competitive Enterprise Institute, Washington. Fiut, B. & Patience, M., Taking a Holistic Approach to Nonrevenue Water. Journal AWWA, 105:10:54. Jernigan, W., Director of water efficiency, Cavanaugh, Winston-Salem, N.C. Personal communication. Mathis, M.; Kunkel, G.; & Chastain-Howley, A., Water Loss Audit Manual for Texas Utilities. Texas Water Development Board, Austin. Thornton, J.; Sturm, R.; & Kunkel G., 2008 (2nd ed.). Water Loss Control. McGraw- Hill, New York. USEPA (US Environmental Protection Agency), Local Drinking Water Information. (accessed Jan. 5, 2015). USEPA, Public Drinking Water Systems: Facts and Figures. infrastructure/drinkingwater/pws/factoid s.cfm (accessed Mar. 16, 2015). USEPA, Control and Mitigation of Drinking Water Losses in Distribution Systems. EPA 816-R , Washington. WRF (Water Research Foundation), Real Loss Component Analysis: A Tool for Economic Water Loss Control. WRF, Denver.