Santa Margarita Water Audit and Water Loss Control Program Audit Period: Fiscal Year Water Loss Control Program Development

Transcription

1 Santa Margarita Water Audit and Water Loss Control Program Audit Period: Fiscal Year Water Loss Control Program Development TECHNICAL MEMO MAY 2016

2 Contents 1 Executive Summary Background Introduction to Methodology High Level Water Balance Results and Performance Indicators Results and Recommendations Water Audit Data Recommendations Water Loss Control Strategy and Recommendations Recommended Water Loss Control Program Timeline Water Supplied Water Supplied Background Water Supplied Definitions SMWD System Input Description Selected System Input Boundary and Data Sources Determination of Water Supplied Water Imported Water Imported - Master Meter and Supply Error Adjustments Water Exported Water Exported Master Meter and Supply Error Adjustments Final Determination of Water Supplied Results and Recommendations Authorized Consumption Authorized Consumption Background Determination of Authorized Consumption Billed Metered Authorized Consumption Billed Unmetered Authorized Consumption Unbilled Metered Authorized Consumption Unbilled Unmetered Authorized Consumption Final Determination of Authorized Consumption Results and Recommendations Apparent Losses Apparent Losses Background Determination of Apparent Losses Unauthorized Consumption... 29

3 Estimating Customer Meter Under Registration Systematic Data Handling Errors Final Determination of Apparent Loss Results and Recommendations Component Analysis of Real Losses Top Down Calculation of Real Losses Bottom-Up Component Analysis of Real Losses Validation of Real Loss Findings Results and Recommendations Pressure Pressure Survey Background System Average Pressure Results and Recommendations Water Loss Control Strategies Real Loss Intervention Strategy Introduction & Background Economic Frequency of Intervention: Proactive Leak Detection Pressure Management Apparent Loss Intervention Strategy Introduction & Background Recommended Small Meter Test Samples Recommended Large Meter Testing Schedule Appendix Water Supplied Analysis Boundary and Data Selection Sensitivity Analysis Justification for Selected System Input Boundary and Data Sources System Input Meter High Resolution Flow Profiling System Input Meter Installation Conditions South County Pipeline Volumetric Testing Authorized Consumption Analysis Evaluating Data Completeness Database Integrity Checks Time Sensitivity Analysis Meter Consumption Range Analysis... 88

4 Meter Consumption Distribution Real Losses Analysis Clerical Errors in Repair Records Reported Leak Clusters Methodology Sensitivity Analysis of Reported Leakage Pressure Analysis Monitoring Critical Infrastructure High and Low Pressures Average Pressures by Location Water Loss Control Strategies Background and Theory Theory of Economic Frequency of Intervention Small Meter Testing: Sample Selection Methodology Large Meter Population Summary Estimated Cost to Test Large Meters The Pressure Leakage Relationship and the N1 Factor List of Tables Table 1: Simple Water Loss Calculation... 9 Table 2: SMWD FY14 Performance Indicators Table 3: Water Loss Control Program Timeline Table 4: Supply Error Adjustment: Water Imported FY Table 5: Water Exported FY Table 6: Water Supplied Summary FY Table 7: BMAC by Meter Read Month Table 8: Summary of Unbilled Unmetered Authorized Consumption Table 9: Estimates for Hydrant Flushing Table 10: Summary of Unbilled Unmetered Authorized Consumption Table 11: Authorized Consumption Summary Table 12: Relationship Between Customer Meter Accuracy and Real Losses Table 13: Final Determination of Apparent Losses Table 14: Water Audit Result Table 15: Reported Leakaged Volume Table 16: Background Leakage Rates Table 17: Background Leakage Volume Table 18 Component analysis resusts Table 19: Logger Deployment Purpose Table 20: Average System Pressure Table 21: SMWD System Characteristics for UARL Formula Table 22: Unreported Leakage from UARL Formula... 51

5 Table 23: Benefits of Pressure Management Table 24: Pressure Reduction Scenarios Table 25: Small Customer Meter Sample (n = 73) Table 26: Medium Customer Meter Sample (n = 127) Table 27: Large Customer Meter Sample (n = 307) Table 28: Example Meter Replacement Schedule Table 29: Meter Testing Frequency Group Table 30: Large Meter Testing Scenario: Annual Loss of 0.50% Table 31: Large Meter Testing Scenario: Annual Loss of 1.0% Table 32: Large Meter Testing Scenario: Annual Loss of 1.50% Table 33: Sensitivity Analysis for Water Supplied Table 34: Source Meter High Resolution Flow Profiling Table 35: Installation Conditions for Supply Meters Table 36: Regulating Reservoir Volume Per Foot Table 37: Measurement Uncertainty Table 38: Drawdown Test Results Table 39: Adjusted Volumetric Drawdown Test Results Table 40: Count of Records per Location ID Meter ID Pair Table 41: Duplicate Readings Table 42: Location IDs with Negative Consumption Table 43: Consecutive Zero Readings by Location Meter ID Pair Table 44: Count of Locations, Customers, and Meters with No Consumption Table 45: Meters with Multiple Sizes Table 46: Meters with Multiple Manufacturers Table 47: Summary of Meter Reading Estimate Codes Table 48: Consecutive Meter Read Estimates Table 49: Comparison of Apportioned and Read Month BMAC Totals Table 50: BMAC by Meter Read Month and Apportionment Table 51: BMAC Time Lag Analysis Table 52: Meter Size Summary Table 53: Meter Make Summary Table 54: Meters Experiencing High Average Flow Rates Table 55: Errors in Leakage Repair Records Table 56: Reported Leakage Volume Calculated by Individual Leakage Events Table 57: Reported Leakage Volume Calculated byawwa 4372 model with defaul flow rates Table 58: Reported Leakage Volume calculated by AWWA 4372 model with average flow rates Table 59: Summary of methods used to calculate Reported Leakage Volume Table 60: Locations with Suspected High and Low Pressure Table 61: Summary Statistics by Location Table 62: Small Customer Meter Testing Sample Size Calculation: 50 Meter Target Table 63: Small Customer Meter Testing Sample Size Calculation: 100 Meter Target Table 64: Small Customer Meter Testing Sample Size Calculation: 250 Meter Target Table 65: Large Meter Summary by Size Table 66: Large Meter Summary by Manufacturer Table 67: Large Meter Summary by Rate Class

6 Table 68: Count of Large Meters by Rate Class and Size Table 69: Revenue from Large Meters by Rate Class and Size Table 70: Average Cost to Test and Repair Large Meters List of Figures Figure 1: SMWD Water Balance (AF) Figure 2: Relationship between System Input Volume, Water Exported, and Water Supplied Figure 3: International Water Association s Standardized Components of Annual Water Balance Figure 4: Basic System Input Diagram Figure 5: Selected System Input Boundary and Data Sources Figure 6: System Input Diagram Figure 7: Position of Authorized Consumption within the Water Balance Figure 8: Billed Metered Authorized Consumption by Meter Read Month Figure 9: Unbilled Metered Authorized Consumption by Meter Read Month Figure 10: Position of Apparent Losses within the Water Balance Figure 11: Distribution of Meter Installation Year Figure 12: Types of Real Losses Figure 13: Position of Real Losses within the Water Balance Figure 14: Feet of Water Mains Installed by Year Figure 15: Main Failure Frequency Literature Comparison Figure 16: Service Connection Failure Frequency Literature Comparison Figure 17: Pressure Survey Summary Map Figure 18: Four Interventions against Real Losses Figure 19: Economic Intervention Frequency for Proactive Leak Detection Figure 20: Meter Make by Installation Year Figure 21: Large Meter Revenue Distribution Figure 23: Selected System Input Boundary and Data Sources Figure 23: Plaza CM-12 Flow over Time FY Figure 24: SC-1 Flow over Time FY Figure 25: SC-2A Flow over Time FY Figure 26: SC- Flow over Time FY Figure 27: SC-4A Flow over Time FY Figure 28: SC-5A Flow over Time FY Figure 29: OC-82 Flow over Time FY Figure 30: OC-88 Flow over Time FY Figure 31: OC-88A Flow over Time FY Figure 32: El Toro Flow over Time FY Figure 34: Installation Conditions for Venturi Style Meter Figure 34: Unintended Discharge from Regulating Reservoir (Sample A) Figure 35: Unintended Discharge from Regulating Reservoir (Sample B) Figure 36: SCP Static Test Regulating Reservoir Level... 77

7 Figure 37: Count of Customers by Meter Read Month Figure 38: Count of Records by Meter Read Month Figure 39: Billed Metered Authorized Consumption by Meter Read Month Figure 41: Explanation of Apportionment Methodology Figure 41: BMAC Apportioned and by Meter Read Month Figure 42: Count of Meters by Size Figure 43: Total Consumption by Meter Size Figure 44: Meter Consumption Range Analysis Figure 45: Meter Consumption Range Analysis Log-Linear Scale Figure 46: Consumption Distribution Figure 47: Cluster of Plastic Main Service Leaks Figure 48: Large Transient: Antonio Parkway Figure 49: Complete Pressure Profile at Antonio Parkway Figure 50: Long Duration Transient at Talega Pump Station Figure 51: Pump Operation without Transients Figure 52: Economic Model for Regular Leak Detection Survey Figure 53: Rate of Rise of Leakage Concept Figure 54: Large Meter Revenue by Size Figure 55: Large Meter Revenue by Manufacturer Figure 56: Large Meter Summary by Rate Class List of Attachments Attachment A AWWA Water Audit Attachment B 4372 Real Loss Component Analysis Attachment C Authorized Consumption Tables Attachment D Large Meter Testing Frequency List of Acronyms ABI: Annual Budget for Intervention AF: acre-feet AWWA: American Water Works Association BMAC: Billed Metered Authorized Consumption CARL: Current Annual Real Losses CCF: hundred cubic feet CI: cost of intervention, referring to the cost of leak detection survey intervention CV: cost of Real Losses EIF: Economic Intervention Frequency ELL: Economic Level of Leakage EP: economic percentage of the system that should be covered by leak detection survey each year GPM: gallons per minute ICF: Infrastructure Condition Factor ILI: Infrastructure Leakage Index IWA: International Water Association KGAL: thousand gallons

8 MET: Metropolitan Water District MNF: Minimum Night Flow NRW: Nonrevenue Water PSI: pounds per square inch RR: rate of rise of leakage SCADA: Supervisory Control and Data Acquisition SIV: System Input Volume SMWD: Santa Margarita Water District UARL: Unavoidable Annual Real Losses UBL: Unavoidable Background Leakage WSO: Water Systems Optimization

9 1 Executive Summary 1.1 Background Water System Optimization (WSO) worked with the Santa Margarita Water District (SMWD) to conduct a thorough water audit and develop a water loss control program. The audit period examined was fiscal year 2014 (FY14), beginning July 1, 2013 and ending June 30, Through the completion of the water audit, WSO quantified SMWD s baseline volume of non-revenue water. From there, WSO identified opportunities that could reduce the system s costs and non-revenue water, and increase revenue. It s important to note that SMWD has extremely low levels of Water Losses as a result many of the recommendations provided are intended to further reduce uncertainty and improve data quality. Currently there are relatively few economically justifiable Real Loss reduction strategies available to SMWD because of the very low levels of Water Loss. 1.2 Introduction to Methodology The American Water Works Association (AWWA) Water Balance uses methodology developed by the International Water Association (IWA) to account for all water entering and leaving a distribution system. This water audit utilizes the IWA/AWWA standardized Water Balance methodology to disaggregate and validate components of System Input Volume, Consumption Volume, Apparent Loss Volume, and Real Loss Volume in order to identify strategies to reduce Water Loss Volumes. The basic component volumes of the Water Balance for SMWD are as follows: System Input Volume (SIV) includes water imported, changes in reservoir storage, and corrections for known source meter inaccuracy. System Input Volume determination is detailed in the section, Water Supplied. Authorized Consumption includes metered and unmetered water consumed by customers and other uses that SMWD authorizes. The main component of Authorized Consumption is Billed Metered Authorized Consumption (BMAC). Other components of Authorized Consumption include water for system flushing and fire-fighting, and SMWD facility consumption. The components of Authorized Consumption are calculated and explained in the section, Authorized Consumption. Water Losses are calculated by subtracting Authorized Consumption from System Input Volume. This calculation is shown below in Table 1. Water Losses are then divided into two main categories, Apparent Losses and Real Losses. Simple Water Loss Calculation Volume FY14 Volume (AF) A System Input Volume 28, B Authorized Consumption 27, = A - B Water Losses 1, Table 1: Simple Water Loss Calculation Apparent Losses are non-physical losses that occur due to customer meter inaccuracies, data handling errors, and water theft. The term apparent is applied because water is consumed but is not properly measured. Apparent Losses are explained further and calculated in the section titled, Apparent Losses.

10 Real Losses are physical water losses such as leaks, breaks and overflows. The Real Loss Volume is the remaining volume after Authorized Consumption and Apparent Losses are subtracted from System Input Volume. Real Losses are characterized as Reported Leaks, Unreported Leaks, and Background Leaks. The section, Component Analysis of Real Losses, elaborates on the component analysis of Real Losses in which Real Losses are determined for each leak that occurred during the audit period. An analysis of economically-efficient Real Loss reduction strategies was performed based on the component analysis of Real Losses and is presented in section Water Loss Control Strategies. 1.3 High Level Water Balance Results and Performance Indicators Figure 1 below shows SMWD s complete Water Balance for the 2014 audit period in acre-feet. Each full column is equal in volume. Please note that the Water Balance is not to scale (i.e. the area of each box is not proportional to its respective volume). Water Supplied 28, Authorized Consumption 27, Water Losses 1, Billed Authorized Consumption 27, Unbilled Authorized Consumption Apparent Losses Billed Metered Consumption 27, Revenue Water Billed Unmetered Consumption 27, Unbilled Metered Consumption Unbilled Unmetered Consumption Unauthorized Consumption Non-Revenue Water Customer Metering Inaccuracies 1, Systematic Data Handling Errors Real Losses 1, Figure 1: SMWD Water Balance (AF) Using the water balance, basic infrastructure information, and a handful cost parameters, the AWWA water audit software calculates a suite of standard performance indicators, displayed in Table 2 on the following page.

11 FY 2014 Performance Indicators Score (FY14) Units Financial Performance Indicators Non-Revenue as percent by volume of Water Supplied 4.90% Non-Revenue as percent by cost of operating system 3.30% Annual cost of Apparent Losses $254,809 Valued at customer retail unit cost Annual cost of Real Losses $1,087,619 Valued at variable production cost Operational Efficiancy Performance Indicators Apparent Losses per service connection per day 2.41 gal / conn / day Real Losses per service connection per day gal / conn / day Real Losses per service connection per day per PSI of pressure 0.22 gal / conn / day / PSI Unavoidable Annual Real Losses (UARL) MG / yr Current Annual Real Losses (CARL) MG / yr Infrastructure Leakage Index (CARL/UARL) 1.0 ratio Data Validity Performance Indicator Data Validity Score 72 Table 2: SMWD FY14 Performance Indicators To assess performance, WSO recommends that SMWD consider all performance indicators, with a particular emphasis on Apparent Losses per service connection per day, Real Losses per service connection per day, and the Infrastructure Leakage Index (ILI). SMWD s water audit results indicate that the system is performing very efficiently, close to the technical minimum level of Real Losses. WSO interprets SMWD s FY14 performance indicators as follows: The ILI value of 1.0 indicates that SMWD s distribution infrastructure is leaking at the predicted technical minimum level for a system with similar physical characteristics. WSO recommends that SMWD continue to monitor annual Real Losses to maintain an ILI around 1.0. Real Losses of gallons per service connection per day indicates that SMWD s system is one of the tightest systems in the state. The top-performing 20% of California utilities lose less than 25 gallons per service connection per day. 1 Real Losses at SMWD s level demonstrates very low levels of leakage. Apparent Losses of 2.41 gallons per service connection per day suggests that SMWD is receiving revenue for almost all of the water it delivers. However, the assessment of apparent loss was estimated in the absence of customer meter test data. Given the annual cost of additional Apparent Losses, studying and then targeting improvement of the accuracy of customer meters could recover additional revenue without incurring undue expense. An overall Data Validity Score of 72 communicates that SMWD s water audit data is reliable enough to serve as the foundation of an informed water loss control program. As for all systems, room for data improvements remains, which at times might not be cost effective to pursue. 1 Data retrieved from Water Research Foundation (WRF) project 4372B, which examined a large data set of water audits. These audit results were not validated, so the data must be interpreted cautiously.

12 1.4 Results and Recommendations Given SMWD s very low water loss levels and overall good quality data the following recommendations are geared towards maintaining these low water loss levels and minor improvements to SMWD s water audit data. Water Audit Data Recommendations Because SMWD s distribution system has very low levels of water losses, any uncertainty in the volume of Water Supplied and Authorized Consumption can have a large impact on the overall findings of the water balance. For this reason, many of the recommendations presented below represent opportunities to further reduce uncertainty in the determination of these volumes. Increased certainty will facilitate more accurate tracking of Water Loss performance over time which can provide advanced warning of any increases in Water Loss in future years. WSO s recommendations to further improve data reliability are presented by volume below Water Supplied Recommendations WSO recommends using daily production data to determine water supplied in future audits because of billing adjustments on the South County Pipeline (SCP), and the additional resolution it provides. One exception to this recommendation is OC-82, where SMWD should use monthly data until issues with the data transfer protocol from MET are resolved. Almost 65% of water delivered to the SMWD distribution system comes from the South County Pipeline (SCP). WSO recommends continuing to track differences between the sum of flows from OC-88 and OC-88A and the volumes recorded by the takeout meters on the SCP. While SMWD is already doing this, continuing to collaborate with districts that also have takeouts on the SCP will facilitate proactive error detection in OC-88, OC-88A and SCP takeout meters. WSO recommends that where feasible SMWD begin an annual volumetric or comparative testing program for all input and export meters in its jurisdiction. Tests can be conducted by passing a known reference volume through the meter, or by comparing the volume recorded by an insertion meter installed in line with the tested meter. SMWD might consider urging MET and MWDOC to test their meters annually in addition to ongoing calibration efforts. All system input meters were installed according to MET specifications. However, based on the drawings provided, industry-standard installation conditions were not met for four of five of the system input meters. This contributes to an increased likelihood that these meters may have inaccurate registration. SMWD should investigate the as-built drawings further to determine how accurately they reflect the actual installation conditions. Authorized Consumption Recommendations Where feasible, SMWD should record estimates of Unbilled Unmetered Authorized Consumption, such as sewer and water mains flushing, using flow rate and time parameters. SMWD should also continue to record estimates for water used to flush fire hydrants. As outlined in the appendix Time Sensitivity Analysis, meter reading lag time appears to have a negligible impact on the calculation of BMAC. However, SMWD should be aware of the potential introduction of uncertainty in future water audits if meter reading cycles are changed.

13 Water Loss Control Strategy and Recommendations SMWD s water losses are very low which means there is limited room to reduce water losses any further. The recommendations and water loss strategy developed for SMWD are designed to help maintain such low levels of water losses, counteracting an increase in water losses caused by aging infrastructure. These recommendations were developed to provide a strategy for the next five years (see Table 3) after which it is recommended to re-evaluate additional data to develop the strategy for the next five years. Apparent Losses Recommendations In order to evaluate the optimum replacement cycle for SMWD s small meter population, WSO recommends that SMWD begin testing random samples of small customer meters. A small meter testing program is outlined in the section, Recommended Small Meter Test Samples. It is recommended that SMWD conducts random testing of small meters for the first two years of this five year water loss control program. In year three, SMWD should have enough data to evaluate if the current small meter replacement schedule is providing SMWD with the desired return on investment or if the replacement schedule should be revised. The last two years of the water loss program SMWD should implement the refined replacement schedule (if applicable) and continue testing small samples of randomly selected meters. To further optimize revenue generation from SMWD s large meter population it is recommended that SMWD test between 5 and 21 large meters more than once every three years to ensure minimal loss in accuracy and revenue. A large meter testing program is outlined in the section, Recommended Large Meter Testing Schedule. This component of the water loss control program should be initiated in year one and continued throughout the duration of the five year program. Pressure Management Recommendations SMWD should continue to review pump operation practices to emphasize soft starts and stops of all pump stations to avoid or limit transients. Most importantly in the short run is to adjust the start and stop sequence at each pump station when system demand changes significantly (changes in demand between summer and winter) to avoid the occurrence of transients. This should be an ongoing strategy throughout the first five years of SMWD s water loss control program. SMWD should continue to collect further pressure data throughout the system to gain a more detailed picture of pressure fluctuations, transients, areas with excessive pressure, and areas experiencing low service pressure. It is recommended to implement the pressure data collection over the first three years of the water loss control program. In year four of the program SMWD should have sufficient data to evaluate the potential for further pressure management strategies and to conduct a feasibility study on implementing operational pressure zones in certain parts of the system. In year five and subsequent years, if cost effective, SMWD should implement formal operational pressure zones zones where pressure can be controlled in isolation of the rest of the distribution system. Current pressure zones are defined by hydraulic gradient, or elevation, and do not provide the same management benefits as operational pressure zones.

14 Leak Detection Recommendations Given the very low levels of Real Losses (leakage losses) in the SMWD system, WSO recommends that SMWD not conduct leak detection in year one of the water loss control program. However, SMWD should conduct a two year pilot leak detection program of approximately 20% of the distribution system each year to recover Hidden Leakage Losses currently running in the pipe network and field validate the low level of Real Losses suggested by the water balance. This pilot leak detection effort should focus on areas with corrosive soil and regions with clusters of reported leakage. The leak detection strategy for year four and five should be based on the findings of the two year pilot leak detection program. Recommended Water Loss Control Program Timeline WSO recommends that SMWD implement the recommendations outlined above over the next 5 years. The program timeline, presented in Table 3, acknowledges capacity to implement recommendations and provides additional opportunity to collect information and refine interventions. Year 1 Year 2 Year 3 Year 4 Year 5 Leak Detection Pilot: 20% of system Pilot: 20% of system Informed by Pilot Results Informed by Pilot Results SMWD Water Loss Control Program Timeline Small Customer Pressure Meters Data Collection Data Collection Data Collection Ongoing Review of Pump Control Practices Ongoing Review of Pump Control Practices Ongoing Review of Pump Control Practices Review of Pressure Data + Feasability Study to Establish Operational Pressure Zones Implimentation of Pressure Management Program Based on Feasability Study Random Test Sample Random Test Sample Refine Replacement Strategy Based on Random Test Results Continue Replacement Schedule + Random Testing Continue Replacement Schedule + Random Testing Table 3: Water Loss Control Program Timeline Large Customer Meters Ongoing Targeted Testing, Repair, and Replacement * SMWD should urge MWDOC and MET to test meters and share the results in addition to current annual calibration efforts

15 2 Water Supplied 2.1 Water Supplied Background Water Supplied Definitions The determination of Water Supplied is the first step and the foundation of the water balance. Because Water Supplied is the largest volume in the water balance, error will have an especially large impact on audit findings. Therefore, it is critical to determine this volume thoroughly. Water Supplied: The volume of treated and pressurized water input to the retail water distribution system. For SMWD s Water Audit, Water Supplied is calculated according to the following formula: Water Supplied = Water Imported +/- Water Imported Master Meter and Supply Error Adjustments Water Exported +/ Water Exported Master Meter and Supply Error Adjustments 2 Master Meter and Supply Error Adjustments describe any over- or under-registration detected from supply meter test results. For example, WSO and SMWD conducted draw-down testing on several takeout meters on the South County Pipeline. The results of that test will constitute an adjustment to the volume of water recorded by the Master Meters that measure Water Imported to account for metering inaccuracy. Figure 2 below illustrates the relationship between Water Supplied and other key volumes. Volume from Own Sources (corrected for known errors) Water Exported (corrected for known errors) System Input Volume Water Supplied Water Imported (corrected for known errors) Figure 2: Relationship between System Input Volume, Water Exported, and Water Supplied 2 In the AWWA model, positive entries into Master Meter and Supply Error Adjustments are treated as a reduction in the total volume, and negative values as an addition. (+/-) adjustments to water imported and exported must be informed by test results.

16 Figure 3 shows the position of Water Supplied within the context of the Water Balance. (The sizes of each box in the figure are not to scale.) 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 Unauthorized Consumption Customer Meter Inaccuracies Data Handling Errors Revenue Water Non- Revenue Water Real Losses Figure 3: International Water Association s Standardized Components of Annual Water Balance SMWD System Input Description SMWD provides potable water to its customers solely from imported water. To estimate Water Supplied we must draw a discrete boundary around the distribution system. The boundary must allow SMWD to quantify all water entering and leaving the system using data from water meters or rigorous estimation. First, we will describe the possible system input meters. In the next section, we will outline how we ultimately defined the system boundary and why. Figure 4 shows the input system for SMWD and a brief description for each possible system input meter follows. Aufdenkamp CM-12 OC-82 El Toro OC-88 OC-88A Plaza PS Lakeside SCP SMWD Distribution System Other Agencies Portola Plano TCWD Storage Figure 4: Basic System Input Diagram Available system input meters: OC-88 and OC-88A: The Metropolitan Water District of Southern California (MET) operates both OC-88 and OC-88A. The Municipal Water District of Orange County (MWDOC) facilitates delivery of water from MET to SMWD. OC-88 and OC-88A feed each of the downstream takeouts on the South County Pipeline (SCP). SCP Takeouts: The South County Pipeline (SCP) distributes water to several independent water districts including Santa Margarita, San Clemente, Moulton Niguel, San Juan Capistrano, and South Coast. SMWD receives water from five takeouts equipped with custom-built Venturi meters: SC-1, SC-2A, SC-3, SC-4A, and SC-5A.

17 Aufdenkamp CM-12: MET operates a meter upstream from the Plaza Pump Station. Plaza Pump Station (Plaza PS): Laguna Beach County Water District operates an electromagnetic meter downstream from MET s Aufdenkamp CM-12. OC-82: The Metropolitan District of Southern California (MET) operates OC-82. OC-82 is upstream of two SMWD owned insertion meters at the Lakeside Pump Station. Lakeside Pump Station: SMWD operates two insertion meters at the Lakeside Pump Station. SMWD staff feel the data from these meters could be improved. However the total cost of improvement was estimated by SMWD staff as approximately $300,000. El Toro R-6: SMWD owns capacity on the El Toro reservoir in the El Toro Water District. As needed, SMWD accesses this capacity to supplement other water sources. For example, in January, SMWD drew AF to supplement supply when OC-82 was off. The configuration of SMWD s system inputs presents a number of options for defining the system boundary. In addition to selecting the physical meters that define the boundary, there are two primary sources of data for each meter SMWD s daily production data and MWDOC billing data. Available Data Sources: Daily Data: SMWD logs daily production totals for most system input meters through SCADA. WSO has reviewed daily data for all system input meters except the Lakeside Pump Station. Lakeside was omitted from review because SMWD staff do not trust it. In general, daily data is preferable because it provides a higher-resolution view of production patterns over time. This higher resolution makes it easier to detect any anomalies in the data. Daily data is also recorded directly from the meter. Therefore, no intermediary financial adjustments have been made to these data. MWDOC Invoices/Monthly Data: MWDOC invoices provide the billed monthly total for each system input meter managed by MET (SCP Takeouts, Aufdenkamp CM-12, and OC-82). MWDOC invoices do not provide the same level of resolution as daily data, so any anomalies may run undetected. Monthly invoices for takeouts on the SCP are subject to volumetric adjustments to align the sum total of readings from takeout meters with the total registered by OC-88 and OC- 88A. Sensitivity analyses exploring the impact of boundary and data choices can be found in the appendix, Boundary and Data Selection Sensitivity Analysis. Selected System Input Boundary and Data Sources WSO ultimately selected the system input boundary based on meter proximity to the distribution system, confidence in meter accuracy, and SMWD ownership. WSO used daily data for all selected system input meters, except for OC-82. Monthly MWDOC invoices were used for OC-82 because of a known data transfer issue with the daily MET transmitter. A detailed discussion of the selected data sources and system boundary can be found in the appendix, Justification for Selected System Input Boundary and Data Sources.

18 Figure 5 below shows WSO s recommended system input boundary and data sources. In the next section, we will calculate the total volume of water supplied for the audit period using this boundary and set of data sources. Aufdenkamp CM-12 OC-82 El Toro OC-88 OC-88A Plaza PS Lakeside Daily Production Data SCP SMWD Distribution System MWDOC Billing Data Other Agencies Portola Plano TCWD Storage System Boundary 2.2 Determination of Water Supplied Figure 5: Selected System Input Boundary and Data Sources Water Imported WSO used data provided by SMWD to determine water imported during FY14. The values below reflect the system boundary and data sources discussed in the section, Selected System Input Boundary. WSO has included changes in distribution system storage in the calculation of Water Supplied. We have included all potable reservoirs located within the distribution system to calculate changes in storage. This excludes Upper Chiquita Reservoir (UCR), which is located outside the audit boundary; any changes in storage at UCR are taken into account by the takeout meters on the SCP. Per AWWA methodology, changes in reservoir storage apply to Water Supplied. When water is stored, it is effectively taken out of the system and should be treated as a decrease in Water Supplied. However, when water is released from storage it re-enters the distribution system and can therefore be treated as an increase in Water Supplied. Positive values in Table 33indicate increases in storage (a decrease in Water Supplied) while negative values indicate decreases in storage (an increase in Water Supplied). Table 33below summarizes Water Imported for each takeout meter during FY14 including changes in storage. Water Imported FY14 Source AF Percent + SC-1 4, % + SC-2A 8, % + SC-3 2, % + SC-4A % + SC-5A 1, % + Plaza CM % + AMP OC-82 10, % + El Toro R % - Changes in Storage % = TOTAL 29, % Table 3: Water Imported FY14

19 Water Imported - Master Meter and Supply Error Adjustments It is important to consider any known source meter inaccuracies when determining Water Supplied. SMWD and WSO conducted volumetric testing on four of the five takeout meters on the South County Pipeline. A detailed review of the results and methodology employed to conduct this testing is provided in the appendix, South County Pipeline Volumetric Testing. During the drawdown test, there was an unintended discharge that may skew the interpretation of test results. WSO recommends using the unadjusted test results to correct the volume of Water Imported because it is uncertain if the unintended discharge was running during the tests themselves. Table 4 outlines WSO s adjustment to Water Imported based on the unadjusted meter test results. Meters that have not been tested were assumed to be 100% accurate. Where feasible, SMWD should test additional system input meters, either volumetrically or with a comparative insertion meter. Meters that record the largest portions of imports should be prioritized for testing. Source Supply Error Adjustment: Water Imported FY14 Total (AF) Percent of Total Meter Accuracy Adjusted Volume (AF) Table 4: Supply Error Adjustment: Water Imported FY14 Difference (AF) + SC-1 4, % 96.2% 4, SC-2A 8, % 98.9% 8, SC-3 2, % 96.4% 2, SC-4A % 100.0% SC-5A 1, % 98.5% 1, Plaza CM % 100.0% AMP OC-82 10, % 100.0% 10, El Toro R % 100.0% Changes in Storage % 100.0% = TOTAL 29, % 98.7% 29, Analyzing high-resolution flow data and evaluating meter installation conditions can provide deeper insight into the potential for uncertainty in the calculation of Water Supplied. These additional analyses are provided in the appendices, Water Supplied Analysis. Water Exported WSO used data provided by SMWD to determine FY14 Water Exported. FY14 Water Exported amounted to AF, or approximately 4 % of total Water Supplied. Table 5 shows water exported for FY14.

20 Water Exported FY14 Source AF Percent - Portola Lake Fill Exp 1, % - Plano TWDC Export % = TOTAL -1, % Table 5: Water Exported FY14 Water Exported Master Meter and Supply Error Adjustments Export meters are not regularly tested or calibrated. For this reason, WSO did not adjust FY14 volumes associated with these meters. These meters recorded a significant volume of water, 4% of Water Supplied. Small inaccuracies in the volumes registered by export meters may have a minor impact on the volume of Water Supplied. However, the compounding effect of small inaccuracies in export volumes results in proportionally greater potential error in the ultimate derivation of Real Losses. Final Determination of Water Supplied Water Supplied including master meter and supply error adjustments is determined in Table 6. Water Supplied FY14 Category Volume (AF) Water Imported 29, Water Imported Master Meter and Supply Error Adjustments Water Exported 1, Water Exported Master Meter and Supply Error Adjustments - = TOTAL 28, Table 6: Water Supplied Summary FY14 Figure 6 on the following page shows the total system input configuration along with all the recorded volumes. Note that Master Meter and Supply Error Adjustments have been applied to the totals for each system input meter presented in the diagram.

21 SMWD System Input Diagram OC-88 & OC-88A Billed volumes adjusted based on SC takeout flows. Aufdenkamp CM-12 Suspect data transfer protocol OC-82 MET Meters Upper Chiquita Reservoir SC-1 SC-2A 4,312.6 AF 8,121.4 AF Plaza CM-12 Mag meter AF Lakeside Insertion meter 10,921.1 AF AF El-Toro R6 Capacity on the El Toro reservoir SC-3 SC-4A 2,889.4 AF AF SMWD Distribution System Net recorded input: (28,403 AF) -1,106.5 AF AF 26.2 AF MET/Other Meters SMWD Meters Other SCP Takeouts SC-2B, SC-4, SC-5B SC-5A 1,830.6 AF Portola Lake Fill Export TCWD Export Trabuco Canyon Water District Storage Distributed capacity throughout system Distribution System System Boundary Figure 6: System Input Diagram

22 2.3 Results and Recommendations Because SMWD s distribution system has very low levels of water losses, any uncertainty in the volume of water supplied can have a large impact on the overall findings of the water balance. For this reason, all of the recommendations presented below represent opportunities to reduce uncertainty in the determination of Water Supplied. Increased certainty will facilitate more accurate tracking of Water Loss performance over time which can provide advanced warning of any increases in Water Loss in future years. Data Management 1. WSO analyzed two potential sources of data to determine Water Supplied daily production data from SCADA and monthly billing data from MWDOC invoices. WSO recommends using daily production data for future audits because of billing adjustments on the SCP, and the additional resolution it provides. One exception to this recommendation is OC-82, where SMWD should use monthly data until issues with the data transfer protocol from MET are resolved. 2. Almost 65% of water delivered to the SMWD distribution system comes from the South County Pipeline (SCP). WSO recommends continuing to track differences between the sum of flows from OC-88 and OC-88A and the volumes recorded by the takeout meters on the SCP. While SMWD is already doing this, continuing to collaborate with districts that also have takeouts on the SCP will facilitate proactive error detection in OC-88, OC-88A and SCP takeout meters. Operations 1. WSO recommends that where feasible SMWD begin an annual volumetric or comparative testing program for all input and export meters in its jurisdiction. Tests can be conducted by passing a known reference volume through the meter, or by comparing the volume recorded by an insertion meter installed in line with the tested meter. SMWD might consider urging MET and MWDOC to test their meters annually in addition to ongoing calibration efforts. 2. As outlined in the appendix, System Input Meter High Resolution Flow Profiling, Almost 30% of the non-zero flow readings at SC-1 during the audit period were outside of the manufacturers recommend range. During these times, SMWD intended for SC-1 to be off when it was actually registering very low flows. Because these low flow rates fall outside of the meter s accurate range of measurement, it is uncertain exactly how much water passed through the meter during that time. Operational consequences are minimal, if any. However, for the purposes of completing a water audit, flows recorded outside of the meter s accurate range introduces uncertainty. SMWD should minimize long durations of low flows to increase certainty for the determination of Water Supplied. 3. All system input meters were installed according to MET specifications. However, based on the drawings provided, industry-standard installation conditions were not met for four of five of the system input meters. This contributes to an increased likelihood that these meters may have inaccurate registration. SMWD should investigate the as-built drawings further to determine how accurately they reflect the actual installation conditions.

23 3 Authorized Consumption 3.1 Authorized Consumption Background Authorized Consumption is the volume of water used by all registered consumers including residential, industrial, commercial, and agricultural users, and the utility itself. Water used for firefighting and infrastructure maintenance (e.g. distribution main flushing) is also considered Authorized Consumption. Authorization of use can be explicit or implicit. In order to subdivide Authorized Consumption into component volumes, consumption is categorized as billed or unbilled and metered or unmetered. Once quantified, Authorized Consumption is subtracted from Water Supplied, to calculate Water Losses. In order for this determination of Water Losses to be accurate, the validity of contributing data sources must be considered. Figure 7 places Authorized Consumption in the broader context of the water audit. Please note that the boxes are not drawn to scale, the size of each box is not proportional to its 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 Unauthorized Consumption Customer Meter Inaccuracies Data Handling Errors Revenue Water Non- Revenue Water Real Losses Figure 7: Position of Authorized Consumption within the Water Balance 3.2 Determination of Authorized Consumption Billed Metered Authorized Consumption Background The determination of the volume of Billed Metered Authorized Consumption (BMAC) is a major component of the AWWA water balance. According to the AWWA M36 manual: Billed Metered Authorized Consumption includes uses such as residential use, commercial use, municipal use, industrial use, and other uses. According to the AWWA methodology the audit includes only the potable water system. In the following sub-sections, we will calculate Billed Metered Authorized Consumption. In addition to performing this calculation, WSO conducted several analyses to evaluate the completeness and integrity of the billing data set and to gain insight into the appropriate sizing of customer meters.the following analyses are provided in the appendix, Authorized Consumption Analysis :

24 Evaluating Data Completeness: These analyses describe the count of accounts and records per month. WSO also evaluated the number of bills assessed to each location and meter. All results indicate that the raw billing data is complete. Database Integrity Checks: WSO checks for basic data integrity such as duplicate readings, negative consumption volumes, and abberant large readings. All indications suggest a very clean and well-organized billing system. Meter Consumption Range Analysis: Billing data can provide an indication for potential meter sizing issues. WSO examines the billing data to identify any meters experiencing higher-thanaverage flows than similarly sized peers. WSO also tests to ensure that each increasing meter size class is appropriately exposed to increasing average flows.two small groups of meters, 1.5-inch and 2.5-inch meters, are exposed to a broad range of flows and may be incorrectly sized. WSO also identified several meters experiencing very high flows on average compared to their similarlysized peers. WSO also investigated consumption patterns by Rate Class, Pressure Zone, Service Type, and meter reader Comment Code. The results of these investigations did not suggest any significant alterations to operational practices or the calculation of BMAC. Exclusions The raw billing data included data that must be excluded to calculate BMAC: 1. Records with the Service Type WN and IN should be excluded because they reflect consumption on recycled water accounts. This water balance is for the potable distribution system only, therefore recycled water consumption should be excluded. 2. District facility usage is indicated by records with the rate class DF, which must be set aside to calculate Unbilled Metered Authorized Consumption. A discussion of District facility use can be found in section titled, Unbilled Metered Authorized Consumption. 3. Customers that draw directly from the South County Pipeline (SCP), colloquially called bootlegs, must be excluded. The South County Pipeline falls outside of the boundary set by the system input meters. The names and IDs for customers that draw directly from the SCP are LaPeyre: and Christianitos: Determination of Billed Metered Authorized Consumption Table 7 and Figure 39 below show the consumption profile by meter read month. The consumption pattern is typical for California water agencies most consumption occurs during the summer months. The total volume for Billed Metered Authorized Consumption during the audit period is 11,773,356 CCF or 27,026 AF.

25 BMAC by Meter Read Month Total Consumption Month (CCF) Audit May ,882 Out of Audit Jun 13 1,118,998 Out of Audit Jul 13 1,258,352 In Audit Aug 13 1,226,488 In Audit Sep 13 1,250,858 In Audit Oct 13 1,129,126 In Audit Nov ,914 In Audit Dec ,800 In Audit Jan ,570 In Audit Feb ,868 In Audit Mar ,446 In Audit Apr ,786 In Audit May 14 1,052,259 In Audit Jun 14 1,183,889 In Audit Jul 14 1,215,135 Out of Audit Aug 14 1,228,188 Out of Audit Total 11,773,356 Table 7: BMAC by Meter Read Month Figure 8: Billed Metered Authorized Consumption by Meter Read Month Billed Unmetered Authorized Consumption SMWD bills several unmetered condominiums at a flat rate. Table 8 outlines the calculation used to estimate Billed Unmetered Authorized Consumption during the audit period. The total volume for Billed Unmetered Authorized Consumption was AF.

26 Calculating Billed Unmetered Authorized Consumption A Consumption Per Month 140 B Months in Audit Period 12 =A*B*748.05/ Total Consumption (AF) Table 8: Summary of Unbilled Unmetered Authorized Consumption Unbilled Metered Authorized Consumption Records in the billing dataset with the rate class DF represent district facility use. Figure 9 shows the consumption profile for district facility use throughout the audit period it follows a typical seasonal pattern with the majority of use occurring during the hotter and dryer summer months. However, district facility use peaked in October of 2013, after summer. SMWD dramatically decreased its facility use in November of The total volume of Unbilled Metered Authorized Consumption was AF. Figure 9: Unbilled Metered Authorized Consumption by Meter Read Month Unbilled Unmetered Authorized Consumption Unbilled Unmetered Authorized Consumption is typically comprised of sewer and water main flushing in addition to firefighting, fire-fighter training, and hydrant flushing. Often, these uses are not documented. WSO must estimate undocumented uses based on the volume of Water Supplied. During the audit period, SMWD did not document most forms of UUAC however, estimates for hydrant flushing use per month were available. Table 9 below shows the available hydrant flushing records per month. Where data was not available, WSO used the average of the recorded months. The total volume of water used to flush hydrants in FY14 was AF.

27 Hydrant Flushing Recorded Month Volume (gal) * WSO Estimate Jul ,912 Aug ,912 Sep ,912 Oct ,912 Nov ,912 Dec-13 1,513,714 1,513,714 Jan , ,766 Feb , ,291 Mar , ,727 Apr-14 35,500 35,500 May , ,476 Jun ,912 Total (AF) * Months with "-" indicate no data was available Table 9: Estimates for Hydrant Flushing WSO used 0.1% of Water Supplied to estimate uses that are not tracked. This estimate was informed by staff testimony that SMWD has significantly reduced operational uses during drought conditions. The total calculation for Unbilled Unmetered Authorized Consumption is shown in Table 10.The total volume of UUAC is the sum of undocumented uses and hydrant flushing, or AF. Total Estimate for UUAC Operation Category Volume A Water Supplied 28,408 B Hydrant Flushing C = A * 0.1% Other Undocumented Uses Total UUAC Table 10: Summary of Unbilled Unmetered Authorized Consumption

28 3.3 Final Determination of Authorized Consumption The total Authorized Consumption during the audit period was 27, AF. Table 11 below summarizes the four volumes that comprise Authorized Consumption. Authorized Consumption Category Volume (AF) Billed Metered Authorized Consumption 27, Billed Unmetered Authorized Consumption Unbilled Metered Authorized Consumption Unbilled Unmetered Authorized Consumption TOTAL 27, Table 11: Authorized Consumption Summary 3.4 Results and Recommendations SMWD has very low levels of Water Losses. Therefore, any uncertainty in the determination of Authorized Consumption can have a large impact on the overall findings of the water balance. WSO conducted extensive exploratory analysis outlined in the appendix, Authorized Consumption Analysis. These analyses attempted to detect any errors in the billing data the results suggest that no noteworthy inconsistencies are present in the raw billing data. In this context, the following recommendations are intended to minimize uncertainty in the calculation of Authorized Consumption. Recommendations Where feasible, SMWD should record estimates of Unbilled Unmetered Authorized Consumption, such as sewer and water mains flushing, using flow rate and time parameters. SMWD should also continue to record estimates for water used to flush fire hydrants. As outlined in the appendix Time Sensitivity Analysis, meter reading lag time appears to have a negligible impact on the calculation of BMAC. However, SMWD should be aware of the potential introduction of uncertainty if meter reading cycles are changed. As outlined in the appendix Meter Consumption Distribution, 1% of meters are responsible for almost 20% of Billed Metered Authorized Consumption. Special care should be taken by SMWD to ensure the accuracy of these meters. Specific recommendations for large and small meter testing programs are provided in the section, Water Loss Control Strategies.

29 4 Apparent Losses 4.1 Apparent Losses Background Apparent Losses describe all paper water losses during the audit period. It is distinct from Real Losses in that it does not describe physical losses of water. Three categories capture all Apparent Losses in the AWWA Free Audit Software: 1. Unauthorized Consumption, or Theft: Examples of theft include misuse of fire hydrants and firefighting systems on unmetered fire lines, bypassed or tampered consumption meters, and illegal connections 2. Customer Meter Under Registration: Most customer meters under-register as they age that is, they typically record less than 100% of the volume passing through. These losses can add up, and usually make up the largest portion of Apparent Losses. 3. Systematic Data Handling Errors: Through the meter reading, data entry, and the billing process, there are often numerous opportunities for error. Billing system data integrity checks provide an indication for the total volume of loss due to data handling errors. Figure 10below shows how Apparent Losses fit into the broader context of the water balance. Please note that the boxes are not drawn to scale. 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 Unauthorized Consumption Customer Meter Inaccuracies Data Handling Errors Revenue Water Non- Revenue Water Real Losses Figure 10: Position of Apparent Losses within the Water Balance 4.2 Determination of Apparent Losses Unauthorized Consumption In the absence of additional information about theft in the SMWD distribution system, WSO used the AWWA default estimate for theft of 0.25% of Water Supplied, or AF. While it is not recommended in order to improve the accuracy of this estimate, SMWD would have to go to great lengths to detect and document all instances of theft. The overall volume of theft is typically small, and therefore has a relatively small impact on the overall findings of the water audit.

30 Estimating Customer Meter Under Registration In general, small customer meters tend to under-register over time. In the absence of test data, WSO had to estimate the current rate of customer meter under-registration. In order to determine a best estimate, WSO took two known factors into account: 1. Age profile of the customer meter population. 2. The overall findings from the mass balance, which indicated a small volume of Apparent Losses. It is important to acknowledge that meter age is generally not a reliable predictor for small meter performance. Other factors such as total throughput, water quality, meter type, and quality of installation are equally important. In general, a combination of these factors is responsible for small meter accuracy. Ultimately, random and representative test data is required to thoroughly assess the accuracy of the small customer meters in service. Existing Small Meter Testing Practices SMWD tests small customer meters upon request. This practice helps resolve disputes with customers and identify problematic meters for replacement. However, this test data cannot be used to inform an average meter accuracy for customer meters. Meters tested upon customer request do not constitute a random sample. Random sampling is important when the sample is intended to derive conclusions about the entire population of meters in the field. Recommendations for randomized small meter testing program can be found in the memo titled, Water Loss Control Strategies Meter Age Profile SMWD had an old meter stock in the field during the audit period. Meter install dates ranged from 1975 to The age profile of SMWD s customer meter stock (see Figure 11) at the end of the FY14 audit period indicates an average meter age of 13 years. The age profile of meters suggests a historic regular meter replacement schedule of approximately 1,500 meters each year. Regular meter replacement declined through the late 2000 s during the recession, and then picked back up with approximately 3,000 and 2,300 meters installed in 2013 and 2014 respectively. Meters are replaced when flagged during the meter reading and billing process, when customers complain, or on an age-based replacement schedule.

31 Figure 11: Distribution of Meter Installation Year Top Down Evidence from Mass Balance The overall mass balance suggests a very small volume of Water Losses in the SMWD potable distribution system. Water Losses are calculated as the difference between Water Supplied and Authorized Consumption. Water Losses are further subdivided into Apparent Losses and Real Losses. Increasing customer meter under registration increases the calculation of Apparent Loss, and therefore decreases the volume of Real Losses. Table 12 outlines a simplified example that illustrates the impact of customer meter under registration on the calculation of Real Losses. In this example, our fictional utility supplied 110 units of water, and 100 units of water were consumed by all uses. The total Water Loss in both of these scenarios is 10 units. Scenario 2 includes a higher assessment of customer meter under registration (5.00%) and the resulting calculation of Real Losses (-4.74) is lower than that for Scenario 1 (-8.99). In this scenario, if we were to continue to increase our assessment of customer meter under registration, we would reduce the calculated volume of Real Loss below reasonable levels. Simplified Example: Relationship Between Customer Meter Accuracy and Real Losses Scenario 1 Scenario 2 A Water Supplied B Authorized Consumption E = B - A Water Losses C Customer Meter Under Registration 1.00% 5.00% F = B - (B/(1 - C)) Apparent Losses G = E + F Real Losses Table 12: Relationship Between Customer Meter Accuracy and Real Losses

32 Using the overall mass balance for SMWD, estimating a rate of customer meter under registration of greater than approximately 0.25% yields an Infrastructure Leakage Index below 1.0. The Infrastructure Leakage Index is defined as the ratio of calculated Real Losses to the predicted losses of an ideal system with the same physical characteristics. Therefore an ILI of less than 1.0 is better than the technical minimum. It is unlikely, barring uncertainty in the determination of Water Supplied and Authorized Consumption, that customer meters are under-registering more than 0.25% on average. Determination of Estimate for Customer Meter Under Registration Without test data from random small meter tests, WSO used the mass balance to identify reasonable estimates for an average rate of customer meter under-registration. Rates greater than approximately 0.25% yielded an Infrastructure Leakage Index, or ILI, less than 1.0 below the technical minimum. Therefore in the absence of small meter test data, the best estimate for customer meter performance is 99.75% on average, or under reading by 0.25%. Systematic Data Handling Errors SMWD maintains a very well organized and complete billing database. WSO conducted several data integrity checks using FY14 billing data including: 1. Large Suspicious Readings 2. Duplicate Readings 3. Negative Readings 4. Consecutive Zero Readings 5. Accounts with No Consumption 6. Estimated Readings The appendix titled Database Integrity Checks details the findings of each of these analyses. The overall results indicate that there are very low levels of loss from Systematic Data Handling Errors. Therefore, WSO estimated the total loss to Systematic Data Handling Errors as 5 AF during the audit period. 4.3 Final Determination of Apparent Loss Apparent Losses are the sum total of paper losses including Unauthorized Consumption (theft), Customer Meter Under Registration, and Systematic Data Handling Errors. Table 13 shows the final determination of Apparent Losses for SMWD. Apparent Losses Category Method Volume + Unathorized Consumption = 0.25% * Water Supplied Customer Meter Underregistrati= 0.25% * Metered Consumptio Systematic Data Handling Errors = Estimate = Total Table 13: Final Determination of Apparent Losses

33 4.4 Results and Recommendations Apparent Losses include inaccuracies associated with customer metering, errors in consumption data handling, plus any form of Unauthorized Consumption (theft or illegal use). Apparent losses are distinct from Real Losses because they do not describe physical loss of water instead they describe paper losses. The primary results and recommendations informed by WSO s Apparent Loss analyses are: Results: The total volume of Apparent Losses calculated for SMWD is very low, placing SMWD among the top performing utilities in California. The AWWA default of 0.25% of Water Supplied ( AF) was used to determine an estimate for Unauthorized Consumption. In the absence of random small meter test data, WSO studied the age profile of the active meter stock, and ultimately used the top down findings from the mass balance to inform an estimate for customer meter under registration of 0.25% of Billed and Unbilled Metered Consumption ( AF). Based on the overall results from the data integrity checks outlined in the memo titled Authorized Consumption, SMWD has a very clean and organized billing database that lends confidence to calculations of metered use. As a result, WSO has estimated loss due to systematic data handling errors at 5 AF, well below the AWWA default of 0.25% of Billed Metered Consumption ( AF). Recommendations: SMWD has very low levels of Apparent Loss. Therefore, recommendations to better manage Apparent Losses are focused on maintaining such low levels of loss and to further optimize revenue generation where possible. In order to evaluate the optimum replacement cycle for SMWD s small meter population, WSO recommends that SMWD begin testing random samples of small customer meters. The tests should be conducted with three primary objectives: 1. To inform an assessment of average customer meter under registration. 2. To migrate from an age-based meter replacement strategy to a meter tests result based replacement strategy. Random test results can indicate when SMWDs small meter population (or sub populations) has reached levels of under registration where replacement becomes economically justifiable. 3. To identify any poor performing sub populations that should be flagged for accelerated replacement. Small meter testing program recommendations are provided in the memo titled, Water Loss Control Strategies. To further optimize revenue generation from SMWD s large meter population it is recommended that SMWD test between 5 and 21 large meters more than once every three years to ensure minimal loss in accuracy and revenue. A suggested large meter testing program is outlined in the memo titled, Water Loss Control Strategies.

34 5 Component Analysis of Real Losses Background and Definitions Real losses can be broken down with a bottom-up approach using the AWWA 4372 Real Loss Component Analysis model. Whereas the top-down water balance approach determines Real Losses via a process of deduction, the bottom-up method uses a combination of actual break data and modeling to determine certain component volumes of Real Losses. This method recognizes that the annual volume of Real Losses consists of numerous leakage events whose individual volumes are influenced by flow rate and duration. Component refers to the various types of leakage making up the total volume of Real Losses. The component volumes are divided based on the intervention strategies that would best address each type of leakage. Reported Leakage (leaks that have surfaced): Breaks reported by the public or utility staff. Generally high flow rate and of relatively short duration. Unreported Leakage (leaks that have not surfaced): Breaks not reported by the public or utility staff but either discovered or potentially discoverable through leak detection. Unreported leaks that have not yet been discovered are collectively referred to as Hidden Losses or Hidden Leakage. Unreported leaks are generally moderate flow rates with average runtimes dependent on the intervention practices of the respective utility. SMWD does not currently have a leak detection program, so there were no discovered Unreported Leaks for the audit period all Unreported Leakage is therefore Hidden Leakage. Background Leakage: Leaks of low flow rates, continuously running, and not discoverable by leak detection. Typically composed of pin-holes and minor leaks at pipe joints and fittings. Figure 12 below shows the three types of Real Losses. A Note on Terminology: This report, in accordance with AWWA methodology, makes no distinction between the terms leak and break. The terms are used interchangeably. This may be different from the practices of some utilities that draw a distinction based on pipe diameter or flow rate. Figure 12: Types of Real Losses 3 3 Advances in Water Research, vol. 24 no. 3. July-September Water Research Foundation. p. 10.

35 The length of time for which a leak runs is divided into two separate components: awareness time, and shutoff/location and repair time, as summarized below. Awareness Duration: The length of time from when a leak first occurs (which is not necessarily known) to the time when the utility becomes aware of the leak. For reported leaks, this duration is usually short, while for unreported leaks, it is a function of the utility s leak detection program (if one exists). Shutoff/Location and Repair Time: The duration that it takes for the utility to investigate a reported leak, locate its position, and to stop all flow from the leak. By knowing the number of reported and unreported leaks, their average flow rates and total leak duration, it is possible to calculate the volume of water lost due to those leaks. Where such information is missing, industry standards may be assumed. 5.1 Top Down Calculation of Real Losses Real Losses represent the total water physically lost from the distribution system. This includes all types of leaks, breaks and overflows. The Real Loss volume is dependent on the condition of the infrastructure, break frequencies, flow rates, leak duration and system pressure. The volume of Real Losses for a given audit period is referred to as Current Annual Real Losses (CARL). In the Water Balance, the volume of Current Annual Real Losses (CARL) is determined by subtracting the total Apparent Losses from the total Water Losses volume. Real losses for the audit period were calculated to be 1, AF. Figure 13 contextualizes how Real Losses fit into the broader water balance and Table 14 below shows the full calculation. 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 Unauthorized Consumption Customer Meter Inaccuracies Data Handling Errors Revenue Water Non- Revenue Water Real Losses Figure 13: Position of Real Losses within the Water Balance Top Down Calculation of Real Losses Operation Water Balance Category Volume (AF) A Water Supplied 28, B Authorized Consumption 27, C = A - B Water Losses 1, D Apparent Losses = C- D Real Losses 1, Table 14: Water Audit Result

36 5.2 Bottom-Up Component Analysis of Real Losses Reported Leakage Reported leakage is the total volume of water that is lost due to leaks that are reported by the public or utility staff and then repaired. WSO calculated SMWD s reported leakage volume using AWWA 4372 Real Loss Component Analysis model populated with data from SMWD s leakage repair records. Organizing and understanding SMWD s leakage repair records was the first step in calculating reported leakage. SMWD s Leakage Repair Records The information included in Santa Margarita s leakage repair records was used to estimate the total reported leakage for FY14. A few important steps were taken to calculate the total reported leakage: Reported Leak events were divided into three main categories: service connections, appurtenances, and main leaks. Awareness duration was generated using industry-standard estimates for different types of leakage events. Shutoff duration the length of time between utilities awareness of leak and when the leakage flow is stopped was calculated by subtracting the estimated time of water shutoff by the time of initial report, as stated in Santa Margarita s leakage repair records. Some of SMWD s repair records included some clerical errors that WSO corrected. These errors are outlined in the appendix, Clerical Errors in Repair Records Real Loss Component Analysis model WSO used the AWWA 4372 Real Loss Component Analysis model with average shutoff duration and flow rates as calculated in the leakage repair records. A few important steps to apply the repair records to the Real Loss Component Analysis model are described below: Service connection leaks were subdivided into <1 or >=1. WSO classified SMWD s 1 service connection leaks as <1 service leaks in the AWWA 4372 Real Loss Component Analysis model since the average flow rate for 1 service leak as calculated by SMWD leakage repair records closely resembled the default flow rate for service connections <1 in the model. Appurtenance leaks were subdivided into Hydrants, Valves, and Other in the Real Loss Component Analysis Model. In the model hydrant leaks are considered to be hydrants that are leaking either through the standpipe, or hydrant valve. Hit hydrants, which typically have short awareness times and high flow rates, were grouped under Other. Hit Air Vacs outlined in Santa Margarita Leakage data export, were classified as valve leaks in the AWWA 4372 Real Loss Component Analysis model. Awareness duration was assumed equal to that of hit hydrants because these leak events were classified as hit and had very high flow rates that would have been detected quickly. Table 15 outlines the values used in the AWWA 4372 Real Loss Component Analysis model and the total reported leakage volume.

37 Leak Type AWWA 4372 Real Loss Component Analysis Model - Reported Leakage Awareness Duration (days)* Shutoff Duration (days)** Events Flow Rate Avg. (GPM)*** Total Volume (AF) 8" Main Breaks " Main Breaks Service Lines <1" Service Lines >1" Hydrants Valves (hit Air Vacs) Other (hit hydrants) , Total *Awareness time is estimated using industry standards **Shutoff Duration calculated using average shutoff duration from leakage repair records *** Flow Rate calculated using average flow rate from leakage repair records Table 15: Reported Leakaged Volume The total reported leakage volume, calculated using data from the leakage dataset and Real Loss Component Analysis Model, is AF. A sensitivity analysis was performed using different methods to calculate reported leakage. All methods calculated total reported leakage for SMWD to be similar to this value. Detailed results of the sensitivity analysis can be seen in Appendix A. Unreported Leakage The SMWD did not have a leak detection program for FY14. Therefore, the volume of Unreported Leakage was zero for FY14. Background Leakage Background leakage represents the water loss that is not discoverable by standard leak detection equipment (flow rates of less than 2.2 GPM at 70 PSI). Typically, this leakage is composed of minor leaks at pipe joints and fittings. Although flow rates are by definition too small to be sonically detected, the cumulative volume of background leakage can be a significant portion of the total real loss volume. In the AWWA Real Loss Component Analysis model Background Leakage is divided into two main sections, reservoirs and distribution systems. Reservoirs The Real Loss Component Analysis model assumes a default background leakage from covered reservoirs to be 0.25 GPM/MG of total reservoir capacity. However, SMWD s reservoirs are above ground and are equipped with visual underdrain leak detection systems. Therefore, any background leakage would likely be visible and repaired quickly. No leaks were noticed by the SMWD s staff during FY2014, so zero background leakage for reservoirs was assumed.

38 Distribution System The 4372 Real Loss Component Analysis model estimates the Background Leakage of the distribution system by comparing the infrastructure s current condition to the theoretical minimum level of leakage for a system of similar characteristics. AWWA estimates the following minimum rates of leakage for various components of system infrastructure based on international data of minimum night flows after repair of all detectable leaks and breaks. The values pertain to normal distribution systems in good condition with rates of infrastructure replacement around 1.5% to 2% per year. The minimum leakage rate for each component is referred to as the rate of Unavoidable Background Leakage: Background Leakage Rates in Distribution System System Component Unavoidable Background Leakage Rate Units Mains gallons / mile of main / day / PSI Services - Main to Curb-Stop gallons / service connection / day / PSI Services - Curb-Stop to Meter gallons / mile / day / PSI Table 16: Background Leakage Rates To calculate the volume of background leakage in SMWD s system, one must multiply the rates of Unavoidable Background Leakage by a factor that describes the condition of the infrastructure. This factor is called the Infrastructure Condition Factor (ICF). AWWA sets forth the following guidelines for selection of the ICF: System age of < 50 years: ICF of 1.0 System age of years: ICF of 1.5 System age of > 70 years: ICF of 2.5 To determine the ICF for the Santa Margarita, WSO analyzed the installation dates and length of mains from SMWD s mains inventory. Santa Margarita s water distribution system is relatively new - the oldest mains were installed in the early 1970 s. Figure 14 below shows the feet of mains installed per year for the entire water system. The results of the top-down water audit, in addition to the system s age, clearly indicate the infrastructure is in very good condition.

39 Feet of Mains Installed by Year Distance Installed (ft) Year Figure 14: Feet of Water Mains Installed by Year The average age of pipe, weighted by length, is 23.4 years, which suggests an ICF of 1.0. The background leakage volume can now be calculated using the standard Unavoidable Background Leakage Rate and an ICF value of 1.0. System Component Units Quantity Background Leakage Volume in Distribution system Background Leakage Rate* ICF Average Pressure (PSI) Volume of Leakage (AF) Mains miles Services - Main to Curb- Stop number 53, Services - Curb-Stop to Meter miles Total * The background leakage rate is expressed as gal/unit/day/psi at an ICF of 1.0 Table 17: Background Leakage Volume It is important to note that the system component, Services Curb Stop to Meter refers to the total length of service pipe (in miles) between the curb-stop and customer meter. The meter is located directly at the curb-stop, so this distance is zero. The total background leakage for the audit period is estimated to be AF.

40 The results of the real loss component analysis for the Santa Margarita Water District (SMWD) can be seen in Table 1 below. Hidden Losses Operation Type Volume (AF) A Reported Leakage B Unreported Leakage found through Leak Detection NA C Background Leakage D Total Real Losses (From Audit) 1, = D - (A + B + C) Hidden Losses Table 18 Component analysis resusts 5.3 Validation of Real Loss Findings Since Reported Leakage is calculated using field data, independent from the top-down water audit, comparing the values from these two methods can be a form of validation. The top down water audit indicates very low levels of Water Losses, 19 gallons/connection/day for FY2014. Because the volume of Real Losses are so low, one would expect the volume of Reported Leakage as calculated by the bottomup component analysis to be very low as well. The volume of Reported Leakage was calculated to be AF, or 0.37 gallons/connection/day, which is also very low. Furthermore, a main break frequency of 0.7 breaks per 100 miles of main per year and a service connection failure frequency of 1.2 breaks per 1,000 service connections per year was calculated for SMWD. Both of these values are well below the predicted failure frequency for optimized systems, which can be seen in Figure 15 and Figure 16. All this information points to a water distribution system that is experiencing a very low level of leakage.

41 Main Failure Frequency Comparison 30 Main Leaks (number / 100miles / yr) Santa Margarita Water District Average Failure Frequency in North America Based on Literature Review - WaterRF 4372 Failure Frequency for Optimized Distribution Systems (Friedman 2010) Figure 15: Main Failure Frequency Literature Comparison Service Connection Failure Frequency Comparison Service Leaks (number/ 1000 service connections / yr) Service Connection Failure Frequency Santa Margarita Water District AWWA Unavoidable Annual Real Losses (UARL) Component of Reported Service Line Failures Figure 16: Service Connection Failure Frequency Literature Comparison

42 5.4 Results and Recommendations Results The failure data for FY2014 indicates very low failure frequencies on mains and service connections, which is in line with the top down water audit results and confirms that SMWD has very low levels of leakage loss. Real Losses for the audit period were calculated to be 1, AF via the top down approach from the water balance. Using the leakage repair records, WSO calculated Reported Leakage as 1.9% of Real Losses, or AF. Reported Leakage represents breaks and leaks that surfaced and were reported to or observed by SMWD staff. Based on Santa Margarita s distribution system characteristics, including miles of mains, the number of service connections, the age of the infrastructure, and average system pressure, WSO estimates the total Background Leakage to be 77.47% of Real Losses, or AF. By taking the difference between the top down calculation for Real Losses and the volumes calculated for Background Losses and Reported Leaks, WSO estimates that 20.59% of Real Losses, or 279 AF, are Hidden Losses that are recoverable by leak detection. This is equivalent to 24 typical service connection leaks running for the entirety of FY14. Operations and Management A large percentage of Real Losses consists of Background Leakage (77%), which can be reduced only through pressure management and/or infrastructure replacement. The pressure survey performed by WSO identified high pressures occurring within some parts of the system as well as transients that could weaken water pipes. Pressure optimization is recommended to reduce background leakage and increase infrastructure service life. As outlined in the appendix, Reported Leak Clusters, a cluster of plastic service main breaks was identified near the Castadel Sol Golf Course. Transients were also recorded in this area. Installation procedures of plastic service connections in this area and pump scheduling management should be investigated. In two to three years SMWD may consider conducting proactive leak detection pilot program to recover Hidden Losses. Areas with high reported leakage and corrosive soil should be targeted during these surveys. The results of the Real Loss Component Analysis and the failure frequency analysis clearly show that SMWD s network is in excellent condition. Maintaining such low levels of leakage and failure frequencies below industry standards and targets will require a proactive strategy to keep leakage levels at a minimum and failure frequencies low as the network is aging. Data Management WSO recommends the SMWD staff complete the 4372 Real Loss Component Analysis model on an annual basis. This free model adds additional understanding of Real Losses to the AWWA Free Water Audit Software. In order to make best use of the Real Loss Component Analysis Model, WSO offers the following recommendations:

43 As described in the appendix Methodology Sensitivity Analysis of Reported Leakage, the default flow rates in the model matched estimated flow rates very closely. Therefore, default flow rates may be used to complete the Real Loss Component Analysis to improve ease of use. The leakage repair records that WSO used to calculate Reported Leakage had several suspected date/time errors. WSO provided SMWD with examples of parameters that should be recorded as part of a leak work order so that time stamps are available for when the flow of water from a leak is contained. This would facilitate more accurate average shutoff times for different types of leak events for use in the 4372 Real Loss Component Analysis model.

44 6 Pressure 6.1 Pressure Survey Background Pressure management is already employed by SMWD by supplying and pressurizing the distribution network through the use of elevated reservoirs which "float on the system". This provides for smooth pressure profiles throughout the day benefitting the system by avoiding large swings in pressure and pressure spikes. The supply through the elevated reservoirs is augmented by pump stations which pump into the system and fill up the elevated reservoirs. Only during the times of pumping into the system do the pressure profiles change and experience larger swings than when only supplied through the elevated reservoirs. Santa Margarita Water District (SMWD) and Water Systems Optimization, Inc. (WSO) conducted a pressure survey by recording pressure data in 31 different locations throughout the District s water distribution system. High frequency data, sampled once every 250 ms, was collected at 19 locations. General pressure data, sampled each second, was collected at 12 locations. A single input must be used in the free AWWA Water Audit software for the average system pressure. This is a significant simplification of pressure dynamics in the water distribution system. However, in developing a water loss control program, it is important to consider the influence pressure has on leakage volumes and system failure frequency. The higher the pressure in a distribution network, the greater the flow rate will be for any given leak. In order to generate a representative figure for system average pressure, WSO conducted a pressure survey throughout the distribution system using pressure loggers. Twenty-one representative locations total across the five hydraulic gradient pressure zones were selected by SMWD staff for the purpose of calculating system average pressure. WSO recorded pressure at these locations for hours to estimate the system average pressure. The recorded average pressures were then weighted by the miles of mains in each hydraulic gradient pressure zone to calculate an overall average system pressure. A map summarizing the pressure survey can be found in Figure 17 below. Several uncertainties are present in this approach. The weighting used to calculate a system average pressure may not accurately reflect the diverse hydraulic connections between SMWD s five pressure zones. The four sources of uncertainty are: SMWD s pressure zones are hydraulic gradient zones, and not isolated from one another. Given that there are a significant number of interconnections between zones, the average pressure is difficult to estimate with limited pressure logger locations. SMWD s distribution system has significant variations in elevation. These variations in elevation cause significant variations in pressure that can confound the characterization of the entire system with a single average value. This pressure survey was conducted in the winter (January), and SMWD staff have noted seasonal pressure variations. The scope of this pressure survey was limited WSO was able to record pressure in 21 locations, but additional locations would provide a deeper appreciation for pressure variation across the pipe network.

45 Based on the survey data, the calculated weighted average system pressure was 96 PSI. Given the uncertainty of the pressure survey design and the overall findings of the water balance suggest that the system average pressure is better characterized by approximately 90 PSI. This is generally a high system average pressure. However, the cost and feasibility of further reducing pressure in parts of the system needs to be carefully assessed which requires detailed pressure recordings and hydraulic modeling exercises. In addition to calculating a system average pressure, SMWD staff and WSO selected seven locations to record activity near pump stations, two to record suspected high pressures, and one to record low pressures. Table 19 shows the number of locations selected for each purpose. Logger Deployement Purpose Purpose Count of Loggers Evaluate System Average Pressure 21 Monitor Critical Infrastructure 7 Check for High and Low Pressure 3 Table 19: Logger Deployment Purpose Additional analyses including monitoring critical infrastructure and checking for high and low pressures can be found in the section, Pressure Analysis.

46 Figure 17: Pressure Survey Summary Map

47 6.2 System Average Pressure The average system pressure was calculated as 96 PSI, however uncertainty in the pressure survey and the overall findings of the water balance suggest that the system average pressure is more accurately characterized by 90 PSI. Table 20below outlines the average pressures recorded in each pressure zone, the miles of mains in each zone, and the resulting weighted average system pressure. The average system pressure of 90 PSI will be used in the AWWA free audit software and the 4372 Component Analysis Model to model leakage flow rates. Summary statistics for each location where WSO recorded pressure are provided in the section, Average Pressures by Location. Pressure Zone Average System Pressure Count of Elevation (ft) Loggers Miles of Mains Mean (PSI) Zone 1 ~ Zone 2 ~ Zone 3 ~ Zone 4 ~ Zone 5 ~ WEIGHTED MEAN: 96 Table 20: Average System Pressure 6.3 Results and Recommendations It is important to note that 90 PSI, is a relatively high system average pressure. For water districts with significant elevation changes across their service territory, like SMWD, it is often challenging to achieve more moderate average system pressures while providing adequate pressure to both high and low-lying customers. However, excessively high pressure can decrease the operational lifespan of distribution system infrastructure, increase the flow rates of any existing failures or leaks in the system, and increase the incidence of catastrophic failures such as main breaks. The cost and feasibility of reducing pressure throughout the system must be assessed. With strategic pressure management it may be possible to reduce overall average pressure while limiting any negative impacts. A potential long term goal for SMWD to strategically manage pressure would be to develop formal operational pressure zones where pressure can be controlled in isolation. These zones could provide a method for lowering pressure in areas with very high pressures, while maintaining pressures in other areas with relatively low pressure. Additional analysis of the economic impact of pressure management, based on the reduction of the volume of Real Losses, can be found in the section titled, Pressure Management. SMWDs distribution system is primarily gravity fed therefore any immediate reduction in pressure would be achieved by lowering the levels of the reservoirs. The reduction in reservoir level would have the consequence of short cycling pump stations frequently starting and stopping pumps when reservoir levels get too low. This short cycling would increase wear and tear on pumping infrastructure. Furthermore, short cycling pump stations could potentially increase the frequency of pressure transients. A balance between pumping infrastructure and pipe wear and tear should be sought by SMWD but was out of the scope of the current pressure survey.

48 If found cost effective longer term modifications including installing PRVs and establishing operational zones would provide additional options for managing pressure that would not require short cycling pump stations. Recommendations: SMWD should continue to collect further pressure data throughout the system to gain a more detailed picture of pressure fluctuations, transients, areas with excessive pressure, and areas experiencing low service pressure. SMWD should continue to review pump operation practices to emphasize soft starts and stops of all pump stations to avoid or limit transients. SMWD might consider programing longer startup sequences and staggering pumps. A review of detected transients can be found in the appendix, Monitoring Critical Infrastructure. Most importantly in the short run is to adjust the start and stop sequence at each pump station when system demand changes significantly (changes in demand between summer and winter) to avoid the occurrence of transients. In the long term, if cost effective SMWD should consider developing formal operational pressure zones zones where pressure can be controlled in isolation of the rest of the distribution system. Current pressure zones are defined by hydraulic gradient, or elevation, and do not provide the same management benefits as operational pressure zones. 7 Water Loss Control Strategies 7.1 Real Loss Intervention Strategy Introduction & Background Even if it were possible, eliminating leakage altogether would be a wasteful use of resources. The cost of doing so would far exceed the cost of balancing water supply and demand by other means, and that would mean higher bills for customers. The Economic Optimum Volume of Real Losses also known as the Economic Level of Leakage (ELL) represents the most cost-effective level of leakage given the current valuation of water lost. Leakage (Real Losses) costs money, as water lost has an intrinsic monetary value. There are also costs associated with locating and repairing the leak and any damage it may have caused to nearby infrastructure. For all utilities there is a balance between the value of the water that is lost through leakage and the cost of finding and fixing leakage. In simple terms, this balance is achieved upon implementing measures dictated by the Economic Level of Leakage. The ELL (Figure 18) represents the most effective level of leakage given current valuation of resources. The outer blue rectangle represents the current level of real losses. The inner core represents the volume of real losses that cannot be technically removed due to the inherent limitations of current leakage management technologies. The volume is calculated from the published UARL values that are the technically achievable lowest level of real losses for networks operated with best practice leakage management and with infrastructure in good condition. Somewhere between these two rectangles lies the Economic Level of Leakage (depicted by the green rectangle). These three rectangles give rise to three distinct layers of real losses.

49 Figure 18: Four Interventions against Real Losses The ELL is influenced by each of the four main intervention techniques against real losses, highlighted in Figure 18. Short term interventions against real losses include 1) managing the duration of reported leakage by improving repair times and reducing leak run times and 2) conducting proactive leak detection to locate and repair unreported failures. Medium to long term interventions against real losses include pressure management and infrastructure replacement planning. The ELL is a function of the total cost of leakage, which includes both the value of the water lost and the cost of all the leakage control activities. Increasing the quantity of leakage control activity in any given year will increase the annual cost of leakage control but will lead to a decrease in the annual volume and cost of leakage. Whether the increase in leakage control activity leads to a reduction in total cost (the sum of both the cost of leakage control activity and the cost of water lost) will depend on the cost factors associated with the leakage control activity, the cost of water and the effectiveness of the leakage control activity in reducing real losses. To evaluate the cost effectiveness of each of these intervention strategies, it is necessary to assign a monetary value to the volume of Real Losses and to estimate the cost of the intervention tools. The cost for each of the intervention tools will be discussed under the specific intervention option of the ELL analysis. The way in which Real Losses are valued by a utility is crucial to the outcome of any ELL analysis. The higher the value of Real Losses, the more aggressive the recommended intervention. For this analysis, Real Losses for SMWD are valued at $923 per AF. This valuation is the cost of imported water from MWDOC. Because SMWD s source of water is relatively expensive, it is important to develop proactive

50 leakage management strategies to maintain such low levels of leakage in order to counter act the natural decay/aging process of the system. The four intervention tools against Real Losses were evaluated to determine if there is room for improvement in SMWD s current leakage management policy. Additional theory and background relevant to the sections that follow can be found in the appendix, Water Loss Control Strategies Background and Theory. Economic Frequency of Intervention: Proactive Leak Detection Rate of Rise of Unreported Leakage Estimation Using UARL Formula An example methodology and additional background on the Rate of Rise of Unreported Leakage can be found in the appendix, Rate of Rise of Leakage (RR). The Rate of Rise of leakage is defined as the development of new leakage that occurs with time in all systems. The rate of rise of unreported leakage can be estimated using the International Water Association (IWA) approach if other data is not available. IWA sets forth standard estimates for the annual occurrence of unreported leaks. These estimates are based on international data collected after leaks have been located and repaired in District Metered Areas (DMAs). Without active leak detection, new unreported leaks would continue to run undetected each year. Therefore, presumably each year the total volume of unreported leakage would increase by the same amount predicted by the UARL formula. Table 22 shows the calculation for the predicted rate of rise of Unreported Leakage for SMWD using the UARL formula and Table 21 shows the system characteristics used to complete the modeling exercise. Based on the UARL formula, the average rate of rise for Unreported Leakage is 0.31 kgal/mile of main/day. SMWD System Characteristics A Miles Mains Miles B Number of Service Connections 53,240 Count C Length of Service Connections: Curb to Meter 0 Miles D Average Pressure 90 PSI Table 21: SMWD System Characteristics for UARL Formula

51 Accumulating Unreported Leakage from UARL Formula Infrastructure Component Table 22: Unreported Leakage from UARL Formula Since SMWD is currently operating the system close to the technically optimum level of leakage and failure frequencies are half of the UARL based minimum failure frequencies WSO used a rate of rise that is 50% less than the UARL formula based rate of rise. For the calculation of SMWD s leak detection intervention frequency a rate of rise of.15 kgal/mile of main/day was used. Cost of Leak Detection Survey Intervention (CI) The cost for undertaking a comprehensive survey is estimated to be $300 per mile. In a comprehensive survey, an operator listens on all available fittings valves, hydrants, service connections, and other fittings and also listens above the ground using geophones, as opposed to a hydrant and valve survey where the operator listens on available main line valves and hydrants only. WSO has found that in California, most unreported leaks occur on service connections. A comprehensive survey is necessary to reliably identify service connection leaks. The cost to repair the leaks found through a proactive leak detection program are not included in the economic evaluation at this point, according to AWWA recommendations. Proactive leak detection does not introduce the cost to repair the leak, as the leak is already there even though the utility is not yet aware of it. It is assumed that leaks will need to be fixed at some point; proactive leak detection simply brings them to the utility s attention sooner at a generally lower repair cost. Notably, some benefits are also excluded in this simplified model: IWA Constant Unreported Leak Frequency* Proactive leak detection avoids potential for contamination through compromised infrastructure. Proactive leak detection enables repair before a leak grows and/or becomes a catastrophic failure, in which case the repair costs can increase substantially. Results of Economic Frequency Intervention Analysis An example and theoretical outline of the methodology to conduct Frequency Intervention Analysis can be found in the appendix, Theory of Economic Frequency of Intervention. The key results of the Economic Intervention Frequency analysis are presented below (using the formulas and values outlined above) Economic Intervention Frequency (EIF) to find unreported leaks: 23.6 months The Economic Percentage (EP) of the system that should be inspected each year: 51% Accumulating Unreported Leaks (gal/day) E = 0.77*A*D Mains 0.77 gal/mile of main/day/psi 41, F =.03*B*D Service Connection: Mains to Curb Stop.30 gal/conn/day/psi 143, G = 2.12*C*D Service Connection: Curb Stop to Meter 2.21 gal/mile of servconn/day/psi 0.00 H = E + F + G Annual Unreported Leak Volume 184, = H/A/1000 Average Rate of Rise of Unreported Leakage (kgal/mile of main/day): 0.31 * Constants based on international data collected on the rate of rise of leakage for different infrastructure components.

52 According to the EIF model, the entire system distribution system should be surveyed every 23.6 months. Following this logic, about 51% of the network should be surveyed every year. This results in an average run-time of unreported leaks of about 359 days. The Economic Intervention Frequency model is presented in Figure 19 below. Figure 19: Economic Intervention Frequency for Proactive Leak Detection Figure 19 shows the relationships between leak survey frequency, costs of hidden losses, and cost of leak detection. The x-axis shows the average run time of Unreported Leaks (in months). The average run time is half the complete-system survey period. To achieve low average run times, a high frequency of leak surveying is necessary. As leak survey frequency increases (moving left on the x-axis), the cost of leak detection increases (see the green line). At the same time, the average run-time of unreported leaks is reduced. With shorter run-times for unreported leaks, the total hidden loss volume and associated costs are also reduced (see the orange line). It is notable that as the leak survey frequency increases, the annual cost of leak detection increases exponentially (see the green line near the y-axis), whereas the cost of hidden losses decreases linearly. The total cost curve (see the blue line) is the sum of the cost of water lost through undetected leaks plus the cost of leak detection survey. The point at which the total cost curve is at a minimum represents the least cost point for real losses from detectable leaks. The least cost point for SMWD occurs when the leak survey interval is 23.6 months with an average runtime for unreported leaks of 11.8 months. Assuming a cost of $300/mile of detailed leak survey, the optimal annual budget for leak detection is $130,522/year. Annual Budget for Intervention (ABI): $90,792/year Economic Annual Volume of Unreported Real Losses (EURL): 98.3 AF/year

53 It is important to note that the total cost curve (blue line in Figure 19) is very flat between an average run time of unreported leakage of 10 months and 24 months. This indicates that surveying at a less frequent interval than the models optimum interval has a minimal impact on the total cost. In SMWD s case, surveying less frequently than recommended by the EIF model would have a negligible impact on the total incurred cost of leakage. Economic Frequency of Intervention Analysis Recommendations The modeled EIF indicates that SMWD s ideal survey frequency is approximately two years. However, a survey frequency of four years does not significantly increase total costs incurred. WSO therefore recommends that SMWD should not immediately undertake proactive leak detection. The volume of hidden losses potentially recoverable is not large enough to justify an aggressive leak detection program. However, SMWD should consider conducting targeted pilot leak detection in specific areas in two to three years. These leak detection efforts should focus on areas with corrosive soil, and near clusters of reported leak events. A discussion of the reported leak clusters can be found in the appendix, Reported Leak Clusters. Once the results of pilot leak detection efforts are available SMWD s proactive leak detection strategy can be refined. As SMWD s infrastructure ages it will be necessary to increase the survey frequency to maintain the very low leakage levels the system currently experiences. Pressure Management Additional information about SMWD s current pressure management practices and an overview of the pressure survey conducted by WSO in January of 2016 can be found in the section titled, Pressure. The volume of Real Losses is proportional to the average system pressure; as a result, strategizing around the link between pressure and leakage is critical in any water loss control program. Pressure management is the tailored adjustment of pressure throughout a distribution system in order to reduce Real Losses (by decreasing leakage flow rates and the occurrence of new leakage) while not compromising quality or reliability of service. Notably, pressure reduction reduces flow rates across all three types of leakage; background losses, reported losses and unreported losses are all affected by pressure management. Reliable and well-tested models are available for calculating the savings in Real Losses stemming from pressure reduction. Beyond the leakage savings, pressure management has many benefits when used to optimize the delivery of water and service to customers, provided in Table 23: Benefits of Pressure Management. Pressure Management: Reduction of Excess Average and Maximum Pressures Conservation Benefits Water Utility Benefits Customer Benefits Reduced Flow Rates Reduced Frequency of Bursts and Leaks Reduced Consumption Reduced Flow Rates of Leaks and Bursts Reduced Repair Costs, Mains & Services Deferred Renewals and Extended Asset Life Reduced Cost of Active Leakage Control Table 23: Benefits of Pressure Management 4 Fewer Customer Complaints Fewer Problems on Customer Plumbing & Appliances 4 Lambert, A and Fantozzi, M (2010). Recent developments in pressure management. Proc. of IWA Specialized Conference Water Loss 2010, Sao Paolo, Brazil.

54 Pressure management also reduces break frequencies by reducing stress on the infrastructure. The reduction of break frequencies produces the following benefits: Extended asset life Reduced repair cost Reduced visit/inspection cost for reported breaks Reduced risk of interruption to supply Reduced risk of compensation payments Pressure management as a real loss reduction strategy requires investment. For such longer-term real loss reduction strategies, it becomes economic to make the investment to reduce real losses if the value of water saved over the investment period pays for the implementation cost. Since the volume of real losses is proportional to the average system pressure, the benefits of pressure reduction can be calculated as the value of consequent water savings. Therefore, there will be a break-even point at which the additional cost of pressure reduction equals the cost of the real losses reduced. Pressure and Real Loss Reduction Scenarios The potential savings that reduced average system pressure would offer is presented in Table 24 below. Every one psi reduction produces real loss savings of 13 AF per year at a value of approximately $12,085. The table shows cumulative savings as more pressure reduction is achieved. WSO assumed a direct, 1:1 relationship between pressure and leakage. Each reduction in pressure produces a linearly proportional reduction in leakage. This relationship is commonly referred to as the N1. Additional information about the relationship between pressure and leak flow rate can be found in the appendix, The Pressure Leakage Relationship and the N1 Factor. Average System Pressure (PSI) Pressure Reduction Scenarios Real Losses Water Savings (AF) (AF) Table 24: Pressure Reduction Scenarios Financial Savings ($) 90 1, , , , , , , , , , , , ,508 This general analysis shows that a 5 PSI reduction would produce 65 AF/year in Real Loss savings at a value of approximately $60,423. This only accounts for savings in Real Loss reduction; the additional important benefits of reduced break frequency and extended infrastructure lifetimes are not monetized here though they should be considered in a pressure management plan.

55 In order to do a full cost-effectiveness analysis, the costs of implementing pressure reduction are required. Sufficient pressure and infrastructure data (by pressure zone) is not currently available to identify the specific actions that would reduce average system pressure and to define the associated costs. In the short term it is recommended that SMWD further expand on WSO s pressure survey and collect comprehensive pressure data throughout the system. Data should be collected during diverse operational conditions reflecting different levels of demand. Once more data is available the feasibility and cost-effectiveness of reducing pressure in targeted areas can be studied. In the short term SMWD should continue to review pump operation practices to emphasize soft starts and stops of all pump stations. Soft starts and stops will reduce the incidence of harmful pressure transients. Most importantly, in the short run, SMWD should adjust the start and stop sequence at each pump station when system demand changes significantly (changes in demand between summer and winter) to avoid the occurrence of transients.

56 7.2 Apparent Loss Intervention Strategy Introduction & Background Similar to Real Losses, it is not economically justifiable to remove all Apparent Losses. The type of Apparent Loss most suited for utility intervention is customer metering inaccuracy. The two sections below outline recommended small and large customer meter testing programs to address customer metering inaccuracy. Recommended Small Meter Test Samples Random small meter testing is important because it provides an indication of the overall accuracy of the utility meter stock. Meters are the utility s cash register, and without evidence of their current performance, it is difficult to design and maintain an informed meter replacement schedule. Further, estimations of Real Losses often depend on an appreciation of customer metering inaccuracies, since Real Losses and Apparent losses are zero-sum in water audits. WSO has found that manufacturer recommendations for meter lifespans are not as reliable as conclusions derived from a well-conceived randomized testing program. Random testing may also help to identify specific sub populations of meters that are under performing. Sub populations found to be faulty can be slated for an accelerated replacement plan. Small meter testing practices are driven by two important factors: 1. Operational Constraints: The number of meters that can be tested due to practical operational constraints such as staff time and financial resources. 2. Representative Sample: The critical types of meters that represent the diverse sizes, makes, and ages of the meters in service must be present in a complete random sample. The composition of the random sample can be weighted by count and by total throughput. To identify critical groups of meters, it is important to consider the age, make, and size breakdown of the meter population. Figure 20 below shows the age distribution of SMWD s meter stock with the bars colored based on the meter make. Using this figure we can see that the oldest meters are primarily Rockwell meters. From the late 1980s through the early 2000s SMWD installed primarily Sensus meters. After 2002 SMWD shifted to installing Neptune meters almost exclusively. SMWD should include a representative group from the Rockwell sub-population in the test sample because they are quite old. Additional tables summarizing the consumption and count of meters by manufacturer and size can be found in the appendix, Meter Consumption Summary.

57 Figure 20: Meter Make by Installation Year Random Samples WSO designed three small meter testing samples for SMWD to consider implementing. These samples describe a small, medium, and large random sample size. WSO modified the distribution of these samples to ensure that they include meters from the most important manufacturers and sizes. The modifications were driven by three objectives: 1. Ensure representation from the most frequently observed makes and sizes. 2. Ensure the makes and sizes that record the most consumption are adequately represented. 3. Minimum sample sizes for each relevant category of make and size must not be less than 5. A detailed explanation and review of the steps taken to design these samples is provided in the section, Small Meter Testing: Sample Selection Methodology. Small Sample Table 25 provides the number of meters that should be tested in each manufacture and size class to generate meaningful results after two to three years of randomized testing. Medium Sample Customer Meter Testing: Small Sample (73) 3/4 Inch 1 Inch 1.5 Inch 2 Inch Total Sensus Neptune Rockwell Badger Metron Hendey Unknown Total Table 25: Small Customer Meter Sample (n = 73)

58 Table 26 provides the number of meters that should be tested in each manufacture and size class for a medium sample size. Large Sample Customer Meter Testing: Medium Sample (127) 3/4 Inch 1 Inch 1.5 Inch 2 Inch Total Sensus Neptune Rockwell Badger Metron Hendey Unknown Total Table 26: Medium Customer Meter Sample (n = 127) Table 27 provides the number of meters that should be tested in each manufacturer and size class for a large sample size. Customer Meter Testing: Large Sample (307) 3/4 Inch 1 Inch 1.5 Inch 2 Inch Total Sensus Neptune Rockwell Badger Metron Hendey Unknown Total Table 27: Large Customer Meter Sample (n = 307) Recommended Large Meter Testing Schedule Since it is economically infeasible to overhaul each of the 151 large meters on a regular basis it is necessary to identify those meters where potential losses in accuracy would result in the largest losses in revenue generation. This necessitates ranking the large meter population by annual revenue generated. Figure 21 provides a breakdown of percent of meters by percent of monthly revenue. This analysis suggests that only a relatively small number of meters are responsible for annual revenues that would require frequent testing and repair to guarantee accurate metering of water delivered and revenue generated. For example, 5% of large meters (7 meters) are responsible for registering 60% of total monthly revenue from all large meters.

59 Figure 21: Large Meter Revenue Distribution An optimized testing/overhaul frequency for a given meter is achieved when the cost of intervention (regular testing/overhaul of the meter) is less than or equal to the cost of under registration (revenue loss). In other words an optimized point is reached when the cost of regular meter testing/overhaul does not exceed the value of revenue saved by the large meter testing/overhaul policy. The frequency of meter testing/overhaul will depend on the inaccuracy of registration and the consequent revenue loss. A meter s accuracy will be impacted by many factors such as total volume registered, age of meter, quality of meter, metering technology, water quality, consumption patterns, etc. Currently, the degree by which a given large meter deteriorates in accuracy each year is not known. However, this information should be assessed by SMWD over the upcoming years as an updated meter testing/overhaul schedule is implemented. Without this system specific information, WSO decided to run three scenarios where the average under registration per meter is 0.5%, 1.0% and 1.5% per year (based on general experience with large meter accuracy). Next the cost to test/overhaul each meter was estimated and the optimum testing schedule was determined by comparing the potential revenue loss to the cost to test/overhaul the meter. Meter under registration was assumed to be equal to the percent of revenue loss for under registration. This may not always hold true in the case of tiered rate structures, where the marginal value of each unit of water sold changes depending on the total quantity sold. However, the impact of this on the final frequency recommendations is negligible. Table 28 provides an example of how each single large meter account was analyzed in order to determine the appropriate testing/overhaul frequency. A theoretical annual under-registration of 1.5% for this meter would result in an annual revenue loss of about $4, The cost to test/overhaul this meter is about $1,718. This this meter should be tested every 4 months to make sure that the meter is always performing at a maximum accuracy. However, realistically the meter should probably be tested every 6 months. The grouping applied to this analysis is outlined in Table 29.

60 Example Meter Replacement Schedule Location ID Rate Class GB Size 8 inch A Annual Revenue Generated $314, B Assumed Loss In Accuracy 1.5% C = A - (A / (1-B)) Annual Revenue Loss -$4, D Cost to Test and Overhaul $1, E = D / C Ratio of Revenue Loss and Cost to Test and Overhaul 0.4 F = E * 12 Break Even Frequency of Testing and Overhaul 4 = Grouping Assigned Test Frequency Six Months Table 28: Example Meter Replacement Schedule Meter Testing Groups Break Even Frequency Group 0-6 Months Six Months 6-12 Months One Year Months Two Years Months Three Years >36 Months Regular Replacement Table 29: Meter Testing Frequency Group Scenario 1: 0.5% Revenue Loss Under the scenario where it s assumed that each meter is under recording by 0.5%, a total of 5 meters should be overhauled once every one or two years at an annual cost of $5, (annual cost = $2,582 + ($5,154/2)). In this scenario, it is not cost-effective to overhaul 146 meters beyond their regular replacement schedule. Large Meter Testing Scenario #1: Annaul Loss of 0.50% Group Count of Meters Total Annual Revenue Loss* Total Annual Cost to Test and Repair One Year 2 $3, $2, Two Years 3 $3, $5, Regular Replacement 146 $5, NA Total* 151 $12, $5, *The Total Cost is weighted by recommended frequency to show expected annual cost. Table 30: Large Meter Testing Scenario: Annual Loss of 0.50% Scenario 2: 1.0% Revenue Loss Under the scenario where it s assumed that each meter is under recording by 1.0% each year, a total of 12 meters should be overhauled once every six months, or once every one, two, or three years at an annual cost of about $13,000. In this scenario, it is not cost-effective to overhaul 139 meters beyond their

61 regular replacement schedule 1.0% annual under registration for these meters does not result in a revenue loss that would justify the expense. Large Meter Testing Scenario #2: Annaul Loss of 1.00% Group Count of Meters Total Annual Revenue Loss* Total Annual Cost to Test and Repair Six Months 2 $6, $2, One Year 3 $7, $5, Two Years 2 $ $1, Three Years 5 $2, $6, Regular Replacement 139 $7, NA Total* 151 $24, $13, *The Total Cost is weighted by recommended frequency to show expected annual cost. Table 31: Large Meter Testing Scenario: Annual Loss of 1.0% Scenario 3: 1.5% Revenue Loss Under the scenario where it s assumed that each meter is under recording by 1.5% each year, 21 meters should be overhauled once every six months, or once every one, two, or three years at an annual cost of about $20,000. In this scenario, 130 meters should not be tested and repaired beyond their regular replacement schedule 1.5% annual under registration for these meters does not result in a revenue loss that would justify the estimated expense of testing and repairing the meter. Large Meter Testing Scenario #3: Annaul Loss of 1.50% Group Count of Meters Total Annual Revenue Loss* Total Annual Cost to Test and Repair Six Months 3 $14, $4, One Year 2 $6, $3, Two Years 7 $5, $8, Three Years 9 $4, $10, Regular Replacement 130 $7, NA Total* 151 $37, $19, *The Total Cost is weighted by recommended frequency to show expected annual cost. Table 32: Large Meter Testing Scenario: Annual Loss of 1.50%

62 1 Appendix 1.1 Water Supplied Analysis Boundary and Data Selection Sensitivity Analysis WSO performed a thorough sensitivity analysis to identify the consequences of different selections of system boundary and data sources. We generated three scenarios to create a low, best, and high estimate. The low and high estimates represent the combinations of system boundary and data choices that result in the lowest and highest possible total estimates for Water Supplied. Scenario 1 (Low): Boundary: SMWD SCP takeouts (SC-1 + SC-2A + SC-3 + SC-4A + SC-5A), OC-82,Aufdenkamp CM-12, El Toro R-6, Plano TCWD export, and Portola Lake Fill export. Data: Daily SCADA production data for takeouts on the SCP, OC-82, El Toro R-6, Portola Lake Fill, Plano TCWD, and storage. MWDOC billing data for Aufdenkamp CM-12. Discussion: The low scenario is not ideal because the data transfer protocol from Aufdenkamp CM-12 is suspect. Aufdenkamp CM-12 OC-82 El Toro OC-88 OC-88A Plaza CM-12 Lakeside Daily Production Data SCP SMWD Distribution System MWDOC Billing Data Other Agencies Portola Plano TCWD Storage System Boundary Scenario 2 (Best): Boundary: SMWD SCP takeouts (SC-1 + SC-2A + SC-3 + SC-4A + SC-5A), OC-82, Plaza CM-12, El Toro R-6, Plano TCWD export, and Portola Lake Fill Export. Data: Daily SCADA production data for takeouts on the SCP, Plaza CM-12, El Toro R-6, Portola Lake Fill, Plano TCWD, and storage. MWDOC billing data for OC-82. Discussion: See section above for a detailed discussion on the selection of these meters and section for details on selecting the data sources. Aufdenkamp CM-12 OC-82 El Toro OC-88 OC-88A Plaza CM-12 Lakeside Daily Production Data SCP SMWD Distribution System MWDOC Billing Data Other Agencies Portola Plano TCWD Storage System Boundary

63 Scenario 3 (High): Boundary:SMWD SCP derived as (OC-88 + OC-88A SC-2B SC-4 SC-5B), OC-82, Plaza CM-12, El Toro R- 6, Plano TCWD export, and Portola Lake Fill export. Data: Daily SCADA production data for OC-88 and OC-88A, Plaza CM-12, El Toro R-6, Portola Lake Fill, Plano TCWD, and storage. MWDOC billing data for OC-82. Discussion: The highest estimate derives the volume of water entering the SMWD distribution system from the SCP by taking the total volume registered by OC-88 and OC-88A and subtracting volumes from non-smwd takeouts. The MWDOC bills for OC-82 were greater than the daily production data due to a suspect data transfer protocol ultimately WSO chose the MWDOC bills for the best estimate for OC-82. Aufdenkamp CM-12 OC-82 El Toro OC-88 OC-88A Plaza CM-12 Lakeside Daily Production Data SCP SMWD Distribution System MWDOC Billing Data Other Agencies Portola Plano TCWD Storage System Boundary Overview of Boundary and Data Selection Sensitivity Analysis Table 33 below summarizes the different estimations for Water Supplied based on the three scenarios presented above. Note that corrections for import meter inaccuracy are not included in the calculations presented below. The low and high estimates for the unadjusted volume of Water Supplied during the audit period differed by 1,085 AF or 3.87% of the best estimate. This observed difference highlights the importance of carefully selecting the most appropriate system boundary and the most reliable data sources for auditing purposes. Because the water audit calculates real losses through a process of elimination, any variation in the volume of Water Supplied will directly affect the volume of real losses. Water Supplied Sensitivity Analysis FY14 Water Supplied (AF) % Difference from Best Estimate Scenario 1 (Low) 27, % Scenario 2 (Best) 28, % Scenario 3 (High) 29, % High to Low Range 1, *3.87% *Percent of best estimate Table 33: Sensitivity Analysis for Water Supplied

64 Justification for Selected System Input Boundary and Data Sources We will discuss the three major segments of the input system separately below the South County Pipeline, Plaza PS, and Lakeside. Figure 23 below shows the selected boundary and data sources. Aufdenkamp CM-12 OC-82 El Toro OC-88 OC-88A Plaza PS Lakeside Daily Production Data SCP SMWD Distribution System MWDOC Billing Data Other Agencies Portola Plano TCWD Storage System Boundary Figure 22: Selected System Input Boundary and Data Sources South County Pipeline (SCP) Two MET meters, OC-88 and OC-88A, supply the South County Pipeline (SCP). Downstream of these two MET meters, several takeouts supply water to SMWD and four other water districts. During the audit period, OC-88 and OC-88A were significantly over-registering the volume of water entering the SCP. MET corrected the issue in April of 2014, however volumes recorded by OC-88 and OC-88A before April are likely inaccurate. In addition, OC-88 and OC-88A are upstream of the Upper Chiquita Reservoir (UCR). Therefore, if SMWD drew the system boundary at OC-88 and OC-88A, SMWD would need to account for changes in UCR storage, introducing additional complication and potential for error. Because of the known inaccuracy of OC-88 during the audit period and the proximity of the SMWD takeout meters to the distribution system, WSO recommends using the takeout meters (SC-1, SC-2A, SC-3, SC-4A, and SC-5A) as the system boundary on the SCP. Daily SCADA data and monthly data from MWDOC invoices are available for each takeout on the SCP. WSO recommends that SMWD use daily data to calculate the total volume of water registered through the SCP because MWDOC adjusts the monthly totals for each takeout to align with the total throughput of OC-88 and OC-88A. AMP OC-82 AMP OC-82 is a Venturi meter managed by MET located upstream of the Lakeside Pump Station. SMWD operates two insertion meters at the Lakeside Pump Station; however, SMWD staff are not confident in the readings from these meters. WSO recommends using OC-82 to define the system boundary because of the uncertainty of the volumes recorded by the insertion meters at the Lakeside Pump Station. SMWD receives daily data for OC-82 collected by MET through an automatic transfer protocol. However, the values sent to SMWD are different from MWDOC s monthly invoices despite the reads originating

65 from the same meter the total for the audit period differs by 42 AF. MWDOC s transmitter is regularly maintained & calibrated, which makes it the source of choice. WSO recommends using the monthly invoiced volumes for OC-82 as opposed to the daily production data because of these data transfer issues. Plaza PumpStation (Plaza PS) Plaza PS is an electromagnetic meter downstream of Aufdenkamp CM-12, a MET owned meter. WSO recommends using Plaza PS to define the system boundary because SMWD owns this meter and it is proximal to the distribution system. El Toro R-6 SMWD owns capacity on the El Toro reservoir in the El Toro Water District. SMWD accessed this storage in January of 2013 to supplement supply when OC-82 was off. System Input Meter High Resolution Flow Profiling WSO analyzed the flow rates measured by system input meters 5. Typically, WSO checks if a significant proportion of reads during the audit period are outside of the meter manufacturer s recommended operating range 6. All source meters had an insignificant percentage of reads outside of their recommended flow ranges except SC-1 and, while not a system boundary meter, OC-88. We will summarize some primary findings for SC-1 and OC-88 below. SC-1 SC-1 had the most reads greater than zero outside of its recommended flow range (~30%). However, most of the reads outside of the range were small volumes recorded in the spring of SMWD intended to shut off SC-1 during this time. However, the amount of time SC-1 recorded low flows outside of its recommended range introduces some uncertainty about the volume of water that actually passed through the meter. 5 WSO used data provided by SMWD and publicly available data from MWDOC to analyze the flow rates. WSO used the average between the minimum and maximum flows recorded each minute to compare against manufacturers recommendations for SCP takeouts and Plaza PS. Data for OC-82, OC-88, and OC-88A was available in 15-minute increments including a volumetric measurement and an instantaneous flow read. This allowed WSO to examine intervals where the flow was positive and yet no volume was registered, and vice-versa. No significant time intervals exhibited these anomalies. 6 WSO derived the recommended minimum flow rate for the meters on the South County Pipeline (SCP) from maximum flows supplied by SMWD using the 10% rule, a guideline set forth by MWDOC, where a Venturi meter s minimum recommended flow is 10% of its maximum flow. MWDOC supplied recommended maximum flows for OC- 82, OC-88, and OC-88A. The user manual for Rosemount 8705 electromagnetic meters cites minimum and maximum recommended flows for Plaza PS.

66 OC-88 The volume registered outside of the recommended flow range will not affect the overall estimate for Water Supplied because OC-88 is not used as a system boundary meter. Of the registered flow rates greater than zero for OC-88, 2.41% of them fell outside of the recommended flow range. However, because this meter is one of two supplying the South County Pipeline (SCP), it is important to pay close attention to flow rates in the future. Overall Results Table 34summarizes results from the flow analysis of input meters. The column for the percent of minutes out of range is calculated as the percentage of reads greater than zero that fell outside of the recommended flow range. Graphs with accompanying discussions for each meter follow. The graphs do not show the maximum flow range for any of the input meters because none of the meters registered flows above that threshold during the audit period. Meter Flow Profile Graphs Recommended Flow Range (CFS) SMWD Source Meter Flow Analysis Minutes Out of Range Volume Recorded Out of Range (CF) Table 34: Source Meter High Resolution Flow Profiling % of Minutes Out of Range Plaza CM % SC ,651 1,067, % SC-2A , % SC , % SC-4A SC-5A , % OC % OC ,465 4, % OC-88A % El Toro R * Minutes recording zero flow were excluded from this analysis -- the main was assumed to be off during that time. ** The Percent of minutes out of range is calculated as the total number of minutes out of range divided by the total number of minutes above zero. The following figures show the flow rates recorded during the audit period in CFS and the minimum recommended flow rate for all input meters that define the system boundary. The maximum recommended flow rate is not shown because none of the meters registered flows above that threshold during the audit period. Figure 23shows flow rates in CFS for the meter at the Plaza Pump Station during the audit period. Plaza PS had generally steady flows, increasing in late January to supply additional water to the distribution

67 system when OC-88 was offline. Overall, average readings for Plaza PS were out of accurate range of CFS for.10% of the total number of non-zero readings during the audit period. The vertical lines that fall below zero are caused by several, brief,power failures that occurred during the audit period. Figure 23: Plaza CM-12 Flow over Time FY14 Figure 24shows flow rates for SC-1, a takeout on the South County Pipeline (SCP), in CFS over time. Almost 30% of the 375,782 non-zero reads were out of the recommended flow range of 6 60 CFS. However, most of those reads were very low and occurred through March and April. SMWD intended the valve near SC-1 to be off during that time, but water still trickled through. The amount of time SC-1 was outside of its recommended range introduces some uncertainty about the volume of water that actually passed through the meter during the audit period. Figure 24: SC-1 Flow over Time FY14

68 Figure 25shows flow rates for SC-2A, a takeout on the SCP. SC-2A was out of the recommended flow range of 4 40 CFS for 0.02% of the non-zero readings. Figure 25: SC-2A Flow over Time FY14 Figure 26shows flow rates for SC-3, another takeout meter on the SCP. SC-3 was out of the recommended flow range of 8 80 CFS for 0.04% of the non-zero reads Figure 26: SC- Flow over Time FY14

69 Figure 27shows flow rates for SC-4A in CFS. SC-4A was offline for much of the audit period, but was online briefly during the spring for a total of 137,092 minutes. There are no known alarm set points or recommended. Figure 27: SC-4A Flow over Time FY14 Figure 28shows flow rates for SC-5A in CFS. The flow rate seems to stay consistently at 3.5 CFS with somewhat frequent aberrations, presumably when the meter is cycled on and off. SC-5A was outside of its recommended flow range of 2 20 CFS for 0.13% of the non-zero reads during the audit period. Figure 28: SC-5A Flow over Time FY14

70 Figure 29shows flow rates for OC-82 during the audit period in CFS. Only one non-zero read fell outside of the recommended flow range of 5 50 CFS. Figure 29: OC-82 Flow over Time FY14 Figure 30 shows flow rates for OC-88 during the audit period in CFS. OC-88 significantly over-registered during majority of the audit period and was subsequently fixed in April of Of non-zero flow reads, 2.4% fell outside of the recommended range of CFS Figure 30: OC-88 Flow over Time FY14 Figure 31 shows flow rates for OC-88A during the audit period in CFS. OC-88A was off during the majority of the audit period. Of non-zero reads, 0.1% fell outside of the recommended range of 2 20 CFS. Figure 31: OC-88A Flow over Time FY14

71 Figure 32shows the flow rates in CFS for El Toro R-6 during the audit period. SMWD owns capacity on the El Toro reservoir in the El Toro Water District. SMWD accessed this capacity briefly in January to supplement supply when OC-82 was shut down a total of 275 AF was imported during that time period. Figure 32: El Toro Flow over Time FY14

72 System Input Meter Installation Conditions The installation conditions of source meters can have a significant effect on accuracy. Meters should have the equivalent length of five pipe diameters of straight, uninterrupted, pipe upstream of the meter and two pipe diameters downstream. Interruptions such as valves and elbows cause turbulence that may confound meter readings. Figure 34 is an example of the minimum acceptable upstream and downstream pipe lengths for a Venturi source meter. Generally, WSO recommends measuring on-site to evaluate installation conditions; however, on-site measurements were not possible because the meters are located in vaults that prohibit manual measurement. SMWD provided WSO as-built drawings for each of the system boundary meters to approximate the distances of uninterrupted piping upstream and downstream. Table 35 outlines WSO s findings about the installation conditions at each source meter. Pipe Diameter Upstream length: 5X pipe diameter Downstream length: 2X pipe diameter Figure 33: Installation Conditions for Venturi Style Meter Evaluating Input Meter Installation Conditions Recommended Conditions Measured Conditions Installation Secondary Upstream Downstream Reported Upstream Downstream Recs Devices Name Make Size (in) Pipe (in) Pipe (in) Accuracy Pipe (in) Pipe (in) Satisfied? Calibrated? SC-1 Custom Venturi /-0.5% - 34 NO Yes (2013) 5% 95% Confidence Interval (+/-) % SC-2A Custom Venturi /-0.5% - 23 NO Yes (2013) 2% SC-3 Custom Venturi /-0.5% - 16 NO Yes (2013) 2% SC-4A Custom Venturi /-0.5% % SC-5A Custom Venturi 18"x16" /-0.5% - 48 YES Yes (2013) 5% Plaza PS Rosemout Mag /-0.25% - 40 NO - 5% AMP OC /-0.5% % El Toro R % Table 35: Installation Conditions for Supply Meters Though the meters were installed according to MET specifications, all but one of the input meters do not satisfy industry standard installation conditions. SC-1, SC-2A, SC-3, and Plaza PS all have check valves within two pipe diameters of straight length downstream of the meter. Volumetric testing on SC-1, SC-2A, and SC-3 suggests some degree of meter under registration. The as-built drawings showed no upstream obstructions for all input meters, so the absolute length of upstream straight-length is unknown.

73 South County Pipeline Volumetric Testing The AWWA software includes a distinct field for Master Meter and Supply Error Adjustments, which allows the user to make manual adjustments to the registered volume of Water Supplied to account for known inaccuracies. In order to quantify known inaccuracies, WSO and SMWD conducted a volumetric draw down test on four meters located along the South County Pipeline. Volumetric drawdown testing involves flowing a known reference volume form or into a tank through a single meter and then comparing the reference volume against the volume recorded by the meter itself. These meters were selected because they are configured in a way that allows draw down testing. They also supply almost 60% of the total volume of Water Imported. These meters are also especially important because they determine billed amounts for a portion of imports from MWDOC. Volumetric Test Methodology The steps taken to test the meters on the SCP are as follows: 1. Review data availability including the resolution of meter totalizer readings and the accuracy of the tank level measurement. Any uncertainty in these measurements must be accounted for in the presentation of results. Measurement uncertainty was quantified based on the resolution and accuracy of the pressure transmitter and the resolution of the meter totalizer reading. 2. Shut off all takeouts on the SCP and fill the Regulating Reservoir from OC-88 and OC88A. 3. Shut off OC-88 and OC-88A to stop filling the regulating reservoir 4. Allow the level of the regulating reservoir to settle for approximately 30 minutes. 5. While the tank level is settling, sound all valves on the SCP to ensure none are still partially open. 6. Ensure the tank level is stable and record initial readings for the tank level and the meter totalizer. 7. Open the valve on the meter to be tested to allow water to flow. Run water at a flow rate reflective of typical conditions during operation. 8. Allow the test to run for sufficient time such that the impact of measurement uncertainty on the overall test result is minimal. 9. Shut the valve on the tested meter and record the totalizer value. The difference between the initial totalizer reading and this reading is your tested volume 10. Allow the SCP, and the level of the regulating reservoir, to settle for approximately 30 minutes. Record the new tank level. The volume that has left the tank is your reference volume for the test. The volume of any auxiliary structures in the tank must also be taken into account. Table 36 shows the calculation for the volume of water per foot of tank level. 11. Repeat steps 4 10 for each meter to be tested on the SCP.

74 Table 36: Regulating Reservoir Volume Per Foot Measurement uncertainty was quantified using the resolution and accuracy of the pressure transmitter and the resolution of the meter itself. The accuracy of the pressure transmitter (+/ ft) was retrieved from SMAR testing records conducted on the instrument before shipment to SMWD. Table 37 shows the factors used to quantify measurement uncertainty. The error margins on the results of the test were calculated as the percent of the total test reference volume made up by measurement uncertainty. Specifications are provided in +/- feet of regulating reservoir tank height. Volumetric Test Results Regulating Reservoir: Total Tank Volume Per Foot Internal Radius of Tank (ft) Internal Volume of Tank (gal) 94, Radius of Overflow Pipe (ft) 2.00 Radius of Support Columns (ft) 1.00 Number of Support Columns Volume of Auxiliary Structures (gal) Total Volume Per Foot Excluding Auxiliary Structures (gal) 93, Total Measurement Uncertainty Pressure Transmitter Resolution (ft) +/ Pressure Transmitter Accuracy (ft) +/ Meter Resolution (ft)* +/ *Meter Resolution only applicable to test on SC-1 Table 37: Measurement Uncertainty The test was successful in large part due to the coordinated effort of the SMWD operations staff. Table 38 summarizes the results of all the meters tested on the SCP. The reference volume is the volume of water that left the regulating reservoir. This volume was calculated by multiplying the change in tank level by the total volume per foot excluding auxiliary strictures shown in Table 36. The Recorded volume was taken directly from the meter totalizer readings from the Honeywell data loggers on-site. The meter accuracy is calculated as the ratio of the reference volume to the recorded volume, and the meter error is the difference between the accuracy and 100%. Volumetric Drawdown Test Results SC-1 SC-2A SC-3 SC-5A Start Time Jan-5 09:40 AM Jan-6 02:35 PM Jan-5 01:46 PM Jan-6 09:53 AM End Time Jan-5 12:57 PM Jan-6 04:56 PM Jan-5 05:03 PM Jan-6 01:26 PM Duration (hrs) Reference Volume 1,069,095 1,002, , ,059 Recorded Volume 1,028, , , ,286 Average Flow Rate (CFS) Test Uncertainty (+/-) 1.17% 0.50% 0.78% 0.51% Meter Error -3.80% -1.09% -3.62% -1.49% Table 38: Drawdown Test Results

75 Unintended Discharge In between meter tests, WSO staff observed a slight and steady drop in the level of the regulating reservoir. This drop in level indicated an unknown discharge of water leaving the regulating reservoir. WSO staff recorded tank level readings in order to quantify the magnitude of this unintended discharge and to evaluate the impact on the test results presented above in Table 38. Figure 34 and Figure 35 show two independent periods of recorded level drop. Note that the scale of these figures do not start at zero; this is an important consideration because it allows SMWD to see this important and steady decline in reservoir level. It is also important to note that it is uncertain whether this discharge was occurring during the meter tests. The total rate of outflow is estimated to be approximately 126 GPM, or CFS. Even though the rate of discharge appears small, it represents 2.5% of the average flow rate of the tests ( CFS) and will impact the results by approximately that margin Regulating Reservoir Level Drop: Sample A y = x Figure 34: Unintended Discharge from Regulating Reservoir (Sample A) Regulating Reservoir Level: Sample A y = x

76 Regulating Reservoir Level Drop: Sample B y = x Figure 35: Unintended Discharge from Regulating Reservoir (Sample B) Regulating Reservoir Level: Sample B y = x Effect of Unintended Discharge on Test Results Even though it is uncertain if the discharge was occurring during the tests themselves, WSO evaluated the impact of the unintended discharge on the results of the tests. First, WSO calculated an average rate of discharge based on Sample A and Sample B outlined above ( CFS). This rate was then used to calculate a volume of unintended discharge that would have escaped the system during each test. The total volume of discharge must be deducted from the reference volume to calculate the adjusted test results. Adjusted results are presented in Table 39 below.

77 Adjusted Volumetric Drawdown Test Results SC-1 SC-2A SC-3 SC-5A Duration (hrs) Reference Volume 1,069,095 1,002, , ,059 Adjusted Reference Volume* 1,044, , , ,118 Recorded Volume 1,028, , , ,286 Test Uncertainty (+/-) 1.17% 0.50% 0.78% 0.51% Meter Error -3.80% -1.09% -3.62% -1.49% Adjusted Meter Error** -1.50% 0.70% 0.26% 1.26% * The Adjusted reference volume was calculated by assuming the uninteded discharge (0.28 CFS) ran continuously for the duration of the test. ** Adjusted by taking into account an estimate for the unintended discharge. Table 39: Adjusted Volumetric Drawdown Test Results The likely cause of these discharges was use by bootlegs on the South County Pipeline including the Chiquita Water Treatment Plant. A second static test conducted in April of 2016 confirmed that the SCP can be isolated, with no observed tank level drop for several hours. During this second static test, SMWD positioned a camera to take still images of the pressure transmitter each minute from 11:00am until after 5:00pm. As Figure 36 shows, during the period the SCP was isolated, no tank level drop was observed. This highlights an important consideration for conducting future tests the bootlegs must not be overlooked, they must be shut off for the duration of any future test. Level (ft) Static Test: Regulating Reservoir Level Time Figure 36: SCP Static Test Regulating Reservoir Level

78 1.2 Authorized Consumption Analysis Evaluating Data Completeness Count of Accounts and Records per Month Figure 37 shows the unique count of customers by read month. The blue bars fall within the audit period. There are no significant variations in the count of customers per month. This suggests that the billing database provided is complete. Figure 37: Count of Customers by Meter Read Month Similarly, WSO counted the number of records in the billing database for each month during the audit period. Figure 38shows the count of records by read month. Once again, none of the months vary significantly from the others another strong indication that the billing data is suitable for calculating BMAC.

79 Figure 38: Count of Records by Meter Read Month Overall, the count of accounts and records by read month suggests the billing data provided by SMWD is complete. An additional check allows us to further validate the billing database and lend confidence to our ultimate estimate for BMAC The count of bills per location and meter. Count of Bills per Location and Meter Each location and meter should receive a single bill each month with a few exceptions. As we expected, since SMWD bills on a monthly basis, the majority of location-meter pairs were billed 12 times, once for each month in the audit period. This finding lends additional confidence in the completeness of the billing data provided. The counts provided in Table 40 below were generated using unique combinations of location ID and meter ID excluding records flagged as initial or set these codes are used to demark meter swap outs and customer swap outs and do not reflect actual consumption. If a new location was created or the meter was replaced during the audit period, there would be fewer than 12 records in the billing database. Similarly, if a customer left and another moved in during the audit period there would be more than 12 records in the billing database. Finally, if a meter has two registers, as is the case for the three location-meter pairs with 24 records, there may be more than 12 reads.

80 Count of Records Per Location - Meter Pair Records Count of Location - Meter Pairs* Percent of Total % % % % % % % % % % % 12 46, % 13 3, % % % % % TOTAL 54, % *Unique combinations of location ID and Meter ID Table 40: Count of Records per Location ID Meter ID Pair Database Integrity Checks Duplicate Readings We defined duplicate records as records that share the same read date, meter ID, location ID, and consumption volume. We found 2,683 duplicate records in the billing database for the audit period; however, none of these records recorded any water consumption. The majority of the records that were duplicated were associated with initial and final reads that occur when a customer moves in/out. This was expected because these reads typically initialize or close out a customer in the billing database and do not record consumption. However, 38 regular reads were also duplicated. Again, none of these reads reflected consumption so they do not change our calculation for Billed Metered Authorized Consumption. Table 41 shows the breakdown of duplicate reads by read type. A list of the duplicated regular reads is also provided in the excel workbook that accompanies this memo, SMWD_AuthorizedConsumption_FinalTables.xlsx.

81 Duplicate Readings By Read Type Total Consumption Read Type (CCF) Records F 0 1,315 I 0 1,326 O 0 36 R 0 38 S 0 30 Total 0 2,745 Table 41: Duplicate Readings Negative Consumption Volumes WSO identified six records that had negative consumption volumes associated with five location IDs totaling to AF (5,865 gal). Records with negative consumption volumes are often billing corrections that reflect appropriate volumetric corrections, however in some cases these corrections are financial adjustments only and may skew the calculation for BMAC. In SMWD s case, these negative consumption volumes are all meters that had misreads and the variance was 1 to 2 CCF. It is not worth the trouble of processing a cancel/rebill to correct such small variances. Overall, the effect on the calculation of BMAC is negligible. Table 42 shows a list of the five location IDs that have negative consumption volumes. Negative Consumption Volumes Location IDs Table 42: Location IDs with Negative Consumption Large Readings Analysis WSO identified the top 108 location IDs responsible for the largest 500 readings in the audit period. For each location ID, WSO graphed the consumption readings for the data provided. These locations are collectively responsible for almost 7% of Billed Metered Authorized Consumption. After a visual inspection of the consumption trend for all 108 locations, WSO did not identify any outlying or suspicious readings. Consecutive Zero Reads WSO counted the number of consecutive zero reads for unique pairing of location ID and meter ID. Consecutive zero reads can indicate stuck or broken meters or other data handing issues. For the purposes of this analysis, location-meter pairs without any water consumption during the audit period were

82 excluded. Consecutive zero reads had to share the same location ID and meter ID in addition to being adjacent in time. Table 43 below shows the count of unique location-meter ID pairings for each number of consecutive zero reads. For example, 63 account- meter pairs had six consecutive zero reads during the audit period. The results of this analysis suggest that SMWD has robust alarms and response protocols for managing consecutive zero readings. Consecutive Zero Consumption Consecutive Zero Reads Summary Count of Location ID - Meter ID Pairs Percent of Total 0 49, % 1 2, % % % % % % % % % % % % Table 43: Consecutive Zero Readings by Location Meter ID Pair Customers without Consumption 1,092 location IDs, or 2.08% of all location IDs, did not record any consumption during the audit period. These locations may be flagged as inactive in SMWD s billing database, but are still included in an export of the raw data. SMWD has thorough protocols for identifying inactive accounts. Table 44 shows a breakdown of IDs that did not record any consumption during the audit period. A list of the 1,092 Location IDs that did not record consumption during the audit period is also provided in the excel workbook that accompanied this memo. IDs with No Consumption ID Variable Count with No Consumption Percent of Total Location ID 1, % Customer ID % Meter ID 1, % Table 44: Count of Locations, Customers, and Meters with No Consumption

83 Meters with Multiple Sizes and Manufacturers There are 12 meters that are listed as having more than one size in the billing data provided. The most recent size reflects the correct meter size. All of the meters that had multiple sizes changed from ¾ inch to 1 inch. Table 45 below lists the 12 meters that had multiple sizes encoded in the database through the audit period. Similarly, five meters were listed first without a manufacturer and then as Neptune meters. These meters are shown in Table 46. Meters With Multiple Sizes Consumption Meter ID Sizes (CCF) /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch /4 Inch, 1 Inch 57 TOTAL 2,771 Table 45: Meters with Multiple Sizes Meters With Multiple Manufacturers Total Consumption Meter ID Manufacturers (CCF) Unknown, Neptune Unknown, Neptune Unknown, Neptune Unknown, Neptune Unknown, Neptune 1,471 Total 2,623 Table 46: Meters with Multiple Manufacturers Miscoded Meter Register Count During the audit period, SMWD had five meters with two registers. These meters are associated with 96 records in the billing database. These compound meters have two reads for each read month; one for the high flow side that was flagged as HCCF, and one for the low flow side that was flagged as LCCF. WSO identified two meters, and , that SMWD flagged as having two registers but do not have readings for both the HCCF and LCCF side.

84 Estimated Reads We explored the records that were flagged as estimated in the billing database. SMWD identifies two types of estimates. Estimates that are manually calculated are coded as C while estimates that are automatically calculated are coded as E. As Table 47 shows, the overall contribution of estimated reads to total BMAC is very small, about.05% of the total. Nevertheless, the average consumption volume for reads marked as E is about twice that of the rest of the billing database. SMWD intended to estimate these reads using the average of the last three months for that location. While any error here would have a very small effect on the overall calculation of BMAC, WSO recommends periodically checking the algorithm that calculates these estimated reads. Estimate Code Summary Consumption (CCF) Average Consumption (CCF) Table 47: Summary of Meter Reading Estimate Codes Six location ID-meter ID pairs had three or more estimated reads in a row. These ID pairs consisted of five construction meters that inspection staff were not able to locate or get reads and one custom home that had an installation issue. The IDs in addition to count of consecutive estimates are provided in Table 48 below. WSO conducts this analysis because consecutive estimates may indicate a more pervasive problem with maintenance of the active meter stock. However, there are so few location ID-meter ID pairs with consecutive estimates, that this is not an issue for SMWD. Table 48: Consecutive Meter Read Estimates Percent of Consumption Estimate Code Records Blank 633,582 11,768, % C 115 2, % E 82 3, % Location ID-Meter ID Pairs with Consecutive Estimates Consecutive Estimates Consumption (CCF) Location ID Meter ID

85 Time Sensitivity Analysis Since we have determined that the billing data is complete, and we have evaluated the integrity of the data itself, we can next calculate Billed Metered Authorized Consumption (BMAC) by read month. Figure 39 below shows the consumption profile by meter read month. The consumption pattern is typical for California water agencies most consumption occurs during the summer months. The summer demand in 2014 did not decline significantly from 2013 drought pressure in the region did not appear to significantly decrease metered consumption until FY16. Figure 39: Billed Metered Authorized Consumption by Meter Read Month One potential source for error when water consumption is aggregated by meter read date is the time lag between consumption and the meter read date. Figure 41 shows how consumption that actually occurred over several months can be incorrectly allocated to the month in which the meter was read. This lag time can introduce uncertainty in the month-to-month estimate for BMAC. Of course, lag time in the middle of the audit period will not affect our overall estimate for BMAC. However, consumption that straddles the edges of the audit period, July 2013 or June 2014, will affect our estimate for BMAC. In order to compensate for this potential error, WSO apportioned consumption across the months when it actually occurred. Month 1 Month 2 Month 3 Previous Read Date Consumption Meter Read Date Figure 40: Explanation of Apportionment Methodology

86 For example, using Figure 41, we can see that the consumption volume would be allocated fully to month 3 by simply summing the records grouped by read month. By apportioning this volume, we can divide the consumption volume proportionally between month 1, 2 and 3 when it actually occurred. Figure 41 below shows the comparison of our estimates for BMAC by read month and after apportioning consumption volumes. Audit Boundary Figure 41: BMAC Apportioned and by Meter Read Month In order to estimate the overall effect of time lag on our estimate for BMAC, we can compare the apportioned total and the total by read date. If the difference between these two estimates is more than 1% then the lag time has a significant effect on our estimate for BMAC. Table 49 below shows the estimates for BMAC by read month and by apportioned consumption. Estimates for BMAC using Read Month and Apportionment Percent Read Month Apportioned Difference 11,773,356 11,753, % Table 49: Comparison of Apportioned and Read Month BMAC Totals As we can see in Table 49, the estimate for BMAC varies by 0.17% between the two approaches. This is a relatively small difference and lends confidence to using meter read date to estimate BMAC in future

87 audits, without apportionment. Table 50 below shows the numeric total for BMAC each month of our audit period aggregated both by read date and by using apportionment. BMAC by Meter Read Month and Apportionment Month Total Consumption (CCF) Apportioned Consumption (CCF) Audit May ,882 1,080,541 Out of Audit Jun 13 1,118,998 1,169,667 Out of Audit Jul 13 1,258,352 1,262,409 In Audit Aug 13 1,226,488 1,277,487 In Audit Sep 13 1,250,858 1,187,552 In Audit Oct 13 1,129,126 1,019,030 In Audit Nov , ,171 In Audit Dec , ,444 In Audit Jan , ,320 In Audit Feb , ,276 In Audit Mar , ,788 In Audit Apr , ,799 In Audit May 14 1,052,259 1,148,992 In Audit Jun 14 1,183,889 1,188,463 In Audit Jul 14 1,215,135 1,238,732 Out of Audit Aug 14 1,228, ,046 Out of Audit Total 11,773,356 11,753,730 Table 50: BMAC by Meter Read Month and Apportionment To explore the sensitivity of BMAC to shifting the bounds of the audit period, we calculate BMAC for several contiguous 12-month periods. Again, if shifting the bounds of the audit period changes our estimate for BMAC by more than 1% the bounds of the audit period might have a significant impact on the total calculation for BMAC. Table 51 shows the estimates for BMAC by shifting the audit boundary by one and two months ahead and back. Time Lag Analysis Estimate of BMAC by Read Date Table 51: BMAC Time Lag Analysis % Difference from Audit Period Interval May April ,648, % June May ,708, % July June ,773, % August July ,730, % September August ,731, %

88 Shifting the audit period bounds by two months back, the April 2013 May 2014 test case, caused our estimate for BMAC to change by more than 1%. This is likely because consumption patterns in the summer months are more erratic month-to-month than during the winter. In future audits, SMWD might consider defining the audit period boundary during the winter months when consumption is less erratic. However, SMWD can continue to define the audit boundary by the fiscal year, keeping in mind the potential for increased uncertainty due to the volatility of consumption in the summer months. Meter Consumption Range Analysis Meter Consumption Summary Figure 42 shows the total consumption that different sized meters recorded in the field in CCF. Similarly, we plotted the count of meters by size to show the distribution by quantity in Figure 43. ¾ inch meters are responsible for more than 50% of the total consumption recorded during the audit period. 1 ½ meters only represented 4% of consumption while 2 meters represented more than 20%. On average, 3 meters recorded the most water per meter during the audit. These findings can help inform meter replacement strategies and testing plans. Table 52 provides a numeric summary of consumption patterns and counts by meter size, while Table 53 replicates that analysis by manufacturer. Please note that in the instance a unique meter ID is recorded with more than one size or make, they will be counted once in each category. The count of Meter IDs with more than one size or make is provided in the appendix, Meters with Multiple Sizes and Manufacturers. Figure 42: Count of Meters by Size

89 Figure 43: Total Consumption by Meter Size Meter Size Summary Count of Meters* Volume (CCF) Records Table 52: Meter Size Summary Average Consumption (CCF)* Percent of BMAC Size 3/4 Inch 42,148 6,044, , % 1 Inch 7,932 1,986,593 92, % 1.5 Inch ,408 10, % 2 Inch 2,644 2,824,327 31, % 2.5 Inch 92 24,410 1, % 3 Inch , % 4 Inch ,791 1, % 6 Inch ,586 1, % 8 Inch ,722 2, % 10 Inch % Total 54,227 11,773, , % * Meters with more than one size will be counted once in each category in which they appear. ** Average calculated per record

90 Meter Make Summary Meter Make Count of Meters Volume (CCF) Records Table 53: Meter Make Summary Average Consumption (CCF)* Percent of BMAC ABB % Badger ,142 5, % Hendey 2 2, % Neptune 22,309 4,571, , % Other , , % Rockwell 4,310 1,012,548 53, % Sensus 27,047 5,969, , % Metron 26 25, % Unknown 30 17, % Total 54,220 11,773, , % * Meters with more than one manufacturer will be counted once in each category in which they appear. ** Average calculated per record

91 Small Meter Flow Range Analysis per Meter Some customer meters in the field may be experiencing flows much higher or much lower than their manufacturer may recommend. These meters are more likely to be inaccurate. Monthly billing data can give us an indication for which meters might be inappropriately sized, however, additional investigation with high resolution data loggers in the field would provide a more complete picture of the actual range of flows observed. Meter flow range analysis allows us to compare the average flow rate for all meters in the distribution system by meter size. This helps us identify particular meters that might by miss-sized for their flow conditions. To properly evaluate the size of a customer meter, instantaneous flow data is required, therefore, low resolution, monthly billing data can only provide an indication for potential meter misssizing. WSO took the following steps to conduct the flow range analysis: 1. For each meter, we took the total days that SMWD billed for the meter in the audit period and the total consumption volume registered during the audit period. 2. We then calculated the average flow per day for each meter by dividing total consumption by the total number of billed days. One meter, for Lake Mission Viejo, was excluded from Figure 44, because it had an extremely high average flow rate of CCF per day. 3. We excluded fire service meters, meters with service type FS, because they are not sized for the same consumption patterns as domestic water or irrigation meters. 4. Finally, we identified the top 100 meters responsible for the most consumption during the audit period. Figure 44 shows the results of this analysis. The y-axis is the average consumption per day during the audit period while the x-axis clusters points by meter size. Each point represents a single meter in the billing database. Meters responsible for the greatest total consumption during the audit period are colored red. The figure shows how meters of certain sizes tend to have similar average consumption. Some meters are registering very high flows for their size. For example, a single 1-inch meter has a much greater average flow rate than other meters of the same size. Meters like this one may be inappropriately sized and should be investigated further. The meters that stand out as experiencing high flows compared to their similarly sized peers are listed in Table 54 below.

92 Meter ID Total Consumption (CCF) Figure 44: Meter Consumption Range Analysis Meters Experiencing High Average Flows Average Flow per Day (CCF) Meter Size Table 54: Meters Experiencing High Average Flow Rates Service Type Percent of BMAC , Inch ID 0.09% , Inch ID 0.08% , Inch ID 0.08% B 8, Inch ID 0.08% D 8, Inch ID 0.07% , Inch ID 0.07% , Inch ID 0.05% , Inch ID 0.04% , Inch ID 0.04% , Inch ID 0.04% , Inch WD 0.04% , Inch ID 0.03% , Inch WD 0.03% , /4 Inch WD 0.01% , /4 Inch WD 0.01% , /4 Inch WD 0.01% , /4 Inch WD 0.01% Total 92, %

93 While Figure 44 provides a good indication of which meters are experiencing higher average consumption per day, it does not help us identify the distribution of flow conditions for meters that have medium to low flows. Small Meter Flow Range Analysis: Log-Linear Scale Using the same flow analysis conducted above, we can visualize the counts of meters exposed to different flow ranges on a log-linear scale. The log-linear scale allows us to study the distribution of meters experiencing medium to low flows more closely. If meters are appropriately sized, each increasing size class should be exposed to higher flow rates as we know, larger meters are typically designed to accurately measure greater flows. Again, to properly evaluate the size of a customer meter, instantaneous flow data is required; therefore, low resolution billing data can only provide an indication for meter misssizing. Figure 45 shows the count of meters exposed to different flow ranges. Note that the categories on the x-axis are log-linear. In other words, each range is twice as large as the previous. The steps we took to conduct this analysis are as follows: 1. For each meter, we took the total days that SMWD billed for the meter in the audit period and the total consumption volume registered during the audit period. 2. We then calculated the average flow per day for each meter by dividing total consumption by the total number of billed days. 3. We excluded fire service meters because they are not sized for the same consumption patterns as domestic water or irrigation meters. While there are fewer 1.5 and 2.5 inch meters than other sizes, meters in these size classes are exposed to a broad range of flow conditions, overlapping significantly with typical flow rates for 1 Inch and even ¾ Inch meters. This suggests that 1.5 Inch meters and 2.5 Inch meters may not be appropriately sized. SMWD should investigate these size classes to ensure they are sized appropriately to measure flows on site. There are very few six and eight inch meters, so few general conclusions can be drawn from these size classes. However, a single 8 inch meter, ID , recorded the highest average flow rate of all meters during the audit period. This meter is located at Lake Mission Viejo. Because this meter registers more water consumption than any other, particular attention should be paid to its accuracy.

94 Figure 45: Meter Consumption Range Analysis Log-Linear Scale

95 SMWD Water Loss Control Program Water Supplied Meter Consumption Distribution Often, the majority of Billed Metered Authorized Consumption is consumed by only a small percent of the meter population. The distribution of consumption can provide an indication for how much attention SMWD should pay to its largest consumers. Figure 46 shows the distribution of BMAC by meter percentile. For example, from this visualization, we can conclude that the top 1% of SMWD s meters are responsible for recording almost 20% of the total Billed Metered Consumption. Because a small percentage of meters are consuming the most water, SMWD should focus efforts on ensuring these few meters are accurate. A more detailed discussion of large meter testing can be found in the memo titled, Apparent Losses and Large Meter Testing. Figure 46: Consumption Distribution 1.3 Real Losses Analysis Clerical Errors in Repair Records Upon review of the leakage repair records, some possible errors in the raw data were identified. These data errors could be incorrect leakage start, shutoff and repair times as well as leakage volumes. Where date errors were encountered, WSO used best estimates to create realistic total leak time. All of these corrections simply required adjusting date times from AM to PM or vice versa. Where reported leakage volumes did not match calculated reported leakage using flow rate and leak duration, WSO used the calculated volumes. The leakage volume errors, although numerous (40%), were relatively small with a total difference between calculated and reported leakage volume equaling 0.58 AF. Table 55 summarizes the suspected data errors encountered in the leakage repair records provided by Santa Margarita.

96 SMWD Water Loss Control Program Water Supplied Suspected Errors in Leakage Repair Records Error Number of Occurrences Percentage Suspected incorrect Leakage start date 2 2% Suspected incorrect shutoff date 6 7% Suspected incorrect fix date 1 1% Suspected incorrect volume calculations 34 40% Total Suspected Errors 43 51% Table 55: Errors in Leakage Repair Records Reported Leak Clusters The leakage repair records provided by SMWD included the address of the leak location. Using this information, WSO plotted the reported leak events for FY2014. A cluster of service main breaks was discovered near the Castadel Sol Golf Course with the majority of these leaks occurring on plastic service connections (Figure 47). SMWD staff has informed us that this area is considered to have aging infrastructure but other causes may also be contributing to this leak cluster.

97 SMWD Water Loss Control Program Water Supplied Casta Del Sol Golf Course Figure 47: Cluster of Plastic Main Service Leaks The transient pressure logger temporarily located at Calle Neruda and hydraulically connected to the temporary pressure logger at St Elena, recorded pressure transients with amplitudes greater than 25 psi. These pressure waves can reduce the service life of pipes in this area, eventually leading to increased leak frequency. It is recommended to investigate the pump shut-off/start-up processes at the pump station servicing this area. Furthermore, installation procedures for plastic service connections in this area should also be investigated.