AMR for Water Distribution System Leak Detection. Dave Hughes Infrastructure Engineer American Water

Transcription

1 AMR for Water Distribution System Leak Detection Dave Hughes Infrastructure Engineer American Water 1

2 American Water individual Largest investor-owned water service services areas provider in North America 0 44,000 miles of distribution mains Serves 16.2 million people 0 84 surface water Operations in 32 states and treatment Canada plants 0 7, groundwater employees treatment plants Owns and operates or Operates 789 water treatment plants and 56 wastewater treatment plants Hydrants wells 0 175,000 Fire Utility Only O&M Only Both 2

3 Acoustic Monitoring Device 0 Using a low-cost, waterproof sensor installed near the water meter 0 Easily strapped to service pipe 0 Can detect noise into system, about 500 feet through metallic pipe 300 feet through plastic pipe 0 Maintenance-free, 10+ years (battery) 0 Read through AMR communication interface or as independent radio unit Graphics courtesy of Flow Metrix/Itron 3

4 AMR Fixed Network and Acoustic Monitoring WEBSITE ACCESS COMPUTER Acoustic (leak) Data Meter Reads, Meter Status Data Collector Units Billing Meter reads Meter Transmitter Units Acoustic (leak) Data Billing, and recent Meter reads Selected Reports Water Meter Water Meter Water Meter Acoustic Monitor American Water Computer Selected Meter reads Schematic courtesy of Hexagram 4

5 Piloting Site 0 Connellsville is located along the steep Youghiogheny River valley forty miles south of Pittsburgh. 0 System dates back to 1880 s. Mostly cast iron and galvanized steel but also AC, ductile, & plastic. About 38 of 57 miles of main are over 100 years old. 0 NRW was over 25% and the cost of water (purchased from an adjacent system) is $1.90 per thousand gallons 5

6 How Does It Work? 0 Monitor listens and identifies the minimum sound in 10 minute intervals between 12:30 and 4:30 AM when water use and other noise sources are minimal. 0 The single nightly data point is broken down into a spectrum of sound associated with leak noise and shows low, medium and high frequencies as well as a sum of the leak noise frequencies for ferrous and plastic. 0 The monitor also reports on the highest differential between the lowest noise and the highest (red). 0 The software interprets changes and magnitude of sounds to rate the location as a possible source of a leak. 6

7 Snapshot of Leak Survey Data The acoustic monitor data is organized and highlight the most likely candidates for leak noise to check 7

8 Continuous Leak Survey Data Scan of 1810 for entire monitoring period back to June 2005 Scan of 1810 for last month showing leak start 5/5/07 There is a typical brief initial noise peak as leaking pipe smoothes and moves soil away from pipe The software displays a history showing the noise level at each day. 8

9 Tracking Leak Sound History 0 An MLOG located near a cast iron main leak indicates a leak running April 1, 2006 until November 23,

10 Initial 2005 Connellsville Results MLOG leak detecting sensors were installed in Connellsville in Spring Five one week correlation studies were conducted by Flow Metrix and American Water staff in June, July, August, October, December A total of 129 locations were visited in From June to December leaks were reported in Connellsville (18 the same period in 2003, 12 in 2004) of the 46 leaks were identified by acoustic monitors and repaired in advance of surfacing. Another 10 were MLOG identified before surfacing but appeared before repair made. The remaining 12 surfaced and were repaired. 0 With the reduction of blow-off flow and leaks, flow is consistently down about 255, ,000 gpd in

11 Finding the Leak that Never Surfaces This leak was destined to flow to the nearby river or into the storm sewer above without coming to the surface for years. Found by acoustic monitoring 11

12 Piloting Results of the 40 leaks were identified by acoustic monitors and repaired in advance of surfacing. Another 6 were MLOG identified before surfacing but appeared before repair made. The remaining 15 surfaced and were repaired. 0 Link to water temperature became apparent in the fall. Each point represents the start of a main or service in Connellsville We can actually anticipate leak starts occurring mostly after a water temperature drop in surface supply systems. Water temperature of surface water near plant 12

13 1/26/2006 1/27/2006 1/28/2006 1/29/2006 1/30/2006 Piloting Results In a year of very cold temperatures 18 of the 68 leaks were identified by acoustic monitors and repaired in advance of surfacing. 0 Another 12 were MLOG identified before surfacing but appeared before repair made. The remaining 38 surfaced and were repaired including at least 18 breaks which are believed to have surfaced immediately and could not be detected before repair. 0 Many of the immediate failures were splits and circumferential failures of cast iron pipe. The monitors continue to detect about half of the smaller pipe failures that are more likely to run for extended time though at lower flows /3/ /15/ /16/ /17/ /18/ /19/ /20/ /21/ /22/ /23/ /24/ /25/ /26/ /27/ /28/ /29/ /30/ /1/ /2/ /14/ /2/2006 2/3/2006 2/4/2006 2/5/2006 2/6/2006 2/7/2006 2/8/2006 2/9/2006 2/10/2006 2/11/2006 2/12/2006 2/13/ /25/2006 1/6/2006 1/7/2006 1/8/2006 1/9/2006 1/10/2006 1/11/2006 1/12/2006 1/13/2006 1/14/2006 1/15/2006 1/16/2006 1/17/2006 1/18/2006 1/19/2006 1/20/2006 1/21/2006 1/22/2006 1/23/2006 1/24/2006 1/5/2006

14 winter 2007 spring fall 2006 summer 2006 Changing the Pattern of Leaks This line shows typical cyclic pattern of leaks before MLOG small steel main leaks - 2 inch cast iron main leaks all service line leaks all leaks typical cycle winter 2003 spring 2003 summer 2003 fall 2003 winter 2004 spring 2004 summer 2004 fall 2004 winter 2005 spring 2005 summer 2005 fall 2005 winter 2006 spring MLOG operational late Spring 2005

15 Not All Leaks Found by Acoustic Monitoring 0 MLOGs have missed some long standing leaks in Connellsville for a variety of reasons: MLOGs placed out of range of some leaks (more units will be added) MLOG failures in meter pits (a more resilient model is being installed) Pipe material (repair) transitions mask out sound Background sounds hide leak noises Potential for loss of contact with pipe 0 AMR system continues to reliably transmit meter readings so communication network is intact but there can be incomplete data sent causing MLOG data to skip. 15

16 AMR Piloting at American Water 0 Research in Connellsville. AwwaRF AMR Project and AWWARF Cross Connection Project will examine AMR and alarms detecting metered backflow 0 Other pilots with AMR fixed network system: Itron in Irvington, NJ system also part of the AwwaRF AMR Project with emphasis on identifying customer leaks Hexagram/Gutermann with leak noise correlation in Metro Chicago Future testing ahead for advanced systems from Datamatic, Neptune 0 Evaluation of alternative acoustic monitoring compatible with mobile AMR. Permalog systems active in PA and other states 0 Radio MLOGs will be combined with mobile AMR. 16

17 Questions Being Answered 0 What Leaks is are the detected effective usually range up of the to 400 acoustic feet but monitors have, for on various occasion combinations been detected of in pipe? the range of 1000 feet. (What Leak sound are the does adjustments dissipate over for leaks distance in various especially pipe where over pipe varying materials distances?) changes from metal to plastic and back. 0 Will User the detection user learn skills to have differentiate improved even with more experience. effectively Sound history between and knowledge leak noise of and background extraneous noise noise? sources helps discern between leak and other noise. 0 Can Pinpointing pinpointing leaks of can leaks be accomplished based upon suspected by one man and leak noise MLOG correlator target in between areas be accomplished minutes. Correlators quickly? can be compromised What by daytime tools help noise. and Mixed when? results from overnight correlation. 0 Will Definite the operators higher noise continue extremes to find of leaks heat and effectively cold. as There the system are patterns. cycles Optimum between time noisy for and leak quiet? detection appears to be the fall. 17

18 Questions Answered in Assessing Continuous Leak Detection as an AM tool If the utility is performing this analysis, a weekly or biweekly analysis 0 What is the effective frequency for monitoring? What is the value of seems to be effective. The vendor is setting up a web site that could every day monitoring versus every 30 days, every week? provide daily alerts on significant changes. 0 What It appears types that and about ratios 50-75% of main of bursts leaks start as small leaks. detectable Splits leaks? and Can circumferential we learn something breaks are by more exhuming likely to pipes surface when rapidly. a leak Many is first of the cast detected? iron pipe breaks are not readily detected slow evolving breaks. 0 How Leaks long become does surfacing it take for bursts a leak in to a become time range a from immediate to burst? months, What maybe dynamics years. Drainage, may play a surface role. conditions play a role. 0 What To date, is the leaks economic detected value and repaired of rapid before repair? surfacing cost 3/4 of other surfacing leaks (and even less compared to leaks we could have detected). 18

19 The AMR BOP a three step process of evaluation for American Water 0 Level 1 Screening eliminates obvious locations where AMR change not practical. 0 Level 2 Screening provides each district an opportunity to gain reasonable view of potential of AMR in systems to include mobile and fixed network and use of acoustic monitoring for leak detection. 0 Level 3 Screening more sophisticated economic analysis designed to set the table for making a business case for AMR. 19

20 Detailing Cost Savings Areas 0 Meter reading labor, meter reading supplies, insurance. 0 Customer service labor, reduced regulatory complaints. 0 Reduced liability claims with breaks. 0 Reduced NRW (savings in chemicals, electricity, etc.). 0 Reduced cost of service and main repairs. Scheduling repairs Reduced damage Improved system knowledge No need for system wide leak surveys 20

21 Level 1 Screening Process 0 The deal killers. It makes no sense to look at AMR if billing is not going to be monthly or meter reading is highly efficient because it is the major (but not the only) economic driver. DON'T LOOK AT AMR IF Readling is not monthly and there is NO intent to implement monthly reading the number of meters read by staff exceeds 500/day and estimated reads average below 1.5% there is already a successful (mobile) AMR system being started and is at least 20% complete a substantial meter changeout program has just been completed (50% in last 3 years) 21

22 Modeling Benefits versus Costs 0 AMR equipment including encoder meter conversions. 0 AMR end point (best performed with meter change-out) and collector installations. 0 Acoustic monitoring equipment and installation. 0 Software changes, training, troubleshooting expenses. 0 Increased leak pinpointing expense. 0 Early repair of all leaks does not mean that the total cost of repair is reduced (more leaks to fix initially). Spending today s dollars for leak repair may be more than spending tomorrow s dollars plus cost of lost water. Permalog is another acoustic monitor being connected to fixed network 0 Potential depreciation loss and meter purchase costs to accelerate meter change-out to accommodate AMR. 22

23 Level 2 Cost Model 0 Side by side comparison of mobile AMR and Fixed Network AMS. 0 Minimal input from district or system under study. 0 Relative value of options expressed in years for payback (less than 5 years is good, 5-8 years is reasonable and more than 8 years increasing less practical at the present time. QUICK AFTER INPUT THIS QUICK SUMMARY PROVIDES A QUICK LOOK AT THE POTENTIAL PAYBACK AND COSTS INVOLVED simple payback time calculation SUMMARY for Fixed Network 8.54 years for Mobile AMR 8.30 years 23

24 Level 2 Cost Model - INPUTS 0 Straight forward data entry. USER INPUTS Northwest Indiana district name/system 74,500 number of customers to be served by AMR (assumes 100% of study area) IN two letter abbrevation of state 10% percentage of meters in pits 25% percent of meters to be changeout at same time as AMR installation number of full-time meter readers (or equivalents i.e., 50% of 2 workers time on 7.00 meter reading = 1 meter reader) number of full-time meter readers to be retained through the project and not counted 1.00 toward reduction in cost of project 40% percent of customers served that is high density 50% percent of customers served that is medium density 10% percent of customers served that is low density average day MGD cost to produce water (including chemicals, electricity, water allocation fees, sludge costs per 1000 gallons $ % percentage of Non Revenue Water $25,000,000 annual metered income - needed only if billing is not monthly 12 number of reads for billing per year 0% percent of meters already part of a mobile AMR system 4 time in years to be used to install AMR transmitters 24

25 Level 2 Cost Model - OUTPUTS 0 Summarizes costs, provides for investment/payback profile COST SUMMARY Fixed Network AMS MOBILE AMR SUMMARY OF CAPITAL COSTS AMR system $ 5,831,688 $ 4,450,788 AMR installation $ 1,160,338 $ 908,900 LEAK MONITORING system $ 814,844 $ 801,806 LEAK MONITORING installation $ 26,075 $ 18,578 TOTAL CAPITAL COSTS $ 7,832,944 $ 6,180,072 SUMMARY OF ANNUAL OPERATING COSTS AMR operating and maintenance $ 141,550 $ 25,330 LEAK MONITORING operating and maintenance $ 15,319 $ 16,297 SUBTOTAL ADDITIONAL OPERATING COSTS $ 156,869 $ 41,627 SUMMARY OF ANNUAL OPERATING SAVINGS Meter Reading/Customer Service $ 724,580 $ 505,783 LEAK MONITORING savings $ 350,023 $ 279,996 SUBTOTAL OPERATING SAVINGS $ 1,074,603 $ 785,778 NET ANNUAL OPERATING SAVINGS $ 917,734 $ 744, , 0 0 0, , 0 0 0, a n n u a l o p e r a t i o n s a v i n g s c u m u l a t i v e c h a n g e 6, 0 0 0, Payback 8.5 years and 8.3 years 4, 0 0 0, , 0 0 0, ( 2, 0 0 0, ) ( 4, 0 0 0, ) ( 6, 0 0 0, ) ( 8, 0 0 0, )

26 AMR COST MODEL Understanding and Building the Business Case 1,000 0 The model is flexible for Districts to 500 input their own data as AMR costs vary between American systems as 0 function of system size, density, (500) staffing and location. $ (thousands) AMR Annual Income Impacts (2,435) (3,168) 391 2,910 free cash flow net income 0 The comprehensive model assesses MLOG as savings from reduction of NRW will vary depending on cost of water, amount of NRW and effectiveness of MLOG. 0 Model will look at cash flow and depreciation in format for Finance and Capital Planning offices to analyze. (1,000) (1,500) SUMMARY OF ANNUAL OPERATING COSTS AMR operating and maintenance $ 63,549 MLOG operating and maintenance $ 9,426 $ 72,975 SUMMARY OF ANNUAL OPERATING SAVINGS Meter Reading $ 886,791 Customer Service $ 66,020 MLOG savings $ 1,157,506 $ 2,110,318 NET ANNUAL OPERATING SAVINGS $ 2,037,342 simple payback time calculation 3.76 years 26

27 Discussion: Leakage, AMR and Customer Service 350 Meter use Elm Street Leaks on the customer side of the meter can be captured and confirmed quickly with the fixed network AMR, continuous flow sensing meters and/or acoustics. Alerts may occur well before bills are issued. 0 This provides the opportunity to notify the customer before that next bill of a leak and before the bill gets extremely large. American Water then becomes proactive rather than reactive defending high use and large customer bills. daily meter use /7/05 3/14/05 3/21/05 3/28/05 gallons Saturday Sunday Customer Usage Monday Tuesday Wednesday 4 weeks of Customer daily meter reads showing possible leak Thursday week1 week2 week3 week4 Friday 27

28 Innovation Continues creates issues related to best timing 0 Several AMR firms have developed mesh networks for water utilities: Mesh network permits AMR units to transmit between units and reduce the number of collector units a major savings for AW. Mesh network allows for two way communication. 0 At least two firms have a working shutoff valve that is operated by fixed network and mobile AMR. Shutoff capability would have major impact on transfer of properties. 28

29 Innovative Continues 0 New interfaces with other leak detection firms to offer other leak sensing options. 0 An integrated transmitter inside a standard street valve box. 0 A new pilot in Connellsville to detect backflow via meters will send an immediate alarms in the event of backflow as well as tampering and continuous metered flow. 29

30 Innovative Continues 0 An AMI actuated shutoff valve being tested by PA American Wide Area Network (WAN) Gateway Range Extender evo evo Mesh topology evo 0 Some AMI vendors are introducing mesh technology that allows units to relay from one transmitter to another reducing the number of data collectors evo evo 0 Some AMI firms are working towards the single solution that migrates from touch pad to mobile to fixed network 30

31 Act or Await Innovation 0 Mesh technology could provide a major cost reduction associated with data collectors and obtaining sites for collectors that might ordinarily be located on private property 0 Actuated valves could have serious cost implications for seasonal use meters and high rates of delinquent customers 0 Migrating from one technology to another allows for easy transition and an alternative approach in case of system breakdowns Utilities must decide whether to wait for the next improvement or jump in now and possibly miss an important development 31