Presentation is loading. Please wait.

Presentation is loading. Please wait.

Field Validation Techniques Paul Fanner. IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 2 Overview What is validation? Why is it important? Validation.

Similar presentations


Presentation on theme: "Field Validation Techniques Paul Fanner. IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 2 Overview What is validation? Why is it important? Validation."— Presentation transcript:

1 Field Validation Techniques Paul Fanner

2 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 2 Overview What is validation? Why is it important? Validation Techniques System input meters Revenue meters –Accuracy –Stopped meters Data handling errors –Meter reading –Billing processes Unmetered consumption –Billed –Unbilled Unauthorized consumption

3 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 3 New Providence Water Balance With Error Margins

4 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 4 Range of New Providence Performance Indicators After Validation

5 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 5 Trinidad PI’s Before Validation

6 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 6 95% Confidence Limits of Trinidad Real Losses Before Validation

7 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 7 95% Confidence Limits of Trinidad Real Losses After Validation

8 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 8 System Input Volumes

9 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 9 System Input Meters Large diameter, expensive Installed when plant was constructed Often hard to find manufacturer’s specifications Not designed to be flow validated Rarely, if ever, flow validated But regular calibration of electronics normal Output often fed to SCADA Reported volumes often taken from SCADA SCADA systems can introduce errors in data Many different technologies: Electro-magnetic, Ultrasonic, Venturi, Dall tube, Orifice plate, Insertion meters, Vortex shedding, Turbine, Propeller etc. Errors in the accuracy of these meters result in much larger errors in estimates of real loss volumes Particularly if small number of inputs

10 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 10 Large System Input Meter

11 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 11 Factors Affecting Accuracy Changes to meter geometry Corrosion Corrosion products Other Deposits Debris Mechanical wear Installation problems Pipework configuration Alignment Not running full Signal processing Square root extraction Calibration factors Span errors

12 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 12 System Input Meter Validation Options Test in-situ! Volumetric tests – by far the best option, if practical Drawing down upstream tank / clear well Filling downstream tank Mass balances – useful first step If multiple sets of meters Test meter in series Insertion meter – better than ultrasonic –Insertion mag or turbine meter Clamp-on ultrasonic meter – least preferable option But be aware of potential swirl problems Portable test rig (small meters only)

13 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 13 Swirl in a Long Straight Pipe Source: Dr. Richard Furness, 2008

14 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 14 Common Issues Test across normal operating range of meter 3 flow rates, if operationally possible Record local (and remote) integrator readings at start and end of each test Log meter output (4-20mA / pulse) from primary sensor Obtain data for period of test from SCADA Calculate error range of test If meter is found to have a systematic error Correct system input volumes for error

15 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 15 Data Chain Analysis Does the SCADA system read the same as the primary and secondary device?

16 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 16 Volumetric Test Requires tank large enough to provide test accuracy required Must be able to isolate tank for period of test Requires accurate measurement of tank dimensions and water depth Ultrasonic or high accuracy pressure transducer level gauge Conduct static test on tank to check for leakage / valves passing

17 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 17 Comparative Volume Test Understand the resolution we can measure to 1mm =? volume

18 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 18 Mass Balances – Multiple Meters Typical meters Raw water meters Treatment process meters Production / system input meters Bulk purchases – dual meters Waste lines not normally metered Requires temporary metering for period of evaluation Insertion meter or ultrasonic If inconsistencies found, other tests required.

19 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 19 Insertion Meters Do not put meters within 50D of pumps, fittings or bends Above DN600, distances should double Above DN1200 add additional lengths Be careful with probes: must have 2 plane data or a minimum of two pairs of ultrasonic meters above DN1200 Always determine velocity profile Undertake test with insertion meter at pipe centreline

20 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 20 Tests using ABB Aquaprobe

21 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 21 Velocity Profiling

22 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 22 Velocity Profile

23 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 23 Clamp-on Ultrasonic Meters Do not put meters within 50D of pumps, fittings or bends Above DN600, distances should double Above DN1200 add additional lengths Be careful to use a minimum of two pairs of ultrasonic meters above DN1200 Always measure pipewall thickness using ultrasonic tool Always use silicon grease to ensure good contact

24 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 24 Clamp-on Ultrasonic Meter

25 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 25 Revenue Meter Inaccuracies

26 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 26 Small Revenue Meters Up to 25mm (1”) If regular random samples are removed for testing (recommended), analyze test data Otherwise select totally random sample of meters 50-200 meters Either test in-situ or remove and replace and test on meter test bench. Test at 3 specified flow rates (AWWA M6) or 4 flow rates (ISO 4064) Ideally also log meters for use profiles to calculate average efficiency result from tests (very important with storage tanks)..Or use M6 recommended weightings (15/70/15% of time, 2.0/63.8/34.2% of volume) Determine average meter under-registration of small meters

27 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 27 Hopefully Your Meters are Better!

28 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 28 Medium and Large Revenue Meters Over 25mm (1”) If regular meter tests are undertaken (recommended), analyze test data Otherwise select totally random sample of meters 50 + meters Test in-situ using a portable or mobile test rig May require installation of test tappings and valves Test at 3 specified flow rates (AWWA M6) or 4 flow rates (ISO 4064) Ideally also log meters for use profiles to calculate average efficiency result from tests..Or use M6 recommended weightings (15/70/15% of time, 2.0/63.8/34.2% of volume) Determine average meter under-registration

29 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 29 Testing Large Customer Meters With Portable Test Unit

30 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 30 Data Handling Errors

31 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 31 Data Handling Errors – Meter Reading Examine meter reading and checking processes How is data collected? What scope for manual errors –Validation and high / low / zero consumption checks Is there scope for “café” meter reading? Is there scope for meter reading fraud? How are readings transferred to billing system? What validation checks are run in the billing system? What follow-up is done with exception reports? What follow-up is done with meter reading error codes? How are zero consumptions checked? Check a sample of readings through these processes

32 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 32 Review Meter Reading Process

33 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 33 Data Handling Errors – Billing Processes How good is customer cadastral? Are customers located on GIS? How are estimates calculated? How are over-estimates reversed? What happens to consumption data when bills are adjusted? Customer complaints Leakage allowances Etc. How does the system manage meters “going round the clock” What happens to consumption of accounts that aren’t billed?

34 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 34 Billing Extract Take full extract of data re. meter reading, meter, consumption, customer information, customer class, flags for estimated, no bill etc. Year of study, plus meter reading interval either side, based on meter reading date Check samples of records to confirm correct operation of processes Determine consumption by month from data, based on meter reading date. Apply reversals! Compare with reported consumption Adjust for meter reading lag time Evaluate average number of months for stopped meters Identify any systematic errors

35 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 35 Check Against Reported Consumption Volumes

36 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 36 Adjust for Meter Reading Lag Time

37 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 37 Unauthorized Consumption (Fraud)

38 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 38 Identify Key Types of Fraud Regular connection mis-coded Inactive connection reactivated Bypassed meter Tampered meter Illegal connection (urban and rural) Illegal use of hydrants or fill points for tankers Illegal use of fire lines Illegal use of new connections for construction water Others?

39 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 39 Where there is power consumption there is usually water consumption…

40 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 40 Illegal Connections (and poor materials!)

41 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 41 By-passed Meter The daily newspaper is used to prove the date the fraud was located

42 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 42 Water Meter Tampered

43 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 43 Water Meter Tampered

44 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 44 Water Meter Seal Broken

45 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 45 Tankers Often Fill in an Unauthorized Manner

46 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 46 Unauthorized Hydrant Use

47 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 47 Illegal Connections in Rural Areas

48 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 48 How to Identify Fraud? Cross reference power consumption with water consumption records Cross reference municipal land use records with water consumption records and rate types Plot active and inactive connections onto municipal parcels identify parcels with no apparent connection Utilize GIS and aerial photos or Google earth to have a first look Random field investigations to identify frequency Estimate number and frequency of tankers filling up at hydrants Road cleaning? Sewer and storm drain flushing? Water trucking? Others?

49 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 49

50 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 50 February/07 Unidade de Negócio Centro – MC Estimation of volumes lost through fraud

51 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 51 Mooca: 0.98% of the test sample had fraudulent activity Paulista: 0.65% of the test sample had fraudulent activity Jardim Sao Pedro: 4% of the test sample had fraudulent activity Simple average: 1.6% 1% adopted due to consensus of group Results from random inspections of active connections in three test zones

52 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 52 1) Active connections: 1% of active connections assumed to be fraudulent Category Nº of active connections 1% Recovery (m3/month) Estimated Volume (m3/month) Residencial 540,3075,40344.65241,247.14 Comercial 123,7771,23849.4061,145.80 Industrial 15,298153119.0518,212.59 Mixed 14,60314694.4413,790.97 Public 1,391000.00 TOTAL 695,3766,940334,397 Estimate of volume lost for water balance

53 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 53 Estimate of volume lost for water balance Nº of inactive conections 2% Recovered volume (m3/month) Volume (m3/month) 56,2641,12519.2121,617 2) Inactive connections: 2% were re-activated 3) Theft from hydrants: 9.3% (study from September 2002) Nº of hydrants9.3% Volume per hydrant (m3/month) Volume (m3/month) 3,733347300104,151

54 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 54 Estimate of volume lost for water balance Total Estimated (1+2+3): 460,164 m3/month Average monthly supply volume 36,711,358 m3/month Fraud  1,3 % 1.5% of supply volume adopted

55 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 55 Billed Unmetered Consumption

56 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 56 Trinidad & Tobago Customers 2007 UnmeteredMetered Stand Pipes54,953 Yard Taps38,232 Domestic248,4383,811 Charitable Organisations1,63848 Industrial118542 Commercial2,8324,480 Cottage2,083452 Agricultural636461 Total348,8059,794

57 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 57 Problem For Water Loss Assessment! System Input Volume? Authorized Consumption? Revenue Water? Non Revenue Water?? Billed Authorized Consumption? Unbilled Authorized Consumption Apparent Losses Real Losses??? Water Losses? Billed Metered ConsumptionUnbilled Unmetered Consumption Unauthorized Consumption Customer Meter Inaccuracies and Data Handling Errors Leakage on Transmission and Distribution Mains Billed Unmetered Consumption?Unbilled Metered Consumption Leakage on Service Connections up to point of Customer Meter Leakage and Overflows at Storage Tanks

58 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 58 Trinidad Unvalidated Water Loss Assessment Performance IndicatorBest Estimate 95% Confidence Limit +/-% Lowest Estimate Highest Estimate NRW as % of System Input32.839.8%19.845.1 NRW % of System Input by Value 59.639.8%35.983.3 Apparent Losses % of Water Supplied 7.233.9%4.89.7 Real Losses litres/connection/day 1,28754.0%5921,981 Real Losses ILI 23.954.1%11.036.8

59 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 59 Methods for Estimating Unmeasured Consumption Issue addressed in UK UKWIR Best Practice Report (1999), Best Practice for Unmeasured Per Capita Consumption Monitors Tynemarch Engineering (2007), Leakage Methodology Review: Variation in Per Capita Consumption Estimates, Final Report for Quinquepartite Group Small area consumption monitors Individual consumption monitors

60 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 60 Comparison of Methods Individual Monitor Install meters on sample of customers But difficult to obtain representative sample Not charged on meter But presence of meter could still influence consumption Not influenced by distribution leakage Small Area Monitor Install meters feeding small representative areas of customers But difficult to obtain representative sample Growth of metering Must remove all leakage in area Background losses Cannot influence consumption

61 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 61 T&T Consumption Monitor Small area monitors not feasible Leak detection very difficult in intermittent supply areas Individual consumption monitors selected But intermittent supplies – air discharged through service lines Register air volumes Meter damage Smartmeter selected Data logged Tank overflows Consumption patterns Random sample of 270 unmeasured residential customers planned 95% confidence limit +/-5%

62 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 62 T&T Consumption Monitor Results Meters installed and read by WASA 200 meters installed within the project time frame Useable readings from 166 meters 95% confidence limit of +/-7.6% Data from 18/9/2008 and 12/2/2009 Spanning part rainy season / part dry season Believed to be representative But monitoring continuing Average consumption 1.29 m3/connection/day Occupancy rate 3.67 Per capita 351 litres/capita/day (lpcd)

63 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 63 Estimating Other Unmeasured Consumption Classes Previous JICA study Stand pipes 36 lpcd Yard taps 305 lpcd Others based on corresponding metered consumption class Assumed that the meters installed included highest consumption users Based on lower 90%

64 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 64 Trinidad Draft Water Loss Assessment Performance IndicatorBest Estimate 95% Confidence Limit +/-% Lowest Estimate Highest Estimate NRW as % of System Input35.210.4%31.538.9 NRW % of System Input by Value 63.510.4%56.970.1 Apparent Losses % of Water Supplied 5.341.5%3.17.5 Real Losses litres/connection/day 1,54312.4%1,3521,734 Real Losses ILI 28.615.9%24.133.2

65 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 65 Impact of Scheduling in Trinidad Volume of real losses (102,005.8 Ml/year) is constrained by scheduling (intermittent supply) If there were sufficient water resource available to supply all customers 24/7 Government plans 5 new RO plants (360 Mld) Volume of real losses would increase to 191,380.5 Ml/year, an increase of 89,374.9 Ml/year (245 Mld) …But of course there is enough water to meet demand – by efficiently managing water losses!

66 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 66 Scope of Water Efficiency Survey Telephone survey of 425 randomly selected unmetered residential customers In-house audit of 270 properties selected from the telephone survey (188 actually undertaken) Analysis of scope for efficiency improvements

67 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 67 Water Efficiency Survey Findings Toilets 10.4 litres per flush – scope to reduce Showerheads 5.2 litres per minute – low Taps – 4.8-6.4 litres per minute – low Laundry – 82% top loading machines – scope to reduce Dishwashing – only 2% had machines. 54% households wash in running water – average 15 litres/person/day – scope to reduce to 7 Leakage – 7 litres/person/day – mainly from taps and showers – scope to reduce Outdoor use – minimal 2 litres/person/day – low Storage Tanks – 83% households have tanks. 7% overflowing. 27% have no level control – scope to reduce!

68 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 68 Water Efficiency Survey Results

69 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 69 Unbilled Unmetered Consumption

70 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 70 Typical Types of Unbilled Unmetered Consumption Fire fighting Fire training Unmetered fire lines Fire sprinkler tests Fire flow tests Mains chlorination / flushing / testing Reservoir emptying / cleaning Water quality flushing Community gardens / parks Etc.

71 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 71 Estimating Unbilled Unmetered Volumes Build up from estimates of individual use events Temporary metering to obtain estimates Good news – small component of water balance ~1.25% of system input +/- 100% So not worth spending too much time on

72 IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 72 To Conclude: All that I have covered may be required To determine a water balance with a low margin of error However the most important thing it to get started! A totally unvalidated balance With honest assessments of margins of error Will help win support for validation work Will highlight where work is most important to reduce the margins of error Tackle the most important issues first Margins of error will quickly reduce

73 Thank You - Any Questions? paul@veritec-us.com www.veritec-us.com


Download ppt "Field Validation Techniques Paul Fanner. IDB NRW Workshop, Georgetown, Guyana 19th-21st May 2009 2 Overview What is validation? Why is it important? Validation."

Similar presentations


Ads by Google