ACE 2016 June 22, 2016
Leveraging AMI Meter Data to Reduce Non-Revenue Water Ed Hackney
Director, Revenue Management Suez
Jon Varner
Practice Director EMA, Inc.
About SUEZ
The Challenge: • Reduce NRW in NJ – trending up for years…
29%
NRW 2012
• Reduction Tactics – For Real Loss:
• Leak Detection • DMA
– For Apparent Loss: • AMI Data
– Reduce Theft – Improve Data
• AMI and Apparent Loss focus began late 2011
3
Why go after NRW?
And The Challenges…
• NRW is a Business issue
– Drives revenue DOWN – Drives operating expenses UP
• NRW is a Sustainability issue – Wastes water resources – Wastes energy and chemicals
• NRW is a Reputation Management issue
– NRW is a universal measurement to compare performance / competence
• Challenges of NRW reduction strategies? – – – –
Needed data is often missing or incomplete Implementations can be costly ….difficult…risky… Strategies often address one component of NRW at a time Requires “change” for employees and customers…
4
AMI Focus Areas
5
Apparent Losses – Causes and Solutions Vary • Meter Inaccuracy • Unmeasured Low Flow • Fire Service Line Use • Meter Tampering • Unauthorized taps • Data handling/”Lost” Customers 6
SUEZ NJ Apparent Loss Investigation (2011-2012)
Goal: Estimate the Following: Unauthorized Consumption: • Meter Tampering • No Record of Service • Illegal Use of Fire Services • Improperly Connected Irrigation Systems
Data Handling Issues: • “Lost” Customers • Intermittent stopped/stuck meters • Meter Inaccuracy
Apparent Loss Investigation Results - 2012 Investigation Results: • Unauthorized Consumption – Meter Tampering: • Up to 1% Loss in most areas • ~ 2% Loss in areas where the water bill is used to calculate the sewer bill
– No Record of Service – 0.07% Loss – Improper Use of Fire Services – 1.0 % Loss – Improperly Connected Irrigation Systems – 0% Loss (Less than expected…)
• Data Handling Issues – “Lost” Customers in Billing System – up to 0.4% Loss – Meter Inaccuracies – 3 to 4 % Loss
• Total % of potential revenue NOT billed > 7 %
8
Tampering Type 1 - Periodic Register/Meter Removal
• Apartment Building with 1” meter • Uses ~6000 gallons per day • Back-billed $61,000
Tampering Type 2 – Periodic Reversing Meter
• Apartment Building with 3/4” meter • Uses ~2500 gallons per day • Back-billed $27,000
4/2/2013 10:16 4/9/2013 10:13 4/16/2013 10:50 4/23/2013 11:05 4/30/2013 10:42 5/8/2013 10:53 5/14/2013 10:54 5/20/2013 15:08
313700 316005 318405 316030 313790 311280 309260 310700
17 23 24 -24 -22 -25 -20 14
The Challenge: We Lacked Analysis Tools for AMI Data • AMI Vendor tools did not incorporate data from our Customer Billing System, including: • METER details • SERVICE CONNECTION details • CUSTOMER ACCOUNT details
• Other Barriers: – Limited historical data (three months) for most AMR/AMI vendors – Limited query, filter and sorting options
11
The Solution: Develop AMI Analytics Tools Focus
Year
1
Develop Requirements and Approach
2012
2
Capture and Process the AMI and Customer Data
2012
3
Build AMI Analytic User Tools
2013
4
Improve the Tools – Iteratively
2014
5
Incorporate Reads from 2nd AMI Vendor
2015
6
Implement DMA Analytic Tools
2016
Step
12
Meter Analytics Solution Overview Customer Billing Mobile Collector
Meters
Fixed Collectors
MDM
Vendor 1
Database
Meters
Fixed Collectors
MDM
Vendor 2
User Tools
Meter Analytics 13
Data Processing – Step 1
Link the Reading to a Service Connection and Customer….
• Analysts need to know:
– Meter details – Location details – Account details …to determine meter tampering 14
Step 1 Processing Example: Linking Readings to a Meter, Location and Account RF Device
Meter Location
Meter Reading: Transmitter ID: 1234 Timestamp: Reading:
1/1/2015 12:00 AM 051234
Register ID: 3223 Meter ID: 5555
* Make, model, size, UOM, etc.
Look up The Meter and Service Connection Associated with the RF Unit ( at the time of the reading…)
Service Connection ID: 6577 Premise ID: 78321
* [Integration point to GIS]
Customer Account
Account ID: 23453
* Status, Type (Res, Comm, Fire)
Look Up the Customer Account for That Service Connection
Customer ID: 78231 (John Smith) > Customer details...
15
Data Processing – Step 2 Calculate and Store ‘Tampering’ Metrics • Metrics:
– Consecutive Days of Zero Consumption – Positive vs. Negative Readings
16
Analysts Use These Metrics (Plus 32 Other Data Fields) to Detect Possible Tampering.
Investigation FLAGS Help Analysts Manage Account Review Actions
18
Who Makes the Best Analyst ? (…for
Meter Tampering)
• Staff members that understand:
– How the utility’s water meters work and how they are physically installed – How the utility tracks meter installations and replacements – The hydraulics of water distribution (e.g. pumping, thermal expansion, etc.)
– Human nature…
• Has the ability to think like a … thief 19
The Addition of Water Production Data Enables DMA Analysis Customer Billing
Water Production
Mobile Collector
Meters
Fixed Collectors
MDM
Vendor 1
Database
Meters
Fixed Collectors
MDM
Vendor 2
User Tools
Meter Analytics 20
Data Processing Steps for DMA Analysis 1. Calculate (and estimate if necessary) hourly consumption for each customer meter 2. Totalize consumption by DMA zone and subzone – Pilot DMA zone (PD 40) has ~30k customer meters
• Challenges:
– New meters for PD 40 (with hourly reads) have not been fully deployed – thus the Meter Analytics system is interpolating many results (data quality is not good yet) •
Of concern: We are not receiving true hourly reads from the newly installed AMI meters 21
DMA Tool Layout (with ‘Massaged’ Data…)
22
Allowing Users to Drill Down and Identify Data Problems Improves Analytics Results
• Example: Users drill down to Hourly Consumption by Service Point, sort by Consumption Gallons descending to identify abnormally high values 23
Real Results
• Data gathering analysis action = significant results… – Apparent Loss a significant component
• Combined with Real Loss remediation work, NRW down: – From: 29 % at the end of 1012 – To: 21.5% (April 2016)
• Reducing apparent losses yields an additional $3M+ annual revenue • AMI and Meter Analytics Project ROI … off the charts! • Sustainable results: problems fixed & the word is out
24
© 2016 EMA, Inc.
Questions To learn more, please contact : Ed Hackney at
[email protected] or Jon Varner at
[email protected]