BIG DATA – ANALYTICS Theft Identification and AMI By Brent Hughes, Consultant, MBH Consulting Transformer metering and recently developed theft analytics provide a dramatic new capability to use AMI meter data to detect, locate and characterize virtually all energy theft without field investigations. It provides a continuous, system-wide view of theft. Adding the revenue recovered from systematically eliminated theft can turn a marginal AMI business case into a very compelling one. Electricity theft is a world-wide problem. Every utility has it to some extent. The energy theft rate can range from a benign 2% or less all the way up to a malignant 25% or more. Figure 1 shows a compilation of theft rates by country from various published sources. Very high theft rates, say 20% and above, may be indicative of social problems that need to be addressed before energy theft can be. At the recent Metering Latin American 2013 conference in Sao Paulo Brazil, for example, the emphasis was on how to deal with gangs, militia, and armed homeowners when it came to disconnecting customers for theft. Locating the theft was reportedly not a problem. At the other extreme, utilities in countries with very small theft rates, say 1 or 2%, do not see energy theft to be a big enough problem to warrant their attention. The losses are tolerable and their utility commissions typically allow them to recover energy theft through their rates. They therefore have no incentive to go after energy theft. The lowest cost approach to theft, even for the honest rate payers here, may very well be to do nothing. Most utilities are somewhere in the middle though, with energy theft rates that are worth pursuing and without widespread social issues that would impede their revenue protection efforts. They have the will to stop the energy theft but don’t have the tools. Most utilities are also moving to Smart Metering. New analytics are now emerging that work with smart meter data and have the potential to revolutionize revenue protection. We can now move from a tip- centric approach to revenue protection to one that covers a utility’s entire distribution system and locates virtually all energy theft right down to the individual customer. This new theft identification approach has a hardware and software component to it. The hardware layer consists of adding a revenue meter to every pole-top or pad-mount distribution transformer in the system or, in sparse or low customer density areas, adding revenue metering to distribution lines themselves. These new meters allow us to do an energy inventory with the customers served from the distribution transformers or between adjacent meters on the feeder. This is a system-wide net that detects theft in any and all inventory “zones”. It identifies which inventory zones have no theft and require no further action, and which have theft and need further investigation. Initially it was thought that field investigators could take it from here and find the customers that were stealing. However, experience has shown that this frequently does not work. Investigators “follow the electrons”, so to speak. Thieves have found that if they turn off their theft load, investigators have nothing left to track and go home empty-handed. We need analytics, and this is where an understanding of theft mechanisms comes in. Figure 1 - Energy Theft Rate by Country 64 METERING INTERNATIONAL ISSUE - 4 | 2013