The promise (and application) of data analytics
One early use case in smart grid era: revenue protection
As utilities begin a new era of using data analytics for myriad reasons, the industry's thirst for peer-to-peer knowledge may be slaked by some insights from a new study of the subject.
"Every employee and customer will be touched by analytics," said H. Christine Richards, senior analyst with the Utility Analytics Institute, underscoring the importance of bringing fresh insights to bear on the topic.
Richards conducted the study, "Utility Analytics Institute: Annual Market Outlook & Forecast," featured in yesterday's Energy Central webcast. (Replay the webcast by clicking the link; or download the accompanying slide deck.)
You may be surprised to learn that old-fashioned factors drive the need to turn data streams into actionable intelligence. The perennial pursuit of performance improvements, cost savings and improved customer service ranked higher as drivers for analytics than smart grid-related activity, according to the study. Smart grid as a driver ranked low, according to respondents. (Survey respondents, whose views created the report's raw data, were utility managers (46 percent), technicians/staff (37 percent) and executives (17 percent).)
That may change, in my view, as utilities just completing rollouts of advanced metering infrastructure and implementing meter data management programs get up to speed with deriving value from that data source. In fact, as Richards defined overlapping phases of the "smart grid path," 2012 will mark the end of the "infrastructure phase" while 2010 marked the beginning of the "value phase."
In the latter phase, extending out a decade from today, much of the value derived from grid optimization and asset management work and the integration of renewables and electric vehicles will be driven by analytics.
In any case, Richards found that utilities' approaches to analytics have begun to change. The industry's traditional approach to analytics often reflected an ad hoc embrace by an individual business unit looking to improve its performance. Today such efforts are more likely to have a significant amount of support from a variety of executives, including CEOs, CFOs and CIOs, as well as vice presidents of transmission and distribution operations and customer service.
The study also forecast that spending on analytics will be a "robust" 29 percent per annum, taking that industry segment from a half-billion in spending this year to nearly $2 billion by 2016, with investor-owned utilities leading and public utilities (municipalities and cooperatives) not far behind. So utilities "get" the value of data analytics and the spending forecast reflects the interest in deriving business intelligence from it.
This will take time, apparently. The study also found that the biggest hurdle to moving forward on analytics was the lack of foundational systems, not necessarily budget constraints. However, once analytics initiatives are in the pipeline, budget constraints are cited as an impediment to implementation. That is followed by concerns about having the staff and skills to implement such an effort.
With data analytics potentially touching every aspect of utility operations, a specific use case delivered on yesterday's webcast offered a concrete example that addressed the top line, aka revenue.
Mark Schweiger, a senior business analyst at Florida Power and Light, described how he helped implement a data analytics program around revenue protection, commonly referred to as electricity theft.
Knowing that theft detection often fell to visual inspection by meter readers or field investigators, FPL knew that an advanced metering infrastructure system would provide data that could flag suspicious accounts for closer examination. So in 2009, as it began its AMI rollout (due for completion in 2013), FPL also implemented a meter data analytics program in concert with vendor DataRaker.
FPL feeds its vendor AMI and CIS (customer information system) data and the vendor crunches that data to create "meaningful flags" for certain accounts, while winnowing out false positives for meter-related monkeyshines. That effort has taken two years to develop valuable output, yet Schweiger said that FPL is in "the infancy of its test strategy." Improvements to the program should help it detect when a premise is using an unauthorized meter, is bypassing an approved meter, is employing a powerful magnet to suppress usage (and thus billing) data and has reconnected service without authorization.
Among FPL's lessons learned:
- Event notifications (tamper flags) from meters alone are not good indicators of theft. Usage analysis must be brought to bear as well.
- In meter bypass detection, for instance, its first version of testing was less than 10 percent effective; by versions 4 and 5 effectiveness was 70 percent. Thus, improving testing methods is a never-ending iterative process.
- The MDA program is also useful to the utility's meter operations group to gauge the health of the meters and the two-way communications network that connects them to the utility.
I can't think of a more fundamental application for data analytics than revenue protection, which Schweiger said was initiated by that business unit and quickly found executive buy-in. It will be fascinating to see what use cases are presented in this context in a year or two from now or—less likely to surface—what use cases do not see daylight for lack of support.
Intelligent Utility Daily