Using analytics to do more than crunch data
IT'S THE OLD CHICKEN-AND-EGG CONUNDRUM, UPDATED. New grid sensors, from smart meters to phasor measurement units and more, are providing a slew of new data for utilities to use to increase the reliability of their product - electricity - while also enabling them to shape electricity consumption and better optimize the grid's performance.
The science of data analysis and analytics has been successfully applied to the banking, oil, insurance, telecommunications, travel and retail industries, to name but a few. Now the electric utility industry is turning its eye to data analytics, as well, in order to best mine the new information now coming in in droves.
Building an information factory
For OGE Energy Corp., it's a brave new world the company is exploring with a vengeance.
"The initial questions we asked were: where are we going to store all this data, how will we organize it and how can we utilize the information to improve our business?" said Craig Johnston, vice president, corporate strategy and marketing, for OGE Energy Corp.
OGE is the parent company of Oklahoma Gas and Electric Company (OG&E), a regulated electric utility, and Enogex LLC, a midstream natural gas pipeline business. OG&E serves more than 780,000 retail customers in Oklahoma and western Arkansas, and a number of wholesale customers throughout the region.
OG&E has approximately 6,800 MW of capacity. Its electric transmission and distribution systems cover an area of 30,000 square miles.
Johnston noted that other utilities are going through a similar exercise with regard to analyzing their new data. "If you can capture and analyze that data, and use it in a predictive fashion, what can we do with it to improve our operations and enhance our customer experience?" he asked, summing up the crux of OG&E's new "Information Factory." Working with OGE chief information officer Reid Nuttall's enterprise architecture team, the Information Factory is adopting those tools other industries have already found invaluable. "What information do you want, and how would you use that data?" defines the initial focus.
Star schema and beyond
This goes far beyond the star schema approach to data, which is the simplest form of a dimensional model in data warehousing and business intelligence. In it, data is organized into facts and dimensions, with a fact being an event that is counted or measured, and a dimension containing reference information about the fact. As a simple utility example, a fact could be a usage of energy within a certain time frame, and the dimension, then, might include the date and time the energy was used, and the customer who used it. A star schema works well if you know what you want to measure.
But what if you aren't sure? In his book Data Mining: Concepts, Models, Methods, and Algorithms, Mehmed Kantardzic writes: "Data mining is an iterative process within which progress is defined by discovery, either through automatic or manual methods. Data mining is most useful in an exploratory analysis scenario in which there are no predetermined notions about what will constitute an `interesting' outcome. Data mining is the search for new, valuable and nontrivial information in large volumes of data." Or, to put it more simply, as many within the industry have explained it: "This is: `we don't know what we don't know.'"
Held back by regulation and lack of detail
According to authors Rasheed Joshi and Diego Klabjan, who wrote a white paper about the subject in late 2009, "Such approaches have been dormant in utilities mostly due to the regulated environment and lack of fundamental detailed data streams." They noted that smart sensors, including smart meters, will provide what they call "power streams" that can be mined for customer relationship management, targeted energy efficiency programs, local forecasting and dynamic pricing.
"Techniques such as data mining, pattern recognition, consumer choice behavioral modeling, and optimization form fundamental blocks in market segmentation, targeted marketing, price optimization, supply chain management, management reporting, and various other applications," Joshi and Klabjan said.
The utility values available from analytics-based applications for smart grid include superior customer service, operational efficiency, optimized delivery of power, smart energy procurement, demand response and dynamic pricing, as well as being able to predict equipment failure and outages, using pattern recognition to detect revenue leakage and theft, and optimization techniques for Volt/VAR control.
An analytical journey
On the ground, what this means for OGE is an utterly new approach to data.
"I look at the journey," Johnston said. "Today, we're pretty happy if we get standard reports and occasional customer reports." But, given that OGE has espoused a goal of not adding new fossil-fuel generation at least until 2020, using detailed data analytics will definitely provide an active assist in reaching that goal.
Johnston said that, for the company, it means looking at how to start to apply statistics on top of data, and use that to improve forecasting. And then, he said, "we can take that and do some `what ifs,' or predictive analysis.
"Utilities are heading into a period of increasing costs [associated with upgrading aging infrastructure and adding new technology]. We need to understand what those drivers of elasticity are," he said. Add to that the question of "how well do we know our system?" This will be imperative to the utility when deciding how to support electric vehicles and other changes on the distribution system.
"Once you start using the smart system, you can look over the entire system and better manage load and demand and generation," he said.
Detailed data analysis will provide the structure for looking at trends, as well as at asset management based upon the history of the asset.
Building short-term wins
Johnston says the first step for OGE - getting the Information Factory up and running - is complete. Along with standing up the technologies, it included information that allows a geospatial look at the utility's customers, revenue and load. The next step? "We have to build some short-term wins: what can we do on a measured scope and scale within the next six months?" Some data analytics business cases within the company will have a clear win, and others, Johnston said, will be more marginal, but will serve to build upon the business case necessary for other projects.
The final step in the project, what Johnston calls the "key missing element," will require longer-term thinking and focus throughout the industry, not just at OGE and its utility companies. "We have to have an analytical skill set. It's not the same as reporting," he said.
"Analytics is a field of very specific talent. You have to be able to use the tools and be able to look at specific scenarios.
"We as business leaders are not really familiar with storing frequency data." Then there's the issue of migrating data from legacy systems to one central source, as well as the issues of data stewardship and data governance.
It's a tall drink of water for any electric utility to swallow. "As we continue to develop our capabilities, we will be recruiting a different talent set - data analytics. It's an exciting place to work, and it resonates with a lot of the college recruits," Johnston said.
No discussions yet. Start a discussion below.