Hydro One: data analytics requires lead time, legwork
Multi-year effort paves way for predictive analytics
A focused program, four years in the making, is enabling Hydro One to move its analytics efforts from basic reporting to informing better business decisions by leveraging key metrics, its Director of Business Architecture Brad Bowness told me recently.
Hydro One's journey provides some lessons learned for a power industry eager to reap the benefits of predictive analytics and improved business intelligence.
Despite the allure of a vision where such gains are available simply for a price, Bowness in an interview described a multi-year effort that spanned the organization and required intense focus and patience before his utility's effort approached fruition. Even now, Hydro One is just approaching a point where it can use analytics to inform predictive asset management, business planning and other forward-looking outcomes.
Hydro One, owned by the province of Ontario, runs one of the largest delivery systems in North America with 1.3 million customers across a service territory both urban and rural.
"At Hydro One we've had a comprehensive program we refer to as our `Cornerstone Program,' Bowness told me. "We've been working since 2007 to implement a four-phase program to adopt industry standards and processes enabled by commercial, off the shelf technology.
"We've embarked on this program to run a more effective, efficient business with the ultimate goal of making better decisions," he said. "This is where business intelligence really comes to the forefront."
Hydro One began its four-phase program with enterprise asset management, which went live in 2008. The utility implemented enterprise resource planning components the following year, which also marked the acquisition of tools to enable fundamental reporting of traditional reports, metrics and key performance indicators.
Where asset management involved cost take-outs and effectiveness, the arc of Hydro One's efforts aimed toward predictive analytics that inform better decisions, Bowness said.
"The business intelligence (BI) space is really about analyzing the business to make better decisions," he told me. "If you look at what we embarked on in 2008-09, it really was about reporting. To quote models on BI, `reporting' is sort of `what happened?' Analytics is `why is that happening?' And predictive analytics is `what's going to happen?' `What's the best possible outcome based on a decision I make today?'
"We went into this with our eyes wide open, making sure that we have the ability down the road to enable analytics and to enable predictive analytics," Bowness continued. "But right out of the gate we felt we needed to get the underlying reporting capability stable and well rolled out across the business."
Several years of work to stabilize the reporting platform, rationalize traditional reports (reducing 1,500 down to 140) and get users accustomed to the tools and outputs has positioned Hydro One to move ahead.
"Stabilizing the platform" included the core goal of identifying a "single version of the truth," which could then be used by various parties for myriad purposes. That means ensuring quality data and properly integrated systems to enable the development of accurate, meaningful reports and metrics.
"Getting the underlying data quality accurate is paramount," Bowness emphasized. "Without it, when you roll-up data into KPIs or metrics, they're not going to hold true."
All this takes time and legwork, Bowness said.
"We had to build up staff's capabilities for using those reports and the new tool sets, understand the data and the new terminology, transition from our legacy systems to our new systems," Bowness told me. "So it took us quite a bit of time to build-out the technology and people elements in order to have people utilize the business intelligence tools up to their full capability in the reporting context.
"So when we got people comfortable leveraging the tool set and the data in a consistent and stable way, we could focus on expanding it and building on our analytical perspective," he added. "Summary reports, dashboards, KPIs - none of that stuff makes sense unless you get the underlying data and reports accurate. Right now, because we have stabilized all the reporting capabilities, we're at that cross-over point where we're moving into that analytics space."
One example: a comprehensive asset analytics initiative that takes into account all the different risk factors that the utility assesses its assets on, including economics, performance, utilization, condition, obsolescence, criticality, and health, safety and environment.
"We break our user community up into three groups: super users, reporting analysts and end users," Bowness told me. "The majority of those end-users are really looking for a structured report. They want to look at, say, work order information, accounts payable, asset information, depending on their role. So those are in the 140 reports I mentioned.
"We started with those 140 reports to serve our business functions in order to allow a user to go in and look at a structured output, with some basic analysis," Bowness said. "The initial view is a standard view, available in public folders. It would have a familiar look and feel, standard rows and columns, with certain filters.
"The reporting analysts and super users are really trying to use the data to figure out what they don't know, to do some heavier analysis. Ad hoc data mining is an area we want to look into in the near future."
Ad hoc data mining may be applied, for instance, to honing a sense of the health of Hydro One's assets.
"We want to look at different factors to determine whether with this vintage an asset, in this type of location, and with this type of condition and maintenance history, we're looking at a higher probability of failure," Bowness said. "If so, we could increase preventative maintenance."
Among lessons learned in Hydro One's analytics work so far, Bowness cited perennial factors such as executive sponsorship and a well-defined scope of work. But he emphasized an additional point.
"This can't be an exercise where the business defines a set of requirements and sends it over the fence for IT to develop," he told me. "One of our keys to success is having a technically savvy business team that lived and breathed that initial implementation. Then we carried forward part of that effort to serve in a centralized support group so we stay true to the original business model."
Other recent articles on data analytics we've written:
Intelligent Utility Daily