Implementing MDMS: lessons learned
Oncor, JEA and PPL recall pitfalls, best practices
On one hand, implementing a meter data management system (MDMS) is just like any other internal initiative that requires an adequate budget, staffing, planning, stakeholder involvement and executive buy-in.
On the other hand, from the testimony of three quite different utilities in yesterday's webcast, "Managing the Data Explosion: The Reality Behind MDM Implementation," it would appear that MDMS implementation is not like any other project you've been involved in.
(Click on the title for a replay of the event or to download the slide deck.)
First off, an MDMS builds upon the deployment of advanced metering infrastructure (AMI) and it's best to envision both systems concurrently, even if implementation is sequential, according to Louise Gross, AMI system operator for PPL Electric Utilities.
Secondly, Gross continued, the level of coordination needed in an MDMS implementation is broader and deeper than usual.
Not only do you need to coordinate MDMS project management with vendors, business unit leaders and executive sponsorship, you may have to lock them all into the same room for six months to a year, according to Jon Pettit, AMI manager for Oncor Electric Delivery, a Texas transmission and distribution provider, as well as Brian Novak, manager of advanced metering systems at municipal utility JEA in Jacksonville, Florida. JEA called theirs "the jungle" for its ubiquitous potted plants.
Both Oncor and JEA designated entire floors to enable all participants to sit shoulder to shoulder until the project went live.
On the depth of coordination with stakeholders, Gross said that because the advent of interval data changes business processes across the enterprise, every business unit has to be engaged on how that data will change their missions. Each business unit must envision how they can use interval data to answer essential business questions and anticipate how an MDMS implementation will change their processes.
In that sense, business process change must be anticipated, envisioned, mapped and documented, so that as system iterations occur, all ripples of change are understood and documented.
While AMI/MDMS offers the promise of automation, at least on the surface, and that means a vastly reduced need for meter readers, there's a concomitant need for business analysts and data analytics experts to make use of the resulting Big Data, all panelists affirmed.
As for how smoothly the process is likely to go, Novak offered a graphic with data sources entering an MDMS process via straight arrows, then "better data" leaving the MDMS with straight arrows. Don't believe it, Novak said. That's the vision. The reality is "never trust a straight arrow" in a diagram depicting a vision, he said.
Because so many systems feed into an MDMS and multiple outputs address operational and customer-facing needs, integration is a major challenge, all panelists agreed. Synchronization stands out among integration challenges, according to Gross.
PPL's MDMS implementation took three years, from fall 2006 to fall 2009. What was the most time-consuming aspect?
"Getting the synchronization right," Gross said. "Ensuring that what we're bringing in from our head-end systems is going into the MDMS properly, and testing that. Each project has its own set of head-end systems. At times we got mired down in 'getting the data right.'"
Oncor's MDM implementation was go-live only six months after vendor selection. How did Oncor make it go that fast?
"As we filed our rate case in May 2008, approved in August 2008, we made a commitment for full implementation [of everything in rate case] within 18 months," Pettit said. "As a result, we put the MDMS vendor on a six-month timetable. Another aspect was 'cohabitation.' We had a full floor in our building designated for AMI, with our MDMS vendor sitting next to our system integrators, sitting next to our head-end vendor. All of our vendors essentially sat next to each other, in rows of tables, chairs and telephones."
Novak also highlighted integration work as a challenge. Did that also highlight decisions around in-house staff versus third-party integrators?
"It's never 'one person knows everything,'" Novak said. "And the more people you have involved, the more you need to 'cohabitate' for the entire duration of the project. That just makes things flow a lot easier, so when questions come up, you just shout across the room, 'What am I seeing here?'"
Did in-house staff acquire SME as a result of MDMS implementation?
"They have to, and that's another key point," Novak said. "You're left holding responsibility for the system when the implementation project is over. Unless you plan on having the vendor sitting at your site forever and ever, which gets expensive. You have to have people who can live and breathe it. Then, in the end, they can support the system going forward."
And going forward?
MDM is "still relatively new," Pettit concluded. "Business intelligence and analytics are probably the big thing. You've got Big Data. What do you do with it?"
Hope that gives you a flavor of the challenges ahead, folks. Click on that link for a replay of this worthwhile discussion.
Intelligent Utility Daily