[vc_row][vc_column width=”2/3″][vc_column_text]It’s one of MISO’s most technology-intensive and mathematically complex challenges: figuring out how to select and schedule the various electricity-generating resources in our footprint so we can serve our customers’ energy needs in the lowest-cost manner possible.
And while we’ve already made great strides on this front, we’re now aiming to revolutionize the industry standard for this crucial task with the help of a sizeable federal research grant and some of the fastest computers.
On November 23, the U.S. Department of Energy (DOE) announced it would award MISO and three partner organizations $3.1 million for an R&D project that’s designed to dramatically reduce the time it takes to clear, or “solve,” MISO’s Day-Ahead (DA) and Real-Time (RT) markets. The project will utilize high-performance computers at DOE’s Pacific Northwest National Laboratory (PNNL) in Richland, Washington, one of MISO’s three partners in the initiative.
Yonghong Chen, a principal advisor in MISO’s System Operations & Market Services division, said the project will help MISO to implement important enhancements to the DA and RT markets that will position us for future market growth, such as Virtual Spread Product and Combined Cycle Modeling in our market roadmap. The R&D initiative will also help MISO to adapt its markets to pressing industry challenges such as EPA’s Clean Power Plan, the growth of distributed generation and storage, and the not-always-synchronized practices of the electric and natural gas sectors, Yonghong said.
“If we don’t address those issues proactively, we’re going to have a problem later for sure,” Yonghong said. “We can’t just wait; we need to get software and our market system ready for these changes.”
Currently, MISO generally solves the DA market in four hours or less, except in cases where system conditions are stressed and we must number-crunch large number of binding transmission constraints coupled with an unusually large number of financial activities. We’ve committed to reduce our DA solve time to three hours by the end of 2016 as part of our approach to complying with a FERC order that’s designed to better harmonize the operations of the electric and gas industries.
But we may not be able to meet that three-hour commitment if we want to implement new market enhancements such as the Virtual Spread Product and Combined Cycle Modeling, because we’d need to use more complex and larger size optimization models that would require more time for number-crunching. And things like the Clean Power Plan could impose even more demands on our models, increasing the chances that we wouldn’t be able to consistently clear the DA market in three hours.
That’s why we’re hoping that this R&D initiative will help us to make paradigm-shifting improvements to our market-clearing process, as opposed to just improving it incrementally. Our target is to be able to clear our DA market in as little as 20 to 60 minutes—a giant leap forward compared to our current 240-minute benchmark. When applied to a 100-GW power grid, that kind of revolutionary improvement could reduce costs by $10 million to $100 million per year, MISO and its partner organizations have estimated.
“It’s very ambitious,” Yonghong said.
The bold scale of the project reflects the philosophy behind the ARPA-E program, the branch of DOE that’s providing the $3.1 million in federal funding. An acronym for Advanced Research Projects Agency-Energy, ARPA-E bankrolls only energy projects have the potential to “radically improve”—the organization’s own words—on the status quo. APRA-E is modeled after the Defense Advanced Research Projects Agency (DARPA), which develops James-Bond-like weapons systems and other technologies for the U.S. military.
Essentially, the process of committing units in MISO’s DA market entails solving a mathematical optimization problem, which will become an increasingly complex and time-consuming endeavor as MISO implements new market enhancements and grapples with challenges such as the Clean Power Plan. Yonghong said it will be a huge advantage for MISO to work on this math problem using PNNL’s high performance computers, which utilize a technique called parallelization and are much faster than MISO’s in-house systems.
“That’s the benefit of working with the lab,” she said. “The challenge for unit commitment is there’s no off-the-shelf algorithm that can take advantage of parallel computing. The software needs to be able to take advantage of that.” A group of people from MISO Market Service and IT Strategy will work closely with the research partners on developing the state of the art new high performance computing based algorithms.
While the details are still being worked out, Yonghong said MISO will probably use PNNL’s high performance computers via some kind of remote-access channel instead of working at the lab itself.
In addition to PNNL, MISO’s other partners in the initiative are GE Grid Solutions (formerly known as Alstom Grid) and Gurobi Optimization. MISO and the other organizations are currently negotiating the details of the R&D contract with ARPA-E.
Jeff Bladen, the executive director of MISO’s Market Design team, said the initiative illustrates how MISO is leading the industry in terms of technological innovation.
“We have a responsibility to our customers and stakeholders to be a forward-looking company,” Jeff said. “MISO works proactively to develop innovative solutions to emerging challenges.”[/vc_column_text][/vc_column][vc_column width=”1/3″][vc_single_image image=”4681″ img_size=”medium” add_caption=”yes”][/vc_column][/vc_row]