Peak power: Predicting the future through analytics
December 2020 – Predicting peak power demand on the electrical grid is not easy to do! In fact, it takes a great team of diverse experts to get it right – something Jim Lytle, Lead of the Electrical Engineering Team at Cenovus Energy, knows well.
“Many large operators spend tens of thousands of dollars annually to forecast peak power as a way to cut electricity costs, and at Cenovus we are challenging the status quo,” explains Lytle. “We have been actively lowering electricity costs at our oil sands facilities and predicting peak power for a fraction of the cost.”
Through years of analyzing and refining our cost breakdown structure we have pinpointed one electricity cost driver called the coincident peak. The coincident peak concept is simple; every month there is a 15-minute chunk of time where Albertans use a maximum amount of electricity. During this 15-minute window, every unit (or megawatt) of electricity we purchase from the grid costs about 210 times more than during the average usage period in the month.
“Drawing from our on-site experts, we aim to avoid importing electricity during the coincident peak by maximizing electrical production and ensuring our cogeneration units are operational and generating power for the system,” says Tim Vandermey, Cenovus Electrical Engineer. “Our success is driven by a coordinated and collaborative effort with our Engineering, Maintenance, Data Analytics and Operations teams.”
Unfortunately, it is extremely difficult to predict when the coincident peak will occur using traditional analyses and the challenge is magnified by regulatory and maintenance constraints related to operating these complex units. This is where the new visualizations and our Artificial Intelligence Robot, named JABI, come in handy.
“JABI is a great example of how we can bring together data and analytics techniques like artificial intelligence and specialized knowledge with interdisciplinary teamwork for excellent results,” explains Brett Wiens, Senior Advisor, Data Analytics. “These solutions have already enabled cost savings for the company while avoiding last minute scheduling changes, and it will just continue to get better as we use it.”
Moving forward, Cenovus will continue to refine and expand these capabilities to predict power price. These revisions will better enable the company to anticipate when high electricity costs will occur, including two weeks out, one month out, and three months out.
With teamwork central to this project, we want to recognize and thank the dedicated inhouse experts who made this possible; Brent Huynh, JV Jamalapur, George Kohn, Jim Lytle, Tim Vandermey and Brett Wiens.