A couple of months ago, I wrote about some of the experiments focused on using electric vehicles as a storage mechanism that could help balance power demand across the grid.
IBM has been applying a similar idea in an industrial setting: It just released the results of a project called FlexLast with Swiss supermarket company Migros that studied how the energy used to keep three food warehouses super cold (minus 18 degrees Fahrenheit) might be better metered during certain seasons and certain times of days to help reduce local power demand.
The idea is pretty simple: it takes a lot of power to get these 200,000-square-meter warehouses cold — approximately 500,000 kilowatt-hours per month (or roughly the amount of power needed to run 1,500 average U.S. homes). That amount varies slightly according to all sorts of factors like how many deliveries are going in and out in a given day, headed for Migros’ 990-plus stores. The number of palettes and their density can play a role as, of course, does the outside temperature.
The pilot in Switzerland (which ended late in 2013) was specifically focused on how the warehouses could help stabilize the integration of renewable solar and wind energy into the grid. Aside from IBM and Migros, it involved BKW FMB Energy (the local utility in Bern) and Swissgrid, which operates the national grid.
The idea was to synchronize when the coolers run with the renewable supply: programming them to run at maximum levels when more clean power was available, and at lower speeds or not at all when it isn’t available. Realistically, smart companies have been doing this manually for many years. What the test was meant to determine was whether or not analytics and business intelligence software can be used to automate this process.
“We asked ourselves, How can we do this smarter?’ ” said Roland Stadler, head of energy purchasing for Migros, when the pilot was started. “ We know when our trucks arrive and depart, and we know the schedule of our employees. Therefore, if we integrate this data with our energy needs based on the availability of renewable energy, we can maintain the temperature of the warehouses and simultaneously contribute to the future stability of the grid.”
Switzerland has a national goal of producing 10 percent of its current electricity needs with renewable energy by 2030; the amount we’re talking about is around 5,400 gigawatt-hours.
All things considered, the experiment went pretty much as expected, said Doug Dykeman, manager of systems management with IBM’s Zurich Research Center. “They needed to cool the warehouses five or six hours per day, but it didn’t really matter when,” he said. “You can think of this as load shifting.”
So what did IBM, Migros and their smart grid partners learn?
For one thing, there are certain times of day (mainly between 6 PM. to about 11 PM) when it wasn’t possible to achieve the desired power levels necessary for shifting (this result was expected). In the early morning hours of between 1 AM and 5 AM however, it’s easier to control the cooling without using much electricity. The measured difference in the energy used during those timeframes was significant: about 1.2 megawatt-hours in the late evening, compared with 200 kilowatt-hours during the early morning, according to results that IBM shared with me.
“This is a significant difference and can be achieved easily and repeatedly,” IBM notes. “Therefore, the value in compensating for energy renewable sources or utilizing lower-cost generation or buying energy at lower costs … is significant.”
It adds: “By pooling loads, a significant total flexibility and finer granularity can be achieved.”
One of the big challenges involved in managing the power loads is the responsiveness of the equipment: it takes some time for compressors to respond to automated requests for more cooling. (It’s not like flipping a light switch.)
IBM is planning to start a similar test in the near future, although it wouldn’t say where. Realistically, this isn’t something that would be done with individual industrial sites rather with grid operators that come up with standardized programs. “Somebody has to pool these flexible resources,” Dykeman said. “We don’t want to be going into every retailers or steel company.
Nor would it make sense to think of using big industrial loads to offset small variations in grid demand – it just can’t happen quickly enough, given the nature of the equipment involved. “That would not make sense nor would it be possible because of the startup times and shutdown times,” he said.