Home air conditioning units can balance the electrical grid without impacting comfort

While adding wind and solar energy to the electricity supply can help reduce carbon emissions, their intermittency poses a challenge to traditional methods for maintaining electricity supply and demand.

Balancing the electrical grid typically involves adjusting the power output of traditional large power plants. Alternatively, balancing the grid by coordinating the power production and consumption of a variety of smaller distributed energy resources can improve grid reliability and support the integration of wind and solar.

Leveraging air conditioning units in Austin, Texas homes as distributed energy resources actively balances the grid without sacrificing comfort, according to a study led by the University of Michigan. The research is published in the journal IEEE Transactions on Smart Grid.

Home indoor temperatures take a while to change after AC units are switched on or off. This makes AC units a great fit for frequency regulation—a type of reliability service in which power plants adjust production to drive the grid frequency back to the 60 Hz standard.

An AC unit switches on or off to maintain indoor temperature within a small range around the setpoint. By slightly changing the timing of those switches, and coordinating those changes across a large number of AC units, the ACs collectively can adjust their power consumption to help drive the grid frequency back to 60 Hz.

“Residential air conditioners can provide high-value grid balancing services without impacting the homeowner. This resource is enormous and already available, we just need to harness it,” said Johanna Mathieu, an associate professor of electrical engineering and computer science at U-M and corresponding author of the study.

Up to this point, most grid balancing strategies with “flexible loads” like this have only been tested in simulation. To put the ideas to the test in the real-world, the U-M team partnered with Pecan Street Inc. to recruit Austin, Texas homeowners to participate in the study.

“There are several challenges that arise when such approaches are tested in the real world. For example, data availability may be significantly limited compared to what is typically assumed in purely simulation studies. Overcoming such limitations is a really exciting endeavor that requires creative solutions,” said Ioannis Marios Granitsas, a doctoral graduate of electrical and computer engineering at U-M and lead author of the study.

Throughout the study, around 100 homes participated in four frequency regulation events lasting one hour each. The researchers supplemented the relatively small sample size with simulated AC units to boost the size of the collection.

Smart home thermostats do not allow on/off control by a third party, creating a hurdle for frequency regulation. To bypass this, the researchers installed custom hardware control boards and sensors in each home. This allowed a master control algorithm to distribute commands through the cloud to each home’s air conditioner, responding to grid changes within 10 seconds.

The system successfully orchestrated AC units to provide frequency regulation (i.e., track a target power signal that varies every 2 seconds), earning a performance score well above the industry standard thresholds.

Importantly, frequency regulation did not impact comfort. Homeowners were given the ability to override controls, but the mechanism was not used at all in most cases and used twice in one case. According to temperature sensors, home temperatures only deviated a maximum of 1.6 F from their set point.

“Over the years, we’ve learned that participants in HVAC control studies want some way to opt out of programs. It’s key for recruitment. These are lived-in homes, and we have to have the trust of our participants. We’ve also learned that if you do the control actions correctly, very few if any participants need to opt out of the day’s event,” said Scott Hinson, Chief Technology Officer of Pecan Street, Inc. and contributing author of the study.

To ensure scalability, the control approach uses a model of the dynamics of the collection of ACs. The model size is not a function of the number of AC units managed, offering the potential to scale up. However, smart thermostat restrictions create a roadblock as installing hardware for every connected home is not practical on a large scale. Smart thermostat companies opening up their application programming interfaces (APIs) to accommodate third-party control for frequency regulation would speed up the process.

Researchers from Pecan Street Inc. and Los Alamos National Laboratory also contributed to the study, and researchers from the University of California, Berkeley were also part of the overall project.