Carnegie Mellon Amps Up Battery Research with Commute Data

Kourosh drove a commute of 19.81 miles in West Menlo Park, CA in his BMW; Erik in Caledonia, WI drove 6.5 miles in a Volkswagen GTI; on October 26, 2009 Josh and Chris drove a total of 20.31 miles in a 2001 Toyota RAV4 EV in Pittsburgh. All of these volunteers have submitted their commute data as part of a research project at Carnegie Mellon University that's investigating how to convert gas-powered cars to battery-powered vehicles designed to meet an individual driver's specific commuting needs.

Headed by Illah Nourbakhsh, associate professor in the Robotics Institute, the team of researchers is examining how an electric vehicle's efficiency can be boosted and its battery life extended by using artificial intelligence to manage power. ChargeCar has already converted a 2001 Scion xB into an electric commuter vehicle that researchers will use to refine their conversion and power management techniques.

"Most electric cars today are being designed with top-down engineering to match the performance of gas-powered cars," Nourbakhsh said. "Our goal is to revolutionize urban commuting by taking a different approach--by first analyzing the needs, conditions, and habits of the daily commutes of actual people and then using this 'commute ecology' to develop electric vehicles suited to each unique commute." The project calculates that a typical Pittsburgh commuter, for example, might save 80 percent of energy costs by switching from a gas car to an electric car.

ChargeCar isn't developing new vehicles but is focused on developing a knowledge base that can be used to convert gas-powered vehicles using existing technology. The team is working with Pittsburgh mechanics to develop community-level expertise in vehicle conversion, as well as a set of conversion "recipes."

An important aspect of the project is a vehicle architecture called smart power management, which uses artificial intelligence to manage the flow of power between conventional electric car batteries and a device called a supercapacitor. Supercapacitors, as the university explained, are electrochemical capacitors with unusually high energy density. Typically, they've been used to start locomotives, tanks, and diesel trucks. Because it can store and rapidly release large amounts of electrical power, a supercapacitor can serve as a buffer between the battery pack and the vehicle's electric motors, improving the vehicle's responsiveness while reducing the charge/discharge cycling that shortens battery life.

"Many people have talked about using supercapacitors as buffers on a battery, but we also will use artificial intelligence to manage how power is discharged and stored," Nourbakhsh said. "Based on a driver's route and habits, the smart power management system will decide whether to draw power for the electric motors from the batteries or the supercapacitor and decide where to store electricity produced by the regenerative braking system as the car slows down or goes down a hill."

Power management is one of ChargeCar's broad objectives. The researchers calculate that an intelligent electric car controller could recapture 48 percent of the energy during braking and that a supercapacitor could reduce 56 percent of the load on the batteries and reduce heating of the batteries--which shortens battery life--by 53 percent.

"The number one cost of electric vehicle ownership is the batteries," Nourbakhsh said. "Smart power management will save money initially because it pairs a low-cost battery pack with a small supercapacitor. And it will continue to save money by increasing efficiency and extending battery life." By customizing each vehicle to the owner's specific commute, ChargeCar will save money for some owners, he said, by allowing them to purchase the minimum number of batteries necessary.

ChargeCar has invited people from around the United States to participate in the project in two ways. First, volunteers can store their commute data via GPS and upload it to the site. The research team will use the database to compute how electric cars would perform on the same trips and in the same traffic conditions. Second, volunteers can participate in a contest in which they submit their own algorithms based on a Ruby program provided by the researchers that would minimize battery duty for all trips for a specific area, initially, Pittsburgh.

About the Author

Dian Schaffhauser is a former senior contributing editor for 1105 Media's education publications THE Journal, Campus Technology and Spaces4Learning.

Featured