April 10, 2013

Intel and Hadoop Help Power Pecan Street

Ian Armas Foster

The advent of electric cars has introduced a new energy usage animal for utility companies and consumers to deal with. With the help of Intel distribution and Hadoop processing and the Texas Advanced Computing Canter, the Pecan Street Project in Austin has been set up to assist utilities and customers make informed decisions on energy usage in homes and across communities, with a special interest in how solar panels and electric cars affect that usage.

“Using data process from Intel distribution of Apache Hadoop, energy analysts and customers can explore energy use patterns during peak hours, compare day use versus night use, and determine overhead for charging electric cars,” noted Intel in their video on the Pecan Street Project.

The video below provides a truly intriguing look into the availability of these energy metrics and how power consumers can utilize them to inform their usage habits.

The community studied in the video is relatively progressive from an energy standpoint, with a fair amount of solar panels, electric cars, and smart meters. These things, while curtailing greenhouse gases, make for more work for the utility, which has to process and analyze all of that data. Not doing so would likely lead to shortages.

According to Intel, the project tracks energy usage over time in homes with smart meters. However, the smart meters are not simply making multiple measurements per day, which was the initial improvement smart meters provided over standard meters. They also track which appliances contribute to the energy usage.

Typically, refrigerators, air conditioners, and water heaters contribute significantly to a home’s energy usage. All of these appliances require changing the temperature of water, an energy-intensive process that leads to significant power bills.

Through the Pecan Street Project, it was realized that the charging electric cars consumed more power than any of the aforementioned appliances. It is typical to see an energy spike at around the time that people come home from work as they turn on lights, activate the HVAC system and perhaps exercise the refrigerator and oven.

However, since that process may now also include charging the electric car, the utility can expect an even larger power spike.

The consumer benefits from this project as they are made aware of their energy habits. For example, Intel focused in on a particular household that held both solar panels and an electric car. On a random day in November, their solar panels generated superfluous energy during the day that was then sold back to the utility grid. Some of that solar power offset the power required to charge the car, but not a lot since the sun sets relatively early in November.

Again, such information would not be readily available to both the utility and the customer without significant data processing investments, which in this case took the form of Intel-distributed Hadoop processing.

Related Articles

Erecting Operational Intelligence Using Machine Data

Preventing Brain Injuries with Predictive Analytics

Baldeschwieler: Looking at the Future of Hadoop