Startup Uses Big Data To Cut Energy Usage

The vast majority of companies use some sort of energy-saving technology to cut energy costs, even if it's something as simple as replacing incandescent light bulbs with compact fluorescent lamps. But a Silicon Valley startup has a far more sophisticated solution that mixes big data, predictive analytics, and battery storage to reduce a commercial building's energy costs by 20% to 30%, the company claims.

Stem, an energy optimization firm based in Millbrae, California, sells a turnkey power system for businesses that analyzes large amounts of data, including historical weather information and industry usage patterns, to help companies avoid power usage spikes and higher energy fees during peak demand times.

In a phone interview with InformationWeek, Stem founder and CEO Salim Kahn made a comparison between his company's energy-saving technology and that employed by hybrid vehicles. "A hybrid car balances what's used by the EV battery and the internal combustion engine," says Kahn. "It's all automated -- the driver doesn't need to worry about it."

[ Read about Microsoft's plans for an environmentally friendly data center. See Microsoft's New Data Center: The Straight Poop. ]

Similarly, Stem's system also uses battery power and doesn't require any change in user behavior. However, many aspects of its design, including cloud-based analytics, are more advanced than those of hybrid cars.

More than 50 business customers have signed contracts to use Stem's technology, says Kahn. The first two locations, both InterContinental Hotels in San Francisco, began using Stem's Energy Optimization service just last month.

"At the InterContinental Mark Hopkins and InterContinental San Francisco, we have already benefitted from traditional energy conservation and efficiency measures, and are excited about Stem's ability to deliver even smarter energy use without compromising the comfort of our guests," said Harry Hobbs, engineering director for both hotels, in a statement.

Stem's hardware consists of a refrigerator-sized cabinet, which contains the company's intelligent storage devices. The InterContinental installations include EV batteries from electrical vehicle manufacturer Coda, although Kahn notes that Stem's system could use other batteries as well. "We are battery agnostic. We can use any supplier's battery. Right now, the world is seeing over supply in electric vehicle batteries. Battery costs are coming down very rapidly, and battery efficiency and cycle times are going up very rapidly," said Kahn.

The Stem plugs into the business' electrical system, but also stores power internally. It automatically determines when to use electricity provided by the grid, and when to should switch to battery power to save money. "Much like a hybrid car, the Stem system will automatically balance when to use battery power, and when to use grid power for maximum energy savings for that enterprise," Kahn said.

A lot of the company's intellectual property is in its modeling, forecasting and simulation software. Stem's predictive algorithms help companies determine the best times to pull power from the grid. It uses weather data, for instance, to develop of profile of changes in temperature, wind velocity, and humidity and any moment in time. "We run millions of simulations in a day to do that," said Kahn.

By studying a business's energy load profile, as well as the energy prices charged by the local utility, Stem can switch to battery power when prices are at their peak. Many variables play a role in predicting how electricity demand will change, Kahn said. "And as the demand changes, so does the consumption of electricity."

Cost savings vary by installation, with the average ranging from 20% to 30%, according to Stem.

Predictive analysis is getting faster, more accurate and more accessible. Combined with big data, it's driving a new age of experiments. Also in the new, all-digital Advanced Analytics issue of InformationWeek: Are project management offices a waste of money? (Free registration required.)