Energy Efficiency’s Next Great Leap: Predicting the Weather

Mark Hopkins Hotel in SF
The Mark Hopkins Hotel’s PG&E bills run $60,000 a month.

A small metal box in the basement of the InterContinental Mark Hopkins Hotel in San Francisco is thinking about the weather. And what time of day it is. And when the sun will rise and set.

This box, part of the first installation of the new Stem Energy system, is paired with a battery, and will be using all that atmospheric and environmental data to make a second-by-second decision about whether the hotel should pull from the meter for its energy, or run on battery power. It’s a new online, data-driven energy management system designed to reduce peak usage based on the type of multi-variable modeling used in the financial industry.

And best of all, these meter-battery hookups require no action by users to maintain it or to hook it up to the HVAC system – nor any behavioral change, no turning lights off or the thermostat down. They just sit there (there’s no charge to rent them) and mint money, the savings split between Stem and the customer. (In the case of InterContinental, a 50-50 split.)

On Thursday night, Stem CEO Salim Khan, founder Brian Thompson, and InterContinental‘s Harry Hobbs led tours and had a kickoff event for the installations in the two hotels in SF, one of them a razor-sharp new Gold LEED hotel, and the other an 85-year old masterpiece. Under Hobbs’ leadership, both hotels have instituted major efficiency measures over several years, but the Stem system “has really filled a gap,” according to Hobbs: Specifically, knowing to the second when to switch to battery power to save on peak usage charges.

For a hotel like the Mark Hopkins, with an average monthly electricity bill of $60,000, this is no small matter. (They get calls from the utility to turn down the AC during peak demand in the city). Hobbs said he expected the Stem system to cut their bill by 15 percent.

Stem was founded in 2009 by Thompson, whose background is in e-commerce and investment banking. One of the main problems in renewables is predicting when energy will be available – based on weather patterns, cloud cover, etc. – and Thompson sought to apply the sophisticated financial models developed for e-commerce to the field of energy. Reports of past demand in the buildings are combined with data from Weather Underground’s API, for example, to create a predictive model to run the system.

cloud computing energy efficiency

What he came up with was a cloud-based algorithm that applied not just to solar, but to all energy consumption, regardless of source. They partnered with LA-based CODA, manufacturer of EV car batteries and storage cells; based on mostly work of mouth, they are almost to their capacity for orders in 2013, mostly to mid-size commercial buildings like hotels, restaurants and gas stations.

“The Stem system… well, it’s a battery…and…” said Thompson, mic in hand at the event, speaking to a Greenbuild crowd. “This feels like the early days of the Internet when no nobody knows what it is and I am trying to explain what Mosaic is.”

Stem employs 45 people currently, and found its first customer through the SF Community Power, a nonprofit that runs efficiency programs with groups like small businesses and low-income residents, Thompson said.

The systems in the InterContinentals required permits and were treated like a solar system by PG&E, which has been very supportive, according to Stem staff.

Hobbs, who is on the Sustainability Committee for the InterContinental chain, hopes that sustainability and luxury can be seen as something to go together.

“I’m encouraging the rest of the hospitality industry to get on board,” he said. “There is a lot of room for improvement.”

Photo by kennejima on Flickr. Used with permission.

Hannah Miller is a writer, ecologist, and adventurer living in Colorado. She is interested in everything, but particularly in creative sustainability practices, the Internet, arts and culture, the human-machine interaction, and democracy. She's lived in Shanghai, New York, L.A., Philadelphia, and D.C., and taught English, run political campaigns, waited tables, and written puppet shows. She definitely wants to hear what you're up to. You can reach her at @hannahmiller215, email at golden.notebook at or at her site:

3 responses

  1. C’mon guys! When you say the STEM system decides “…whether the hotel should pull from the meter for its energy, or run on battery power” are you really being serious? Or is this a joke?
    How big a battery does it take to run a 380 room luxury hotel? That’s news!

    1. They don’t necesarily need to run the entire hotel on batteries but probably manage the peak demand charge and shave the peak during the summer when prices are the highest. A 15% cost reduction isn’t necesarily a 15% reduction in energy consumption.

  2. I understand the idea of hosting an enterprise server utilising complex algorithms to predict energy consumption based on various measurements and data inputs, but if costs are recovered through “savings on a 50/50 split”, how do you factor in the hardware and installation costs? Also, the switching between grid and battery must involve other equipment such as SCADA, PLC’s etc. Surely there’s more to it than the installation of standard lithium-ion batteries and inverters? What’s the catch?

Leave a Reply