Algorithms for Sustainability

Monday, December 10, 2018

Mechanical engineering professor Igor Mezić comes to sustainability by way of algorithms, particularly as they relate to "smart" buildings and, peripherally, tracking oil spills in open ocean.

In an era when energy efficiency is a major topic of concern, buildings account for forty percent of the energy and seventy percent of the electricity used in the United States. According to Mezić, roughly twenty to thirty percent of that energy is wasted, because buildings aren’t smart.

Mezić is a world expert on smart buildings — the systems they need to be energy-efficient, and the systems needed to monitor those systems and the building generally and receive data that can inform corrective action. He directs UCSB’s Center for Energy-Efficient Design and leads the Buildings and Design Solutions Group within UCSB’s Institute for Energy Efficiency (IEE). Most importantly, he develops algorithms that are embedded in software that drives a building’s capacity to be smart and informs human managers what is needed for it to run at optimal efficiency.

“It’s a problem that has global significance, and it’s not easy,” Mezić says. “Buildings are very dynamic.”

A building changes constantly over a 24-hour cycle, depending on the level of occupancy, the temperature and humidity, and the number of lighting, air-conditioning, heating, and other systems that are running, and are often left running even though they are not in use

While many large, modern buildings are loaded with sensors that deliver data related to every element of a building’s moment-to-moment energy profile, historically, the vast majority of that data has not been monitored or used until a system malfunctions, largely because humans had to sift through the data to recognize useful patterns, and there simply weren't enough people for the job. However, in the past twenty years, Mezić says, “We’ve replaced the level of analysis humans once did with an automated algorithmic layer that didn’t used to exist at a very profound level.”

Buildings have wised up.

Mezić’s first effort in this area came at the LEED Gold–certified UCSB Student Services Building, just to see what he could discover. He and his lab researchers added many new sensors to the building, hand-positioning them in places where measurements did not exist to avoid fusing the new data with existing measurements. They targeted hundreds of data points, and the data from them was captured every five to ten minutes.

Then, Mezić recounts, “We passed those massive sets of data through our algorithms to get a global picture of how energy was being used, what was working and what wasn’t, and where you could get savings. We tried to squeeze as much efficiency as we could from the systems that we already had.”

The result was impressive: energy efficiency was increased by twenty percent in a building that was already fifty percent better than average.

From there, Mezić and his colleagues continued to develop the mathematics. Some of their resulting algorithms have received patent protection over the past ten years, and in 2011, the UCSB Office of Technology Transfer licensed some of them to a company called Ecorithm, which incorporated them into a web-based application that offers real-time indicators of a building’s performance.

While many early adapters of the system are companies that do business in a technology-related space, Mezić expects that to change. “The whole world is moving toward achieving greater energy efficiency and using less energy, and the energy savings pay for the cost of the software quickly, in some cases within a year,” he says. “Adaptation will happen over time.”

Sustainability concerns related to energy can easily include purely environmental matters, such as those that arose from the Deepwater Horizon disaster that occurred in April 2010, when the platform exploded, caught fire, and sank, releasing an estimated 5.9 million barrels of crude oil into the Gulf of Mexico.

In June, Mezić was watching news about the spill and the spread of the oil. Hearing people talking about the oil and what seemed to be the impossibility of predicting where it would go made him think, With so much knowledge available about currents in the ocean, we should be able to say something about this.

“I thought my community needed to get involved just from the purely scientific standpoint,” he recalls.

Mezić creates algorithms that apply not to a single challenge but, with some adjustments, to multiple challenges, and so, with “a few theories on the shelf,” he ran some calculations and shared them with an attorney friend in Los Angeles, who passed them on to some government officials. Shortly thereafter, Mezić received a call requesting that he travel to the gulf and, about a week later,   found himself boarding a flight that was carrying oil-cleanup business executives to the site of the spill. Once in Louisiana, he started testing algorithmic simulations of oil movement against what people were measuring on the water and shared the results with the Coast Guard and anyone else who was interested.

Mezić's simulations turned out to be highly accurate, and his work caught the attention of Jane Lubchenko, then administrator of the National Oceanic and Atmospheric Administration (NOAA), who invited him to write a paper on the topic, which was published in Science.

After the spill had been capped, he joined UCSB colleague David Valentine, professor at the UCSB Marine Science Institute and an expert on the microbial ecology of hydrocarbons, to co-author a paper published in the Proceedings of the National Academy of Sciences. The paper resolved a number of scientific questions related to the spill, especially how water movement in the Gulf of Mexico caused  the oil to spread, both on and below the surface, and how the spill had created a bloom of oil-eating bacteria that helped in some small measure to mitigate the disaster.

Three years later, when the Refugio oil spill occurred, releasing nearly 143,000 gallons of oil along the Gaviota Coast, Mezić used his existing algorithms again to predict, correctly, that the oil would reach Manhattan Beach in Los Angeles.

By accurately simulating reality, Mezić’s model can enable those responding to actual spills in the future to better understand where the oil will go and, thus, direct their cleaning efforts most effectively.

Professor Igor Mezić's algorithms have wide-ranging application, from energy efficiency to environmental protection.