- Buildings account for more than 40 percent of US energy consumption
- Cities struggle to assimilate building energy data to comply with benchmarking laws
- Urban Informatics Laboratory helps settle urban policy and program recommendations
For the first time in history, more than half the world’s population lives in urban areas.
Swelling urbanization poses problems for urban governance, infrastructure, and environmental management. To meet these challenges, researchers in the Urban Informatics Laboratory (UIL) have been developing integrated and data-driven solutions.
Big city, big data
Cities and urban areas occupy three percent of the world’s total land surface, but account for 60-80 percent of global greenhouse gas emissions, while consuming 75 percent of natural resources and 80 percent of the global energy supply.
Cities across the US are turning to benchmarking to understand the energy performance of their buildings and identify opportunities to reduce energy usage.
Benchmarking measures a building’s actual energy consumption against a performance baseline, and derives information about how buildings compare to similar buildings and the potential impact of energy efficiency improvements.
Across the US, new laws mandating the collection and disclosure of building energy-use data have sprung up. The US Environmental Protection Agency estimates that energy benchmarking of buildings results in seven percent annual savings, roughly $10 billion in reduced energy costs.
Mo' data, mo' troubles
However, numerous municipalities are struggling to translate this data into actionable insights and identify which buildings are the best candidates for energy efficiency interventions. Results from implemented benchmarking initiatives indicate that merely enacting benchmarking laws does not result in savings.
Currently, few tools exist that allow governments to quickly and accurately translate benchmarking data into recommendations for effective energy efficiency policy and programs or track the implementation of such recommendations.
For example, New York City passed its landmark Local Law 84 (LL84) in 2011 mandating that all owners of buildings over 50,000 ft2 report energy consumption data to the city for performance benchmarking. The law was envisioned as a mechanism to arm policymakers with data to develop municipal energy efficiency policies and programs.
To achieve energy independence, the US must drastically improve the energy efficiency of buildings.
However, as a data analytics research consultant for the LL84 program, Rishee Jain, director of the UIL at Stanford University, saw first-hand how city officials struggled to analyze, visualize and, most importantly, translate the data from 15,000+ buildings into policy and program recommendations.
“Even today city officials rely on ad-hoc analysis modules and are constrained to issuing static yearly reports that fail to drive energy efficiency policies and programs,” Jain says. “There is a tremendous amount of data to be generated about the buildings by the benchmarking laws, and we as researchers should help the decision makers think about how they can effectively and efficiently use those data.”
To meet this need, the UIL have developed a novel integrated and data-driven method. This method successfully analyzes benchmarking data collected by cities across the US.
Specifically, the UIL researchers analyze the relationships between high-dimensional building characteristics and building energy use obtained based on governmental benchmarking mandates and separate the impacts of random errors from inefficiency sources.
Instead of comparing a target building with nationwide reference buildings, UIL compares it with local peers to calculate its potential maximum level of efficiency.
These efforts help policy makers and city planners visualize assorted ranks of building energy efficiency and quantities of energy saving potential.
Housed at the Stanford Linear Accelerator Center (SLAC), this cluster consists of 781 compute nodes, 12,320 CPU cores, 444 GPUs and 570 teraFLOPS of computing power. These tools allows UIL to runs tens of thousands of simulations to determine energy saving potential.
“We should accurately calculate the levels of energy efficiency for buildings at the city scale and maintain the interpretability necessary to facilitate municipal policy and program design,” says Zheng Yang, researcher at UIL. “Data science is the bridge to link the two objectives.”
For instance, data such as when fuel oil is delivered to buildings each month, building size, and building value enable Yang’s team to build a scalable process. This helps UIL to examine communities of buildings collectively and find hidden patterns of their features and energy use.
Another feature of their benchmarking method is its high generalizability, Yang says, and it can be easily applied to other cities with different benchmarking laws through the UIL server. In the midst of rapid urbanization, this data-driven benchmarking method is expected to play a significant role in realizing energy savings from energy intensive urban buildings and is integral to meeting global emissions goals.
Buildings account for more than 40 percent of the total energy consumption in the United States. To achieve energy independence, the US must drastically improve the energy efficiency of buildings, use the exact amount of energy when required, and reduce all unnecessary urban energy consumption.
UIL’s work highlights how public disclosure laws can reduce the information gaps that limit investment in energy efficiency improvements.
Their pioneering benchmarking research represents a first-step toward applying data-driven analysis and support to the sustainability challenges facing urban operation and management.