The US Department of Energy (DOE) is going to harness the power of supercomputers in two new initiatives from its Vehicle Technologies Office (VTO) that aim to improve transportation energy efficiency across the country.
The DOE’s 17 National Laboratories are home to 32 of the world’s 500 fastest supercomputers, which are used by the agency’s scientists and researchers to accelerate research by creating models from complex data sets.
Now, two new VTO initiatives – High Performance Computing for Mobility (HPC4Mobility) and Big Data Solutions for Mobility – will use the supercomputing capabilities of the national labs to find solutions to real-world transportation energy challenges. These initiatives are part of VTO’s Energy Efficient Mobility Systems (EEMS) Program that aims to conduct early-stage research at the vehicle, traveler and system levels to create knowledge, tools and solutions that increase mobility for individuals and businesses, while improving transportation energy efficiency.
VTO’s EEMS program has launched a US$2m multilab research initiative to develop new algorithms and big data tools that can model urban-scale transportation networks using real-world, near real-time data. The initiative will develop the data science approaches and HPC-supported framework for next-generation mobility systems modeling and operational analytics.
This will deliver an understanding of transportation system efficiency opportunities that is not attainable with current approaches. Modeling informed by real-time data will allow transportation systems to respond to events such as accidents, weather, and congestion in such a way that optimizes the overall energy use of the system.
The Big Data initiative includes researchers from the Lawrence Berkeley, Pacific Northwest, Argonne and Oak Ridge National Laboratories, as well as partners from academia and industry.
With initial VTO funding of US$500,000, HPC4Mobility will provide cities, companies, transportation system operators and others that qualify, access to national laboratory resources, including supercomputing facilities, data-science expertise and machine-learning capabilities, which can be applied to emerging transportation data sets. The first year ‘seed’ projects include:
• Lawrence Berkeley National Laboratory will work with the Los Angeles County Metropolitan Transportation Authority (Metro) on HPC-enabled computation of demand models at scale to predict the energy impacts of emerging mobility solutions. Possible applications include modeling the impact of autonomous vehicles on transportation energy use and the hour-by-hour impact of ride hailing services on traffic congestion;
• Oak Ridge National Laboratory will work with intersection management systems developer Gridsmart Technologies on reinforcement learning-based traffic control approaches to optimize energy usage and transport efficiency.