By Rob Mitchum // December 18, 2014
The cities and urban developments of the future will dwarf what architects and city planners have done in the past, creating an urgent need for new large-scale tools and approaches. Lakesim, one of the first projects of the CI’s Urban Center for Computation and Data seeks to address this need, combining modern design tools with scientific supercomputing to create a new platform for testing and modifying different plans. Working with architecture firm Skidmore, Owings & Merrill and developers McCaffery Interests, a team of researchers from the Computation Institute and Argonne National Laboratory have built a prototype system that simulates the energy demands and transportation needs of the Chicago Lakeside Development, a 600-acre transformation of a former US Steel site on Chicago’s South Side.
In her The Science of Cities column, Next City reporter Rebecca Tuhus-Dubrow talked to these researchers, “some of the most hard-core scientists in the world,” about the creation of LakeSim and how these tools could help ensure that tomorrow’s cities are both sustainable and pleasant for residents.
“From our point of view it’s about using high-powered science and computing to make cities more livable and efficient,” Argonne scientist and CI Senior Fellow Jonathan Oziktold Tuhus-Dubrow. “Design processes tend to be pretty ad hoc, tend to be pretty time-consuming, tend to be pretty unsatisfying … If you are creating this site that has 600 buildings, to be able to understand the impact on water, energy, economic development is really important.”
The article highlights what is truly different about LakeSim and similar programs — the ability to design at enormous scale, and use supercomputing power to run thousands, even millions, of scenarios to describe a broad range of scenarios and understand how the complex interplay of different urban systems might play out over decades of time.
Transportation modeling has been around for decades, and energy modeling, too, is widespread. What the Argonne scientists add is the ability to model in much greater detail and at much greater speed than the typical methods. What might take days for other computers, they can complete in a matter of hours. They also integrate the simulations into one framework, in order to see the interplay among various factors.
“The output generated by the models is massive,” says [Vadim] Sokolov [Computational Transportation Engineer at Argonne]. “It’s not something you can open in your Excel spreadsheet.” The scientists then have to interpret it and visually represent it in order to communicate it in a meaningful way to the developers.