University of Minnesota Duluth

GPU Plume

Researchers: Dr. Pete Willemsen | Andy Norgren

Project Overview

Real-time Environmental Simulation

Human populations are increasingly concentrated in urban regions, which, in turn, are becoming major centers of pollution and energy use. Our research aims to give urban planners tools to evaluate the impact of building location and form on pollution dispersal and energy needs.

Image from GPU Plume

Our research utilizes GPUs to accelerate dispersion simulation and heat energy transfer to solve problems related to minimizing pollution and energy usage in urban environments.

While prior research enabled single simulations that could predict the impact of a particular design on pollution concentrations, by harnessing the multi-core power of computer graphics hardware (i.e. Graphics Processing Units or GPUs) to quickly run many thousands of simulations, this research is building tools that can suggest best-case designs to urban planners, potentially enabling cities that trap less pollution and require less energy to heat and cool.

Image from GPU Plume Virtual Reality Interaction

We also utilize interactive and immersive virtual environment technology to provide unprecedented understanding and refinement of the complex physical processes associated with the energy balance and pollutant dispersion in an urban setting. We expect that the modeling capabilities that will be developed through this work will aid urban planners in developing useful and novel planning strategies to improve the sustainability of modern cities. Through this tool, we are addressing the following questions aimed at evaluating our hypothesis:

  1. Can urban form be optimized to minimize air pollution while simultaneously minimizing building energy use?
  2. Can interactive, immersive visualization and simulation assist engineers, urban planners, and students in understanding and refining complex environmental processes in the urban setting?

Related Publications

Funding Source

This material is based upon work supported by the National Science Foundation under Grants No. IIS-1162131 and IIS-1162617 (Utah). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.



The project is supported by the National Science Foundation and is a highly collaborative project including colleagues from the University of Utah’s Mechanical Engineering Departments and the University of Minnesota Duluth’s Department of Computer Science.