Computational engineering team receives $1.2 million to model phenomena occurring at multiple scales
Nov. 8, 2005
AUSTIN, Texas—The U.S. Department of Energy has granted $1.2 million to University of Texas at Austin engineers to improve the accuracy of computer simulations of processes that occur in different lengths of time, different scales of size or in other multi-layered circumstances.
J. Tinsley Oden, professor of aerospace engineering and engineering mechanics and director of the university’s Institute for Computational Engineering and Sciences will direct an interdisciplinary team of university researchers for the project.
|Dr. J. Tinsley Oden stands in front of a computer nicknamed “Bevo,” which is a 90-processor device. The computer is housed within ICES in the Center for Subsurface Modeling run by Professor Mary Wheeler.|
|Photo: Jennie Trower|
Oden, who also is an associate vice president for research, compared the goal of the simultaneous simulations to viewing a landscape portrait in its entirety while still noticing its distinct features.
“We want to develop computer simulations that allow us to look at the blade of grass when it’s important and the landscape when it’s important,” he said, noting that this is a major challenge for all areas of science in the 21st century.
For example, designing the next-generation of faster, cheaper computer chips involves manipulating the motion of handfuls of atoms during a billionth of a second. Yet that manipulation is affected by the fact that the manufacturing process can last minutes and involve millions of atoms. Similar multi-scale processes exist in areas such as physics, biology and biomedicine. A drug, for example, may work its magic by briefly interacting with a protein on the surface of cells in one organ, but instigate the release of cell chemicals over minutes that affects the entire body.
“An enormous collection of mechanical and biological processes now being studied by scientists and engineers involve events that take place at the molecular or atomistic scale,” said Oden, who holds the Cockrell Family Regents’ Chair in Engineering No. 2, “and we’re at a point in the history of science and engineering where we cannot rely on traditional methods that fail to account for effects that involve different scales.”
With the three-year grant, Oden and colleagues will turn theoretical models of a process into mathematical representations of the process that can be fed into a computer. To incorporate multiple states of one process into a final computer simulation, the engineers will combine models representing the different scales of the process and develop interfaces between those models.
A key feature of this effort will be insuring the final simulation is more mathematically rigorous than the conceptual models often used to explain how biological or other systems function. To make the models rigorous, they will be meticulously analyzed using a process that the university’s Institute for Computational Engineering and Sciences pioneered called adaptive modeling. The process involves taking into account the amount of error that might exist in a theory underlying a model, and adapting the model to minimize that error.
“There’s a tremendous strength in having a validated computational modeling tool in which you can have confidence,” Oden said, “because you can then explore things you would never hope to explore just by experimental observation.”
As a reality check, initial simulations will focus on the real-life process of producing advanced semiconductor devices. The simulator will be used to guide the development of new materials and processes in the laboratory of co-investigator C. Grant Willson. The early emphasis will be on applications related to imprint lithography, a technique the holder of the Rashid Engineering Regents Chair developed.
“With the simulator,” Willson said, “we hope to explore the effects of literally thousands of changes in imprint materials and processing conditions to get guidance for the design of optimum materials. The simulator will allow us to rapidly and efficiently achieve an optimum result without having to do the expensive and time consuming real tests normally required to develop such processes.”
The findings from the computer simulations can then be used to improve the semiconductor manufacturing process, Oden added, noting that “we hope that the simulation methods we develop will be applicable to a broad range of applications.”
For more information contact: Becky Rische, College of Engineering, 512-471-7272.