Researchers are using some of the world’s fastest supercomputers operated by the DOE’s Office of Science to study and develop alternate fuels that can allow for the reduction of dependency on oil and the pollution it generates.
|
Large Eddy Simulation an automotive high-pressure fuel injector design for use in a hydrogen fuelled internal-combustion engine |
It’s no secret that the global supply of quality, easily extractable, light crude oil is dwindling while demand for fuel continues to grow in this increasingly industrialized world.
With cars and trucks accounting for about two-thirds of all oil use and a quarter of greenhouse gas emissions, it's becoming more important than ever to find alternative fuels that can help reduce both our dependency on oil and the pollution that they generate. But it's not that simple. Studying and developing alternate fuels presents science and government researchers with some very complex challenges.The use of fossil fuels has been widespread since the mid-1850s, providing industry experts with a long history and thorough understanding of how the internal combustion engine functions with petroleum-based fuels like gasoline and diesel. The legacy of this research cannot simply be extended to the study of new fuels.
Each fuel responds differently to diverse temperatures and pressures. Gathering data and building new alternative-fuel models necessary to study each requires extremely reliable, complicated, and nuanced computational models to generate new data appropriate to the fuel.
Researchers,like Jacqueline Chen and Joseph Oefelein at the Combustion Research Facility of Sandia National Laboratories in Livermore, CA, are conducting some exciting new studies to aid the U.S. Department of Energy (DOE) in its search for viable alternative fuels. Leaders in computational combustion science, Chen and Oefelein have developed some high-fidelity simulation approaches in order to effectively address the complexities and enormous data gathering needs of their research and to take full advantage of some of the world’s fastest supercomputers operated by the DOE’s Office of Science.
Specifically, Chen and Oefeleinare using these supercomputers to the study the burning processes associated with a variety of alternative fuels. By understanding more thoroughly how each fuel works, engineers hope to design internal combustion engines that burn it for maximum performance and minimal emissions. To quickly interpret this computationally intensive data, Oefelein and Chen rely on Tecplot 360, a computational fluid dynamics (CFD) visualization and analysis software.
Specifically, Chen and Oefelein are looking at the complex thermo-chemical interactions in internal combustion engines using carbon-neutral fuels like biofuels, and alcohols such as ethanol and dimethyl ether.
“You always want a clean-burning, highly efficient system,” says Oefelein. “And you want it to be a stable system, meaning that there are no combustion instabilities or transient types of processes that will damage the engine.”
To develop predictive models for designing clean and fuel-efficient engines, Oefelein and Chen use two computational techniques. One, called “large eddy simulation,” is a numerical technique used to solve the partial differential equations governing turbulent fluid flow. With this approach, the energy containing eddying motions that are dependent upon the geometry of the combustor are resolved numerically, and the dissipative small-scale turbulence and combustion scales require closure models.
The other, known as “direct numerical simulation,” examines the entire range of spatial and temporal scales of turbulence and flames, and therefore is restricted to a limited dynamic range of scales. Often this approach is well-suited for studying the micro-scales of turbulence chemistry interactions where turbulent mixing scales interact with the reactive flame and ignition scales. These two approaches then complement one another with large-eddy simulation characterizing the large-scale entrainment and mixing processes, and direct numerical simulation providing sub-grid information regarding micromixing and reaction.
Both techniques are computationally-intensive, in many cases generating petabytes of data. To put it into perspective, that's more than six times as much data as contained in the U.S. Library of Congress archive.
“Ten years ago, this research was at the terascale level in terms of computational speed. Now it’s at petascale,” says Chen. “In a few years, we’ll be at exascale.”
As computing power continues to increase, Chen and Oefelein can simulate a wider dynamic range of scales. Because the scaling of turbulence with Reynolds number (the ratio of inertial to viscous forces) is so challenging, researchers can only simulate a ten-fold increase in dynamic range of turbulence scales for every 1,000 fold increase in computing power. As a result, no one method can resolve the entire range of scales relevant to practical combustors in the foreseeable future. Instead, a multi-scale approach is required in which different, well-suited methods, like those that Chen and Oefelein are using, are necessary to bridge the gaps between the differing ranges of scales.
Interpreting with Tecplot
Oefelein and Chen discovered Tecplot 360 as a substantive and relatively fast way to interpret and visualize the massive amounts of data their research requires. They often perform calculations on supercomputers located offsite. Using a computer dashboard with a user interface and intuitive drop-down menus, they send Tecplot images such as isocontour plots, 2D and 3D plots, and animations back to their home facility, which allows them to quickly hone in on key statistics and identify critical trends.
Key to their work is that Tecplot 360 helps them to accurately analyze and visualize large volumes of data. “It’s a lot easier to examine visual images than it is to wade through numbers or statistics,” says Chen. “By looking at isocontour plots or volume renderings, we can see what’s going on from a broad perspective and then zoom in and gather statistics at a finer level.”
Seeing the data visually also provides a better diagnostic view. “If there’s a problem with the way the code or numerical parameters were set up, it’s much faster to spot these anomalies visually than by printing out rows and rows of numbers,” says Chen.
The visualization made possible by Tecplot 360 is important because it helps present results in a way that is easy to interpret. “Visualization software guides both our research and our presentation of results,” says Oefelein. “When we’re done and we’re ready to publish a paper, it helps us to tell our story and summarize what we saw in a concise way.”
Optimizing the internal combustion engine to work with alternative fuels is an important challenge – one that could significantly affect the long-term health and stability of our planet. By using Tecplot 360 to help them sort through their computationally-intensive data, Oefelein and Chen are getting closer to the goal of developing the underlying fundamental understanding that is required to ultimately develop predictive models, which will help reduce the nation’s reliance on fossil fuels.
Tecplot, Inc. delivers visualization software for engineers and scientists to analyze, discover, and communicate results. For more information visit www.tecplot.com.
