Liquid transportation fuels—and combustion engines that burn them—are “going to be with us for quite awhile, a number of decades,” said Dr. Steven E. Koonin, Under Secretary for Science, US Department of Energy in his plenary talk at the 16th Directions in Engine-Efficiency and Emissions Research (DEER) Conference in Detroit.
However, he noted, “we need to learn how to use liquid fuels more efficiently than in the past.” In the quest for more efficient engines to meet that goal, Koonin said, “computer modeling and simulation loom large”.
...it’s primarily a physics issue about energy density; the energy density—whether volumetric or gravimetric—of a liquid fuel is about 50 times better than the best batteries we can make now. For transportation, that is a primary consideration. However, we need to learn how to use them much more efficiently than we have been in the past. The department, to that end, is first of all supporting the development of the next generation of internal combustion technologies. There is a lot of headroom here.
We are trying to support that through the Combustion Research Facility in California run by Sandia National Laboratories, jointly funded by two parts of the Department of Energy, the science part and the vehicle efficiencies part, which is providing laser- and optical-based diagnostics for the fundamentals of combustion. Also through facilities such as the Advanced Proton Source at Argonne used for visualization and understanding.
As I look out over the department’s portfolio and ask what could make a difference in the next 5 to 10 years in terms of practical impact, computer modeling and simulation looms large in my thinking...It’s not widely known, but roughly 15 years ago when the US government decided to end underground nuclear testing of our stockpile, it embarked on a deliberate program to accelerate high-performance computing and to develop the methodologies to combine that computing with experimental data and historical test data that has resulted in the last 15 years in a truly remarkable predictive understanding of what goes on inside a nuclear weapon.
At the same time...for the last decade, the open, unclassified science part [of the DOE] has been funding and making available the same high-performance computing to many other uses: climate modeling, protein folding, materials science...it’s now time, we think, to take that expertise and apply it in a much more concerted and accelerated way to energy systems. If we can do that properly, we can optimize designs which shorten design cycles and facilitate the transition to scale from the lab bench to full scale deployment, accelerating it and making it much more economical.
...We are looking to build a collaboration between the national labs and the OEMS, centered on the Combustion Research Facility that I think can be a real competitive advantage for US industry as we work to develop a validated in-cylinder computational model that can accelerate design and optimize it as well. I think you will see us working toward that goal over the next year or two.—Under Secretary Koonin
A subsequent paper presented at DEER by Daniel Flowers from Lawrence Livermore National Laboratory highlighted the magnitude of the computation challenge. Flowers and his colleagues have been researching chemical kinetics of diesel fuels, and have developed mechanisms for complex long-chain species, enabling more representative diesel surrogates.
They have developed 2-methyl alkane mechanisms up to C20— branched iso-alkanes are significant components in gasoline and diesel fuels—bringing them up to about 7,900 species with some 27,000 reactions.
To run a full 3D cylinder model with 7,900 species would require about 42,000 Peta flops—about 30 hours on Oak Ridge Laboratory’s Jaguar supercomputer (currently the fastest), or about 60 years on a conventional workstation, Flowers said. Flowers outlined ongoing work in exploring new computing architectures (such as the optimized use of graphical processing units) as ways to improve simulation and modeling efficiency.
In his talk, Koonin noted that DOE was looking at pushing from Peta-scale computing to Exa-scale computing (1018 flops) in ten years—also targeted as necessary to be support the development of an eventual “500-mile battery” (earlier post).
The 2017-2025 phase of CAFE. In a separate talk at DEER, Tom Cackette, Chief Deputy Executive Office of the California Air Resources Board, shared ARB’s observations on the imminent release of the Technical Assessment Report (TAR) being developed by the Environmental Protection Agency (EPA) and the National Highway Traffic Safety Administration (NHTSA), with input from ARB, on the targets for the next round of light-duty vehicle fuel economy and greenhouse gas regulations for the years 2017-2025.
California is developing its own set of standards for the follow-on to the current Pavley requirements, but is seeking harmonization with the upcoming Federal rules (as is the case with the regulations through 2016).
According to Cackette, the TAR will outline a number of different annual reduction scenarios (3%, 4%, 5%, and, 6%) with the resulting CO2 grams/mile (tailpipe) in 2025 ranging from 190 g/mi under the 3% scenario to 143 g/mi under the 6% scenario. The 2016 target is 250 g/mi. This would equate in miles per gallon equivalent to a 2025 range from 47 mpge to 62 mpge under test and from 37 mpge to 50 mpge under use, respectively on the different percentage scenarios.
ARB’s initial observations on these targets, according to Cackette:
- Weight reduction is the most cost-effective means to reduce consumption and emissions.
- Further greenhouse gas emission reduction from conventional internal combustion engines is achievable and cost effective: “A lot more can be done.”
- Hybrids are necessary, ranging from current levels of a few percentage points in sales at less stringent standards to 50% sales with higher annual CO2 reductions
- Plug-in vehicles (including fuel cell vehicles) are only necessary by 2025 for higher annual CO2 improvements, because other technologies are pretty promising and have more favorable costs. Even with a 6% annual improvement, Cackette said, the fleet would need only a few—less than 10% in the higher scenario. It doesn’t mean that they won’t happen or be desirable from a consumer standpoint, just that they wouldn’t be necessary in large numbers to meet the target, Cackette said.
Daniel Flowers et. al. Computationally Efficient Simulation of High-Efficiency Clean Combustion Engines (DEER 2010)