Advanced Computing in the Age of AI | Tuesday, March 19, 2024

A New Era in Automotive Engine Development Driven by HPC 

<img style="float: left;" src="http://media2.hpcwire.com/dmr/ICE_Car.png" alt="" width="95" height="81" />Thinking electric vehicle, hydrogen, or maybe something solar? Think again. The internal combustion engine (ICE) is going to be around for a while. Oak Ridge National Lab, with its Jaguar supercomputer, is helping companies like Ford and GM design the next generation of advanced ICEs.

Love it or hate it, it’s likely that the internal combustion engine will be a part of your life for many years to come.

So says the National Research Council: “…the internal combustion engine will be the dominant prime mover for light-duty vehicles for many years, probably decades…” (From Review of the Research Program of the FreedomCAR and Fuel Partnership: 3rd Report, NRC 2010).

The Department of Energy concurs: “The performance, low cost, and fuel flexibility of ICEs (Internal Combustion Engines) makes it likely that they will continue to dominate the vehicle fleet for at least the next several decades.  ICE improvements can also be applied to both hybrid electric vehicles (HEVs) and vehicles that use alternative hydrocarbon fuels.” (DOE Quadrennial Technology Review, 2011).

Not Your Father’s ICE
But the ICEs you’re accustomed to today are undergoing a rapid metamorphosis – in fact, tomorrow’s multi-cylinder engines are being designed to meet current and future CO2 regulations in many countries that are even more stringent than those in the U.S.  In 2025, the combined Corporate Average Fuel Economy (CAFE) regulation in the U.S. will be 54.5 mpg.

To achieve these goals, in the next generation of ICEs, automotive researchers are exploring low temperature combustion modes as well as new directions in post-combustion emissions controls. However, these new combustion and control technologies may have a few surprises for their designers.

For example, more efficient engines have lower exhaust temperatures – a major challenge for aftertreatment systems.  Also, gasoline direct-injection systems are raising concerns about gasoline-based particulates.   The need to simultaneously improve engine efficiency while reducing tailpipe emissions will require a systems approach.

This is why researchers at the Department of Energy’s Oak Ridge National Laboratory like Dr. Robert M. Wagner, Director of the Fuels Engines and Emissions Research Center, and Dr. Sreekanth Pannala, Senior Research Staff Member in the Computing and Computational Sciences Directorate, are exploring new approaches to accelerating the development and optimization of internal combustion engines.

There have been a number of technology advances recently that are leading to a major shift in engine management.  For example, advances in component technologies permit more complete control of engine boundary conditions and the combustion process.  New sensor developments provide the potential for high speed pressure and temperature monitoring as well as onboard emission diagnostics.  The combination of improved components, new sensors, and faster onboard computers enables more active control over the engine, pushing the boundaries of highly efficient operations.

In fact, the technologies now available are creating new opportunities for engine development by using advanced prediction and control of the various components to bring about the forced stabilization of inherently unstable systems.  In other words, bringing order out of chaos.

HPC is Key


High performance computing (HPC) plays an essential role in these investigations. Oak Ridge is home to Jaguar, a Cray XK6 system that's currently number six on the Top500 list. It is being revamped with a hybrid architecture that will get a significant boost from the installation of NVIDIA's next-generation Kepler graphical processing units (GPUs). This will take the current Cray system — operating at a max of 3.3 petaflops or 3.3 million billion mathematical calculations per second — up to potentially 20 petaflops in the coming months.  

Says Wagner, “Engine technology has grown incredibly fast in the past few years – engine design is more complex but also more flexible, there are many new technologies such as variable valve systems, high pressure injection systems, and sophisticated on-board computers. Advanced sensors that once were cost prohibitive are also now becoming affordable.

“Engine parameter space is growing rapidly and conventional design methods may not generate optimal solutions,” he continues.  “Also, this massive calibration space is difficult to manage.  However, these recent advances afford us an opportunity to enable new, high efficiency strategies for real-world applications.  For example, we will have the potential for improved fuel flexibility – the ability to make use of a more diverse fuel supply.”

He cites as an example advances in reactivity controlled compression ignition (RCCI) combustion.  Two fuels with different reactivity – e.g. gasoline and diesel – are blended in-cylinder to improve control of the premixed combustion event.  A key benefit of this approach is brake thermal efficiency that is equivalent to or better than a conventional diesel approach.  It also provides lower nitrogen oxide and particulate emissions, which, in turn, means less aftertreatment requirements.

However, realizing the benefits of RCCI combustion is not easy.  It requires two fuel systems that are precisely controlled.  In some conditions the fuel mix must be highly diluted and intake conditions must be accurately controlled.  Combustion feedback control is also essential.

Coping with Complexity

Comments Wagner, “Advances such as RCCI are accelerating engine development, but they also make for a high degree of complexity.  It takes an amazing amount of computer resources to tackle these problems.  Fortunately we have the resources and significant in-house expertise. This is what prompted Ford and GM to collaborate with us.”

Working through the lab’s Industrial HPC Partnerships Program, Ford Motor Company and the Oak Ridge scientists and engineers are using the massive supercomputer resources at ORNL to improve and accelerate the design of internal combustion engines.  This Program provides a pathway for companies to collaborate with the expert computational scientists at Oak Ridge as well as gain access to DOE’s leadership HPC resources at the lab.  The project uses high-resolution computational fluid dynamics (CFD) modeling based on Large Eddy Simulation (LES is a mathematical model for turbulence) to understand the effects of higher exhaust gas recirculation (EGR) rates on the combustion process.  

Says Pannala, “Hundreds of LES simulations of in-cylinder combustion running in parallel using innovative sampling algorithms on massively parallel computers such as Jaguar, provides the researchers with unprecedented data related to cycle-to-cycle variations, allowing us to better comprehend the factors that cause these variations. This open research will provide Ford with knowledge the company can use to create enhanced engine systems designs that enable robust combustion at high EGR rates.  The bottom line – greater fuel efficiency and reduced emissions.”

The second project underway at Oak Ridge through the Industrial HPC Partnerships Program is a collaboration with General Motors. Its objective is to improve the understanding and design optimization of gasoline fuel injector-hole patterns for improved engine efficiency and reduced emissions. The researchers are using OpenFoam, a C++ toolbox for the development of customized numerical solvers, and pre-/post-processing utilities for the solution of continuum mechanics problems, including CFD – to look at injector optimization and fuel flow.

Multi-hole injectors used by spark-ignited direct-injection (SIDI) engines have a number of advantages, including flexibility.  During manufacture, nozzle holes can be placed at various orientations to create a variety of spray patterns.  If the design is less than optimal, the SIDI engine will exhibit emission issues and poor thermal efficiency.  With the help of high performance computing, the GM and Oak Ridge researchers can gain a better understanding of the spray formation and the complex fuel-air mixture inside the combustion chamber – knowledge that is critical in designing higher efficiency engines.

(Figure courtesy: Neroorkar, Mitcham, Plazas, Grover, Schmidt, “Simulations and Analysis of Fuel Flow in an Injector Including Transient Needle Effects”, ILASS-Americas 24th Annual Conference on Liquid Atomization and Spray Systems, San Antonio, TX, May 2012)

A New Era in Engines    
“Engines are undergoing a paradigm shift in flexibility,” says Wagner. “We are looking at an amazing variety of new control opportunities that will push the boundaries of high efficiency operations.  To optimize the next generation of engines, we have to develop new approaches to calibration, sensing and control – and to achieve these goals, predictive simulation is essential.”

“Large scale computing capabilities exemplified by Jaguar are absolutely essential to making the breakthroughs necessary for advanced engine design and the ability of the automotive manufacturers to meet and surpass tomorrow’s stringent emission regulations,” he concludes.

-----

This research is being supported by the U.S. Department of Energy through the Vehicle Technologies Program and the Advanced Combustion Engines R&D sub-program under the direction of Gurpreet Singh.

EnterpriseAI