Aggressive national goals for significantly reducing petroleum use and greenhouse gas emissions require major improvements in all aspects of our nation’s energy use. Combustion processes have historically dominated electrical power production and transportation systems. Despite significant advances in improving the efficiency and reducing the costs of alternative energy sources, combustion-based systems are projected to dominate the marketplace for decades, especially for hard-to-electrify sectors including aviation. Consequently, these systems must be optimized for energy efficiency and reduced emissions.
Diesel and gas-turbine engines drive much of the world’s industry and transportation. However, burning fossil fuels to power these engines strengthens the climate-warming greenhouse effect and emits harmful pollutants—which, according to the World Health Organization, kill more than 13 people every minute due to lung cancer, heart disease, and stroke. Scientists have spent the past half-century in search of safe and economically viable fuels, but a poor understanding of the complex physics inside combustion chambers has hindered their development efforts. High-performance computing is a key tool in overcoming this obstacle; researchers can begin to understand the best conditions for burning new and existing types of fuel by simulating mixing and combustion processes such as fuel-air mixture formation and ignition processes in extreme detail. Information from these simulations will be used to develop and test new fuel types and combustion conditions, ultimately leading to low-emission and high-efficiency combustion engines and a more sustainable standard in global transportation and power generation.
The Exascale Computing Project’s Combustion-Pele application is designed to apply the power of the world’s fastest supercomputers to recreate complex combustion environments in unprecedented detail. The Combustion-Pele team focused on developing new software architecture for exascale modeling of key scientific unknowns such as low-temperature combustion dependent on mixture formation, and the effects of multistage ignition. Understanding these processes will lay the groundwork to build new and much more efficient compression ignition combustion engines and will support the development and implementation of viable alternatives to fossil fuels.
The physics of ordinary combustion engines are already well understood, but new designs require far more precision to ensure that they function safely and consistently. These designs are difficult to implement without understanding the unique fluid and combustion dynamics that result from altering factors such as fuel types which affect, ignition temperatures and fuel injection timing. Researchers cannot efficiently address these unknowns without extremely granular models of the dynamic processes within combustion chambers, and creating these models requires previously unattainable computational power combined with software developments to greatly improve computational efficiency.
The Combustion-Pele application addresses these needs, delivering a system capable of the most precise physical models of combustion processes to date. The application boasts several innovative software improvements—including adaptively refined simulation mesh for improved computational efficiency and increased fidelity in critical regions; more realistic physics models for multi-component sprays, soot formation, gas dynamics, and thermal radiation effects; and a new portable code that functions on a wide range of computing hardware and scales up to more than three-quarters of Frontier, the world’s largest exascale computer. The Combustion-Pele team has used these improvements to simulate the interaction of two unique fuels with varying reactivity under a multipulse injection strategy. This work provides a functional baseline for future simulations to test the most promising biologically derived fuels and hydrogen fuel blends for use in key sectors such as aviation and power generation.
Future simulations using Combustion-Pele will capture the complex physical and chemical processes in realistic engine environments, augmenting the limited measurements that can be made in laboratory-scale engine configurations. Researchers can use this high-fidelity data to train artificial intelligence systems for more complex and efficient simulations in the future and to develop predictive models for engineering simulations. The impending rapid improvement in combustion simulation capabilities will allow scientists and engineers to develop much more efficient fossil fuel combustion engines that can utilize sustainable fossil fuel alternatives, thereby greatly reducing the negative impact of transportation and power generation on our environment.
The motivating problem underlying this project is a sufficiently realistic simulation of the in-cylinder processes in an internal combustion engine using low-temperature combustion for which reactivity-controlled compression ignition (RCCI) is the exemplar. The enabled exascale-era simulations will address key scientific questions regarding fuel mixture effects, the multistage ignition of a diesel surrogate fuel, lifted flame stabilization, jet entrainment affected by cylinder-wall geometry, and emissions.
The simulation accounts for the isentropic compression, subsequent injection of the high-reactivity fuel, and combustion processes in a compression ignition engine. Necessary physics include gas compression and models of fuel injection process, spray vaporization (i.e., injecting liquid fuel sprays into high-pressure conditions), and mixing. Combustion processes include autoignition, flame propagation, and soot/thermal radiation, all in a nontrivial engine geometry. The scenario involves kinetically controlled processes in turbulent combustion, including ignition, extinction, and emissions. The application used for this project, Pele, implements a hybrid large-eddy simulation (LES)/direct numerical simulation (DNS) approach in both the compressible and low-Mach limits at which the project team will refine the mesh to the DNS limit where necessary to capture turbulence/chemistry interactions by using the machinery of adaptive mesh refinement while restricting grid resolution to that required for a high-fidelity LES model far from the flame.
This project is structured around providing a combination of first-principles DNS and near-first-principles DNS/LES hybrid simulations to advance the scientific community’s understanding of fundamental turbulence-chemistry interactions at device-relevant conditions. The exascale motivating problem is to perform high-fidelity simulations of the relevant processes in an RCCI internal combustion engine. The relevant processes include turbulence, mixing, spray vaporization, low-temperature ignition, flame propagation, and soot/radiation. RCCI is thermodynamically favorable relative to existing engines and thus holds the promise of groundbreaking efficiencies while operating in a regime that limits pollutant formation. The roadmap toward this exascale-era motivating problem includes simulations of a multi-injection low-temperature diesel jet into an open domain with a large alkane fuel undergoing two-stage ignition processes, dilute spray evaporation and mixing, and multi-injection with fuels of varying reactivity in a geometry that influences the mixing field. The multi-injection simulation forms the challenge problem to demonstrate new exascale capability.
The specific science-based challenge problem was derived from the roadmap toward the motivating exascale-era problem. Specifically, the challenge problem demonstrates the ability to simulate the interaction of two fuels with varying reactivity under a multi-pulse injection strategy into an engine-relevant geometry. It also serves as a baseline for future simulations utilizing a diverse range of sustainable aviation fuels in aero-engine applications where the impacts of spray vaporization on mixture formation, flame stabilization and fuel-lean flame blowout in complex swirl-stabilized confined flames can be studied with Pele. As with the RCCI problem, the widely differing reactivity of SAF fuels poses challenges during take-off and cruise conditions in aero-engines. High-fidelity simulations will be used to inform the development of predictive subgrid combustion and spray models used in CFD as part of the co-design of sustainable aviation fuels (SAF) with aero-engines. Besides reducing an aircraft’s carbon emissions, using a SAF blend can also result in much lower ice crystal contrails at cruising altitudes, further reducing aviation’s impact on the climate. Ice crystal formation occurs through its interaction with exhaust soot particles and supercooled water vapor, and soot particle size and morphology depend upon SAF properties. Exascale simulations with multi-physics and geometry capabilities will enable detailed analyses of aircraft ‘turbulence-chemistry’ interactions and characterization of exhaust soot particles, potentially offsetting the expense of ‘flying laboratory’ which currently trails behind aircraft, sampling and analyzing gases and particles in its wake.