While the transportation field is mostly characterized by the use of liquid fuels, gaseous fuels like hydrogen and natural gas have shown high thermal efficiency and low exhaust emissions when used in internal combustion engines (ICEs). In particular, high-pressure direct injection of a gaseous fuel within the cylinder overcomes the loss of volumetric efficiency and allows stratifying the mixture around the spark plug at the ignition time. Direct injection and mixture stratification can extend the lean flammability limit and improve efficiency and emissions of ICEs.
Compared to liquid sprays, the phenomena involved in the evolution of gaseous jets are less complex to understand and model. Nevertheless, the numerical simulation of a high-pressure gas jet is not a simple task. At high injection pressure, immediately downstream of the nozzle exit the flow is supersonic, the gas is under-expanded, and a large series of shocks occurs due to the effect of compressibility. To simulate and capture these phenomena, grid resolution, computational time-step, discretization scheme, and turbulence model need to be properly set.
The research group on hydrogen ICEs at Argonne National Laboratory has been extensively working on validating numerical results on gaseous direct injection and mixture formation against PIV and PLIF data from an optically accessible engine. While a good general agreement was observed, simulations still could not perfectly predict the mixing of fuel with the surrounding air, which sometimes led to significant under-prediction of fuel dispersion. The challenge is to correctly describe the gas dynamic phenomena of under-expanded gas jets. To this aim, x-ray radiography was performed at the Advanced Photon Source (APS) at Argonne to provide high-detail data of the mass distribution within a high-pressure gas jet, with the main focus on the under-expanded region.
In this paper, the numerical simulation of high-pressure (100 bar) injection of argon in a cylindrical chamber is performed using the computational fluid dynamic (CFD) solver Fluent. Numerical results of jet penetration and mass distribution are compared with x-ray data. The simplest nozzle geometry, consisting of one hole with a diameter of 1 mm directed along the injector axis, is chosen as a canonical case for modeling validation. A sector (90°) mesh, with high resolution in the under-expanded region, is used and the assumption of symmetry is made. Results show good agreement between CFD and x-ray data. Gas dynamics and mass distribution within the jet are well predicted by numerical simulations.