Natural gas as a fuel for internal combustion engines is a combustion technology showing great promise for the reduction of CO2 and particulate matter. To demonstrate the potential of natural gas direct injection, especially in combination with supercharging, some experimental investigations were carried out using a single-cylinder engine unit with lateral injector position. For this purpose different injection valve nozzles, piston crown geometries as well as operating strategies were investigated. First experimental results show that it is also possible to better support the combustion process by providing a late injection of a part of the fuel, near ignition point, so that the additional induced turbulence can speed up the flame propagation 1
Mixture formation with gaseous fuels due to its low mass density is more critical than in gasoline engines, because even high injection velocities still produce very low fuel penetration. For this reason mixture homogenization in the cylinder depends much more on charge motion than common for liquid fuels. In an internal combustion engine charge motion and mixture processes are extremely spatial and time dependent and experimentally scarcely detectable. For this reason, the 3D-CFD-simulation is one of the most appropriate approaches in order to investigate these complex phenomena under many different engine operating conditions and then find solutions for enhancing the engine performance.
In this research project the 3D-CFD-simulations that accompany the experimental investigations have been performed by the fast 3D-CFD-tool QuickSim. This tool has been released a few years ago by the research institute FKFS in cooperation with the IVK / University of Stuttgart. QuickSim is a 3D-CFD-tool that uses the commercial code Star-CD in the background. It introduces a new concept in the 3D-CFD-simulation of internal combustion engines (SI-manifold-injection, SI-GDI, CNG and soon also Diesel-engine), that drastically reduces the CPU-time in comparison to a conventional 3D-simulation. Numerous validation tests on different engines during the last six years show that this tool is able to predict the engine behavior of widespread applications with a very high accuracy.