Search Results

The air-fuel mixture formation in an optical direct-injection internal combustion engine is investigated by numerical simulations for the two biofuels Ethanol and 2-Butanone. The gas phase in the internal combustion (IC) engine is predicted by a large-eddy simulation, in which the fuel phase is determined by a spray model based on Lagrangian particle tracking. A hollow-cone injector is used for which the primary breakup is modeled by a series of small full-cone injections, while the Rosin-Rammler initial droplet size distribution is used. The secondary spray break-up is modeled by the Kelvin-Helmholtz-Rayleigh-Taylor (KHRT) model, and the evaporation of the fuel is determined by the Bellan-Harstad model. The gas phase simulation is based on a finite-volume method formulated for hierarchical Cartesian grids, in which the immersed moving boundaries are resolved using a multiple level-set/cut-cell approach.

New combustion processes require an understanding of the highly three-dimensional flow field to effectively decrease fuel consumption and pollutant emission. Due to the complex spatial character of the flow the knowledge of the development of the flow in an extended volume is necessary. Previous investigations were able to visualize the discrete three-dimensional flow field through multi-plane stereoscopic PIV. In this study, cycle resolved tomographic particle-image velocimetry measurement have been performed to obtain a fully resolved representation of the three-dimensional flow structures at each instant. The analysis is based on the measurements at 80°, 160°, and 240° after top dead center(atdc) such that the velocity distributions at the intake, the end of the intake, and the compression stroke at an engine speed of 1,500 rpm are discussed in detail.

The interaction of biofuel sprays from an outward opening hollow cone injector and the flow field inside an internal combustion engine is analyzed by Mie-Scattering Imaging (MSI) and high-speed stereoscopic particle-image velocimetry (stereo-PIV). Two fuels (ethanol and methyl ethyl ketone (MEK)), four injection pressures (50, 100, 150, and 200 bar), three starting points of injection (60°, 277°, and 297° atdc), and two engine speeds (1,500 rpm and 2,000 rpm) define the parameter space of the experiments. The MSI measurements determine the vertical penetration length and the spray cone angle of the ethanol and MEK spray. Stereo-PIV is used to investigate the interaction of the flow field and the ethanol spray after the injection process for a start of injection at 60° atdc. These measurements are compared to stereo-PIV measurements without fuel injection performed in the same engine [19].

Direct injection (DI) of compressed natural gas (CNG) is a promising technology to increase the indicated thermal efficiency of internal combustion engines (ICE) while reducing exhaust emissions and using a relatively low-cost fuel. However, design and analysis of DI-CNG engines are challenging because supersonic gas jet emerging from the DI injector results in a very complex in-cylinder flow field containing shocks and discontinuities affecting the fuel-air mixing. In this article, numerical simulations are used supported by validation to investigate the direct gas injection and its influence on the flow field and mixing in an optically accessible ICE. The simulation approach involves computation of the in-nozzle flow with highly accurate Large-Eddy Simulations, which are then used to obtain a mapped boundary condition. The boundary condition is applied in Unsteady Reynolds Averaged Navier-Stokes simulations of the engine to investigate the in-cylinder velocity and mixing fields.

Numerical analyses of the liquid fuel injection and the subsequent fuel-air mixing in a high-tumble, long-stroke direct injection engine at operation conditions of 2000 RPM are presented. The Navier-Stokes equations are numerically solved with a finite-volume method for compressible flow based on a hierarchical Cartesian mesh. The solid wall boundaries are represented by a conservative multiple cut- and split-cell method, where a semi-Lagrange level-set solver is used to track the location of the individual moving boundaries. To determine the fuel vapor before ignition, a two-way coupled large-eddy simulation of the turbulent flow field with the spray droplets is conducted. Due to the large number of spray droplets, a Lagrangian Particle Tracking (LPT) algorithm is used to predict the liquid spray penetration and evaporation.

The influence of in-cylinder flow on the propagation of 2-Butanone and Ethanol sprays is studied. To solely evaluate the interaction of air flow and fuel, high-speed Mie-Scattering Imaging of hollow cone sprays is conducted both in a single-cylinder optical engine with tumble movement and in a pressure vessel with negligible air flow. The direct comparison reveals an improved spray propagation of 2-Butanone due to the engine’s air flow. The lower viscosity of 2-Butanone causes an enhanced jet breakup compared to Ethanol such that the spray consists of more and smaller droplets. Small droplets possess a lower momentum, which allows the droplets to be more efficiently transported by the air flow. Consequently, the fuel distribution across the cylinder is enhanced. As the liquid fuel is distributed to a larger volume, improved convection accelerates evaporation.

Numerical analyses of the liquid fuel injection and subsequent fuel-air mixing for a high-tumble direct injection engine with an active pre-chamber ignition system at operation conditions of 2000 RPM are presented. The Navier-Stokes equations for compressible in-cylinder flow are solved numerically using a hierarchical Cartesian mesh based finite-volume method. To determine the fuel vapor before ignition large-eddy flow simulations are two-way coupled with the spray droplets in a Lagrangian Particle Tracking (LPT) formulation. The combined hierarchical Cartesian mesh ensures efficient usage of high performance computing systems through solution adaptive refinement and dynamic load balancing. Computational meshes with approximately 170 million cells and 1.0 million spray parcels are used for the simulations.