Modeling of a DaimlerChrysler Truck Engine using an Eulerian Spray Model

Transcription

1 Modeling of a DaimlerChrysler Truck Engine using an Eulerian Spray Model C. Hasse, S. Vogel, N. Peters Institut für Technische Mechanik RWTH Aachen Templergraben Aachen Germany c.hasse@itm.rwth-aachen.de Abstract Simulations of a large bore DI diesel engine using an Eulerian spray model close to the nozzle and a flamelet model for the turbulence-chemistry interaction are presented. Pressure data and 2D images showing soot radiation from an optically accessible engine are compared to numerical results. The prediction of the penetration of the gas phase and the mixing is evaluated. Introduction The testing of new spray and combustion models for engine simulations is usually done in simple geometries such as injection chambers where optical access can be achieved relatively simple. This is a very important step in model development. However, further testing and verifying under real engine conditions including moving pistons and complex geometries is much more difficult and usually only global results such as pressure or heat release are investigated. In this paper we will use data from an optical engine to evaluate the mixing and combustion process using a recently developed spray model which has been tested in injection chambers [1 3] previously. Numerical Model A new spray model, the Interactive-Cross-Sectionally-Averaged-Spray model, describing the liquid phase close to the nozzle with Eulerian transport equations, has been developed. At detailed description and derivation can be found in [1, 2] and only a short summary is given here. Eulerian boundary layer equations for the gas and the liquid phase are formulated and integrated over the cross-section of the spray. This leads to the instationary, one-dimensional CAS-equations. These equations are solved on a separate grid which has usually a higher resolution than the CFD grid. This avoids some of the resolution related problems especially close to the nozzle where very small structures occur. The CAS-code communicates with the CFD code by exchanging boundary conditions and source terms for momentum, energy and mass. A first application of this model along with a flamelet model was presented by Hasse and Peters [3] for a diesel spray in a high pressure, high temperature injection chamber. The turbulence-chemistry is described by the Representative-Interactive-Flamelet (RIF) model. The model and the coupling to the CFD code has been described in detail in a number of papers [4 7] and only a brief description is given here. The flamelet parameters scalar dissipation rate and pressure are passed to the flamelet code which computes the instationary flamelet solutions for that time step. These solutions, being a function of the mixture fraction variable, along with the probability density function of the mixture fraction can be 1

2 used to compute the mean temperature and species profiles in the 3D field. The chemistry which also includes pollutant formation processes is completely solved within the flamelet code. It is important to note that the soot model is also completely incorporated into the framework of the flamelet concept. Soot radiation which is used to compare against experimental data is modeled according to [8]. Experimental Setup Figure 1: Experimental setup of the optical engine. The results presented here were obtained on two different test benches. The pressure data was obtained with an one-cylinder engine without optical access. All the optical measurements were obtained on a separate, optically accessible one-cylinder version of the same engine. Figure 1 shows the optical engine setup. The laser on the left side is used for LII (Laser-Induced Incandescence) and LIF (Laser-Induced Fluorescence) measurements. The copper vapor laser on the bottom and the optical fiber are used for the visualization of the liquid phase by light scattering. The data is collected by a high speed camera system. Global soot radiation can be measured with a 2-color detector which can be used to calculate the soot concentration. For the case presented here both systems were equipped with a PLN (Pump Line Nozzle) injector system. The engine runs at 1130 rpm and the displacement volume is 2 liters. The start of injection is 3 degrees before TDC and the injected amount is 131 mg which corresponds to 50% load. Results Figure 2 is a comparison between the experimental and the numerical pressure curves showing good agreement. This result was obtained using one flamelet. Figures 3-9 show a comparison of soot radiation for seven different crank angles. The experimental data was obtained on the optical test bench described above. The perspective of all pictures is from the bottom through the quartz piston. The luminescence can be associated with the soot radiation. The postprocessing of the simulation data plots all the cells that are in a certain mixture fraction range, typically , which was found to be the range in which radiation occurs. The cells are plotted in yellow and the intensity scales with 2

3 90 80 Pressure [bar] Experiment ICAS Crank angle [deg.] Figure 2: Comparison between experimental and simulated pressure curve. the magnitude of soot radiation. It is important that this color encoding is done in a transparent manner such that all cells that lie along a line can be seen. Otherwise the first radiating cell would block the view and all cells that lie behind the first cell would be hidden. Figure 3: 6.5 degrees after TDC 3

4 Figure 4: 9.8 degrees after TDC Figure 5: 15.3 degrees after TDC Figure 6: 20.5 degrees after TDC 4

5 Figure 7: 25.3 degrees after TDC Figure 8: 29.6 degrees after TDC Figure 9: 35.1 degrees after TDC 5

6 When comparing this sort of images it has to be kept in mind that the simulation yields ensemble-averaged results whereas the experiment show a single event. This means that some of the structures seen in the experiment cannot be expected to be seen in a RANS calculation. This would require an LES approach. Considering that the simulation of soot and consequently the soot radiation is a very complex problem the overall agreement can be considered good. A general trend of underprediction of the gas phase penetration can be seen. Especially in the squish region close to the cylinder less radiation is seen in the simulation. At the edge of the piston bowl the experiment shows almost no luminescence. On the one hand this can attributed to optical errors due to high curvature of the bowl in this region. On the other hand this may be due to quenching at the cold walls. The temperature drops locally such that no radiation is visible anymore. This effect has not been accounted for in these simulations but Hergart [9] showed that this phenomenon can be incorporated into the flamelet modeling framework. Summary A new Eulerian Spray which has been developed previously and tested in injection chambers was used to simulate a large bore DaimlerChrysler DI diesel engine. Simulation and experiment were compared for a part load case giving good agreement for the pressure curve. In order to compare the penetration and the mixing inside the engine experimental 2D images showing soot radiation are compared to numerical results. The agreement shows similar behavior although the simulation seems to underpredict the penetration in the squish region above the piston. Acknowledgments The authors wish to thank Gerhard König from DaimlerChrysler for providing the experimental data. This work was partially financed by the European Commission in the framework of the Competitive and Sustainable Growth Programme, D-Level Project (GR3RD-CT ). References [1] Wan Y. and Peters N., Atom. Sprays 9: (1999). [2] Krüger C., Validierung eines 1D-Spraymodells zur Simulation der Gemischbildung in direkteinspritzenden Dieselmotoren, Ph.D. thesis, RWTH Aachen, [3] Hasse C. and Peters N., 12th International Multidimensional Engine Modeling User s Group Meeting, Cray Inc, [4] Hasse C., Barths H., and Peters N., Paper No. SAE [5] Barths H., Pitsch H., and Peters N., Oil & Gas and Technology-Rev, vol. 54, IFP, 1999, pp [6] Barths H., Hasse C., and Peters N., Journal of Engine Research 1: (2000). [7] Barths H., Hasse C., Bikas G., and Peters N., Proc. Combust. Inst., vol. 28, The Combustion Institute, Pittsburgh, 2000, pp [8] Hubbard C. and Tien C., Prog. Energy Combust. Sci. 10: (1982). [9] Hergart C.A., Modeling combustion and soot emissions in a small-bore direct-injection diesel engine, Ph.D. thesis, RWTH Aachen,