International Multidimensional Modeling User s Group Meeting at the SAE Congress March 4, 2001 Detroit, MI FIRE A Generic CFD Platform for DI Diesel Engine Mixture Formation and Combustion Simulation INTRODUCTION R. Tatschl, H.-P. Gabriel, P. Priesching Advanced Simulation Technologies AVL LIST GmbH A-8020 Graz AUSTRIA In recent years CFD has been successfully established for three dimensional simulation of fluid flow, mixture formation, combustion and pollutant formation in direct injection diesel engines. The accuracy of the simulation results and hence their contribution to design analysis and optimization, however, strongly depends on the predictive capabilities of the models adopted for simulation of injector flow, spray propagation, combustion and pollutant formation. In the last decade, intensive worldwide research has led to the development of a large number of models for simulation of each of the above mentioned processes and their implementation into the different in-house and non-commercial CFD codes. The increasing demands to provide fast and reliable answers to everyday engineering problems related to DI diesel injection and combustion system characteristics and performance, however, in many cases cannot any longer be met by adopting in-house and non-commercial CFD solutions. Workflow integration, application specific user interfaces, sophisticated pre- and post-processing capabilities are nowadays considered of being mandatory for the applicability of CFD within the engine development process. In the context of transient IC engine simulations, FIRE has already been recognized of being the leading CFD solution in terms of the above mentioned capabilities. With its open software architecture and well defined model interfaces, FIRE serves as a generic CFD platform for easy integration of all kinds of alternative spray, mixture formation and combustion / pollutant formation models that might be available at the user s site. Additionally, the CFD code FIRE also provides its own broad range of validated spray, combustion and pollutant formation models. The present article provides an overview of the various models available in or interfaced to FIRE for simulation of the flow in diesel injectors, spray propagation, auto-ignition, combustion, and pollutant formation. The individual model results are validated against selected experimental data obtained in different idealized configurations, full model validation is achieved via comparison of simulated results with measured data obtained in an AVL DI diesel engine. INJECTOR FLOW The knowledge of the injector mass flow rate and the flow conditions at the nozzle exit is a key issue for a successful simulation of all the subsequent processes of mixture formation, combustion and pollutant formation. In order to take into account the impact of geometrical details on the highly transient nature of the cavitating injector flow, only a comprehensive multidimensional two-phase flow model is able to provide the relevant information required as input for an IC engine spray simulation. In FIRE, the mathematical model used for injector flow calculations is based on the 1
two-fluid formulation of the conservation laws for two-phase flows [1, 2, 3, 4]. As an example, Figure 1 shows calculated results of the injection pressure influence on the shape and extension of the cavitation region in a diesel injector. Figure 1: Fuel vapor distribution in a VCO nozzle at 53 bar back pressure; 300 bar injection pressure, 1300 bar injection pressure; color scaled from 0.0 to 1.0 volume fraction level MIXTURE FORMATION The FIRE spray model is based on the well established Lagrangian discrete droplet method [5] and offers a broad range of sub-models for turbulent dispersion [6], coalescence [7], evaporation [8], wall interaction [9], primary and secondary spray / droplet break up [10, 11, 12, 13]. Alternative model formulations or additional models can be easily implemented via well defined interfaces to the basic droplet tracking algorithm and the gas-phase source term set-up routines. 0.08 0.08 penetration [m] 0.06 0.04 0.02 measured, liquid measured, vapor calculated, liquid calculated, vapor penetration [m] 0.06 0.04 0.02 measured, liquid measured, vapor calculated, liquid calculated, vapor 0.00 0.0000 0.0005 0.0010 0.0015 0.0020 time [s] 0.00 0.0000 0.0005 0.0010 0.0015 0.0020 time [s] Figure 2: Liquid and fuel vapor penetration evolution; ( 800 bar injection pressure, ( 1500 bar injection pressure Figure 2 shows the results of spray bomb simulations for a chamber density of 25 kg/m³ and a temperature of 800 K together with the experimental data taken from [12]. For the injection pressure variation study all parameters of the spray and injector flow models have 2
been kept constant. It can be seen that the agreement between experiment and simulation is very good for both the vapor and the liquid penetration characteristics. COMBUSTION The reaction mechanism used for simulation of the auto-ignition of diesel fuel in FIRE is based upon an extended version of the well known SHELL model [13, 14]. The high temperature hydrocarbon oxidation process in the fully established premixed / diffusion flame is expressed by a global reaction step or based upon equilibrium chemistry assumptions. The turbulent combustion model supposes that the reactions proceed instantly to completion once mixing occurs at the molecular level in the fine-scale turbulent structures of the flow [15, 16]. Alternatively, the ERC (University of Wisconsin) combustion and pollutant formation models have successfully been interfaced with FIRE on a fully modular basis [17]. Flamelet and presumed PDF models can be easily interfaced in the same way. Additionally, the availability of a transported joint-scalar PDF approach and the corresponding Monte-Carlo solution algorithm provides the basis for the development of the next generation IC engine combustion models. c) d) Figure 3: Calculated temperature field evolution; 1000 1/min, 75% load; ( 10, ( 15, (c) 20, (d) 25 ATDC; color scaled from 800 K to 2200 K temperature level; Mixing Time Scale Model The multidimensional simulation provides a detailed space and time resolved picture of the governing in-cylinder flow and flame evolution characteristics. Figure 3 shows the calculated spatial and temporal evolution of the temperature field in an AVL DI diesel engine which is a turbo-charged 2 liter per cylinder DI diesel engine with a bore and stroke of 123 mm and 164 mm, respectively. The engine configuration that forms the basis for all the results shown in this section has a swirl ratio of 1.8 and is equipped with an eight-hole nozzle and a common-rail injection system. Based upon the locally resolved flame and heat release data, a number of global data can be extracted, such as rate of heat release and cylinder pressure traces. Figure 4 shows calculated vs. measured cylinder pressure traces for variations of engine operating conditions and start of injection. The results demonstrate the degree of coincidence between measurement and calculation that can be expected from the presently adopted models. Moreover, the results clearly show the required sensitivity of the simulations with respect to the variations of the engine operation parameters. 3
Figure 4: Calculated vs. measured cylinder pressure traces; ( engine operation variation, ( SOI variation POLLUTANT FORMATION Thermal NO formation is accounted for via the Zeldovich mechanism (Figure 5). In the present case a single NO formation rate equation based upon partial equilibrium and steady state assumptions for the hydrogen radicals and molecular nitrogen is used [17]. Figure 5: Calculated temperature and NO concentration fields at 20 ATDC; 1000 1/min, 75% load; color scaled from minimum to maximum temperature / NO level; ERC Characteristic Time Scale Model, Equilibrium Chemistry The soot formation / depletion model adopted in FIRE is based upon a combination of chemical / physical rate expressions for representation of the processes of particle nucleation, surface growth and oxidation. The processes of particle formation and surface growth are taken to be functions of the local fuel and soot nuclei concentration, respectively, and the predominating flame temperature governing the Arrhenius rate coefficient of the particle mass addition term. The particle oxidation process that actually determines the soot emission level is modelled according to a hybrid chemical kinetics / turbulent mixing controlled rate expression. Oxygen partial pressure, local flame temperature as well as actual soot 4
concentration and local turbulent mixing time scale, obtained from the solution of the two equations k-ε turbulence model, contribute to the soot oxidation source [15, 16]. Alternatively, NO as well as soot formation processes can be calculated adopting the ERC combustion and pollutant formation models [17]. 1.0 3000 0.8 Normal Injection Split Injection 2500 Split Injection Normal Injection Normalised Soot Emission [-] 0.6 0.4 0.2 max. Temperature [K] 2000 1500 0.0-10 10 30 50 70 CA ATDC [deg] 1000 0 20 40 60 80 CA ATDC [deg] Figure 6: of soot formation and maximum flame temperature with conventional and split injection; ERC combustion and soot formation model As a result of the CFD simulation, details of the local interaction of mixture formation, combustion and pollutant formation are provided. Based on the temperature and species composition distributions in different sections across the combustion chamber the governing processes can be easily interpreted. The impact of combustion system parameter variations on the details of the soot and NO formation mechanisms can be assessed on a local space and time resolved basis. Figure 5 shows representative calculation results in a section across the spray axis. The results clearly show the pronounced interaction of the swirling gas flow with the temperature and species concentration distributions. The extraction of global pollutant formation data then serves as the basis for further assessment of the combustion system behavior under parameter variations (Figure 6). This enables the individual study of, for example, the influence of start of injection and injection pressure on NO and soot formation trends. Figure 7 shows representative results of the relative NO and soot formation trends for a CR pressure variation. NO Soot 1200 [bar] 1400 [bar] 1200 [bar] 1400 [bar] Figure 7: Calculated NO and soot formation trends; 1000 1/min, 75% load; 5
Exhaust gas return and injection rate optimization are, besides a number of other options, two possible strategies for targeting towards reduced NO formation. The sensitivity of the adopted models to reflect the impact of small changes of the injection rate and EGR rate on the NO formation trends is demonstrated in Figure 8. The comparison of the calculated with the corresponding measured data proves the applicability of the present method to the relevant injection / combustion system development targets. NO NO 2,0 2,0 Cam A Cam B Cam C 0% EGR 3% EGR 10% EGR Figure 8: Calculated NO formation trends; 1000 1/min, 75% load; ( injection rate variation, ( EGR rate variation SUMMARY AND CONCLUSION The present status in the applicability of advanced models for the simulation of cavitating nozzle flow, spray propagation, combustion and pollutant formation in DI diesel engines has been described. The individual model capabilities and accuracy have been demonstrated via comparison of calculated results with the corresponding experimental data. The application of the full set of models to the simulation of spray combustion and pollutant formation in the AVL research DI diesel engine for different operating conditions demonstrates that the overall characteristic features and trends of the governing processes are well reproduced. The mixture formation, combustion and pollutant formation models have proved to be sensitive to variations in engine operating conditions, such as engine speed and load, start of injection, injection rate and amount of exhaust gas return. Comparisons of the calculated results with measured data show good overall agreement for the cylinder pressure traces and pollutant formation trends. ACKNOWLEDGEMENTS Parts of this work have been funded by the German Automotive Research Association (FVV) and the Austrian Research Fund (FFF). REFERENCES [1] Drew, D.A., Ann. Rev. Fluid Mech., Vol. 15, pp. 261, 1983 [2] Drew, D.A., Passman, S.L., Theory of Multicomponent Fluids, Springer, New York, 1998 [3] Alajbegovic, A., Grogger, H.A., Philipp, H., of Transient Cavitation in Nozzle Using the Two-fluid Model, ILASS-99, Indianapolis, 1999 [4] Lee, S.J., Lahey, R.T., Jr., Jones, O.C., Jr., Japanese J. Multiphase Flow, 3, 335, 1989 [5] Dukowicz, J.K., A Particle-Fluid Numerical Model for Liquid Sprays, Journal of Computational Physics, Vol. 35, pp. 229-253, 1980. 6
[6] Gosman A.D. and Ioannides, E., Aspects of Computer Simulation of Liquid-Fueled Combusters, J. Energy, 7, pp. 482-490, 1983 [7] O Rourke, P.J., Modeling of Drop Interaction in Thick Sprays and a Comparison with Experiments, IMechE - Stratified Charge Automotive Engines Conference, 1980 [8] Dukowicz, J.K., Quasi-steady Droplet Phase Change in the Presence of Convection, Los Alamos Report LA-7997-MS, 1979 [9] Naber, J.D., Reitz, R.D., Modeling Engine Spray / Wall Impingement, SAE 880107, 1988 [10] Liu, A.B. and Reitz, R.D., Modeling the Effects of Drop Drag and Breakup on Fuel Sprays, SAE 930072 [11] v. Kuensberg Sarre, Ch., Tatschl, R., Spray Modelling / Atomisation Current Status of Break-Up Models, Turbulent Combustion of Gases and Liquids - Leading Edge Technologies, ImechE Seminar, Lincoln, 1998 [12] Habchi, C., Baritaud, T. et al., Modeling Atomization and Breakup in High-Pressure Diesel Sprays, SAE 970881 [13] Tatschl, R., Kuensberg Sarre, Ch., Alajbegovic, A, Winklhofer, E., Diesel Spray Break-Up Modeling Including Multidimensional Cavitating Nozzle Flow Effects, Proceedings: ILASS Europe 2000, Darmstadt [13] Halstead, M.P., Kirsch, L.J., Prothero, A. and Quinn, C.P., A Mathematical Model for Hydrocarbon Autoignition at High Pressures, Proc. R. Soc. Lond. A., 364, pp.515-538, 1975 [14] Theobald, M.A., Cheng, W.K., A Numerical Study on Diesel Ignition, ASME Energy-Source Technology Conference and Exhibition, Dallas, Texas, 1987 [15] Tatschl, R., Pachler, K., Winklhofer, E., A Comprehensive DI Diesel Combustion Model for Multidimensional Engine Simulation, COMODIA 98, pp. 141-148, Kyoto, 1998 [16] Tatschl, R., Riediger, H., Bogensperger, M., Multidimensional Simulation of Spray Combustion and Pollutant Formation in a Medium Speed Marine Diesel Engine, 1998 FISITA World Automotive Congress, Paris, 1998 [17] Song-Charng Kong, Zhiyu Han and Rolf D. Reitz, The Development and Application of a Diesel Ignition and Combustion Model for Multidimensional Engine Simulation, SAE 950278 [18] Heywood, J.B., Internal Combustion Engine Fundamentals, McGraw - Hill Book Company, 1988 [19] Kennedy, I.M., Kollmann, W., Chen, J.Y., A Model for Soot Formation in a Laminar Diffusion Flame, Comb. and Flame, 81, pp.73-85, 1990 [20] Hiroyasu, H. and Kadota, T., Models for Combustion and Formation of Nitric Oxide and Soot in DI Diesel Engines, SAE 760129, 1976 [21] Nagle, J., Strickland-Constable, R.F., Oxidation of Carbon between 1000-2000 o C, Proc. of the Fifth Conference on Carbon, Pergamon Press, p.154, New York, 1962 7