THE USE OF Φ-T MAPS FOR SOOT PREDICTION IN ENGINE MODELING

Transcription

1 THE USE OF ΦT MAPS FOR SOOT PREDICTION IN ENGINE MODELING Arturo de Risi, Teresa Donateo, Domenico Laforgia Università di Lecce Dipartimento di Ingegneria dell Innovazione, 731 via Arnesano, Lecce Italy INTRODUCTION The simulation of direct injection diesel engines requires accurate models to predict spray evolution and combustion processes. Several models have been proposed for traditional injection strategies characterized by single injection pulse close to top dead center and no EGR. Unfortunately, these models show some limits when applied to different injection strategies and EGR so that a correct simulation of engine performance and emissions cannot be achieved without changing variables included in spray and combustion models. The aim of the present investigation is to check the prediction capability of the modified version of the KIVA3V code developed at the CREA Research Center of Lecce Italy [1]. This modified version of the code was obtained by eliminating the hypotheses of constant fuel density and constant spray angle and by using an improved version of the Shell model. The modified version of the shell model, including two more radicals and three new chemical steps, has been applied to simulate the ignition process of both homogeneous mixture and diesel engine. The results of previous investigations showed that the new version of the KIVA code is also able to follow the changing in injection strategy and EGR as far as ignition delay and NOx emissions are concerned. However both the original and the modified versions of the KIVA3V code were found unable to predict the strong increase of soot level and the following drop when high values of EGR are used. The present investigation focuses on the improvements made to the combustion and soot emission submodels in order to advance the prediction of the pressure traces and soot trend when increasing EGR up to 57%. Soot emissions were calculated by using ΦT maps obtained with the CHEMKIN code. EXPERIMENTAL DATA In the present investigation, experimental data obtained with high EGR levels for a singlecylinder diesel engine were used to test the code. Engine specifications are reported in the Table 1. Bore [mm] 84. Stroke [mm] Connecting rod lenght [mm] 136. Compression Ratio 17. Intake valve closing (start of simulation) [ BTDC] 148 Injection system Common Rail Hole diameter [µm] 16 Number of holes 6 Spray con angle 16 Table 1 Engine specification The available experimental data refers to the operating condition of Table 2 and include pressure traces and emissions levels. 1

2 Engine speed [rpm] BMEP [bar] IMEP [bar] Injection Pressure [bar] Table 2 Engine operating condition In the experiments fuel was injected with a standard single pulse strategy and the injection timing was adjusted to take into account the slower combustion rate. In particular, the timing was changed for each EGR level so that the 5% of the injected fuel burned at the same crank angle (ALPHA) as shown in Fig. 1. burned mass EGR LEVEL Fig. 1 Crank angle corresponding to 5% of burned mass (experimental values) The corresponding operating conditions and measured emissions are reported in Table 3. % EGR Injection timing [CA ATDC] NOx [g/kgf] Soot [g/kgf] Table 3 Experimental data NUMERICAL MODELS FOR COMBUSTION AND SOOT The present investigation focuses on the submodels for combustion and soot emissions in order to improve the prediction of the pressure traces and soot trend when increasing EGR up to 57%. The combustion phase was simulated with the laminar and turbulent characteristic times combustion model. In the laminarandturbulent characteristictime combustion model the local time rate of change of the partial density of species, m, due to conversion from one chemical species to another, is given by: d Ym Y Y dt τ * = m m c where Y m is the mass fraction of species m, Y* m is the local and instantaneous thermodynamic equilibrium value of the mass fraction, and τ c is the characteristic time to achieve equilibrium. ( 1) 2

3 The characteristic time is the sum of a laminar and a turbulent timescale, τ c = τl + f τt The laminar timescale τ l is derived from an Arrhenius type reaction rate (Bergeron and Hallett, [3]) with the preexponential constant A=7.68x1 8 and the activation energy E=77.3 kj/mol : τ l = A [ C H ] [ O ] exp( E/ RT) ( 3) The turbulent characteristic time is assumed proportional to the eddies turnover time: ( 2) k τ t = cm2 ε where C 2 =.1 (Kong et al., [2]), and k and ε are calculated from the modified RNG k ε turbulence model. ( 4) The delay coefficient, f, simulates the increasing influence of turbulence on combustion after ignition has occurred, r 1 e f =.632 and r is the ratio of the mass of products to that of total reactive species: r = ( YCO Y CO2) + ( YH O Y H2O ) + ( YCO Y CO ) + ( YH Y H2) Y N2 ( 5) ( 6) For soot prediction, two different approaches were considered. The first one consisted in using the Hiroyasu formation model [4] and the Nagle and StricklandConstable oxidation model [5]. The second approach was the use of ΦT maps. Such maps were obtained with CHEMKIN by considering a detailed scheme for dodecane oxidation [6]. A set of homogeneous reacting systems, characterized by seven values of the equivalence ratio Φ and six of the mixture initial temperature T, was simulated. The results of CHEMKIN simulations were used to predict the percentage of mixture converted to soot for each value of Φ and T. This percentage was assumed to be the conversion factor in the production of soot as shown in Fig.2. In this way the map indicate not only the regions of the ΦT plan where emissions are formed but also how much soot is formed in those regions φ Temperature [K] Conversion factor 2.E32.2E3 1.8E32.E3 1.6E31.8E3 1.4E31.6E3 1.2E31.4E3 1.E31.2E3 8.E41.E3 6.E48.E4 4.E46.E4 2.E44.E4.E+2.E4 Fig. 2 Sooting maps obtained with CHEMKIN 3

4 USE OF THE ΦT MAP The ΦT map obtained with CHEMKIN was used to predict the soot emission obtained from a diesel engine with the following procedure: 1) Each case of Table 3 was simulated with the KIVA3V code; 2) During the simulation, the mass of mixture at each value of Φ and T was recorded; 3) At the end of each simulation, the total mass of mixture for all Φ and T values was calculated; 4) For each evaluated Φ and T combination, the local mass of soot was calculated by multiplying the mass mixture with the corresponding conversion factor; 5) The total mass of soot was obtained by summing the soot mass over the whole map. The results of the use of this approach were compared with the soot levels predicted with the Hiroyasu formation model and the Nagle and StricklandConstable oxidation model. The soot trends predicted with the two approach are shown in Fig. 3 together with experimental data. Note that in the case of low EGR the two approaches give similar results. But, when increasing EGR over 44% the standard soot model fails to predict the soot trend while the method based on the ΦT maps shows a good agreement with the experimental data. Fig. 3 Calculated and experimental results of soot emissions with high levels of EGR Note that the use of the Φ and T map to predict soot emissions is possible only when the local distribution of the equivalence ratio and temperature is correctly simulated. For this reason it is mandatory to improve the capability of the combustion models to correctly predict the combustion phenomena in diesel engines when using non standard injection strategies and high EGR levels. In the present investigation, the laminar characteristic time has been adjusted according to the EGR level to match the experimental pressure traces reported in Fig. 4. 4

5 NO EGR 27% EGR % EGR 44% EGR % EGR 54% EGR Cylinder Pressure (bar) Crank Angle (deg) Cylinder Pressure (bar) Crank Angle (deg) 56% EGR 57% EGR numerical simulations experimental data Fig. 4 Experimental and numerical pressure traces 5

6 An adjustment of the laminar characteristic time with EGR levels was also used by Xin et al. [6] to take into account the effects of residual gas. However, in the cases considered in the present investigation a linear correlation between the laminar timescale and the EGR level could not be found, therefore the correction factor reported in Fig. 5 has been introduced. This means that the laminar characteristic time shows a linear dependency from the EGR levels lower than 44% and greater than 52%. However, it has to be noticed that the slope of the linear fitting is very different for the two regions. For EGR levels in the range between 44% and 52% the correction factor was to be reduced from.2 to Correction factor CF = xEGR.2 CF =.12.2xEGR % 1% 2% 3% 4% 5% 6% EGR level Fig. 5 Correction factor for the laminar characteristic time parameter (A) CONCLUSIONS The present investigation focuses on the submodels for combustion and soot emissions in order to improve the prediction of the pressure traces and soot trend when increasing EGR up to 57%. The comparison with experimental data available for a single cylinder direct injection diesel engine revealed that: The ΦT map can be useful to predict soot emissions obtained in the case high levels of EGR when combustion is correctly simulated; To take into account the slower combustion rate due to EGR a correction to the laminar characteristic time was made; A better understanding of the combustion processes in the case of high EGR is needed to improve the combustion model of the KIVA3V code. REFERENCES 1. de Risi,A., Donateo, T., Laforgia, D., (24), CFD Modeling of Pilot Injection and EGR in DI Diesel Engines, ICEF24837, Fall Technical Conference Long Beach, CA, USA, October 2427, 24; 2. Kong, S.C., Han, Z., Reitz, R.D., (1995), The Development and Application of a Diesel Ignition and Combustion Model for Multidimensional Engine Simulation, SAE paper 95278; 3. Bergeron, C.A. and Hallett, W.L.H. (1989) Ignition Characteristics of Liquid Hydrocarbon Fuels as Single Droplets, Canadian J. of Chem. Engng, 67, , Hiroyasu, H., Nishida, K., (1989), Simplified Three Dimensional Modeling of Mixture Formation and Combustion in a D.I. Diesel Engine, SAE Paper 89269; 5. Nagle, J., StricklandConstable, R. F., (1962), Oxidation of Carbon between 12 C, Proc. of the Fifth Carbon Conference, Vol. 1, Pergammon Press; 6. Nordin N., Golovitchev, V.I., (1997) Numerical Evaluation of nheptane Spray Combustion at Diesellike Conditions, KIVA Users Workshop 97223; 7. Xin, Montgomery, Han and Reitz (1997) Multidimensional Modeling for a SixMode Emissions Test Cycle on a DI Diesel Engine, Journal of Engineering for Gas Turbine and Power; 8. Akihama, Takatori, Inagaki, Sasaki, Dean, (21), Mechanism of the Smokeless Rich Diesel Combustion by Reducing Temperature, SAE PAPER Kitamura, Senda, Fujimoto, (24), Mechanism of smokeless diesel combustion with oxigenate fuels base on the dependence of the equivalence ratio and temperature on soot particle formation, Int. J. Engine Research, Vol. 3 No. 4, pp