《Multi-Zone DI Diesel Spray Combustion Model for Thermodynamic Simulation of Engine with PCCI and High EGR Level.pdf》由会员分享，可在线阅读，更多相关《Multi-Zone DI Diesel Spray Combustion Model for Thermodynamic Simulation of Engine with PCCI and High EGR Level.pdf（21页珍藏版）》请在三一文库上搜索。
1、 1 2009-01-1956 Multi-Zone DI Diesel Spray Combustion Model for Thermodynamic Simulation of Engine with PCCI and High EGR Level A.S. Kuleshov Bauman Moscow State Technical University Copyright 2006 SAE International ABSTRACT A multi-zone, direct-injection (DI) diesel combustion model, the so-called
2、RK-model, has been developed and implemented in a full cycle simulation of a turbocharged engine. The combustion model takes into account: transient evolution of fuel sprays, interaction of sprays with swirl and walls, evolution of near-wall flow formed after spray-wall impingement depending on impi
3、ngement angle and local swirl velocity, interaction of Near-Wall Flows (NWF) formed by adjacent sprays, influence of temperatures of gas and walls in the zones on evaporation rate. In the model the fuel spray is split into a number of specific zones with different evaporation conditions including zo
4、ne on the cylinder liner and on the cylinder head. The piston bowl is assumed to be a body of revolution with arbitrary shape. The combustion model supports central and non-central injector as well as the side injection system. NOx formation model uses Detail Kinetic Mechanism (199 reactions with 33
5、 species). Soot formation model is phenomenological. The general equation for prediction of ignition delay period was derived as for conventional engines as for engines with PCCI where pilot injection timing achieved 130 CA deg. before TDC. The model has been validated by experimental data obtained
6、from high-speed, medium- speed and low-speed engines over the whole operating range; a good agreement has been achieved without recalibration of the model for different operating modes. General equations for prediction of spray tip penetration, spray angle and ignition delay for low temperature comb
7、ustion and high temperature combustion were derived and validated with the published data obtained for different diesels including diesels with multiple injection system and injection timing varied from very early up to after the TDC. To make a computational optimization of multiple injection strate
8、gy possible, the full cycle thermodynamic engine simulation software DIESEL-RK has been supplied with library of nonlinear optimization procedures. INTRODUCTION The combustion process as well as NOx and PM emissions formation in direct-injection diesels are very sensitive to engine parameters such a
9、s the sprayer design and location, the piston bowl shape, swirl intensity, strategy of EGR use, and injection profile including multiple electronic controlled injection. One area that still holds potential for lowering NOx emissions within the engine is the modification of the combustion process so
10、that more of the combustion occurs under lean conditions, which reduces local combustion temperatures and NOx formation. Lowering combustion temperature allows simultaneously reduce NOx and soot formation. One method of achieving overall lean combustion is Premixed Charge Compression Ignition (PCCI)
11、 or Homogeneous Charge Compression Ignition (HCCI) where the entire fuel and air charge is premixed prior to the start of combustion. PCCI/HCCI has been the focus of numerous researches due to its potential for extremely low NOx without increases in PM or fuel consumption. Due to the high rate of pr
12、essure rise and problems with controlled start of combustion there are challenges of use the PCCI/HCCI operation over the full engine operating range. So at computational analysis of an engine the mathematical model and computer code have to support both the conventional engine combustion and PCCI/H
13、CCI combustion without discontinuity to make it possible the computer optimization of engine control parameters over the whole engine operating range. Due to a great number of available options and their combinations, a search for the optimal solution is a long and labor-consuming task. Simulation a
14、s well as computational optimization can effectively to point out trends and effective ways of engine improvement. There are nowadays three categories of diesel engine simulation models: 2 Zero-dimensional single-zone models; Quasi-dimensional multi-zone models; Multi-dimensional models (CFD). Altho
15、ugh CFD possesses, in principle, the ability to simulate all fluid-flow/heat-transfer/chemical-reaction phenomena, in practice it is limited by: a) the limited capability of current computers to handle small-scale phenomena in an acceptably short time; b) the inadequacy of current scientific knowled
16、ge about turbulence, two-phase flow and chemical reaction, especially when all three are simultaneously presented. So, the attempt to optimize diesel-engine processes by way of pure CFD cannot be succeeded at present. The choice of the thermodynamic engine model together with quasi-dimensional multi
17、-zone model of combustion accounting peculiarities of conventional and PCCI diesel combustion is conditioned by requirements of high accuracy and the high computational speed, because the number of engine simulation sessions in optimization over the whole operating range comes up to few thousands. T
18、he earlier published multi-zone diesel spray combustion model 1, 2, 3, named the RK-model includes three independent emission submodels: the NOx formation submodel based on the Zeldovichs scheme 4 developed by Zvonov 5; the soot formation submodel developed by Razleytsev 6; the advanced NOx formatio
19、n submodel based on Detail Kinetic Mechanism (DKM) for correct prediction of NO emission in engine with large EGR, multiple injection and PCCI/HCCI. DKM includes 199 reactions with 33 species, one was built on the kinetic scheme by prof. Basevich V.J. 5, 7, 8, 9. Soot and NOx formation processes are
20、 simulated with separate procedures after combustion modeling. The main equations of the RK-model were derived by Razleytsev in 1990-1994. This method afterwards was modified and complemented by Kuleshov 1, 2, 3. The RK-model takes into account conditions of evolution of each fuel spray and near-wal
21、l flows generated by sprays; and also interaction between sprays and swirl, as well as between near-wall flows, formed by adjacent sprays. These features of the RK-model allow prediction of diesel combustion and emissions over the whole operating range for the following varying parameters: piston bo
22、wl shape and sprayer location; swirl intensity; number, diameter and direction of sprayer nozzles; injection profile shape and any multiple injection. A more detailed description of the RK-model was given in 1, 2, 3, 53. The present paper describes the same model with peculiarities being implemented
23、 to improve accuracy of heat release prediction in diesels with PCCI. GENERAL PRINCIPLES OF SPRAY MODELLING The theoretical background of the model is based on the method developed in 6 in which the fuel spray injected into the combustion chamber of the engine is split into a number of specific zone
24、s. This is caused by the necessity of detailed account for following fuel spray evolution peculiarities: difference of fuel droplet evaporation conditions in different spray zones; fuel redistribution among zones in the process of free spray movement and in its interaction with a wall; interaction o
25、f fuel sprays with the walls of the piston bowl and possibility of fuel hitting the cylinder liner and the cylinder head surface; effect of the wall temperature on the fuel evaporation rate in near-wall zones; interaction of adjacent sprays in near-wall flows. The spray evolution passes through thre
26、e stages: Initial formation of dense axial flow. Main stage of cumulative spray evolution. Period of spray interaction with the combustion chamber walls and fuel distribution on the walls. Liquid spray breakup takes place close to the nozzle. High-speed fuel portions move quickly towards the spray t
27、ip pushing apart, impacting and thus coalescing with droplets formed earlier. In the spray cross section the distribution density of droplets and their diameter reduce rapidly with growing distance from the spray axis. In this context, near the jet boundary droplets are slowed down more rapidly than
28、 those near the axis and are gradually delayed and separated from them. At the main evolution stage later on, the axial flow is slowed down and concentrated on the forward front side because of the surrounding gas resistance. New flying fuel portions reach the axial flow, penetrate inside, push and
29、condense it at the rear. This results in an extended axial core with increased density and droplet velocity being formed in the middle of the spray 10. This core is surrounded by a relatively dilute outer sleeve made of delayed droplets. Corresponded scheme of diesel fuel spray is presented in Fig.
30、1. Figure 1. Scheme of diesel fuel spray, concentration of fuel droplets and its tracks. The border between the initial and main stages of spray evolution corresponds to the moment when the axial flow close to the spray tip starts to deform and break up, forming a condensed mushroom-shaped forward f
31、ront. As the spray moves on, constant breakup of the spray forward part 11 takes place and the front is renewed by new flying fuel portions 12, 13. The delayed droplets move from the breaking front to the environment. The moving spray carries the surrounding gas with it, the gas 3 velocity in the en
32、vironment being rather low. Meanwhile gas in the axial core is rapidly accelerated to the velocity close to that of droplets 14. The core diameter in the cross section is about 0.3 of the spray outside diameter. In accordance with 6, the current position and the velocity of an Elementary Fuel Mass (
33、EFM) injected during small time-step and moving from the injector to the spray tip are related as mo l l U U = 1 2 3 (1) where: l is the current distance between the injectors nozzle and the EFM; k ddlU/= is the current velocity of the EFM; k is the travel time for the EFM to reach a distance l from
34、 the injectors nozzle; Uo is the initial velocity of the EFM at the nozzle of the injector and lm is the EFMs penetration distance. As an illustration, Fig. 2 presents the variation of spray evolution parameters l, lm , U and Um as functions of time for a medium speed diesel engine. a) b) Figure 2.
35、The simplified sketch of the spray (a) and variations of spray evolution parameters: l, lm, U and Um as functions of time (b). Upon termination of fuel evolution there is still some fuel in its axial core supplied at the final injection stage. At the initial combustion stage the flame is unable to b
36、reak up the fuel spray condensed core 15, 16, 6 which explains why the sprays continue to move to the combustion chamber sidewalls during injection even after fuel ignition. By the end of fuel supply a considerable fraction of the fuel cycle portion is accumulated near the walls. This takes place bo
37、th in diesels with compact combustion chambers and in engines with wide piston bowls (Hesselman). The interaction of fuel sprays with chamber walls was studied in numerous literature sources 17-24. Having analyzed various data, Razleytsev proposed the following model of fuel spray interaction with a
38、 wall. On reaching the wall, the spray is spread over its surface in every direction. The upward flow over the wall gets quickly into a clearance between the piston and cylinder head and under constricted conditions spreads along the piston crown as well as the cylinder head surface (Fig. 3). A part
39、 of fuel can reach the cylinder liner. Analysis of experimental data has shown that characteristics of flows moving along the wall in different directions are similar to those obtained for a freely moving spray, but the velocity level is lower and depends on the flow direction. Reduction in the flow
40、 velocity along the wall is caused by the hydrodynamic resistance of the nearby wall. . Figure 3. Photo-record obtained by Koptev, Gavrilov 25, Plotnikov. Fuel was injected into the bomb with the piston model. Nozzles: 7 x 0.4 mm. Speed of shooting is 3700 frame/sec. The numbers correspond to frames
41、 of shooting. The mentioned similarity in evolution between near-wall flows and free sprays provides the basis for the application of the same computational procedures to near-wall flows and free sprays. The assumption that the velocity of an elementary fuel portion flying along the wall 4 is analog
42、ous to the fuel outflow velocity from a sprayer makes it possible to apply conventional criterion relations for simulation of the flows in the near-wall zone. As the spray hits the wall, the forward front fuel enters the near-wall flow zone. The spray trajectory and, therefore, the time, the place a
43、nd wall impingement angle are determined in view of the swirl effect. The process of spray interaction with a wall is rather complicated. The following scheme of spray and NWF evolution is proposed. During the spray forward front impingement with a wall a conical condensed gas-fuel layer is being fo
44、rmed there within the borders of a stain formed at the intersection of a conical spray with the wall surface (zone 4 in the fig. 10). On fast fuel spray front impingement with the wall the fuel spreads out of the initial stain limits. The high-speed axial flow of a spray in its hitting the wall thic
45、kens the near-wall layer, draws apart its borders and a part of the flow moves to its periphery above this layer. The shape of the near-wall stain and its spreading rate in various directions depend on the spray-with- the wall impingement angle and the air swirl effect. A typical photo-record of the
46、 evolution of a spray in the combustion chamber is shown in Fig.3. The partial solution of the differential equation 1 is: 0113 0 333 . 0 = k m m U l l l (2) wherek is the time of EFM movement from the nozzle up to l. When EFM is stopped in a spray tip l= lm, k = m and lm = Uo m / 3. (3) From equati
47、ons 1, 2, 3 it follows: ()2 0 1 mk UU= ; (4) () 3 11 mkm ll= . (5) Parameters of the spray tip are calculated with empirical equations of Lyshevsky 26. These relations use dimensionless criteria: ffnmd UWe 2 0 =; (6) () fnff dOhM 22 = ; (7) () 32 nffs d= ; (8) fair = (9) where: U0m is the average in
48、jection velocity, dn is the nozzle hole diameter, f is the fuel density, air is the air density, f is the fuel surface density, f is the dynamic viscosity coefficient of fuel, s is the current time from the injection beginning. Evolution of a free spray consists of two main phases: a) initial phase of pulsing evolution; b) basic phase of cumulative evolution. The border between these phases is marked as lg (length) and g (time): ;MWedCl . nsg 6040250 = (10) ;B/l sgg 2 = (11) ();D/MWeUdB s mns 2 160210 0 = (12) where Cs = 8.258.85; Ds = 4.55 for di