Scholarly article on topic 'Evaluation of the Mixture Formation Process of High Performance Engine with a Combined Experimental and Numerical Methodology'

Evaluation of the Mixture Formation Process of High Performance Engine with a Combined Experimental and Numerical Methodology Academic research paper on "Mechanical engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Energy Procedia
OECD Field of science
Keywords
{"High Performance Engine" / Injection / "Mixture Formation"}

Abstract of research paper on Mechanical engineering, author of scientific article — Claudio Forte, Gian Marco Bianchi, Enrico Corti, Buono Michele, Fantoni Stefano

Abstract The combustion process of a Port Fuel Injection (PFI) engine is deeply influenced by the mixture formation process. The needs of reducing engine emission and fuel consumption push the engine manufactures to implement new advanced experimental and numerical techniques to better control the mixture quality. The local mixture air-index at the spark plug is closely related to combustion instabilities and the Indicated Mean Effective Pressure (IMEP) Coefficient of Variation (CoV) well correlates with the variability of the flame kernel development. The control of air to fuel ratio is especially critical for high performance engines: due to the low stroke-to-bore ratio the maximum power is reached at very high regimes, letting little time to the fuel to evaporate and mixing with air. The injector located upstream the throttle causes a lot of fuel to impinge the throttle and intake duct walls, slowing down the dynamics of mixture formation under part load conditions. The aim of the paper is to present a multi-cycle methodology for the simulation of the injection and the mixture formation processes of high performance PFI engine, in order to evaluate both the quality of combustion and the fuel dynamics in the intake duct. The phenomena involved in the process are highly heterogeneous, and particular care must be taken to the choice of CFD models and their validation. In the present work all the main models involved in the simulations are validated against experimental tests available in the literature, selected based on the similarity of physical conditions of those of the engine configuration under analysis. The lagrangian spray is initialized with a semi-empirical methodology, based on available experimental data, and its interaction with the wall is simulated by means of the Kuhnke model, so to take into account the high wall temperature of the intake valves during impingement. The dynamics of the wall film are accurately represented by the activation of momentum equation of wall film and a validation of its dynamics is accomplished against proper test cases. The methodology is applied in two different engine configurations with separate objectives: a part load condition, where the dynamics of fuel in the intake duct is crucial to ensure drivability, a full load configuration, for the evaluation of mixture quality and combustion stability.

Academic research paper on topic "Evaluation of the Mixture Formation Process of High Performance Engine with a Combined Experimental and Numerical Methodology"

Available online at www.sciencedirect.com

ScienceDirect

Energy Procedia 45 (2014) 869 - 878

68th Conference of the Italian Thermal Machines Engineering Association, ATI2013

Evaluation of the mixture formation process of high performance engine with a combined experimental and numerical methodology

Claudio Fortea*, Gian Marco Bianchia, Enrico Cortia, Buono Micheleb, Fantoni Stefanob

aDIN - Department of Industrial Engineering - University of Bologna, Via del Risorgimento 2, 40136 Bologna, Italy bDucati Motor Holding SpA, Via Cavalieri Ducati 3,40132 Bologna, Italy

Abstract

The combustion process of a Port Fuel Injection (PFI) engine is deeply influenced by the mixture formation process. The needs of reducing engine emission and fuel consumption push the engine manufactures to implement new advanced experimental and numerical techniques to better control the mixture quality. The local mixture air-index at the spark plug is closely related to combustion instabilities and the Indicated Mean Effective Pressure (IMEP) Coefficient of Variation (CoV) well correlates with the variability of the flame kernel development. The control of air to fuel ratio is especially critical for high performance engines: due to the low stroke-to-bore ratio the maximum power is reached at very high regimes, letting little time to the fuel to evaporate and mixing with air. The injector located upstream the throttle causes a lot of fuel to impinge the throttle and intake duct walls, slowing down the dynamics of mixture formation under part load conditions.

The aim of the paper is to present a multi-cycle methodology for the simulation of the injection and the mixture formation processes of high performance PFI engine, in order to evaluate both the quality of combustion and the fuel dynamics in the intake duct.

The phenomena involved in the process are highly heterogeneous, and particular care must be taken to the choice of CFD models and their validation. In the present work all the main models involved in the simulations are validated against experimental tests available in the literature, selected based on the similarity of physical conditions of those of the engine configuration under analysis. The lagrangian spray is initialized with a semi-empirical methodology, based on available experimental data, and its interaction with the wall is simulated by means of the Kuhnke model, so to take into account the high wall temperature of the intake valves during impingement. The dynamics of the wall film are accurately represented by the activation of momentum equation of wall film and a validation of its dynamics is accomplished against proper test cases.

The methodology is applied in two different engine configurations with separate objectives: a part load condition, where the dynamics of fuel in the intake duct is crucial to ensure drivability, a full load configuration, for the evaluation of mixture quality and combustion stability.

© 2013TheAuthors.Publishedby ElsevierLtd. Selectionandpeer-reviewunderresponsibility ofATINAZIONALE

Keywords: High Performance Engine; Injection ; Mixture Formation

* Corresponding author. Tel.: +39-051-2093-314 ; fax: +39-051-2093-313. E-mail address: claudio.forte@unibo.it

1876-6102 © 2013 The Authors. Published by Elsevier Ltd. Selection and peer-review under responsibility of ATI NAZIONALE doi: 10.1016/j.egypro.2014.01.092

1. Introduction

The mixture composition heavily influences the combustion process of PFI engines: the local composition at the spark plug is closely related to the combustion instabilities. The cycle-by-cycle variability of the indicated work well correlates with the variability of the early stages of combustion. The relationship between the variability of IMEP and the fluctuation of local lambda was clearly showed in the work of Ikeda et al. [1] where the chemiluminescence technique was used to analyze the mixture homogeneity and composition at the spark plug for a high performance engine. Unfortunately, Ikeda [1] could not draw conclusions on the cause of such mixture variability. The mandatory requirements for improving the combustion efficiency in racing applications has been fulfilled by operating the engine with high Air to Fuel Ratios. In a previous work [2], the authors investigated the root causes of the cycle-by-cycle variability increase with leaner combustion, by means of a joint numerical and experimental approach: the authors showed that the combustion sensitivity to the initial perturbation of the mixture air index at spark location and to the level of in-cylinder air index homogeneity increases, due to the lower laminar combustion speed of leaner mixtures. The authors concluded that efficient mixing processes are mandatory any time the engine operates with suboptimal air indexes (far from those giving the maximum laminar speed).

The effect of injection targeting and timing of a PFI engine has been investigated by McGee et al.[3] . The authors found that the open valve injection strategy minimized the A/F excursions and caused lower film mass modeled by the X-tau model. On the contrary, the closed valve injection strategy resulted in lower hydrocarbon emissions and lower cycle-by-cycle variability of IMEP thanks to the better homogeneity of fuel distribution. A combined experimental and numerical investigation of injection parameters for a PFI motorcycle application has been performed by Kato et al. [4] . A good correlation has been found between the mixture homogeneity in the chamber of the CFD results and the variability of IMEP of the engine. Such a combined experimental and numerical methodology for the evaluation of combustion stability allowed a good reconstruction of the main phenomena involved in the mixture formation process: it is noticeable that at 4000 rpm some configurations showed a non-convergence condition of the wall film, with continuum accumulation, thus suggesting the needs of a multi-cycle methodology for the high speed range.

Henriot et al. [5] used an improved version of KIVA2 code for the evaluation of the mixture formation process of a F1 racing engine. They showed the importance of a multi-cycle simulation for the convergence of fuel dynamics in the intake duct and compared the results of combustion simulation for two different injection cases with experimental pressure traces in the chamber.

The aim of the paper is to present a CFD methodology for the evaluation of the mixture dynamics and quality by means of a multi-cycle 3D engine simulation. The accuracy of results are evaluated againts experimental data of the engine on the test bench and all the main physical models are validated with respect to test cases selected because in condition similar to those under the analysis. The innovation of the methodology lies in fast response of the multicycle simulations, which is desirable for racing applications, with no loss in accuracy of the mixture information.

The engine under analysis is an high performance 4 cylinder Ducati race engine. The configuration analyized are a part load and a full load maximum power condition. Due to confidential agreement with Ducati all the data in the present paper will be referred to a conventional condition.

2. Descritpion of the simulation flow-chart

An accurate evaluation of the physical and chemical conditions inside the combustion chamber at ignition is still a challenging task in the simulation of SI high performance engines. The need for a correct representation of turbulence can lead to use numerical models (LES, DNS) which require high computational efforts, not yet compatible with industrial applications. Vermorel et al. [6] computed a multi-cycle LES simulation for the evaluation of cyclic variability, simulating nine consecutive engine cycles. The results were interesting in terms of variability of the velocity field, but the engine was supposed to be homogeneous, being fuelled with a gaseous propane-air mixture. As it will be shown, the simulation of the injection process needs higher number of engine cycles to reach a converged solution. The engine under analysis is a Ducati high performance engine, here evaluated in the configurations of Tab. 1

The simulations are performed with FIRE v2011 (AVL-AST) code: the turbulence model used is the standard k-e, with hybrid wall function [7], thus taking into consideration both viscous sub-layer and turbulent log-layer wall

Table 1. Engine configuration

Engine parameter Configuration 1 Configuration 2

Regime > 10000 rpm > 16000rpm

Load Part load Full load

A rich mixture rich mixture

conditions. Fig. 1 shows the main steps for the CFD RANS methodology here proposed. The simulated domain is a half of the total domain thanks to the geometric symmetry.

The methodology consists of three simulation steps:

1. The first step is the simulation of the exhaust stroke. The physical domains activated at this point consist of cylinder and exhaust duct. The initial conditions are taken from 1D simulation of the whole engine, while the mass flow boundary condition is imposed at the exhaust outlet. The aim of this part of the methodology is an accurate representation of the velocity field at the beginning of the overlapping period (i.e. at Intake Valve Opening).i

2. In the second step the cycle-to-cycle convergence of the intake duct gas dynamics is achieved. The intake port domain is always active during this portion of the simulation, while the exhaust and cylinder volumes are activated just as they connect to the intake. The simulation of the second step starts at IVO, after having mapped fluid dynamic results of the first step on cylinder and exhaust domains. In this phase the mesh resolution is fine (1 mm cell size) and four engine cycles are needed for the full load configuration to reach a stationary condition in the intake duct Fig. 2(b), while three engine cycles are enough for the convergence of pressure in part load conditions Fig. 2(a).

Pressure traces at a control section in the intake duct

Pressure traces at a control section - Full Load

Fig. 1. Scheme of the cyclic simulation methodology

Crank Angle [deg]

(a) (b)

Fig. 2. Pressure traces evelution in the intake duct: (a) Part load ; (b) Full load

The simulations are performed in mpi-mode 4-cpus on a modern workstation and the whole engine cycle needs about 48 h to get finished. The activation of spray and wall film models at this step would likely increment the time request for the simulation, with an estimation of 4-6 months of calculation to get a converged solution (i.e. 80 engine cycles).

3. The needs for achieving short times to result of the simulations suggest to use an optimized robust and accurate methodology based on coarser meshes. The results of the second step can be used as fluid dynamic constraints for the simulation. The three dimensional evolution of fluid dynamics physical parameters at the inlet valve curtain area over the fourth engine cycle are then stored in a file and used for the third step of the methodology: the multi-cycle simulation of the injection process. The computational grid resolution is now coarser than the previous, with a total number of 26000 cells. The correct representation of the dynamics is guaranteed by the mapped boundary condition imposed on the valve curtain area and the simulation of each engine cycle can be

360 720

Crank angle [deg]

accomplished in about 3 h. The simulation of the injection process is iterated until a stationary condition of the liquid fuel film is reached in the intake duct (60-80 engine cycles).

The whole methodology, able to reproduce the injection over more than 60 engine cycles, takes about ten to fifteen days to get completed for each configuration analyzed.

3. Injection simulation: description of the CFD models

3.1. Injection setup

The spray is simulated by means of a lagrangian approach (Discrete Particle Method).

The primary break-up models are closely related to the instabilities arising inside the injector, whose design specifications are not available. In the present work it was chosen not to activate the primary break-up models, by using a proper initialization of the spray droplet size based on the injector experimental characterization available.

Many empirical models are proposed in the literature in order to evaluate of the characteristic droplet break-up time over the different regimes basing on the Weber number. In this work the WAVE model [8] is used as secondary break-up model, with the C1 constant set equal to 8. The choice is performed on a physical base according to the experimental correlations proposed in [9] in order to accurately predict the non-dimensional breakup time order of 5.5 over the whole range of droplet Weber numbers. It must be noted the Wave model provides the droplet non-dimensional breakup time at large Weber numbers for inviscid liquid right equal to the C1 constant (also referred to as B1 in the literature). The fuel is considered as a single-component fuel and the Spalding model [10] is used for the droplet heating and evaporation processes.

The simulation of the spray liquid portion with Lagrangian approach suffers from strong grid dependency. Li et al [11] proposed the application of a strategy of AMR (Adaptive Mesh Refinement) for the multi process simulation of injection on KIVA4 code, based on the execution of the algorithm on a single root processor every ten time-steps. The criterion used for mesh refinement is based on a threshold imposed on the sum of liquid mass and fuel vapor in a cell.

The injector used in the engine under analysis is a 6-holes, 30deg outer cone angle. The absolute pressure of injection is 10 bar. A full experimental characterization is given by the injector supplier in terms of droplet size characterization by PDPA technique and Tomography Image analysis. The injection images analysis in a constant pressure vessel allows defining the shape of the spray (cone angles) and the information about mechanical actuation, Fig. 3 shows the comparison between the experimental and simulated spray pattern in the injection configuration under study. The hydraulic delay of the injector is estimated in 1 ms from the electrical start of the injection. The reconstruction of the penetration of the spray is good.

Fig. 3. Spray penetration

The initial guess droplet distribution is based on the droplet size measurements recorded on a section located as near as possible to the injector tip. The primary breakup is deactivated, and the initial droplet size is imposed following

Spray granulometry - Plane 15 mm

0 d Experimental (transverse directio _ _ ■ d Computational (transverse directi □ d Experimental (transverse directio directi

d Computational (transver

_ _ i. - -V • •

□ O--M O O

• * V ~

J_i_i_i_i_i_i_i_L-

-!-!-!-!- ---* : : : .vH - -*- CS""tail

.......:.......:.......:...... - : : :■»••: .......:.......:.......:.....> • .......:.....

: : : .......:.......:....... r •.......:.......:..... : / • : : f m

....................t * ................... / • : .......:....... > •......:.......:.......:..... : : : :

:: i /1 j : c ::

-2 -1.5 -1

1 1.5 2

x 10"'

Fig. 4. Spray granulometry and distribution: (a) Granulometry; (b) Radial mass distribution

Secondary droplet d (traverse line locations)

Fig. 5. Spray wall interaction: Trujillo test case: (a) Secondary droplet granulometry; (b) Splashing cloud

-0.5 0 Positi

a Rosin Rammler probability distribution, based on the averaged values of Algebraic (d10) and Sauter Mean Diameter (d32) experimentally measured on a plane placed 15 mm downstream the injector tip. The balance between statistical representation of the spray and computational efforts in calculation has taken to use 800 parcels/ mg of fuel injected in the system. The results in terms of granulomtery Fig. 4(a) and mass distribution Fig. 4(b) evaluated at a plane of 15mm downstream the injector location are good.

3.2. Spray wall interaction

Because of the complexity in modeling the interaction between an impinging spray and a wetted wall, empirical and phenomenologial models are best used for the simulation of impingement. The model used in this paper is the one proposed by Kuhnke [12], because it takes into account also the effects of the wall temperature. Four different impact regimes are considered depending, on the values of the thermal condition of the impingement and the balance of intertial and viscous strength.

The validation of the models was carried out against the test case of Trujillo et al. [13]. A pintle injector is used in the experimental apparatus with an injection pressure of 275.8 kPa. The injector is mounted on a rotational mount to regulate the angle of impingement.

The results in terms of spray droplet size are plotted in Fig. 5(a). The Kuhnke model exhibit a good prediction in terms of granulometry of the splashed droplets. Fig. 5(b) shows the good reconstruction of the secondary droplets clouds, which means that the velocity of the secondary droplets is good in terms of strength and direction.

Wall film thickness - Sensor 4 Wall film thickness - Sensor 5

Fig. 6. Foucart wall film test case: (a) Experimental setup; (b) Film heigth at sensor 4; (c) Film heigth at sensor 5

3.3. Wall film model

The mixture formation process in PFI engines is strongly influenced by the evolution of wall film in the intake duct. The dynamics of wall film formation and evaporation are different from the spray characteristic time, heavily influencing the control strategy during transients. The wall film model implemented in FIRE v2011 is based on the Eulerian approach under the hypothesis of thin liquid assumption. All the main interactions with gas, wall and spray are modeled in the code.

The evaporation model activated is the combined Sill-Himmelsbach/Diffusion model[14] , which considers both the turbulent evaporation condition and the direct diffusion of fuel. One of the key feature in the wall film model is the solution of the momentum equation. The standard momentum approach in FIRE is based on the velocity profile assumption: as long as the film is thin and no transient forces are taken into account, steady state conditions are reached within fractions of a second. The information of wall film velocity is based on analytical steady state assumptions and the three conservation equations of momentum are not solved. Cazzoli et al. [15] [16] developed a wall film model integrated with a fully explicit method and showed the importance of a complete solution of momentum equation. The University of Bologna started a collaboration with AVL in 2005 [17] for the implementation of a full dynamic wall film momentum equation in FIRE code. In the last version (v2011) it is now possible to solve the complete momentum transport equation by activating the momentum flag. The wall film evaporation process is related to the contact area with the gas, thus a correct representation of the spread of wall film is crucial for accurate simulations of the mixture formation process. The prediction capability of the model has been assessed by simulating the experimental test-case by Le Coz et al. [18], which reproduced injection and liquid-film conditions similar to those occurring in PFI gasoline engines. A pulsed-injection of iso-octane is operated under a constant air flow into a transparent pipe.

The pipe surface is scanned at specific locations and the film thickness is recorded. The results for the stationary and dynamic wall film models are depicted in Fig. 6(b) Fig. 6(c). It is noticeable that the stationary hypothesis causes an over-estimation of the wall film thickness in the impinging zone. The dynamic solution, instead, allows the film to flow downstream the impact zone, with a good representation of the mean value of film accumulation. This dynamic of the liquid film will cause a higher spread of the fuel on the wall, thus increasing the evaporation rate. The peaks of film height clearly visible in the experimental data are related to the pulse caused by the spray in the impinging zone [15] [16] . The actual model version is yet not able to locally reproduce such dynamics, but the overall spread of the film is well reproduced, thus ensuring accurate prediction of the evaporation rate.

4. Simulation of the injection process: results

4.1. Part load condition

Transient operation of engines leads to air fuel (A/F) ratio excursions, thus causing driveability problems. These excursions have been attributed to the formation of fuel films in the intake port, which are caused by a portion of the intake fuel impinging and adhering on the relatively cool port surface. These films act as a source or sink which cause the AF variations depending upon the transient condition. The compensation of fuel accumulation during transients

Lambda Fuel mass accumulation

(a) (b)

Fig. 7. Fuel dynamics at part load condition: (a) Experimental and simluated lambda;(b) Accumlation of fuel in the intake duct

(a) (b)

Fig. 8. Aquino single paddle X e t model: (a) Best fitting of X0 e t 0 with fuel accumulation;(b) Activation of compensator with X0 and t 0

has been fully investigated by several authors. In this work the injection simulation is used for the evaluation of the most widely used X t single puddle model [19].

The evaluation of the mixture dynamics is done on the test bench by imposing a discontinuity in the lambda of the engine. The Ducati engine runs in a stationary condition without injection, then suddenly a fixed mass of fuel is injetcted, corrsponding to the lambda target imposed. The evolution of the lambda signal is recorded at the exhaust of the cylinder and is compared to the results of the injection simulation.

Fig. 7(a)shows the evolution of both simulated and experiment lambda in the chamber: good is the reconstruction of dynamics of mixture formation. The engine takes about four engine cycles to reach an ignitable mixture and the lambda target is reached after forty engine cycles. Fig. 7(b) shows how the accumulation of fuel in the intake duct can be docomposed.

The single puddle model reveals to not be able in reproducing two different dynamics of spray and gas on one side, and liquid wall film on the other side. The best model constant fitting causes an under prediction of fuel accumulation in the early cycles after the step, and an over fuel mass accumulation in the later stages. The best couple of X and tau constants (X0 and Tau0) are then used for the compensation injection strategy on the engine. Fig. 8(b) shows the results in terms of lambda in the chamber for both the simulation and experiment. It is clear that the compensator is not capable in reaching and maintaining the lambda target in the chamber. In the early stages the lack in the modeling of the fast response of spray and fuel gas accumulation causes a richer mixture, that needs almost 40 engine cycles to get converged

4.2. Full load condition

The injection methodology is then used for the evaluation of mixture quality at full load condition with rich mixture target. Three different injection timings are analyzed: a reference condition, with injection laying in the latest stages of intake stroke, and two further timings, with the injection start delayed by 180 deg and 360 deg

For each injection timing almost sixty engine cycles are needed to reach a converged condition: Fig. 9shows how the X-TAU model fits the CFD simulations. Even if the dynamics caught by the CFD simulations are more complex than those representable by means of an X-TAU model, the fitting capacity of this control-oriented model is good. It is noteworthy that the injection phase influences the dynamics and, as a consequence, the estimation of the model constants.

Fig. 9. XeTauFit

A further validation of the methodology is accomplished with the use of the experimental apparatus for the visualization of the spray on the engine running with the analyzed configuration.

A proper apparatus has been designed and setup in order to capture the injection event characteristics under actual running conditions.

The idea behind the developed apparatus is to observe how "actual" injections take place, on the actual engine configuration. To do so, an optically accessible airbox has been used, allowing to see cylinder # 2 injector in action, with the motorcycle running on the rolls dynamometer. An Optronis Camrecord 5000 fast camera has been used to capture the images: the system is able to sample images with 256x256 resolution at a frame rate of 10 kHz, for a maximum duration of 6.4 s. Optical fibers has been used to bring light within the airbox: the light source is a halogen lamp. In order to associate the acquired frames to the crankshaft angular position, an external synchronizer unit has been designed. The synchronizer unit is based on a National Instruments cRIO system, capable of monitoring the digital input lines at 20 MHz, sampling CAN messages at 1 kHz and analog signals up to 1 MHz per channel. All the data can be synchronized and converted from the time domain to the angular domain in real-time, thanks to the FPGA (Field Programmable Gate Array) hardware programmable in LabVIEW. The same HW/SW platform has been previously successfully used in other Rapid Control Prototyping (RCP) and Real-Time analysis applications [20] [21] [22] [23] and it is now part of Ducati standard test cell equipment.

Fig. 10 shows the comparison of the spray evolution for the reference injection timing. The evolution of the spray is well captured by the simulations, thus confirming the initial settings imposed in terms of granulometry and velocity. The sequence of images refers to a complete engine cycle (1 image each 120) and it is interesting to notice that the injected spray is still in the zone above the throttle at the end of the cycle. It is also possible to qualitatively appreciate the interaction between spray and gas dynamics, with a fine pulverization when the droplets encounter the pressure reflection (120 and 240 plots).

The last step of the methodology allows defining quality parameters for the injection process. A crucial factor in high performance engines development is the increase of volumetric efficiency. The latent heat of vaporization of

Fig. 10. Comparison between experimental versus simulated fuel spray

1.01 1.008 >■ 1.006 Ö 1.004

0 1.002

> 0.998

'1 0.996 à

11 0.994 0.992 0.99

Volumetric efficiency

I experimental I computational

REF + 180°

0.02 0.04 0.06 0.0.

(a) (b)

Fig. 11. Full load influence of injection phase: (a) Volumetric efficiency;(b) Standard deviation of IMEP

fuel and the energy exchange between spray, film and wall can lead to different density in the intake duct. Fig. 11(a) compares the estimated and measured volumetric efficiency (normalized by the reference case value). Good agreement is obtained between simulation and experiment for the three considered injection timings.

Finally, the compression stroke simulation of the cylinder is used for the evaluation of mixture quality at ignition. Fig. 11(b) shows that there is a good correlation between the local value of lambda at the spark plug and the standard deviation of the IMEP. As expected, leaner combustions at ignition cause higher variability of combustion. It is noticeable that also the reference case (lowest IMPE variability) has a local lambda leaner than the mean one.

5. Conclusion

A multi-cycle methodology for the simulation of the mixture formation process of an high performance engine is presented. All the main physical phenomena involved are modeled and validated against test cases. The methodology proved to be both fast and accurate, being able to well simulate the fuel accumulation in the intake duct in different engine configurations. The evolution of the spray injection on running engine at very high speed has been showed, by setting up a proper experimental apparatus, estabilishing a further validation of the injection simulation methodology.

REF+360°

REF+180"

REF+360°

loc mear

A part load condition has been simulated to provide information about the capability of the most used compensator model (X t model). The analysis of fuel accumulation highlighted different dynamics in terms of liquid and vapor evolution, thus resulting in the conclusion that a proper multi-puddle model can be more accurate then the single ones.

The influence of the injection phase on the mixture quality at the spark plug at full load condition has been analyzed, showing good correlations with the variability of indicated mean effective pressure. The methodoly allowed reproducing more then 60 engine cycles in about ten days, thus resulting a powerful tool for the mixture formation analysis of industrial applications.

References

[1] Ikeda, Y., Nishiyama, A., Kim, S., Takeuchi, A., Wilnklhofer, E., Baritaud, T.. Cyclic variation of local a/f (lambda) and mixture quality in si engine using local cheminuminescence. 7th International Symposium in Internal Combustion Diagnostics 2006;.

[2] Forte, C., Bianchi, G.M., Corti, E., Fantoni, S.. Combined experimental and numerical analysis of the influence of air-to-fuel ratio on cyclic variation of high performance engines. ASME(ICEF 2008-1668) 2008;.

[3] McGee, J., Curtis, E., Russ, S., Lavoie, G.. The effects of port fuel injection timing and targeting on fuel preparation relative to a pre-vaporized system. SAE Technical Papers SAE 2000-02-2834 2000;.

[4] Kato, S., Hayashida, T., Iida, M.. The influence of port fuel injection on combustion of a small displacement engine for motorcycle. SAE Technical Paper 2007-32-0009 2007;.

[5] Henriot, S., Bouyssounnouse, D., Baritaud, T.. Port fuel injection and combustion simulation of a racing engine. SAE Technical Papers SAE2003-01-1845 2003;.

[6] Vermorel, O., Richard, S., Colin, O., Angelberger, C., Benkenida, A., Veynante, D.. Towards the understanding of cyclic variability in a spark ignited engine using multi-cycle les. Combustion and Flame 2009;156(8):1525-1541.

[7] Popovac, M., Hanjalic, K.. Compound wall treatment for rans computation of complex turbulent flows. In: 3rd M.I.T. Conference on Computational Fluid and Solid Mechanics. 2005, p. 802-806.

[8] Liu, A.B., Mather, D., Reitz, R.D.. Modeling the effects of drop drag and breakup on fuel sprays. SAE Technical Papers SAE Paper 930072 1993;.

[9] Pilch, M., Erdman, C.A.. Use of breakup time data and velocity history data to predict the maximum size of stable fragments for acceleration-induced breakup of a liquid drop. International Journal of Multiphase Flow 1987;13(6):741-757.

[10] Amsden, A.A., O'Rourke, P.J., Butler, T.D.. Kiva-ii: A computer program for chemically reactive flows with sprays. Tech rep, Los Alamos National Laboratory 1989;.

[11] Li, Y., Xue, Q., Kong, S.., Xu, Z., Yi, J., Torres, D.J.. Parallel computing of kiva-4 using adaptive mesh refinement. SAE Technical Papers - SAE 2009-01-0723 2009;:1-10.

[12] Kuhnke, D.. Spray/wall- interaction modelling by dimensionless data analysis. PhD Thesis - Technische Universitat Darmstadt 2004;.

[13] Trujillo, M.F., Mathews, W.S., Lee, C.F., Peters, J.E.. Modelling and experiment of impingement and atomization of a liquid spray on a wall. IntJ Engine Research 1999;1(1).

[14] Cfd solver v2010 - ice physics and chemistry. AVL Manual 2010;Cited By (since 1996):1.

[15] Cazzoli, G., Forte, C.. Development of a model for the wall film formed by impinging spray based on a fully explicit integration method. SAE Technical Paper 2005-24-087 2005;.

[16] Cazzoli, G., Forte, C., Vitali, C., Pelloni, P., Bianchi, G.M.. Modeling of wall film formed by impinging spray using a fully explicit integration method. ASME ICES2005-1063 2005;.

[17] Beretta, C.. Implementation and validation of an improved wall film model for a commercial 3d cfd program. Thesis - Master degreee -University of Bologna 2005;.

[18] Coz, J.F.L., Catalano, C., Baritaud, T.. Application of laser induced fluorescence for measuring the thickness of liquid films on transparent walls. 7th IntSymposium on Application of Laser Techniques to Fluid Mechanics 1994;.

[19] Aquino, C.F.. Transient a/f control characteristics of the 5 liter central fuel injection engine. SAE Technical Papers 1981;.

[20] Corti, E., Cazzoli, G., Rinaldi, M., Solieri, L.. Fast prototyping of a racing diesel engine control system. SAE Technical Papers SAE 2008-01-2942 2008;.

[21] Corti, E., Solieri, L.. Rapid control prototyping system for combustion control. SAE Technical Papers SAEv2005-01-3754 2005;.

[22] Corti, E., Moro, D., Solieri, L.. Real-time evaluation of imep and rohr-related parameters. SAE Technical Papers SAE 2007-68-0024 2007;15.

[23] Cavina, N., Corti, E., Poggio, L., Zecchetti, D.. Development of a multi-spark ignition system for reducing fuel consumption and exhaust emissions of a high performance gdi engine. SAE Technical Papers SAE 2011-01-1419 2011;.