03-05-2011, 10:39 AM
Modeling of Gasoline Direct Injection Mixture Formation Using
KIVA-3V: Development of Spray Breakup & Wall Impingement
Models and Validation with Optical Engine Planar
Laser Induced Fluorescence Measurements.
ABSTRACT
The computational code KIVA-3V has been used as the modeling platform for Gasoline Direct Injection enginesimulations. Improved models for fuel injection, wall impingement and stratified combustion have been implemented.To complement the modeling effort, experimental data of an optical engine has provided extensive data for validatingthe new models. The experimental data include Planar Laser Induced Fluorescence measurements, of the injection,mixing and combustion events in an optical engine. In this work, focus is given on fuel injection, wall impingement andair-fuel mixing.
INTRODUCTION
The design of more powerful, fuel-efficient, and environmentallyfriendly gasoline engines is currently oneof the main goals of engine researchers. With the adventof increasingly stringent fuel consumption and emissionsstandards, engine manufacturers face the challengingtask of delivering conventional vehicles that abide bythese regulations. Many automotive manufacturers areresponding to the challenge of developing highly efficient,Ultra Low Emission Vehicles (ULEV's) by developingGasoline Direct Injection (GDI) systems [1-7].Controlling the mixture formation of a GDI engine undera wide range of engine operating conditions is essentialto reduce smoke and particulate generation and optimizefuel economy. A valuable tool for understanding andcontrolling the mixture formation is the combination ofoptical diagnostics with Computational Fluid Dynamics(CFD) simulations of the in-cylinder processes.In order to perform reliable CFD simulations of aGDI engine, appropriate models for fuel injection, mixingand combustion are required. Accurate predictions ofthe air-fuel mixing process are critical in order to providecorrect initial conditions for the combustion modeling. Itis common practice to validate fuel injection and wallimpingement models against experimental data acquiredunder constant temperature and pressure conditions, inconstant-volume vessels (bombs). These experimentalset-ups allow for highly accurate measurements withlaser diagnostics methods and offer precious insight onthe physical phenomena occurring during injection andwall impingement. However, in an engine environment,temperature and pressure conditions are continuouslyvarying, presenting an additional challenge for the models.Therefore, it is important to validate these modelsagainst measurements acquired in optical engines configurations.Han et al. [8-9] have presented comparisonsof optical measurements with spray evolution and fuelvapor distribution calculations for stratified operatingconditions; however, there exists a void in the literaturefor this type of study at early injection timings. In thecurrent work, focus is given on coupling the individualmodels for fuel injection, wall impingement and combustionand validate them under homogeneous engineoperation, achieved by early fuel injection.The submodels developed for fuel injection, spraybreakup and wall impingement have been extensivelyvalidated against optical diagnostics experiments [10-13]and will be briefly described here. The validation wasperformed with data acquired under atmospheric conditionsand should apply to early injection timings in aGDI engine. Subsequently, comparison with Planar LaserInduced Fluorescence (PLIF) measurements from anoptical single-cylinder engine will be presented to demonstratethe predictive capability of the models. In a laterstudy, these models are coupled to a combustion modelapplicable for both stratified and homogeneous operation[14].
EXPERIMENTAL SETUP
The experimental setup consists of a single-cylinder,four stroke GDI engine with a four valve head, featuringfull optical access, variable swirl [15] and dimensions asgiven in Table 1. The spark plug is located on the exhaustside of the combustion chamber, while the injector islocated on the intake side. The optical access is achievedby a quartz window in both the piston and pent roof ofthe engine, combined with a quartz cylinder. The pistonwindow is smaller than the engine bore, giving opticalaccess to part of the engine’s volume. The swirl is variedby restricting the airflow through one of the two intakevalves. A laser induced fluorescence (LIF) technique, withToluene as a tracer added to the fuel (iso-octane), hasbeen used to provide visualization of the fuel-injection,mixing and combustion processes [15]. The LIF signalwas calibrated by taking images of early injection events90o after top-dead-center (ATDC), at which point the fuelhad sufficient time to mix and evaporate to create a homogeneous,gaseous fuel-air mixture. Dependence of thetoluene signal on fuel concentration, laser energy and gastemperature was corrected for in the measurements.
COMPUTATIONAL FRAMEWORK
For the numerical simulations performed in this work amodified version of the Los Alamos KIVA-3V code [17]has been used. The modifications include the fuel injectionmodel, the wall impingement model and the CoherentFlame Combustion model [14]. The computationalmesh consists of approximately 70,000 cells. An imageof the grid showing the intake and exhaust ports is shownin Figure 1. The adopted mesh has been found to generatea swirl behavior very similar to that in the opticalFig. 1: View of the computational grid for theoptical engineengine. To generate the swirl, one of the valves is deactivatedcompletely.
Fuel Injection Model
A comprehensive model for sprays emerging fromhigh-pressure swirl injectors has been developed, accountingfor both primary and secondary atomization[10-11]. The model considers the transient behavior ofthe pre-swirl spray and the steady-state behavior of themain spray. The pre-swirl spray modeling is based on anempirical solid-cone approach with varying cone angle.First, a solid-cone-like injection is performed, representingthe pre-swirl spray. The cone angle is gradually increased,using a linear profile. At the transition point, thecode switches into a hollow-cone structure, while thecone angle is still increasing, until the steady-state valueis reached.The primary breakup approach adopted here for themain spray is based on the Linearized Instability SheetAtomization (LISA) model [17-18], with appropriateextensions to include a swirl velocity component. ARosin-Rammler cumulative distribution is used for calculatingthe droplet size. For the secondary dropletbreakup the TAB model [19] is being used both for thepre-swirl and the main spray, with its baseline constants.
Wall Impingement Model
A wall impingement model, developed by Grover etal. [12-13] has been used to improve the prediction capabilityof spray-wall interactions. The model conservesmass, tangential momentum and energy of an impingingparcel. This model focuses on spray impact on dry andwet surfaces below the fuel’s Leidenfrost temperature, ascenario encountered under typical engine operatingconditions [20].Three splashing parcels and one wallfilm parcel areused to represent the shattering of a splashing dropletupon impact with the surface. It is assumed that the impulsiveforce on an impinging droplet normal to the surfaceis dominant allowing one to treat the magnitude ofits tangential momentum component constant after impact.The viscous dissipation of an impinging droplet andkinetic energy of the wallfilm are accounted for in theenergy conservation equation.
Download full report
http://me.engin.umich.edu/autolab/Public...004_13.PDF
