lecture_2011_10_27 - Multi-dimensional Modeling for Engine...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Multi-dimensional Modeling for Engine Design Dr. Jim Hilditch Technical Specialist Ford Research and Advanced Engineering Acknowledgements Combustion Systems Simulation and Development Group – Ford Research Laboratories Group Leader: Dr. James Yi CFD Technical Specialist: Dr. Zheng Xu Former Group Leader: Dr. Zhiyu Allen Han Additional thanks to the entire Ford Powertrain Research Department that provide high quality data and technical insight that make model development possible. Outline • Drivers for multi-dimensional modeling • Basic needs and requirements for the tools • Model development and validation • Methodology development • Application • Summary Need for Virtual Development of Future Powertrains Improved Attributes by Sophisticated Functionality • Improved attribute • Sophisticated function • Reduced cost • Decreased cycle time • Improved quality Basic Needs/Requirements of the CAE Tools Provide an accurate prediction • Contains physics • Boundary conditions & geometry representation • Validation Easy to use • Faster than building and testing prototypes • Used by individuals not the code developer • Skilled user General Modeling Approach Predictive Models Design Requirements Identification of Metrics Data from Similar Engine? Upfront Design Iterations Engine build / Experimental Data Design Improvements / Optimization Engine Verification Improved Metrics What is CFD? Incoming fluid Local state P,T,etc. Outgoing fluid • Divide the domain into small cells • Track what happens at the interfaces, e.g., velocity in and out, diffusion, pressure forces, transfer of heat • Calculate what happens in the cell, e.g., chemical reaction, heat radiation • March forward in time Fluid Flow and the Bernoulli Equation What is Bernoulli’s equation? P 1 + V 2 = Const ρ2 How does it work? V = 150 mph Airplane wing Net pressure force is upwards keeping plane in the air. Pressure Force Example – Intake Port Design • Intake Ports Engine depends on ports for breathing • Better breathing = more power • How do you know you have a good design? • Build an engine and test (3+ months, $10Ks) • Build a rapid prototype model and test (3 weeks, $1Ks) • Analyze with CFD (<1 day, computer time) Thought Experiment: Filling the Room Corner Hallway too narrow Too much space between students 1 door open part way 1 door closed Table in the way Students walking too slow Initial Intake Port Design CAD Model CFD Results Restriction Yi, Han, and Trigui, SAE 2002-01-2656 Improved Design Yi, Han, and Trigui, SAE 2002-01-2656 Port Optimization How do you know you have the best design? Quantify effect of parameter change Parameterize the port 98 10 11 12 7 6 5 Change parameters 4 3 2 Test with CFD 1 Optimized Design Yi, Han, and Trigui, SAE 2002-01-2656 Intake Port Optimization 1.2 Normalized Flow Rate 1 0.8 0.6 Original Intake Port Optimized Intake Port 0.4 0.2 0 0 2 4 6 8 10 12 Valve Lift (mm) Yi, Han, and Trigui, SAE 2002-01-2656 Difficult Application: Direct Injection Engine Design What does “direct injection” mean? Fuel is injected “directly” into the cylinder. (Most current engines inject fuel into the intake port.) What makes this interesting for CFD? • Fuel sprays – droplet dynamics, evaporation, wall interaction • Combustion – heat release, emissions Why Use CFD for DISI-SC Engines? • Fundamentally different requirements for fuel-air mixture formation in DI Engines – Part-load: locally stratified for reduced fuel consumption – Full load: homogeneous throughout the cylinder for improved performance • Mixture formation highly affected by engine parameters – – – – Injection strategy Spray characteristics Piston/chamber geometry Flow motion • Optimization of multiple parameters required. Why Use CFD for DISI-SC Engines? • Many technical challenges – Piston shape design must achieve robust stratified-charge combustion with minimal soot emissions, without sacrificing engine power output – Control over fuel injection and flow motion is limited at high speed, full load • Advantages of CFD modeling – Provides physical insights: Engineers must know why a design works or doesn’t work – Enables thorough examination of each key design parameter including piston/chamber geometry, injector positioning, spray specification, port design, etc. – Permits systematic upfront optimization of design parameters (before hardware is built) – Complements experimental efforts in later design stages for high quality design – Inspires creation and innovation Models vs. Methodology Models: CFD approximations of real geometry and physical processes Methodology: Process for using CFD to address specific design questions related to customer requirements Examples Mesh for ICE system Number of mesh layers / mesh density Length of intake runner Injector / spray models Development of injector specifications, spray targeting, and piston design Spray / wall interaction and film models Predictions of smoke emissions and recommendations for improved design Modeling: Building Blocks • Base CFD Capability – Air flow, Geometry, Moving mesh • New physical models for sprays – – – – – – – Heat transfer Evaporation Atomization Spray / wall interaction Film evolution Droplet Break-up … Spray Modeling – Simple Approach Use experimental data to initialize spray model Vessel pressure : 0.1MPa (mm) 0 • Measure droplet sizes and velocities Spray Pattern near the nozzle 50 • Iteratively develop droplet inflow conditions until downstream sizes 100 and velocities match experiment Vessel pressure : 0.35MPa Problems (mm) 0 • Need hardware to test • Sensitive to operating conditions (back pressure, injection pressure, fuel type, etc.) 50 100 Spray modeling – Better Approach Liquid sheet breakup Pressure swirl atomizer Nozzle flow: Mass conservation: Ratio of nozzle area to air core area: Solve for Velocity: where: 5 Equations 7 Unknowns: Q, Cd, X, V, θ, h, Kv Han et al. Atomization and Sprays, 1997 Spray modeling – Better Approach (2) Assume Kv=constant for a given injector (independent of pressure, etc.) What does this imply? • For fully developed sprays, θ does not depend on P (observed*) • Sheet thickness does not depend on P (some CFD evidence*) Spray geometry Where θo is the measured (specified outer cone angle) See Han et al. Atomization and Sprays, 1997 for references Pressure swirl atomizer Additional Details Atomization details come from non-linear stability theory Pressure swirl atomizer Break-up length Mean droplet radius (TAB model, O’Rourke and Amsden SAE 872089) Droplet size distribution Han et al. Atomization and Sprays, 1997 Spray Model Validation Model predictions are compared to spray measurements in a closed vessel. Han, Yang, and Anderson, Ford Tech Report 1999-0045 Spray Model Validation Experiment Model t = 1 ms t = 2 ms Pf = 10 MPa; Pamb = 0.1 MPa Han, Yang, and Anderson, Ford Tech Report 1999-0045 Spray Model Validation Atomization model was validated with Mie scattering images in an optical engine Early Injection Late Injection Han, Xu, Wooldridge, Yi, and Lavoie, SAE 2001-01-3667 Spray Model Validation Detailed comparison of spray model predictions to optical engine spray visualization. Model captures “lifting” of the spray due to uneven air entrainment. This can be important to spray targeting. Han, Xu, Wooldridge, Yi, and Lavoie, SAE 2001-01-3667 Spray / Wall Models Fuel Injection Spray Atomization Vaporization Turbulent Mixing Spray Impingement & Splash Fuel Film Dynamics Heat/Mass Transfer Wall Model What happens when a drop hits a wall? It depends … (Naber & Reitz SAE 8801007) Wen = 2 ρrVn2 / σ If droplets start to accumulate on the wall, how do we describe the fuel film? Current model builds on previous work (e.g., O’Rourke and Amsden SAE 961961, SAE 2000-01-0271) Wall Film Model Formulation • Major Assumptions – Thin film approximation – Film is in direct contact with wall – Film temperature is lower than the fuel boiling temperature • Momentum Conservation Inertial ρh ∂u f ∂t [ pressure gradient gas flow viscous + (u f − uw )⋅ ∇ S u f + h∇ S p f = τ w t − µ l (T f ) [ u f − uw + Pimp − (Pimp ⋅ n )n + M imp (uw ⋅ n ) n − u f + δ p f n + ρ hg momentum source mass source h2 gravity Han and Xu SAE 2004-01-0099 Post-splash Droplet Velocity • The secondary (post-splash) droplet velocity u is given as u = wn + vζ (cos ϕe t + sin ϕe p ) (1) • The azimuthal angle ϕ is statistically chosen according to the distribution suggested by Naber and Reitz (SAE 880107) and shape function ζ is derived from the jet impingement study by Ibrahim and Przekwas (Phys. Fluids, 1991) n et w0 w front θ back ϕ v v0 ep Han and Xu SAE 2004-01-0099 Film Model Validation • The experiment was reported by Mathews et al. (Atomization and Sprays, 2003) • Fuel is injected to a flat Plexiglas plate with three impingement angles • Measurements : – A phase Doppler Analyzer (PDPA) for droplet size – An un-intensified CCD camera for images of the wall film and secondary droplet cloud – An optical, non-intrusive technique for the thickness of the wall film Injector Fuel Injection Pressure (kPa) Pintle PFI injector Iso-octane 370 Injection pulse width (ms) 8.45 Spray cone angle (deg.) 7 Initial drop SMD (µm) 380 Distance to wall (mm) 100 Injection Angle to wall (deg.) 60,45,30 θ Han and Xu SAE 2004-01-0099 Spray Model Results • • The spray shape is seen to become more and more elliptic as the impingement angle is reduced The computed spray normal penetration and tangential penetration versus time after the start of impingement are correlated with the experiment images θ=600 θ=450 12 ms (ASOI) θ=300 Han and Xu SAE 2004-01-0099 Film Evolution Comparison • OA Experimental Images Computed Images θ=600 θ=450 θ=300 Han and Xu SAE 2004-01-0099 Comparison of Film Front 40 M easured, θ =30 deg. M easured, θ =45 deg. M easured, θ =60 deg. Com puted, θ =30 deg. Com puted, θ =45 deg. Com puted, θ =60 deg. 35 Impingement point Film front length (m m) 30 • The film front length increases with the decrease of the impingement angle • Film front motion slows down after the end of impingement 25 20 15 10 5 0 5 10 15 20 25 30 Tim e (ms) Han and Xu SAE 2004-01-0099 Comparison of Film Thickness 80 M easured, θ =30 deg. M easured, θ =45 deg. 70 M easured, θ =60 deg. C om puted, θ =30 deg. Film thickness (µ m) 60 C om puted, θ =45 deg. C om puted, θ =60 deg. 50 40 30 80 20 Measured, θ =30 deg. Measured, θ =45 deg. 70 10 Measured, θ =60 deg. Com puted, θ =30 deg. 0 0 5 10 15 20 Front distance (m m) • Film thickness increases towards the edge the film in both impingement and transverse direction 25 F ilm thickness (µ m) 60 Com puted, θ =45 deg. Com puted, θ =60 deg. 50 30 35 40 30 20 10 0 -15 -10 -5 0 5 10 Distance from centerline (m m) Han and Xu SAE 2004-01-0099 15 20 Impingement Model Validation Spray Impingement model was validated with PDPA measurements (Experimental data from Park et al. SAE 1999-01-3662) Test: Spray impingement on a heated flat plate Case 1: 90o incidence angle, 1 bar, T=20C Case 2: 60o incidence angle, 1 bar, Plate heated to 160C Case 3: 60o incidence angle, 3 bar, Plate heated to 160C Measurement 25 Simulation SMD (micron) 20 15 10 5 0 Han, Yi, and Trigui, SAE 2002-01-2655 case1 case2 case3 Stratified Charge Operation 65o BTDC 55o BTDC 45o BTDC 25o BTDC Injection Timing Effects Han, Yi, and Trigui, SAE 2002-01-2655 Injection Timing Effects Earlier injection (SOI = 75 bTDC) results in a very lean mixture – likely to misfire. Later injection (SOI=65 bTDC) results in higher piston wetting – what effect???. Han, Yi, and Trigui, SAE 2002-01-2655 Piston Wetting / Smoke Methodology 25 Wetting Location Piston I-10, 70 20 Normalized liquid fuel on piston surface (%) Develop a link between piston wetting and smoke emissions from the engine 0 Piston I-10, 60 Piston B-10, 70 0 FSN 1.2 spray spray 0 spray FSN 0.3 15 FSN 0.1 10 5 0 -80 -70 -60 -50 -40 -30 -20 -10 0 Crank angle (degree) Model Prediction Photo of Piston Optical Engine Flood Illuminated LIF (Similar design, idle, 20 BTDC) Han, Yi, and Trigui, SAE 2002-01-2655 Effect of Spray Cone Angle 60 deg Injector As spray cone increases from 60o to 70o, the piston wetting reduces significantly, so does the dyno measured soot emissions. 70 deg Injector 0 60 s pray 20 0 70 s pray 1.5 Computed Piston Wetting History 1.2 15 FSN Normalized liquid fuel on piston surface (%) 25 10 Measured Engine Smoke Number 0.9 0.6 0.3 5 0 60 deg 0 -80 -70 -60 -50 -40 -30 C rank angle (degree) 1500 rpm, 2.6 bar BMEP -20 -10 0 70 deg Effect of Swirl Motion 1500 rpm, 2.6 bar BMEP o At 40 + SOI 1500 rpm/2.62 bar BMEP, Piston I-9, Chamber 12 25 Normalized liquid fuel on piston surface (%) SR=0.94 20 SR=0.0 15 10 5 0 SR=0.94 SR=0. -80 -70 -60 -50 -40 -30 -20 -10 0 Crank angle (degree) Swirl motion enhances fuel vaporization, thus less piston wetting. Han, Yi, and Trigui, SAE 2002-01-2655 Misfire Robustness I-10 B-10 Han, Yi, and Trigui, SAE 2002-01-2655 Unburned Hydrocarbons Methodology Unburned hydrocarbons (UHC) are a regulated emission that can significantly impact engine fuel economy. UHC mechanisms in SC operation are different than mechanisms in homogeneous charge, stoichiometric operation. Objectives Wall Guided DI Design • Present UHC mechanisms • Analyze SC dyno data to understand effects of spark advance, injection timing, and manifold pressure • Predict mixture preparation using CFD • Propose semi-empirical UHC model. • Apply model to new design Hilditch, Han, and Chea SAE 2003-01-3099 UHC Mechanisms Homogeneous Charge Stratified Charge 3 1 Primary Sources [Heywood] 2 1 Primary Sources 1. Crevices (60%) 1. Lean Quenching 2. Deposits/Oil Layer (22%) 2. Liquid Fuel 3. Liquid Fuel (13%) 3. Spark Plug Crevice 4. Quench (5%) Hilditch, Han, and Chea SAE 2003-01-3099 Unique Aspects of DI-Stratified A/F 30 25 20 • 15 Stratified mixture 10 Fuel films • Low bulk temperatures Frequency [%] • 8 7 6 5 4 3 2 1 0 0.05 Ex haus t Temperature [C] 3.5 Liquid Fuel [%] 0.85 1.25 1.65 Equivalence Ratio 2.05 2.45 500 4.0 MAP = 90 kPa MAP = 75 kPa 3.0 2.5 2.0 1.5 1.0 0.5 0.0 640 0.45 660 680 700 Crank Angle [cad] 720 740 450 H o m oge neo us C h arge, la m bda =1 400 350 160 C 300 250 140 C 2 00 1500 RPM 2000 RPM 1.75 bar NMEP 2.81 bar NMEP Stratified C harg e, Lean Ignition Timing Effects 30 18 14 20 15 10 64/18 12 10 68/30 8 6 5 4 0 0 2 15 20 25 30 35 Spark Advance 40 45 15 20 25 30 35 40 45 Spark Advance 32 18 16 12 Peak Pressure [bar] R2 = 0.9324 14 HC E-Index [%] 68/18 16 HC E-Index [%] HC Index [%] 25 68/18 10 8 64/18 6 4 68/30 2 30 28 26 2 R = 0.6829 24 75 kPa MAP Data Linear (75 kPa MAP Data) 22 20 0 20 22 24 26 28 Peak Pressure [bar] 30 32 10 20 30 40 50 Spark Advance Hilditch, Han, and Chea SAE 2003-01-3099 Injection Timing / Mixing Duration Piston wetting More Piston Wetting 0.25 UHC Smoke Stability 10 8 0.2 0.15 6 0.1 4 2 0.05 0 0 50 55 60 65 70 75 ? Mixing Duration 15% Fuel in Lean Regions 12 STD NMEP [bar] HC E-Index [%] & 10*FSN Piston overshoot EOI = 656 10% 5% 0% 650 90 kPa MAP 75 kPa MAP 670 690 710 730 Crank Angle EOI [bTDC] Longer Mixing Duration Piston Overshoot Mixing duration may not be significant for wall guided systems beyond a certain threshold Hilditch, Han, and Chea SAE 2003-01-3099 Manifold Pressure 4.0 Piston Wetting MAP/EOI/SCV 30 85/68/18 75/ 68/18 75/64/18 20 2.5 2.0 1.5 1.0 0.5 0.0 640 75/68/30 15 15% 5 0 20 25 30 Peak Pressure [bar] 35 660 680 700 720 740 Crank Angle [cad] 65/72/18 10 MAP = 90 kPa MAP = 75 kPa 3.0 Fuel in Lean Regions HC E-Index [%] 25 Liquid Fuel [%] 3.5 Mixture Distribution 10% 5% 0% 650 90 kPa MAP 75 kPa MAP 670 690 710 730 Crank Angle Hilditch, Han, and Chea SAE 2003-01-3099 Stratified Charge UHC Model Factors that affect the potential to produce UHC Factors that affect the efficiency of oxidizing the fuel • MAP, EOI, EGR • Spark advance • Piston design: bowl diameter, depth • EGR Other factors • Spark plug crevice • Injector tip? • Injector specifications: cone angle, targeting angle UHC = A*(Fuel in lean regions + B * Liquid fuel) * exp (-C * SA) + D CFD Hilditch, Han, and Chea SAE 2003-01-3099 Stratified Charge UHC Model (2) UHC for Small Displacement Engine* HC E-Index [%] 20 y = 0.4194x + 2.4546 2 R = 0.8651 15 10 5 Corrected for SA 0 0 10 20 30 40 Fuel Fraction [%] UHC = 2.1(Lean fuel + Liquid fuel) exp(-0.05 SA) + 2.5 Coefficients may vary with load or engine. * 3 Piston designs, 2 Injectors, 2 MAP, +1 large displacement point Hilditch, Han, and Chea SAE 2003-01-3099 Run CFD at standard condition: • • 2. 3. 1500 RPM, 1 bar BMEP 64 EOI Get piston wetting and fuel in lean regions from CFD output. Apply correlation. Predictions 90 kPa MAP, 64 EOI, 34 SA HC E-Index = 10% (10% dyno) 75 kPa MAP, 64 EOI, 32 SA HC E-Index = 7% (6% dyno) 4 3 AA10 - 90kPa 2 1 0 640 660 680 700 720 740 Crank Angle Fuel mass in lean regions 1. Liquid Fuel on Piston [%] Application 16% 12% 8% 4% 0% 640 AA10: Phi < 0.2 660 680 700 720 740 Crank Angle [cad] (Phi < 0.3 is used in the model) Hilditch, Han, and Chea SAE 2003-01-3099 Application 30 Design A Features* Design A Cone Angle: 5 degrees wider Targeting: 5 degrees less offset (* Relative to Design B) 25 A/F Ratio Bowl Depth: 1.5 mm deeper Design B 20 15 10 670 680 690 700 710 720 Crank Angle Design A CFD Predictions Design B A/F 30 20 10 Optical Engine Confirmation Intensity 8 0 Hilditch, Han, and Chea SAE 2003-01-3099 Application 12 20 8 Design B HC E-Index [%] % of Fuel Mass 10 y = 0.4227x + 2.1673 2 R = 0.8938 Design A 6 4 2 15 Design B 10 Design A 5 0 0 0 0.5 1 1.5 2 2.5 0 Liquid Fuel Design B: 15.3% 20 30 40 Fuel Fraction [%] Equivalence Ratio Design A: 1.2% 10 Design A Design B Predicted UHC Measured UHC E-Index [%] E-Index [%] 10.2 10.0 17.3 17.5 Hilditch, Han, and Chea SAE 2003-01-3099 Summary • Rapid development time requires fast, accurate modeling to support up-front design before hardware is built. • The best models are based on physical principles and careful observation They must be well validated. • Sound methodology is required to use the models to drive the design process. ...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online