i
ABSTRACT
Goal of this thesis was to build a reliable set of tools able to simulate the behavior of a two- stroke, dual-fuel, direct injected engine, from the point of view of flow field and of combustion. The combustion type applied is the Reactivity Controlled Compression Ignition concept, which prescribes the auto-ignition of a reactivity-stratified charge.
The most appropriate tool to study and optimize the operations of this engine is a CFD model, which has been built make use of a commercial CFD code: FIRE, supplied by AVL. This has been coupled with a detailed chemical mechanism and a detailed chemical equation solver called Chemkin, supplied by Reaction Design.
Than the most important part of the work was the model validation. To do this a comparison between the CFD data against available experimental data has been performed, using a previous thesis by Matthew Labaza at the University of Wisconsin-Madison and experimental runs performed in the fall of 2013, always at the University of Wisconsin-Madison.
Also a first attempt of validation of the spray behavior has been done comparing the simulation with respect to available experimental test of the injectors on a dedicated bench. In this way it is assured that the fuels (two different fuels with two different injectors) are sprayed in a similar way both in the real engine and in the modeled engine.
A 0-D model has also been built in order to supply suitable initial condition for the CFD simulation, for the cases where no experimental data are available, increasing the predictivity capability. This also made possible to reduce the computational time for exploratory parametric study, giving a prompt response at various trial that has been made.
In the end some first study has been performed in order to guide the future experimental research, which will be carried out at the Engine research center of the University of Wisconsin- Madison.
ii
SOMMARIO
Titolo: Studio numerico di un motore due tempi ad iniezione diretta con combustione RCCI
Scopo della tesi è la costruzione di un set di modelli in grado di simulare il comportamento di un motore due tempi, alimentato con due combustibili tramite doppia iniezione diretta. La strategia di combustione applicata è nominata Reactivity Controlled Compression Ignition (RCCI), la quale consiste nell'auto-accensione di una miscela a reattività stratificata. L'obiettivo di tali modelli è di migliorare la comprensione dei fenomeni che accadono all'interno della camera di combustione e dare indicazioni utili per continuare la campagna di prove sperimentali su tale motore.
E' stato quindi costruito un modello 3D (CFD) del motore, utilizzando il software FIRE fornito da AVL. Per la modellazione dei fenomeni inerenti alla combustione tale codice è stato accoppiato con un solutore di cinetica chimica: Chemkin. Il dominio di calcolo considerato comprende la camera di combustione, le luci di aspirazione e il sistema di scarico completo.
Oltre a questo è stato approntato anche un semplice e veloce modello 0-D nel quale il processo di combustione è basato sull'ipotesi di carica omogenea accesa per compressione (HCCI), mentre il processo di lavaggio è basato su un modello a due zone. Lo scopo di questo modello è di fornire risultati di prima approssimazione in maniera immediata e una prima stima delle condizioni iniziali per i calcoli CFD.
La parte maggiore del lavoro è stata quindi la validazione del modello CFD, comparandone i risultati con dati sperimentali raccolti all'University of Wisconsin-Madison, che ha infine consentito di evidenziare il peso di fenomeni come la stratificazione di temperatura causata dal lavaggio e la stratificazione di reattività causata dall'iniezione.
Il modello 0-D realizzato si è dimostrato un valido strumento per il calcolo delle condizioni iniziali delle simulazioni CFD, poiché dopo soli due cicli motore le condizioni iniziali della simulazione sono uguali alle condizioni finali. La velocità di calcolo de tale modello è stata inoltre sfruttata per costruire mappe di funzionamento del motore al variare di svariati parametri come il rapporto tra i due combustibili (diesel e benzina), il rapporto di equivalenza, il rapporto di lavaggio, il regime del motore e il rapporto di compressione. Tali mappe potranno essere utili per guidare i futuri test sperimentali.
iii
PREFACE
The reduction of the pollutant emissions and the increases of the fuel-economy demanded by the actual regulations have raised the interest in new combustion concept, which does not rely on flame propagation or on mixing controlled combustion.
One of the most promising concepts is the Reactivity Controlled Compression Ignition (RCCI) combustion, firstly proposed by researchers at the University of Wisconsin-Madison. This concept is based on the staged auto ignition of the mixture inside the combustion chamber, taking advantages of ignitability stratification. This stratification is obtained with an early injection of a high reactivity fuel (usually diesel) in an almost homogenous mixture of low fuel (such as gasoline). It has been demonstrated that this concept is able to decrease the pressure rise rate, increasing the operating range toward higher load, with near-zero NOx and Soot emissions.
This thesis focuses, from the numerical point of view, on how this combustion is applicable to a two stroke-engine, using a dual fuel direct-injection. The application of the RCCI concept on a two- stroke engine seems very promising mainly thanks to the possibility to have a high internal residual content, without the use of expensive devices (i.e. exhaust gas recirculation systems). This is a very interesting feature for the RCCI concept: the thermal content of the residual gas will help to achieve the auto ignition of the charge also with low compression ratio (with respect to traditional compression ignition engine) and the dilution will be able to help to achieve a smoother combustion (with respect HCCI combustion), helping to reach higher load. The well phased and fast RCCI combustion is also very appropriate to two-stroke engine, in which, because of the very advanced exhaust port opening, the time available for the combustion and the expansion is reduced.
Furthermore the injections of both the fuels directly inside the cylinder will avoid or reduce the short- circuiting of the fuel directly to the exhaust during the scavenging process: this indeed is one of the most important deficiencies of the two-stroke engine which currently leads to high UHC content and poor fuel-economy. In the end this application has the potential to achieve a very clean, simple and cheap engine.
However the optimization of this application requires a deep understanding of what happens
iv
inside the cylinder, for example the interaction between the spray and the flow field and the stratification obtained after the scavenging process and the injections. To do this it has been decided to use a Computational Fluid Dynamics (CFD) software in order to have full access to the fluid dynamics and combustion phenomena inside the cylinder. Furthermore a 0D HCCI model has been built in order to obtain suitable initial condition for the CFD analysis and immediate information to guide the experimental test. Most of this thesis has been focused on the validation of the models, but also interesting findings on engine behavior has been found.
Chapter 1, 2, 3 and 4 are mostly a literary review. Chapter 1 explains briefly the deficiency of the traditional combustion mode, what is the current approach of the industries to answer at the pollutant regulation and how the research has moved towards innovative combustion concepts.
Chapter 2 is a zoom on the innovative combustion concept applied at our engine, the RCCI concept, highlighting the born of this concept, its development and its most important features. Chapter 3 is a very brief explanation of how two-stroke, ports scavenged, crankcase scavenged intake design works, focusing on the particularity represented by the scavenging process. Chapter 4 is a brief summary of the work experimentally conducted by Matthew Labaza on this Engine. The laboratory set up is explained in order to understand the origin of the data used for the sequent validation, as well as the first results obtained.
Chapter 5 shows the models used in this works: the CFD model and the 0D HCCI model. For the CFD simulation all the sub-model used are listed, with an explanation of why this model have been chosen among all the available ones. Than it is explained the purpose of the 0-D HCCI model, how it has been built and what are the assumptions made in order to obtain a very simple and fast tool.
Chapter 6 collects all the results obtained during this work. The first part is obviously focused on the validation of the CFD model in motored and fired condition. The spray validation is showed as well.
Several sensitivity studies has been performed on the variables not experimentally measured (such as temperature in the crankcase) to evaluate their influence. Than the first findings are showed supplied both by the CFD model and by the 0D HCCI model.
v
Chapter 7 gathers the conclusion after all this work and some thought about the deficiency founded (both in the modeling approach and the experimental measurements) and how these can be improved.
vi
TABLE OF CONTENT
1 INTRODUCTION... 1
1.1 MOTIVATION ... 1
Spark-ignited engine (SI) ... 2
Compression-ignited engine (CI) ... 3
1.2 POLLUTANTEMISSIONSFROMINTERNALCOMBUSTIONENGINES ... 4
Carbon dioxide (CO2)... 4
Unburned hydrocarbon emissions (UHC) ... 4
Carbon monoxide (CO) ... 5
Nitrogen oxides (NOx) ... 5
Particulate Matter (PM) ... 6
1.3 CURRENTTECHNOLOGYSTATEOFART ... 6
1.4 CURRENT EXTERNAL DEVICES FOR POWERTRAIN POLLUTANT REDUCTION ... 7
1.4.1 Current technology for in-cylinder pollutant reduction ... 8
1.5 INNOVATIVECOMBUSTION ... 9
Activated Thermo Atmosphere Combustion (ATAC) ... 11
Homogeneous charge compression ignition (HCCI) ... 12
Low temperature combustion (LTC) ... 13
Partially Premixed Charge combustion (PPC) ... 14
1.5.1 Fuel behavior and design ... 14
2 REACTIVITY CONTROLLED COMPRESSION IGNITION (RCCI) ... 17
2.1 INJECTIONSTRATEGYANDCOMBUSTIONPROCESS ... 19
2.2 EFFICIENCY ... 23
2.3 EMISSIONS ... 26
3 TWO-STROKE OPERATIONS ... 27
3.1 STATEOFARTOFTWO-STROKEENGINE ... 27
3.2 TWO-STROKEPORTS-CONTROLLEDOPERATIONS ... 28
vii
3.2.1 Scavenging process ... 31
3.3 DIRECT-INJECTEDTWO-STROKE ... 35
4 EXPERIMENTAL RESULTS ... 37
4.1 EXPERIMENTALSETUP ... 37
4.2 RESULTS ... 40
5 MODELING TOOLS ... 44
5.1 AVLFIRE(CFDCODE) ... 46
5.1.1 Mesh and simulation features ... 47
5.1.2 Combustion modeling ... 50
5.1.3 Spray modeling ... 53
Break-up model ... 54
Evaporation model ... 55
Turbulence dispersion... 55
Coalescence / collision model ... 56
Wall interaction model ... 56
5.1.4 Resume of the used models and time performance ... 57
5.2 0-DIMENSIONHCCIMODEL ... 58
5.2.1 Single zone internal combustion engine model ... 59
5.2.2 0-Dimensional scavenging process model ... 61
5.2.3 Coupling and iteration process... 68
5.2.4 Check of the supplied initial condition and comparison with CFD ... 69
6 RESULTS ... 72
6.1 MODELSVALIDATION ... 72
6.1.1 Motored condition... 72
Inlet temperature... 74
Wall temperature ... 76
Multicycle analysis ... 77
viii
Exhaust pressure sensitivity ... 80
Final results ... 81
6.1.2 Fired condition without combustion and injection ... 83
Exhaust and crankcase pressure ... 84
Inlet temperature and wall temperature ... 86
Final results ... 87
6.1.3 Sprays ... 88
Gasoline injector Bosch HDEV 5.2 ... 88
Diesel injector DELPHI Multec ... 92
6.1.4 Fired condition ... 93
Injection duration ... 94
Final results ... 97
6.2 OTHERRESULTS ... 99
6.2.1 Parametric studies and 0-D HCCI model maps... 99
Effects of the mixture: PRF number and equivalence ratio... 101
Effects of the delivery ratio ... 104
Effects of the compression ratio ... 105
Effects of the engine speed ... 107
6.2.2 Comparison between combustion concept (ideal HCCI, real HCCI, RCCI) ... 108
7 CONCLUSIONS AND RECOMMENDATION ... 115
7.1 CONCLUSIONS ... 115
7.2 FUTUREWORKS ... 116
8 BIBLIOGRAPHY ... 118
APPENDIX A - MODELING APPROACH LIMITS ... 122
ix
List of figures
Figure 1.1 - Energy source and use in the united states, from www.eia.goc ... 1
Figure 1.2 - Diesel and gasoline cars EURO legislation developing ... 2
Figure 1.3 - Phi-T diagrams showing innovative combustion , from[3] ... 11
Figure 1.4 - HCCI combustion with different PRF fuels, from [2] ... 15
Figure 2.1 - Constant volume ignition delay for different fuel, from [13] ... 19
Figure 2.2 - Fuel system for RCCI operation, from [22] ... 20
Figure 2.3 - Heat release shape with different injection timing and strategies ... 22
Figure 2.4 - Comparison between RCCI-optimized piston and stock piston, from [14]. ... 23
Figure 2.5 - Flow of fuel energy over the load sweep for RCCI heavy-duty engine, from [13] .... 25
Figure 3.1 - Schematic of a two-stroke ports-controlled crankcase scavenged engine ... 29
Figure 3.2 - Rotary valve ... 30
Figure 3.3 - Reed valve ... 30
Figure 3.4 - Pressure in the cylinder and in the crankcase during the scavenging process ... 32
Figure 3.5 - Main type of scavenging flow ... 33
Figure 3.6 - Scavenging and trapping efficiency with respect to the delivery ratio ... 35
Figure 4.1 - The original engine and the experimental mounting assembly ... 38
Figure 4.2 - RCCI head and most its most important feature, from [25] ... 39
Figure 4.3 - Schematic of the experimental set-up ... 40
Figure 4.4 - Results for Gasoline and Diesel start of injection ... 41
Figure 4.5 - RCCI phasing control with the variation of gasoline percentage ... 42
Figure 5.1 - The mesh of the RCCI piston used in this study ... 47
Figure 5.2 - Example of the arbitrary connection ... 49
Figure 5.3 - ECFM-3Z scheme, from [26] ... 51
Figure 5.4 - Depiction of the effect inside the spray, from [26] ... 53
Figure 5.5 - Abramson-Sirignano multi-component evaporation model, from [26] ... 55
Figure 5.6 - Droplet behavior hitting a wall ... 57
Figure 5.7 - HCCI/RCCI/PPC comparison at same initial condition, from [17] ... 59
x
Figure 5.8 - Schematic of charge division in the combustion chamber, from[33] ... 62
Figure 5.9 - Comparison between Foudray's models and experimental data, from [33] ... 64
Figure 5.10 - 0-D scavenging model, results on the content of residual gas ... 65
Figure 5.11 - 0-D scavenging model, results on the temperature at the EPO ... 67
Figure 5.12 - Schematic of the 0-dimension HCCI model ... 68
Figure 6.1- Cylinder pressure and Crankcase pressure measured in motored runs ... 74
Figure 6.2 - Schematic of the intake system ... 74
Figure 6.3 - Set of inlet temperature used for the inlet temperature sensitivity analysis ... 75
Figure 6.4 - Exhaust port pressure during the multicycle analysis ... 78
Figure 6.5 - Quantity variations in the multicycle analysis for TP00 and TP58 ... 78
Figure 6.6 - Instantaneous mass-flow to the exhaust and from the scavenge ... 79
Figure 6.7 - Effects of the exhaust pressure changes on motored condition ... 80
Figure 6.8 - Motored validation results, pressure trace on low- and high-pressure phase ... 82
Figure 6.9 - Comparison between the crankcase pressures ... 84
Figure 6.10 - Effects of the crankcase pressure and exhaust pressure ... 85
Figure 6.11 - Effects of the wall heat transfer on the compression curve ... 86
Figure 6.12 - Comparison between experimental and calculated cylinder pressure ... 87
Figure 6.13 - Geometrical features of the Bosch HDEV 5.2 injector ... 91
Figure 6.14 - Spray penetration comparison between simulation and test ... 91
Figure 6.15 - Comparison between the experimental images and the simulation of the spray .. 92
Figure 6.16 - Liquid quantity inside the cylinder for the injection duration sweep ... 95
Figure 6.17 - Apparent rate of heat release comparison between long and short injection ... 96
Figure 6.18 - Final results of the fired validation ... 98
Figure 6.19 - Combustion efficiency with respect the variation of PRF number and ER ... 101
Figure 6.20 - CA50 with respect the variation of PRF number and ER ... 102
Figure 6.21 - Thermal efficiency with respect the variation of PRF number and ER ... 103
Figure 6.22 - Combustion zone with respect the variation of PRF number and ER ... 104
Figure 6.23 - Combustion zone with respect the variation of delivery ratio ER ... 105
Figure 6.24 - Combustion zone comparison between two different compression ratios... 106
xi
Figure 6.25 - GMEP comparison between two different compression ratios ... 106
Figure 6.26 - Thermal efficiency comparison between two different compression ratios ... 107
Figure 6.27 - Combustion zone comparison between two cases with engine speed ... 108
Figure 6.28 - Comparison between ideal HCCI, real HCCI and RCCI combustion ... 110
Figure 6.29 - Instantaneous mass flow during the scavenging process ... 111
Figure 6.30 - Flow field at the EPC ... 112
Figure 6.31 - Wall heat transfer comparison between quiescent and real simulations ... 112
Figure 6.32 - Species stratification at the EPC ... 112
Figure 6.33 - Temperature stratification at the EPC ... 112
Figure 6.34 - Combustion start in real HCCI case, on the right the orientation of the engine... 113
Figure 6.35 - Combustion start in RCCI case, on the right the spray targeting ... 114
Figure A.0.1 - Instantaneous mass flow from scavenge and exhaust ports changing the PEPO .. 123
xii
List of tables
Table 2-1 - Comparison of energy share between RCCI and diesel, from [12] ... 24
Table 4-1 - Original BRP engine feature ... 37
Table 5-1 - Resume of the tunable constants n the KH-RT break-up model ... 54
Table 5-2 - Resume of the model applied in the CFD simulations... 57
Table 5-3 - Time performance of the CFD model on a 4 processor 2.80 GHz ... 58
Table 5-4 - Condition used in the CFD simulation of the scavenging process ... 65
Table 5-5 - Comparison between 0-D HCCI model and CFD simulations ... 70
Table 6-1 - Experimental set-up and results for motored runs ... 73
Table 6-2 - Motored validation, comparison of the most important quantity ... 82
Table 6-3 - Geometrical features of the Bosch HDEV 5.2 injector ... 90
Table 6-4 - geometrical features used for the DELPHI Multec injector ... 93
Table 6-5 - Fired conditions used for the validation, from [24]... 94
Table 6-6 - Injection durations used for the fired validation ... 95
Table 6-7 - Final results of the fired validation : important variables ... 98
Table 6-8 - Initial conditions and results of the combustion concept comparison ... 109
xiii
Nomenclature
Acronym
HCCI Homogenous Charge Compression Ignition
LES Large Eddy Simulation
LTC Low Temperature Combustion
NS Navier-Stokes
PCI Premixed Compression Ignition
PM Particulate Matter
PPC Partially Premixed Combustion
PRF Primary Reference fuel
PRR Pressure Rise Rate
RANS Reynolds Averaged Navier-Stokes equation
RCCI Reactivity Controlled Compression Ignition
RPM Revolute Per Minute
SCR Selective Catalytic Reduction System
SI Spark Ignited
SOI Start Of Injection
SPC Scavenge Ports Closing
SPO Scavenge Ports Opening
TDC Top Dead Center
TP Throttle Position
xiv TWC Three way catalyst
UHC Unburned Hydro-Carbon
VVT Variable Valve Timing
WOT Wide Open Throttle
Symbols and Variable
CA50 = Crank angle at which half of the total heat released has been released
CR = Compression ratio
D = Diameter
EGR % = Residual gas mass / total mass
GMEP = Gross indicated mean pressure
Keff = Efflux coefficient
l = Length
LHV = Lower Heating Value
m = Mass
n = Polytrophic exponent
Nu = Nusselt Number
p = Pressure
Pr = Prandtl Number
PRF number = Vn-heptane / (Vn-heptane + Viso-octane)
Re = Reynolds Number
T = Temperature
xv
V = Volume
v = Velocity
We = Weber Number
ηb = Brake or total efficiency
ηcomb = Combustion efficiency
ηGIE = Gross indicated efficiency
ηm = Mechanical efficiency
ηsc = Scavenging efficiency
ηt = Thermal or indicated efficiency
ηt = Thermal gross efficiency
ηtr = Trapping efficiency
Λ = Delivery ratio
Λ* = Modified delivery ratio
ρ = Density
σ = Surface tension
