FUEL TARGETING AND THERMAL ENVIRONMENT EFFECTS ON SPARK-
IGNITION ENGINE MIXTURE PREPARATION PROCESS by loannis Kitsopanidis
Bachelor of Science in Mechanical Engineering
Aristotle University of Thessaloniki, Greece
(1998)
SUBMITTED TO THE DEPARTMENT OF MECHANICAL ENGINEERING IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE IN MECHANICAL ENGINEERING
AT THE
MASSACHUSETS INSTITUTE OF TECHNOLOGY
JUNE, 2000
D 2000 Massachusetts Institute of Technology
All rights reserved
Signature of Author
Department of Mechanical Engineering
May 15, 2000
Certified by
Wai K. Cheng
Professor, Department of Mechanical Engineering
Thesis Advisor
Accepted by
MASSACHUSETTS I
OF TECHNOLOGY i u it
SEP 2 0 2000
Ain A. Sonin
Chairman, Department Graduate Committee

This thesis is dedicated to my parents
Fi-bpyo and Kwvuravria
3

"You will never do it; nobody would sit above afire"
Thomas Edison to Henry Ford about the potential use of internal combustion engine as a means of transportation
5

FUEL TARGETING AND THERMAL ENVIRONMENT EFFECTS ON SPARK-
IGNITION ENGINE MIXTURE PREPARATION PROCESS by
Ioannis Kitsopanidis
Submitted to the Department of Mechanical Engineering on May 5, 2000 in Partial Fulfillment of the
Requirements for the Degree of
Master of Science in Mechanical Engineering
The thermal environment of the engine intake port has been identified as the most important factor affecting the mixture preparation process in a port-fuel-injection sparkignition engine. The main objective of this study is to quantitatively investigate the temperature effects at different places in the intake port on mixture preparation. The engine behavior is observed at different stages in the warm-up process during which the port temperature changes substantially, and also when specific port regions are locally heated. Thin surface heaters mounted on specific port regions are used to increase local temperatures up to 50K. The relative importance of port and block thermal environment on mixture preparation has been assessed by blocking coolant passages accordingly. All the above experiments have been carried out using two injectors with different spray patterns (single and dual spray), so that fuel targeting effects could also be evaluated.
The quality of mixture preparation has been evaluated by the engine response in throttle transients in which the above effects could be most clearly studied. The residual fuel increment, which is the cumulative value of the injected fuel that is not delivered into the combustible mixture during the transient, is used for comparing the effects of different thermal settings. This parameter has been estimated according to a three dimensional engine map of the gross indicated mean effective pressure, which accounts for the engine response and it is calculated from the pressure data, as a function of the peak pressure location, which accounts for the combustion phasing, and the relative air-fuel ratio.
Fuel targeting has been found to have negligible effect in most cases, while the overall engine temperature improves mixture preparation by a factor of 4 or 5. Port thermal environment has been found to be the most significant contributor to this improvement, while block' s contribution is smaller. Local heating corresponding to the regions of initial fuel impingement has almost no impact on mixture preparation since significant fuel redistribution takes place in the port. When the heaters are located just upstream the intake valves, a moderate effect of the order of 10 to 20% for cold engine has been recorded. The x-tau model, which characterizes fuel transport phenomena has been modified and used. The model predicts the experimental data with sufficient accuracy.
Thesis Supervisor: Wai K. Cheng
Title: Professor of Mechanical Engineering
7

In this short paragraph I would like to acknowledge some people who contributed substantially to this project. First of all, I feel honored to have been working with my advisor Prof W.K. Cheng for'almost two years. His guidance by the time I joined the
Sloan Automotive Laboratory at MIT in September 1998, has been invaluable. Also I would like to thank Prof J.B. Heywood for his occasional advice and the opportunity he gave me to get in contact with the members of the Engine and Fuels Consortium, who provided the right guidelines to my research. Also, I feel grateful to the Ph.D. candidate and friend, Jim Cowart, for his precious and continuous technical and scientific support until the implementation of this study. I do not think that the experimental part of this project would have ever been accomplished without the expertise of Brian Corkum and
John Baron. Finally, I also owe special thanks to my sister Despina, who as an expert in the English language, went through deeply and corrected the text of this thesis.
Ioannis Kitsopanidis, Cambridge, MA, 2000
9

A B ST R A C T .................................................................................
ACKNOWLEDGEMENTS.................................................................
TABLE OF CONTENTS....................................................................
LIST OF TABLES............................................................................13
LIST OF FIGURES.........................................................................14
NOMENCLATURE....................................................................... 16
CHAPTER 1: INTRODUCTION........................................................
1.1 M otivation.............................................................
17
17
1.2 B ackground ............................................................. 18
1.2.1 Fuel and air transport mechanisms in the intake port
1.2.2 Significance of temperature effects
1.3 O bjectives............................................................. 22
CHAPTER 2: EXPERIMENTAL SETUP................................................ 23
2.1 Engine and Dynamometer..........................................23
2.2 Injectors............................................................... 24
2.3 Local heating Equipment.............................................. 29
2.4 Sensors and Actuators...............................................34
CHAPTER 3: EXPERIMENTAL PROCEDURE AND APPROACH.............35
3.1 Fuel and Throttle Perturbations..................................... 35
3.2 Operating Conditions............................................... 38
3.3 Engine C alibration...................................................... 40
3.3.1 Introduction
3.3.2 Air and fuel flow estimation
3.3.3 Method validation
3.3.4 Data analysis
CHAPTER 4: PRELIMINARY TESTS................................................. 49
4.1 Reference Warm-up Tests..........................................49
4.2 Instrumentation Testing............................................... 55
4.2.1 Configuration A
4.2.2 Configuration B
7
9
11
11

CHAPTER 5: EXPERIMENTAL RESULTS........................................... 59
5.1 Introduction ............................................................. 59
5.2 Throttle Perturbations................................................ 60
5.2.1 Configuration A
5.2.2 Configuration B
5.3 Fuel Perturbations.................................................... 71
CHAPTER 6: SIMULATION OF FUEL TRANSPORT PHENOMENA............ 73
6.1 Introduction..........................................................73
6.2 X-Tau Model.........................................................73
6.2.1 Overview
6.2.2 x and tau values
6.2.3 Model modifications
6.3 X-Tau Model Simulation Results.................................... 83
6.3.1 Configuration B
6.3.2 Configuration A
CHAPTER 7: SUMMNARY AND CONCLUSIONS...................................93
7.1 Summary of Research Components...............................93
7 .2 C onclusions..............................................................94
R E FE R E N C E S ................................................................................
A P PE N D IX A .................................................................................
A P P E N D IX B .................................................................................
97
99
103
12

Table 1.1
Table 2.1
Table 2.2
Major-component fuel model for indolene............................ 21
Engine specifications.................................................... 24
Heating equipment specifications......................................30
13

Figure 1.1 Fraction of fuel in vapor phase......................................... 20
Figure 1.2 Gas phase relative AFR............................................ ...
21
Figure 2.1 Single spray injector pattern............................................. 25
Figure 2.2 Dual spray injector pattern................................................. 26
Figure 2.3 Fuel deposition estimation algorithm.................................... 28
Figure 2.4 Fuel deposition estimation: single spray injector..................... 28
Figure 2.5 Fuel deposition estimation: dual spray injector....................... 29
Figure 2.6 H eater and R TD ............................................................ 30
Figure 2.7 Heater configuration A.................................................. 32
Figure 2.8 H eater configuration B ..................................................... 33
Figure 3.1 Throttle versus fuel transient........................................... 36
Figure 3.2 Compression pressure during a throttle transient..................... 37
Figure 3.3 O perating conditions....................................................... 40
Figure 3.4 Air flow calibration.......................................................42
Figure 3.5 Combustion phasing calibration........................................ 43
Figure 3.6 Calculation of residual fuel increment................................. 44
Figure 3.7 Predicted and measured lean spike.................................... 45
Figure 3.8 Combustion variability: MBT spark timing for WOT............... 46
Figure 3.9 Combustion variability: MBT spark timing for 0.7 bar.............. 47
Figure 3.10 D ata analysis.............................................................. 48
Figure 4.1 Temporal variation of temperature: single spray injector.............. 50
Figure 4.2 Temporal variation of temperature: dual spray injector.............50
Figure 4.3 Spatial variation of temperature: single spray injector............... 51
Figure 4.4 Spatial variation of temperature: dual spray injector................. 52
Figure 4.5 Relative AFR during the warm-up......................................53
Figure 4.6 Air flow and intake pressure during the warm-up.................... 53
Figure 4.7 GIMEP at WOT and 0.7 bar during the warm-up.................... 54
Figure 4.8 Heater configuration A testing: single spray injector................. 56
Figure 4.9 Heater configuration A testing: dual spray injector.................... 56
Figure 4.10 Heater configuration B testing: temperature rise..................... 58
Figure 4.11 Heater configuration B testing: heat diffusion........................ 58
Figure 5.1 Configuration A, results: baseline, CVI............................... 61
Figure 5.2 Configuration A, results: baseline, OVI............................... 61
14

Figure 5.3 Configuration A, sample test results: cold and hot engine, dual spray injector, CVI.................................................. 62
Figure 5.4 Configuration A, results: local heating, single spray injector, CVI.. 63
Figure 5.5 GIMEP step for baseline and local heating cases during warm-up, single spray injector...................................................... 64
Figure 5.6 Configuration A, results: local heating, dual spray injector, CVI.....64
Figure 5.7 GIMEP step for baseline and local heating cases during warm-up, dual spray injector...........................................................65
Figure 5.8 Configuration B, results: dual spray injector, CVI..................... 67
Figure 5.9 Sample test results: baseline and no coolant case.................... 68
Figure 5.10 Temperature of region B 1................................................. 70
Figure 5.11 Temperature of region TC1............................................. 70
Figure 5.12 Relative A FR .............................................................. 71
Figure 5.13 Fuel perturbation baseline tests........................................... 72
Figure 6.1 x-tau m odel................................................................ 74
Figure 6.2 Curve fit of x-tau model to engine transient..........................77
Figure 6.3 x value during warm-up..................................................79
Figure 6.4 tau value during warm-up...............................................79
Figure 6.5 Temperature of region B 1 during heater activation and transient.... 84
Figure 6.6 Predicted and measured AFR with configuration B.................... 85
Figure 6.7 Predicted residual fuel increment with configuration B............. 86
Figure 6.8 Predicted engine transient behavior with configuration B.............87
Figure 6.9 Sensitivity analysis chart for engine transient behavior prediction... 88
Figure 6.10 Temperature of region Al during heater activation and transient.... 90
Figure 6.11 Predicted engine transient behavior with configuration B.............91
15

NOx
OVI
PFI
PID
RTD
SI
UEGO
WOT
AFR
BDC
CAD
CO
CVI
ECU
GIMIEP
HC
MIBT
MPI
NIST air-fuel ratio bottom dead center crank angle degree carbon monoxide closed valve injection electronic control unit gross indicated mean effective pressure hydrocarbons maximum brake torque multi point injection national institute of science and technology nitrogen oxides open valve injection port fuel injection proportional integral derivative resistance temperature detector spark ignition universal exhaust gas oxygen sensor wide open throttle
16

1.1 MOTIVATION
The role of the mixture preparation process in Spark-Ignition (SI) engines is to mix the appropriate amounts of air and fuel and to create a homogeneous charge required for robust combustion. Accurate control of the mixture Air-Fuel Ratio (AFR) delivered into the cylinders has a substantial impact not only on the combustion process, but also on engine's emission levels.
The stringent emission control regulations and the necessity for improved fuel economy and driveability dictate continuing developments in fuel metering and engine control systems. The automotive industry has met the challenge of accurate AFR control by modifying the mixture preparation process accordingly: carburetor has been replaced by injection systems controlled by a sophisticated Electronic Control Unit (ECU) of the engine. Especially after the introduction of the three-way catalytic converter, even higher precision in the control of the AFR is required. In order to create the appropriate environment for oxidation of carbon monoxide (CO) and unburned hydrocarbons (HC) and simultaneously for reduction of nitrogen oxides (NOx) at acceptable performance levels, a very narrow window of AFR variation is demanded (roughly ± 2% of stoichiometric) [1, 2]1. Keeping the AFR variation within such strict limits for a wide range of engine operating conditions including both steady state and transients can only be achieved if the physical processes are better understood, so that an appropriate model is used by the control system.
Past research has shown that the mixture preparation process depends on almost all the engine operating parameters [3]. Speed, load, thermal environment and injection
1 Numbers enclosed in square brackets denote references.
17

characteristics (timing and targeting) affect fuel metering in different ways. AFR control is believed to depend on fuel composition and temperature as well. Studying such a multi-dimensional system becomes a very complicated task; therefore the most important parameters need to be sorted out. It is believed that the mixture preparation process is most sensitive to the thermal environment of the engine [3]. Therefore, the current study focuses on temperature and fuel targeting effects on the fuel metering process.
1.2 BACKGROUND
1.2.1 Fuel and air transport mechanisms in the intake port
Almost all modem SI engines use Multi-Point Injection (MPI) systems, i.e. the fuel is injected for each cylinder individually. Most of these engines use Port-Fuel Injection
(PFI) systems, since the fuel is delivered into the port just upstream the intake valves.
Such approach enhances mixture preparation and minimizes cylinder-to-cylinder composition variation.
The fuel transport mechanisms in the intake port from the injector to the combustion chamber are very complicated [4]. A fraction of the injected fuel evaporates when being airborne and the rest impacts on the walls forming a liquid film or puddle. After depositing on the walls the fuel may be transported as a liquid rivulet before evaporating or transported into the cylinder as a liquid. Deposition on the walls, transportation as a liquid and atomization are additional mechanisms that increase the time constant of the fuel delivery process [1]. As a result some of the fuel cannot enter into the cylinder during the cycle event it was injected for. At steady state conditions, the amount of fuel leaving the liquid film and entering into the cylinder, either as liquid or as vapor, is equal to that entering the film since the amount of liquid film should be constant. Therefore the amount of fuel injected is the same as the amount of fuel entering the cylinder. However, because of the finite rate of fuel evaporation and the dynamic behavior of the liquid fuel film in the port, this is not true for transient engine operation. As a result of the so-called fuel lag less fuel enters the cylinder than the amount injected and the rest stays in the port as "residual fuel". Apart from the finite fuel transport time constant in the port, there is
18

also a finite air time constant due to the charging up of the manifold during a transient.
However, the air transport constant is much lower and as a result changes in the air and fuel flow into the cylinder do not occur in phase with each other. This mismatch of fuel and air flow causes significant lean or rich excursions in the mixture entering the cylinder when the engine is working under transient conditions [5]. Under such conditions appropriate modifications in fuel delivery from the injection system should be applied in order to meet driveability and catalytic converter performance issues. However, the socalled fuel compensation requires a very detailed model in order to eliminate these excursions. If the vaporization rate off the wall is sufficiently high, these spikes are expected to be small and vice versa. Hence, the thermal environment of the port becomes important in the mixture preparation process. Furthermore, if the fuel is directed to hotter port regions, less liquid film is expected to build up, so fuel targeting may also affect the whole process [6].
1.2.2 Significance of temperature effects
Before actually determining the experimental method for testing thermal environment and fuel targeting effects on the mixture preparation process, a rough estimation of the temperature impact has been performed. Specifying the temperature and pressure of an air-fuel mixture, the amounts of fuel in the vapor and liquid phase under equilibrium can be specified. Setting the mixture to correspond to typical AFR values in the port (e.g.
stoichiometric) the amount of fuel in the vapor phase and the corresponding amount in the liquid phase can be determined. Even though the conditions in the port are far from equilibrium, such calculations can give a rough indication of the sensitivity of the fuel fraction in vapor phase to temperature (engine thermal environment) and pressure (load).
Such calculation has been performed with SUPERTRAPP, which is a computer program developed by the National Institute of Standards and Technology (NIST) [7]. The program predicts the thermodynamic properties of mixtures when the temperature, pressure and mixture components are given. In Figure 1.1 the output of such calculation has been plotted. The sensitivity of the fraction of fuel in vapor phase to temperature and pressure is shown when a stoichiometric mixture of air and indolene is used. Indolene is a
19

pure hydrocarbon and the compounds used to model it are summarized in Table 1.1. The steep curves at low temperatures imply the significance of temperature especially at cold engine operation. Also as the pressure (load) increases the conditions for evaporation become less favorable suggesting another reason besides fuel lag for poor mixture preparation during hard accelerations. The relative pressure effects however seem to be less important than temperature effects.
100.
S 80
6
_
-
0.4 bar
-- -- 1.0 bar
Stoichiometric AFR (injected)
20
0 10
Figure'L
20 30
Temperature [ C]
40
Fraction of fuel in vapor phase
50 60
Similar calculations with lean mixtures have shown that the fraction of fuel in vapor phase is higher keeping the other conditions constant, which is consistent with past research suggesting that air improves the evaporation process [8]. Figure 1.2 converts the fraction of fuel in vapor phase to an equivalent gas-phase relative AFR. Indolene has been used as a fuel and the relative AFR corresponding to the whole amount of fuel (both liquid and vapor) is stoichiometric. The figure shows that for cold engine operation the amount of fuel compensation needed in order to obtain stoichiometric gas-phase AFR becomes significantly high, especially at high loads.
20

2.5
2.0
Q.)
15I~
0
1.0-
0.5 +-
0.0
0
Stoichiontric AFR (injected)
10
Figure 1.2
20 30
Temperature [ 0C]
40
Gas-phase relative AFR
0.4 bar
------- 0.7 bar
- - 1.0 bar
50 60
Component n-butane isopentane
2,3-dimethylbutane benzene
2,3-dimethylpentane toluene
2,2,4-trimethylpentane meta-xylene
2,2,5-trimethylhexane isopropylbenzene n-decane n-undecane n-dodecane
Indolene
Table 1.1 Major-component fuel model for indolene
Molar fraction [%] Molecular weight Stoichiometric AFR
9.12 58.12 15.5
14.15
8.42
72.15
86.18
15.3
15.2
2.54
8.36
78.11
100.21
13.3
15.2
17.62
23.77
2.10
1.41
6.86
3.76
1.60
0.29
92.14
114.23
106.17
128.26
120.20
142.29
156.31
170.34
13.5
15.1
13.7
15.1
13.7
15.1
15.0
15.0
97.14 14.7
21

1.3 OBJECTIVES
The fuel transport characteristics described in Section 1.2.1 imply that the fuel lag is observed under transient operating conditions. As a result the sensitivity of the mixture preparation process to engine thermal environment and fuel targeting can be better observed when engine operation is somehow perturbed, e.g. during a step increase in load or fuel. The higher the step, the higher the engine torque change during the transient and the easier it becomes to study fuel lag effects. The extent of the fuel lag can be assessed indirectly by keeping track of the AFR of the mixture both injected and entering the cylinder.
Therefore, the objective of this study is to relate the engine transient response during step changes in load to fuel targeting and thermal environment of the engine. The fuel targeting is to be observed by using injectors with different spray characteristics. The thermal environment is studied in many different ways. Tests are carried out during normal engine warm-up, when specific port regions are locally heated and when coolant flow rates to intake port and engine block are accordingly adjusted.
22

2.1 ENGINE AND DYNAMOMETER
A single-cylinder engine has been selected for the experiments of the current study.
Affecting the mixture preparation process and interpreting the corresponding results in a single-cylinder engine are considered to be easier tasks than in a multi-cylinder engine, which has cylinder-to-cylinder interaction effects. The engine used in the experiments is a
Ricardo Hydra Mark III single-cylinder research engine. The intake port and head design for that particular engine has been found to be obsolete since it has only 2 valves per cylinder and a hemispherical combustion chamber. In order to be consistent with the latest technology, the head from the five-cylinder Volvo B5254 FS engine has been used, after being cut and mounted on the Ricardo engine block. Special care has been applied to the sealing of oil and coolant passages. Gaskets, sealing pastes aluminum epoxy and appropriate plugs with o-rings have been extensively used for preventing any leakages.
This engine has a modem cylinder head with 4 valves per cylinder, a pent-roof combustion chamber, and the injector tip is just 10 cm away from the intake valves. All engine specifications are summarized in Table 2.1.
The intake, cooling and lubricating systems are not similar to a common car engine. The intake system has no plenum, therefore there is a relatively small volume between the throttle plate and the intake valves. This feature of the engine plays a significant role in eliminating the air flow dynamics during rapid throttle movements, as will be explained in Section 3.1. The cooling system is divided into two subsystems, one for the cylinder head and one for the engine block. The two subsystems do not communicate inside the engine, but they meet just before the water-cooled heat exchanger and they split again just after it. This allows the implementation of tests studying the sensitivity of mixture preparation to cylinder head and/or engine block thermal environment by preventing the coolant stream from getting into either of the subsystems. The oil system is uniform for
23

the whole engine, and the oil stream also passes through a water-cooled heat exchanger.
Both coolant and oil pumps are electrically driven.
The engine is coupled to a high-speed, air-cooled electromagnetic dynamometer. The
Eaton Dynamatic AF 6360 dynamometer has been used, which can both deliver and absorb power so that the engine can be initially motored at a specified speed before start firing.
Table 2.1
Engine specifications
Volvo B5254 FS
Spark-Ignition, Port-Fuel Injection, Single-Cylinder, 4-Valves per Cylinder, Naturally-Aspirated
Stroke [mm]
Bore [mm]
Connecting rod length [mm]
Clearance volume [cm
3
]
Intake valve diameter [mm]
Intake valve opening - 0.1 mm [CAD BTDC]
Intake valve closing 0.1 mm [CAD ABDC]
Maximum intake valve lift [mm @ CAD ATDC]
Exhaust valve diameter [mm]
Exhaust valve opening - 0.1 mm [CAD BBDC]
Exhaust valve closing - 0.1 mm [CAD ATDC]
Maximum exhaust valve lift [mm @ CAD ATDC]
90
83
158
53.28
31
4
56
8.45 @ 115.5
28
48
12
8.45 @ 252
2.2 INJECTORS
In order to estimate the fuel targeting effects on mixture preparation process, two different injectors have been used throughout the present study. Each of them has different spray characteristics, especially as far as the spray footprint is concerned. The first is Bosch EV1.3A [9], which generates a single hollow-cone spray, and the second is
Bosch EV6E, which generates a four filled-cone spray. The former is referred to as a single spray injector, while the second as a dual spray injector since according to rig tests
24

14
12 -
10-
8-
6
4-
2
0-
-24 -20 -16 -12 -8 -4 0 4 8
Angle from injector axis [deg]
12 16 20 24
Figure 2.1 Single spray injector pattern
25

26
14
12
10
8
S4-
2
-24 -20 -16 -12 -8 -4 0 4 8
Angle from injector axis [deg]
Figure 2.2 Dual spray injector pattern
12 16 20 24

carried out by Volvo, the four cones merge into two, right after the injector tip. Figures
2.1 and 2.2 present quantitatively the spray patterns for the two injectors respectively.
Since the study investigates the importance of fuel targeting on the mixture preparation process, an estimation of fuel distribution in the port is essential in order to interpret the results accordingly. Therefore, an algorithm has been developed which estimates the static fuel targeting on the port given the fuel spray pattern and the port geometry. The static fuel deposition estimation is only an approximation to actual fuel targeting because the air flow may deflect the spray, and the lower pressure present in the intake port and the increased fuel temperature may also distort the spray. However, since most tests are to be conducted with Closed Valve Injection (CVI), at high loads (intake pressure close to atmospheric) and the fuel is to be cooled, static fuel deposition is considered to be a relatively good assumption.
The static fuel deposition algorithm approximates the intake port area with simple geometric surfaces, such as planes and cylinders. Then, each surface is subdivided into finite plane area elements and the mass of fuel impinging on each element is considered to be proportional to the injector fuel spray pattern and the solid angle by which the injector tip sees the element. Therefore according to Figure 2.3 the mass of fuel impinging on each element is given by the following equation:
6AC am= cosO r2 f (P)
271sinq where f((p ) is the injector fuel spray distribution on rings at an angle p from the injector axis as shown in Figures 2.1 and 2.2. The rest of the fuel not impinging on the port is assumed to be landing on the back of the intake valves.
Since it is of interest to roughly quantify the fuel deposition at large regions of the port, names have been assigned to different regions: top, bottom, side and septum. Figures 2.4
and 2.5 show the intake port geometry approximation and fuel deposition estimation for the single and dual spray injector respectively. The discontinuity found in the fuel deposition estimation of the single spray injector is due to the change of port slope, but the effect is minor, because the fraction of fuel impinging there is relatively low. It is
27

obvious that the former injector creates a wide fuel deposition in the port, while the latter directs the fuel very close to the intake valves creating a very high peak.
Injector
28
8A
Figure 2.3 Fuel deposition estimation algorithm
Fraction of fuel per mm
2
(100 % = fuel deposited at each side of septum)
0.1
0.115
40
400
300
200
100
-----
--
4.
Top42Tp
-------
0
.---
Side
4%
%%
Bottom
2
50 100 ectof ip V001Valves:
Septum
7%Septum
1500
42%
Injector axis & tip
1t o
Figure 2.4 Fuel deposition estimation: single spray injector

Fraction of fuel per mm
2
(100 % = fuel from one cone)
3N.----
2,
1,
----- -38%
-
400
300
200
100
400
0 0
To
-
-Septum
50 eV
100
150
Side
35%
Bottom
1% alves: 26%
Injector axis & tip
Top
Septum
Figure 2.5
Fuel deposition estimation: dual spray injector
2.3 LOCAL HEATING EQUIPMENT
The impact of engine thermal environment on the mixture preparation process can be estimated by many different approaches. The first is during normal engine warm-up, the second during warm-up with the coolant from the cylinder head and/or the block removed, and the third by locally affecting the temperature of specific regions in the port on top of the gradual thermal environment change during the warm-up. Part of the investigation is also to determine those regions in the port, which are critical for the mixture preparation process. All above approaches combined together may shed some light on the sensitivity of mixture preparation process to port thermal environment.
In order to affect and control locally the temperature of specific regions in the port, small surface heaters are mounted on the desired locations. On top of them, rapid-response
Resistance Temperature Detectors (RTD) are connected to Proportional Integral
Derivative (PID) controllers for feedback. The equipment has been provided by Minco and the corresponding specifications are summarized in Table 2.2 (heater and RTD
29

locations are explained in the following paragraphs). Figure 2.6 shows one heater and one
RTD used in the tests. A reliable mounting of both heaters and RTD' s on the port has been achieved using the Epoxi-Patch resin from Dexter. The mounting paste and the heater insulation have been found stable both to temperature changes and gasoline effects after extensive bench testing.
Even though the heater power density is relatively small, the desired temperature rise could be achieved, as the instrumentation testing results suggest in Section 4.2. In the same section it is proved that there is minor heat diffusion to surrounding locations, therefore temperature rise can be achieved only locally.
Table 2.2 Heating equipment specifications
Heaters Location Width [in] Length [in] Resistance
[Ohm]
26.1
Model number
Al 0.50 (diameter)
A2
BI, B2, B3
0.25
0.75
Thickness: 0.012 in, Insulation: Kapton
RTD
0.25
0.75
10.0
35.0
Location Width [in] Length [in] Thickness [mm]
Al, A2, B1 0.30 0.30 0.020
HK5537R26.1L12D
HK5565R10.OL48D
HK5578R35.OL12D
Model number
S65
Platinum element: 100 Ohm at 0"C, 0.00385 (Ohm/C)/Ohm, Insulation: Kapton lPDZ24A
Controllers: CT15122, PID control, 1Hz
30
Figure 2.6 Heater and RTD

The PID controllers provided the option for closed-loop control of the heated regions.
This could be applied during a transient, where local port temperatures change substantially on a cycle time scale due to the significant amount of additional fuel landing and evaporating on the port. However, controlling the temperatures at such time scales needs very high controller frequency and heater power output, which are not available.
Since the pure thermal behavior of these regions is investigated, most tests are carried out with constant rate of heating instead of temperature control.
The fuel targeting estimation presented in Section 2.2 suggests that critical regions in the intake port for mixture preparation process could be the initial fuel impingement locations. In other words, it is of interest to investigate whether fuel targeting and local thermal environment effects are coupled. Furthermore, the regions just upstream the intake valves and the valve seats, where probably most of the fuel flows on its way to the cylinder regardless of fuel targeting, may have significant impact on the mixture preparation process. To investigate both the above concepts, two different heater location configurations have been used. Configuration A, which investigates the coupling between fuel targeting and local thermal environment, is shown in Figure 2.7. Heater Al is located inside the estimated fuel targeting region of the single spray injector, while heaters A2 are located in the estimated fuel targeting regions of the dual spray injector, which are symmetric with respect to the septum. All heaters have a relatively small surface area, in order to only affect the region of initial fuel targeting. The temperatures of all heaters are monitored since RTD's are mounted on top of them.
31

Region A2
Region A1
-- TC 1
TC2
Figure 2.7 Heater configuration A
Configuration B, which investigates the impact of the region very close to the intake valves, is shown in Figure 2.8. The heaters B1, B2 and B3 are symmetrically located with respect to the septum and in the circumference of the runners leading to the valves. These heaters make up a relatively large surface area and there is only a small region behind the valve stem left unheated. Since this is close to the exhaust valves and the only way fuel can get there is due to redistribution during the back-flow process, that region is considered to be already too hot to be additionally heated [10]. In this configuration only the temperature of one heater is monitored.
32

TC/
TC2
I el7
7~j
Region B2
Region B3
I
RPegion B1
Figure 2.8 Heat er configuration B
In order to understand port thermal environment better, two more temperature sensors have been added in both configurations. As shown in Figures 2.7 and 2.8 one thermocouple, referred to as TC1, has been mounted between one intake valve and heater
Al and one thermocouple, referred to as TC2, has been mounted on the back of one intake valve, on the part exposed to fuel deposition. The former thermocouple has been mounted with the same procedure as the heaters and RTD' s, while the latter has been welded on the back of the intake valve.
Even though the port has been crowded with all the above equipment and wires, the geometry has been barely affected because the heaters and RTD' s are extremely thin and the wires have been stuck on carefully designed grooves with the same mounting paste.
33

2.4 SENSORS AND ACTUATORS
Apart from the temperature sensors inside the intake port, thermocouples have also been installed at critical points of the experimental set-up in order to control the conditions of the tests as accurately as possible. Therefore, thermocouples are installed in the cooling, oil, intake, exhaust and fuel system. The temperature of the fuel is controlled via a heat exchanger in order to keep it at relatively low levels (less than 25
0
C). Past studies have shown that fuel temperature, influences substantially the spray, which is distorted when the fuel gets hot [11].
The air mass flow rate is monitored by a laminar air flow element, while the intake pressure by a vacuum pressure sensor. Certain corrections are applied to the air flow meter according to ambient temperature in order to incorporate air viscosity change effects.
An NTK Universal Exhaust Gas Oxygen (UEGO) sensor has been used to measure the oxygen content of the exhaust gases and therefore estimate the relative AFR. The sensor is located approximately 1 foot downstream the exhaust port and it is heated in order to give right signals in the very early stages of the warm-up.
The in-cylinder pressure is measured with a Kistler 6121 piezoelectric pressure transducer. The sensor is connected to a Kistler 5004 model charge amplifier. The pressure is recorded every Crank Angle Degree (CAD), and the Bottom Dead Center
(BDC) just before compression is set as a reference point.
The throttle movement is achieved by a high-speed pneumatic actuator provided by SMC.
The purpose of this actuator is to move the throttle adequately fast so that the new position is reached within one cycle at moderate rpm. That kind of actuator along with the plenum absence discussed in Section 2.1 eliminate the air flow dynamics during a transient, as explained in Section 3.1.
The injection and ignition timings are electronically controlled. The fuel pulse width is also electronically controlled and can be changed instantaneously allowing for a different fuel flow rate from one cycle to another.
34

FUEL AND THROTTLE PERTURBATIONS
There are two different ways to perturb the steady state operating condition of an engine: fuel transients and throttle transients.
The fuel transients involve a step change in the amount of fuel injected, while the air mass flow rate is kept constant. These tests are relatively easy to be carried out in the laboratory since there is no throttle movement and they can purely manifest the fuel lag effect under different conditions since no air flow dynamics are involved. However, in order to keep combustion stability at acceptable levels, the AFR cannot vary too much with respect to stoichiometry for both operating conditions. If the fuel perturbation goes from lean to stoichiometric mixture, then the relative AFR cannot be higher than 1.2 in the lean side in order to achieve good cycle-by-cycle repeatability. This is a significant restriction since the expected change in torque is then of the order of only 20% and therefore, engine thermal environment and fuel targeting effects can be hardly observed.
On the other hand, the mixture should not go to the rich side increasing the range of torque variation, because then the relationship between torque and AFR becomes highly nonlinear and the results difficult to interpret.
The throttle transients involve a step increase in load, i.e. both air and fuel flow change, so that the mixture injected is always stoichiometric. This kind of transient is more representative of actual engine operating conditions and the range of torque variation before and after the transient can be substantially high. However, the observed lean spike is due to both fuel lag and air flow dynamics since the transport time constant of air is also finite. These tests are difficult to be controlled, and the amount of fuel injected during the transient should vary according to the instantaneous air flow if the same injected mixture ratio is maintained. Fuel pulse width should change according to the air
35

trapped into the cylinder for every cycle, which is either calculated with a fillingemptying model [1]. All above calculations and measurements provide uncertainties to the system. To overcome these uncertainties, the air flow dynamics may be minimized by achieving a throttle transient in which the air induction changes within one cycle. This is only feasible when a high-speed throttle actuator is used and the capacity of the intake manifold is substantially low. In order to achieve the above requirements a high-speed pneumatic throttle actuator is used and the intake manifold plenum has been removed as discussed in Sections 2.4 and 2.1 respectively.
Figure 3.1 compares the variation of Gross Indicated Mean Effective Pressure (GIMEP), which is the work per cycle divided by the displaced volume, for a throttle and a fuel transient. Obviously the variation is significantly higher for a step change in load, and the thermal environment and fuel targeting effects can be then more clearly observed.
10
8
6
4
2
-20
I-
- -- ---- bar to WOT)
Fuel transient (Relative
AFR: 1.15 to 1.00)
0
Figure 3.1
20 40
Cycle number
60
Throttle versus fuel transient
80 100
36

Figure 3.2 proves that the air flow dynamics are actually eliminated in. the throttle transient configuration used here. In order to estimate the amount of air trapped into the cylinder before and after the transient, the cylinder pressure 1 CAD before spark discharge is plotted. The pressure at that point is called compression pressure and it is approximately proportional to the mass of air trapped [12]. The figure shows that the transient is actually achieved in just one cycle. Therefore, the pure fuel lag effect with significant GIMEP change before and after the transient can be clearly studied.
Most of the tests and results of the current study use throttle transients as a tool for estimating the fuel targeting and thermal environment effects on mixture preparation.
However, some results with fuel perturbations are also included.
18
16
I14 p...
12
U.
10
S 0
8
6
-20 -15
Figure 3.2
-10 -5 0
Cycle number
5 10
Compression pressure during a throttle transient
15 20
37

3.2 OPERATING CONDITIONS
As already stated part of the investigation is to evaluate the mixture preparation process as the engine warms up. Since engine behavior is mostly apparent during transients, the test procedure involves perturbation of normal operating conditions at different stages during the warm-up. Throttle transients are carried out depending on the engine coolant temperature and, in particular, they are actually initiated every 5
0
C from 30 to 70
0
C.
Therefore there is always enough time between two subsequent transients for engine stabilization. The transients are considered as being quasi-static because their time scale is significantly smaller than the time scale of the engine warm-up as explained in Section
4.1. In the same section the volumetric efficiency drop and the associated impact of the conditions of the transients at different stages during the warm-up are also discussed. The tests during a normal engine warm-up are referred to as baseline, as opposed to the tests when local heating effects are present or coolant has been removed.
Each test consists of two parts: the up-transient (tip-in) and the down-transient (tip-out).
The former starts from part load at steady state and finishes again at steady state at Wide
Open Throttle (WOT), while the other way round holds for the down-transients. This study uses only the up-transient for estimating the engine response, while the downtransient just brings the engine back to its starting point. It has been found that when the part load condition (starting point of a transient) corresponds to intake pressure of 0.7 bar and the new operating condition is at WOT (end point of a transient), the transient is sufficiently large to be readily analyzed.
Even though throttle transients are easier to analyze, some fuel perturbations have been also carried out. In that case the intake pressure is kept constant at 0.5 bar and there is a
15% step increase in fuel (the relative AFR goes from 1.15 to 1). Keeping the pressure below the critical value of 0.528 bar, the air flow is choked and therefore the mass of air inducted depends only on the conditions upstream the throttle. The temperature upstream the throttle does not change as the engine warms up due to a long rubber tube between the cylinder head and the throttle plate assembly, which acts as an insulator. Therefore, the volumetric efficiency drop is negligible and the conditions of transients at different stages during the warm-up are almost identical.
38

The speed for all tests is kept at 1500 rpm. That moderate speed has been chosen because it is representative of actual driving conditions and it is low enough for the sensors to respond on a CAD basis.
The current study concentrates on CVI since the impact of port thermal environment needs to be assessed. Injection starts approximately at 30 CAD BTDC while IVO occurs at 356 CAD ATDC; thus enough time is given to the fuel to sit on the port and evaporate with the heat transfer from the walls. However, since the fuel targeting effects are believed to be mostly apparent for Open Valve Injection (OVI), some tests with OVI have been also carried out. The injection timing is important since time should be allowed for the fuel to travel from the injector tip to the cylinder. Early injection timing is chosen in order to match the spray arrival to the valves with the maximum valve lift. Past tests conducted on that particular engine at similar operating conditions with flux sensors on the port have shown that it takes at least 50 CAD (at 1500 rpm) for the spray to travel from the injector tip onto the intake valves [10].
The ignition timing is kept constant both during a transient and the warm-up. As far as the throttle transients are concerned, the Maximum Brake Torque (MBT) spark timing for the part load is 22 CAD BTDC and for WOT is 16 CAD BTDC, when the engine is close to fully warmed-up conditions. As a compromise between engine knock for WOT and late burning cycles for the part load, it has been chosen to set the spark timing at 18 CAD
BTDC. Similar logic is followed to fuel perturbations, and the spark timing is then set at
26 CAD BTDC. In order to eliminate fuel effects, all tests were run with the same fuel, indolene, which is a pure hydrocarbon.
Figure 3.3 summarizes the operating conditions discussed above and also shows the valve lift profiles relative to the in-cylinder pressure trace.
39

30 30
25 -
Pressure
S20 -
15
Dwell angle
----- --
CVI
S10
5 -
0
0
EVO
OVI
------- -
IVO
10
8
6
4
90
2
180 270 360
CAD ABDC
450
Figure 3.3 Operating conditions
540 630 720
0
3.3 ENGINE CALIBRATION
3.3.1 Introduction
As already explained in Section 1.2.1, fuel targeting and thermal environment effects can be quantified by the fuel lag observed during transients. The fuel lag can be estimated by keeping track of both the injected fuel and the fuel that enters the cylinder. They obviously are not the same when the engine operating condition is perturbed. There are a couple of different ways to estimate fuel mass in the mixture that enters the cylinder and bums in one cycle [12].
One approach could be the use of an UEGO sensor, which directly gives the value of the
AFR based on the oxygen content of the exhaust gases. Even though the UEGO sensor is very reliable at steady state, its inherent time response of the order of 150 ms suggests that the output signal is the average value of 2-3 cycles at a moderate engine speed; therefore its signal cannot be used accurately in transient tests [13]. Moreover, due to the
40

pulsating exhaust flow, the signal exhibits an oscillatory behavior and accurate results cannot be obtained [12].
In the current study, the AFR of the mixture entering the cylinder during transients is based on the in-cylinder pressure transducer signal and an appropriate engine calibration.
The method is described in the following section and it yields very accurate results, if the in-cylinder pressure is accurately measured. However, the pressure transducer signal exhibits a drift, especially during the transients, because the sensor experiences a thermal shock due to the rapid temperature increase in the cylinder. Extensive testing showed that the drift is negligible during a single cycle (the signal auto-resets every cycle) to affect the pressure and therefore the pressure measurement is considered to be very reliable.
3.3.2 Air and fuel flow estimation
At steady state operation the air flow into the engine is accurately measured by a laminar flow meter described in Section 2.4. During a fast throttle movement however, the dynamics involved in the process prevent the sensor from measuring accurately the amount of air trapped into the cylinder. Since the amount of air trapped in the cylinder is proportional to the pressure in the cylinder just before the spark discharges (compression pressure), the steady state calibrated value may be used to relate the compression pressure in a cycle during the transient to the air mass trapped in that cycle. Figure 3.4 shows this correlation and since it is a straight line, liner interpolation can be carried out during a transient to obtain the mass of air trapped per cycle by measuring the compression pressure. Past research has shown that this linear behavior is independent of the relative
AFR [12]. As Figure 3.2 suggests, the air flow switches between two values during a transient, and it therefore becomes an easy task to obtain the mass of air trapped in the cylinder. The above method assumes that the partial pressure of air (which is proportional to its mass) is linearly proportional to the compression pressure. However, the partial pressures of fuel vapor and residual gas even though taken into account by the calibration curve, are considered as being constants during a transient. The error associated with this simplification is considered to be very small since the fuel vapor comprises approximately 2% of the total mass trapped at steady state and it becomes even smaller
41

during the transient and the residual gas fraction experiences very moderate changes from
0.7 bar to WOT [14]. The larger uncertainty in this method is the departure of the trapped charge temperature from its steady state value.
600
500
~300
2U !
8 10
Figure 3.4
12 14
Compression pressure [bar]
Air flow calibration
16 18
1
Obtaining the mass of fuel burned for every cycle during a transient is a more complicated task. For a particular (and known as described above) amount of air trapped in the cylinder, the work output per cycle is proportional to (a) the mass of fuel burned at that cycle and (b) the combustion phasing. The latter means that for a certain composition entering the cylinder, the work output is less than the maximum, if the combustion process takes place too early or too late. The work output is calculated from the pressure trace, as the integral of pressure with respect to cylinder volume during the compression and expansion processes. This is the gross work output and is divided by the displaced volume in order to obtain GIMEP, which is the most commonly used engine performance parameter. In order to account for the combustion phasing, the peak pressure location has been used, which is approximately at 15 CAD ATDC for optimum combustion phasing.
Therefore, GIMEP and peak pressure location obtained from the pressure signal can give the relative AFR (and so the mass of fuel burned since the air trapped is also known) after
42

an appropriate engine calibration at steady state, where the UEGO signal is reliable. The calibration for this particular engine is depicted as a chart in Figure 3.5 for the two different intake pressures of interest. The chart has been constructed after running the engine at different relative AFR' s and spark timings. The value of each point on the graph has been calculated as the average of approximately 150 cycles.
10
WOT
WOT
Relative AFR
0.98 to 1.30
0 .7 bar
Relative AFR
0.98 to 1.05
8 12 16 20 24
Peak pressure location [CAD ATDC]
Figure 3.5 Combustion phasing calibration
28 32
Both the air flow and combustion phasing calibrations depicted in Figures 3.4 and 3.5 respectively may shift from one day to another or even during the warm-up. To correct this shift, the steady state values of the laminar flow meter and the UEGO sensor just before and after each transient are taken into account and move the calibration curves linearly. A shift higher than 5%, has never been observed so the linear curve movement is supposed to be a very good approximation.
Figure 3.6 summarizes the procedure of the fuel mass burned estimation in the form of a flow chart. The difference with the amount of fuel injected during a transient can give the amount of fuel not burned. This amount cannot be attributed only to fuel lag effects [12].
The additional fuel injected during a transient and not burned, can follow one of the following paths: (a) enter the cylinder and get partially burned or totally escape the
43

combustion process and therefore be exhausted as CO or HC emissions respectively, (b) flow past the piston rings to the oil sump and (c) stay in the port or cylinder increasing the residual fuel. In practice, the first two possible paths comprise a very small fraction of the fuel not burned especially after cold start [15]; therefore if their contribution is assumed constant before and after the transient, the fuel not burned can be treated as the residual fuel increment (fuel lag effect).
Peak pressure location
UEGO value at steady state before & after transient
Relative AFR
Fuel mass
GIMEP
Fuel mass injected
In-cylinder pressure
Pressure at spark Air trapped
Figure 3.6
Air flow meter value at steady state before & after transient
Calculation of residual fuel increment
Residual fuel increment
3.3.3 Method validation
One way to check how well the method predicts the lean excursions in the mixture entering the cylinder during a transient is to compare it with the UEGO sensor. Even though this sensor has a large time constant compared to the time scale of a cycle as explained in Section 3.3.1 it can provide a rough indication of how well the method predicts the lean spike. This has been done in Figure 3.7 and it is obvious that the
44

agreement is very good (UEGO sensor signal has been shifted to allow for transport delay from the cylinder to the sensor). Due to the sensor's delay the lean spike seems to be underestimated and the whole signal is filtered.
1.3
1
.25 -
1.2 c:
LL
1 15-
1.1
c:
.cu
()
1 .05[
1
0.95 [
0.9
0
I-
UEGO sensor
Engine calibration
20 40
Figure 3.7
60 80
Cycle number
100
Predicted and measured lean spike
120 140
The method has been extensively tested at moderate and small lean spikes, where it appears to work sufficiently enough. However, this is not necessarily true for severe spikes observed at very low coolant temperatures. The calibration chart in Figure 3.5, which shows GIMEP as a function of relative AFR and peak pressure location, does not show a relative AFR contour line higher than 1.3. The reason is that calibration becomes difficult due to significant combustion variability at such lean mixtures. Figure 3.8 shows steady state tests with different mixture composition for the intake pressures of interest when the spark timing is set at 16 CAD BTDC, which is the MBT spark timing for WOT.
It is obvious that combustion variability becomes higher as the mixture gets leaner and this results to many slow burning cycles. In that case the calibration even for a 30% lean
45

spike may lead to spurious and unreal results overestimating these spikes. One way to deal with that problem is to advance the spark timing in order to allow more time for slow burning lean mixtures to bum.
10
-
8
6
* 1.0-WOT
* 1.1 -WOT
-1.2-WOT o 1.3-WOT
0
A 1.0-0.7
bar
0 g
8 WO
%0
0
0
40
0
A
V-
880go J9()
'9,
I..
0o 0
0
#66AJ#AAAA i
-
4-
-5 0 5 10 15 20
Peak pressure location [CAD ATDC]
25
Combustion variability: MBT spark timing for WOT
30 35
Figure 3.8
Figure 3.9 shows again the same steady state tests but when the spark timing is set at 22
CAD BTDC, which is the MBT spark timing for intake pressure 0.7 bar. Even though the combustion variability gets smaller, the calibration is still not very reliable at very lean mixtures. Moreover, engine knock becomes significant for stoichiometric mixtures at
WOT especially as the engine warms-up. As a compromise of cycle-by-cycle variability and engine knock, the spark timing is fixed at 18 CAD BTDC. At this spark timing knock is almost not present even at high temperatures, and significant combustion variability misleading the calibration is present only in the first one or two cycles of the spike for very cold engine (one or two transients). In such cases the calibration is done manually.
46

10
* 1.0 WOT
* 1.1 - WOT
- 1.2 - WOT
* 1.3 - WOT
* 1.0 0.7 bar
0 0
0
-'.2-W.T i II
*
;* V
!I
o u V
06
10 a 1 8 i
0 i
AA
A
a.
0 11
0
Figure 3.9
5 10 15 20
Peak pressure location [CAD ATDC]
25
Combustion variability: MBT spark timing for 0.7 bar
30
3.3.4 Data analysis
In order to assess thermal environment and fuel targeting effects on mixture preparation during transients a parameter should be chosen that characterizes engine transient behavior. Following the flow chart of Figure 3.6 this parameter is the residual fuel increment. This residual fuel does not refer only to the fuel staying in the port, but also inside the cylinder. Past research has shown that the "in-cylinder puddle" is not negligible compared to the "port puddle". Figure 3.10 shows a typical engine response to a step increase in load. The fuel dynamics prevent the engine from reaching the new steady state condition instantaneously. Integration of the area between the injected and burned fuel gives the residual fuel increment as explained in Section 3.3.2. The integration starts at the first cycle after the transient and ends when 95% of total change has been reached:
Residual fuel increment =
95% of total change f
(1hj1 jected
nibumred)dt transient initiation where the time unit is the engine cycle.
47

Due to combustion variability especially during the transient, it is difficult to locate the
95% of the total change. Therefore, an 11-point moving average has been used i.e. it takes into account 5 points before and 5 points after the particular point. The arbitrary selection of both 95% and 11-point moving average have been checked and it has been found that the final results are barely sensitive to them.
35
Residual fuel increment -- -
30
95% of total change
(11 point moving average)
0b
0n
25
20
-Fuel injected
-o-- Fuel burned
15
-40 120 0
Figure 3.10
40
Cycle number
Data analysis
80
48

REFERENCE WARM-UP TESTS
As already explained in Section 3.2, the experiments all focus on the warm-up period.
Tests are carried out at different stages of the warm-up period as the thermal environment continuously changes and local heating effects are added on top of this gradual temperature change. To make sure that the desired local temperature rise is achieved, the heaters are activated approximately 30 sec before each transient (which is sufficient time for them to reach the desirable temperature as shown in Section 4.2). They are turned off again after the engine stabilizes to the new operating condition. The warm-up is not a steady state condition by itself since the thermal state of the engine is not yet stabilized. In order to understand the conditions under which the step changes in load or fuel take place, reference charts of the engine thermal environment history during the warm-up have been constructed. Since the thermal environment is not only a function of engine design and combustion process, but also of fuel targeting, different charts for the two injectors (single and dual spray) have been drawn. Figures 4.1 and 4.2 show the temporal distribution of various temperature sensors during the warm-up for the single and dual spray injectors respectively (all abbreviations are explained in Figures 2.7 and 2.8).
The coolant temperature increase is approximately 2.5
0
C/min, which is very slow compared to a typical car engine (3 or 4 times faster). This slow rate is a result of the low combustion frequency (single-cylinder engine) and the thermal characteristics of the particular cooling system. This slow engine warm-up is particularly desirable since the time scale of the engine warm-up is substantially larger than the time scale of a transient
(less than 5 sec). Therefore, the conditions just before and right after a throttle transient can be treated as quasi-static since the thermal environment is almost identical.
49

140
120
C
&
4)
Cu
Cu
I-
4)
0~
S
4)
H
100
80
60
40
20
0
0
TCl
200
-A
-
Figure 4.1
- - - - C oolant -
400 600
Time after firing [sec]
800 1000
Temporal variation of temperature: single spray injector
A2
1200
140
120
0
&
S
H
U
U
I-
Cu
Cu
U
100
80
60
40
20
0
0
TCl
TC2
A2
Coolant ------
Figure 4.2
200 400 600
Time after firing [sec]
800 1000
Temporal variation of temperature: dual spray injector
1200
Comparing Figures 4.1 and 4.2 there are some differences, which are well worth observing. The valve (sensor TC2) stays approximately 10% cooler for the dual spray
50

injector implying that more fuel impinges on that part of the valve. The charts also imply that more fuel impinges on the regions close to the valves (A2 sensor) and less further upstream (Al sensor) when the dual spray injector is used.
Since there are already many temperature sensors in the port, a rough spatial temperature distribution in the port at different coolant temperatures can be determined. These distributions are presented in Figures 4.3 and 4.4. Even though this temperature distribution depends on fuel targeting, there does not seem to be any significant temperature gradient in the port for neither of the injectors especially at low coolant temperatures. The temperature increases rapidly only when the back of the intake valves is taken into account. The above observation implies that mixture preparation must be substantially enhanced if a lot of fuel impinges on the valves themselves since a substantial temperature difference exists between them and the port region [16].
140
120
U
100
40
20
80
60
0
0
Coolant temperature
- - - -
70
60
50
Figure 4.3
1 2 3 4
Distance from valves [cm]
5 6
Spatial variation of temperature: single spray injector
[0 C]
7
51

140
120
U
100
S 80
60
~4.
200
20
A
0
Coolant temperature [ C]
70
......- 60
---540
------30
1 2 3 4
Distance from valves [cm]
5
Spatial variation of temperature: dual spray injector
6 7
Figure 4.4
In order to complete the picture of a typical warm-up Figure 4.5 shows the variation of the relative AFR as the engine warms-up (at constant fuel flow rate) and Figure 4.6 the corresponding charts for the air flow and intake pressure. Heat transfer from the walls to the mixture forces the volumetric efficiency and as a result the air flow to drop as the temperatures rise. However, due to the particular intake system design (absence of plenum and massive metal parts) the drop is less than 3%. The mixture becomes richer as the coolant temperature increases mainly because of the volumetric efficiency drop. The drop in relative AFR is more than 3% because a larger fraction of fuel ends up in the sump and is exhausted as HC emissions when the engine is cold. Also part of the fuel builds up a liquid film in the port and cylinder especially during cranking [17]. Since the air flow drop is relatively small, it can be assumed that similar trends would be observed, if the throttle position changes, i.e. the drop varies linearly with the throttle position. If this is true, then transients conducted when the engine is cold and hot can be compared since the step increase in load would be approximately the same.
52

1.20
1.15
1.10
1.05
1.00
0.95
0.90
0
L---
200
Figure 4.5
400 600
Tint after firing [sec]
800
Relative AFR during the warm-up
1000 1200
0.75
0.73
Intake pressure
------- Mass air flow rate
4.00
3.90
cd~
0.71
0.69
0.67
--...-.....-.
3.80
S
3.70
3.60
0.65
0 200
Figure 4.6
400 600
Tin after firing [sec]
800 1000
Air flow and intake pressure during the warm-up
1200
3.50
53

Figure 4.7 shows GIMEP as the engine warms up for the two throttle positions of interest and constant speed. Since the mixture goes from lean to stoichiometric and the heat losses decrease as the engine warms up both curves slightly increase and they are almost parallel. Therefore, the step increase in load is almost identical for all tests during the warm-up and the thermal environment effects can be analyzed. However, as the air flow decreases the evaporation rate also decreases and as far as air flow is concerned the evaporation conditions are less favorable for a hot engine.
10
7
6
5
9
8
4
20
-
WOT
0.7 bar
30
Figure 4.7
40 50
Coolant temperature [C]
60 70
GIMEP at WOT and 0.7 bar during the warm-up
80
Figure 4.6 also shows the average intake pressure variation with the engine coolant temperature. As the engine warms up more fuel vaporizes in the port before entering the cylinder and therefore its partial pressure increases [17]. As a result, there is approximately a 4% increase in the intake pressure even though throttle position and engine speed are held constant. Since the rate of evaporation and therefore mixture preparation depend not only on temperature but also on pressure, conditions are more favorable for mixture preparation when the engine is cold as far as pressure is concerned.
Hence, comparing the fuel lag for cold and hot engine apart from temperature we also see pressure effects. However, as the NIST evaporation model in Figure 1.1 suggests
54

temperature effects are more important than pressure effects, and also the temperature increase between cold and hot engine operation is very large, while the pressure increase is only 4%.
To sum up, the temperature effects are by far most prominent than the air flow and pressure effects. The slight decrease in airflow and the slight increase in intake pressure are not considered important factors to bias the study of temperature effects.
4.2 INSTRUMENTATION TESTING
4.2.1 Configuration A
Before actually starting the experiments, the experimental set-up needs to be tested. In particular, it is important to check the impact of heaters on thermal environment, heat diffusion, fuel targeting and so on.
For configuration A, the check is twofold: whether the heaters are located at the regions where the fuel impinges, and also whether they increase the local temperatures at desirable levels. In order to assess the above, throttle transients have been carried out both when the heaters are off and on. Comparing the local temperatures before a throttle perturbation would give the temperature rise, while a temperature drop right after it, would imply a larger amount of fuel evaporating and therefore right fuel targeting.
Figures 4.8 and 4.9 show such transients for the single and dual spray injector respectively. Both CVI and OVI is shown and the engine coolant temperature is 55 0 C for all cases. The heated region for the single spray injector is Al and for the dual is A22
2 For all graphs shown later on with configuration A, the heated region for the single spray injector is A l and for the dual is A2.
55

90
80
&
U
70
0
I-
60
50
40
30
20
-40
i
-20
...... , ... .
Figure 4.8
i -
_--
-1
-
-
i--
OVI heaters on
CVI heaters on
OVI baseline
CVI - baseline
0 20
Cycle number
40 60
Heater configuration A testing: single spray injector
80
90
80 -
U
70-
60
50
40
30
20
-40
-
OVI heaters on
OVI baseline
------- CVI heaters on
--CVI baseline
..........
-20
Figure 4.9
0 20
Cycle number
40 60
Heater configuration A testing: dual spray injector
80
Both figures show a temperature increase of approximately 25K when the heaters are activated, which a significant change in the local thermal environment. Naturally the
56

temperature levels in Figure 4.9 are higher, since the temperature sensor is located just upstream the intake valves and therefore closer to the combustion chamber. The local temperature drop right after the transient implies that more fuel impinges on heaters, therefore the right regions seem to be heated.
On the other hand, the OVI tests imply a significant deflection of the fuel spray due to the air flow, especially for the single spray injector. Even for the baseline case (heaters off) the temperatures are higher and therefore less fuel lands on that particular region. There does not seem to be any temperature drop right after the transient, but a slight increase due to heat conduction from the combustion chamber. When the heaters are activated the temperature rise is of the order of 50K for the single spray and it stays about the same for the dual. The large temperature rise when the heaters are activated implies once again that there is no fuel landing on that particular region to cool it off. All these effects of OVI are mainly observed for the hollow-cone spray, which is believed to be substantially deflected
by the air flow [18].
4.2.2. Configuration B
Configuration B does not study any fuel targeting effects and therefore, the local temperature rise can be checked at steady state conditions. Since the heaters comprise a relatively large surface area in the port, heat diffusion to the rest of the port need to be checked in order to verify that only local heating is achieved.
Figure 4.10 presents the local temperature rise as recorded from sensor B1 for both injectors and different power densities to the heaters. The maximum temperature rise is of the order of 40K for the single and 30K for the dual spray injector. Since no fuel targeting effects are studied with configuration B, most tests are carried out with the dual spray injector, which is more widely used in practice. Also the maximum power density is delivered to the heaters in all following tests.
Figure 4.11 shows the temperature history of region TCI for the first 5 minutes of engine operation both when the heaters are off and on. Since the region TC1 is close to the heaters, its temperature history can indicate any heat diffusion in the port. As Figure 4.11
57

shows, the temperature of this region is barely affected by heater activation and therefore local heating is actually achieved.
90
80
0
&
70
60
0
U
I-
50
40
Single spra
30
20
-40
__Dual spray
0
-
Figure 4.10
40 80
Cycle number
120 160
Heater configuration B testing: temperature rise
12W/in 2
7 W/in
2
4 W/in
2
-12 W/in 2
7 W/in 2 _
~~~4 W/in2
2C 0
60
&
50
40
U
E-
30
20
10
Baseline
------- Heaters on
0
0 50
Figure 4.11
100 150
Tine after firing [sec]
200 250
Heater configuration B testing: heat diffusion
300
58

5
5.1
INTRODUCTION
In Section 3.1 it was explained that the throttle transient is more favorable than fuel transient for assessing the mixture preparation process. Therefore most of the following tests involve a step increase in load while the injected air-fuel mixture is always kept at stoichiometric.
Section 5.2 includes results with throttle transient tests. The first part (Section 5.2.1) concerns configuration A, which studies the impact of engine thermal environment during a normal warm-up, fuel targeting and local heating effects on the mixture preparation process. In this configuration local heating concentrates on the regions of initial fuel deposition and the coupling between local thermal environment and fuel targeting is studied. The second part (Section 5.2.2) concerns configuration B, which studies the effect of local heating of regions just upstream the intake valves regardless of fuel targeting. Moreover, the impact of port and/or block thermal environment alone on mixture preparation is observed by blocking certain coolant passages accordingly.
Section 5.3 includes some fuel transient tests. As explained in that section, results are not obtained as readily as in throttle transient tests, but some sample baseline test results are discussed.
59

5.2 THROTTLE PERTURBATIONS
5.2.1 Configuration A
As already explained in Section 3.2, throttle transients are initiated at various stages during the warm-up. The residual fuel increment as defined in Section 3.3.4 is used as the parameter for comparing engine transient behavior. Therefore, all the following graphs present the residual fuel increment as a function of coolant temperature, which represents the overall engine thermal environment. To gauge the magnitude of the residual fuel increment, the steady state fuel mass injected before and after the load increase have been set at 21 and 34 mg per cycle respectively.
Figure 5.1 presents such graph for the baseline case with CVI and both injectors. The impact of the overall engine thermal environment is substantial due to the decreasing fuel lag effect as the engine warms up. The residual fuel increment drops approximately by a factor of 4 or 5 as the engine reaches the fully warmed-up condition. Even though the method followed in the present study is somewhat different from what past research has followed, both the trend and the absolute levels are consistent [3, 19,20]. The same figure suggests that no fuel targeting effects can be observed. Since CVI is used the fuel has time to flow towards the back of the intake valves before they open regardless of its initial fuel deposition. Furthermore, the relatively flat temperature distribution in the port shown in Figures 4.3 and 4.4 makes fuel targeting less important unless substantially different fraction of fuel impinges on the valves, which is not the case for these two injectors and the particular port geometry.
Figure 5.2 shows the corresponding results if OVI is used. Both the trend and the absolute levels of the residual fuel increment are approximately the same as for the CVI case when the dual spray injector is used. The dual solid-cone spray seems to be hardly affected by the air flow as also Figure 4.8 suggests. However, the single hollow-cone spray is prone to deflection due to air flow as shown in Figure 4.8 and as a result the mixture preparation process deteriorates especially when the engine is cold. The fuel is redistributed in the
60

port and in the cylinder and it may be transported to regions less favorable for evaporation
[19].
120
64
100
80
60
-P
+ , o
_-----
--
+ Dual spray o Single spray
Linear (Dual spray)
Linear (Single spray)
CA
40
0- 0
20
0
20 30 8 )
Figure 5.1
40 50
Coolant temperature [0C]
60
Configuration A, results: baseline, CVI
70
120
100
80
CO
U cj4
60
40
20
0
20 30
. 0
+ Dual spray o Single spray
(Dual spray)
...-- .
Linear (Single spray)
80
Figure 5.2
40 50
Coolant temperature [ 0
C]
60
Configuration A, results: baseline, OVI
70
61

Comparing Figures 5.1 and 5.2, it can be concluded that CVI and OVI do not have any particular difference as far as mixture preparation is concerned if the spray is not affected too much by the air flow. Past research suggests that substantial differences in mixture preparation between CVI and OVI occur during engine cranking and cold start; the current results suggest that the differences are not so large during the warm-up process
[15].
In order to observe more clearly the impact of engine thermal environment on the residual fuel increment, Figure 5.3 demonstrates the engine transient response for cold (30
0
C) and warm conditions (70
0
C). In that particular example CVI and the dual spray injector have been used. The two responses suggest that both the initial response and the time constant of the transient improve substantially as the engine warms-up.
10
7
6
5
9
8
2 -
Hot operation
------- Cold operation
4
-40
Figure 5.3
-20 0 20 40
Cycle number
60 80 100
Configuration A, sample test results: cold and hot engine, dual spray injector, CVI
Figure 5.4 shows the local heating effects for single spray injector and CVI 3 . When the heaters are activated the trend of the residual fuel increment with respect to engine
3 OVI is not used any more since less fuel impinges on the heated regions.
62

coolant temperature is approximately the same. However, local heating does not improve the transient response at all but on the contrary the residual fuel increment seems to be
10% larger. Even though all above tests cannot be strictly repeatable (throttle transients on top of the overall transient nature of the warm-up), this result has been found to be consistent with all experiments.
Figure 5.5 shows the step increase in load in terms of GIMEP as the engine warms up for both cases (heaters on and off), where:
GIMEP step = GIMEPWOT GIMEPO.
7 bar
120
100
80
+ Heaters on
3 Baseline
Linear (Heaters on) -
Linear (Baseline)
Ej-
60
-
M
40
'-.1
20
0
20
Figure 5.4
30 40 50
Coolant temperature [C]
60 70
Configuration A, results: local heating, single spray injector, CVI
80
-.
63

64
4.2
4.0
U
3.8
3.6
I'l 0
c Baseline
Heaters on
+ 0
3.4
Figure 5.5
3.2
20 30 40 50
Coolant temperature [0C]
60 70 80
GIMEP step for baseline and local heating cases during warm-up, single spray injector
120
100
U
U
0
80
60
-o
U
40
20
0
20
., u
+ Heaters on o Baseline
Linear (Heaters on)
Figure 5.6
30 40 50
Coolant temperature [ 0
C]
60 70
Configuration A, results: local heating, dual spray injector, CVI
80

4.2
4.0
3.8
0
0 Baseline
S*Heaters on
01
E0
3.6
3.4
2')
20
Figure 5.7
30 40 50
Coolant temperature [0C]
60 70 80
GIMEP step for baseline and local heating cases during warm-up, dual spray injector
This figure shows clearly that the amplitude of the transient is almost identical for both cases and therefore the result shown in Figure 5.4 is not biased.
The same trends are also observed when the dual spray injector is used as Figure
5.6
suggests. Again the amplitude of the transient is almost the same for both cases (actually
4% higher for the baseline case) as shown in Figure 5.7.
All above results suggest that there does not seem to be any coupling between the local thermal environment and the fuel targeting. Some plausible explanations are: fuel splashing when landing on the port, flow towards the back of the intake valves, redistribution due to the back-flow process and small heated compared to fuel deposition areas. However, the result that when the heaters are activated, the residual fuel increment is consistently 10% higher than in the baseline case should be looked at more carefully.
Even though the difference is not that significant it cannot be simply attributed to experimental variability due to its consistency. Two possible explanations could be (a) an airflow effect due to volumetric efficiency drop when the heaters are activated and (b) the initial condition of the residual fuel just before the transient, which is different when the
65

heaters are on and off. The former issue is discussed in Section 5.2.3, since the heaters in configuration B comprise a larger surface area than in configuration A, so the volumetric efficiency drop (if any) must be then more important. The latter is studied more carefully in Chapter 6, where a mixture preparation model is used to address that issue in more detail.
5.2.2 Configuration B
Heater arrangement in the port according to configuration B examines the impact of the regions just upstream the intake valves on mixture preparation regardless of fuel targeting. Therefore, the regions B 1, B2 and B3 are heated, while the temperature of B 1 is monitored. All tests are done with CVI and they use the dual spray injector. As opposed to heater arrangement of configuration A, heaters occupy now a significant surface area in the port. Therefore, the corresponding volumetric efficiency drop needs to be estimated.
This is done in the Appendix A and the result suggests that the effect on volumetric efficiency can be neglected since it is less than 0.2%. Apart from local heating effects, port and/or block thermal environment alone are also studied by blocking certain coolant passages accordingly. With engine coolant removed, there is no point in measuring coolant temperature any more. Therefore, transients are initiated at different stages during the warm-up based on time passed after the engine starts firing. In the current study throttle perturbations are carried out every 1 minute until the engine completes
5 minutes of running. Running the engine more than 5 minutes without any coolant in the port and/or block has been avoided in order to prevent any engine thermal cracking. Even though running an engine without coolant is not a common practice in car engines and it affects the volumetric efficiency, it is believed that such tests could illustrate very clearly the sensitivity of mixture preparation on engine thermal environment. What is more, the relative effect of port and/or block thermal environment on mixture preparation can be addressed.
66

Figure 5.8 shows the residual fuel increment during the warm-up for different cases. The baseline case is similar to that in Figure 5.6, where the engine thermal environment effects during a typical warm-up are clearly seen.
80
'-"
50
40
70
60
30
20
10
0
0
A- Baseline
-- a-
-- Heaters on
-
-0- No coolant in block
- -<-No coolant in port
---- No coolant
1 2 3
Tinm after firing [min]
4
Figure 5.8
5
Configuration B, results: dual spray injector, CVI
6
When running without any coolant the mixture preparation process is substantially enhanced since generated heat in the combustion chamber cannot be carried away easily and as a result, both port and block temperatures increase. The effect is particularly important after the first 1-2 minutes because the thermal capacity of both port and block prevents the temperatures from increasing very rapidly. Figure 5.9 illustrates the sensitivity of mixture preparation to engine thermal environment by showing the engine transient response 5 minutes after firing for the baseline case and when the coolant has
4 Note that the absolute values of the residual fuel increment are slightly different when compared to the baseline case in Figure 5.6 mainly due to the change in ambient conditions. Tests with configuration A, were run in the summer with high temperature and humidity, while tests with configuration B, were run in the winter.
67

been removed. When coolant passages are blocked the engine stabilizes to the new operating conditions in just 3 cycles.
10
9r-"
8
7
6
5
--Baseline -
-o--No coolant
A
-20 -10 0 10
Cycle number
20 30
Sample test results: baseline and no coolant case
40
Figure 5.9
Tests without any coolant in the port suggest that the thermal environment of the port alone seems to be particularly critical for the mixture preparation process. On the other hand, when running without any coolant in the block, residual fuel is substantially reduced compared to the baseline case but the impact of block thermal environment alone is not as important as the port. This observation can be also interpreted in the following way. Since port thermal environment mainly affects port residual fuel and block thermal environment affects cylinder residual fuel, the fuel lag in the port is more important because there is more residual fuel in the port than in the cylinder. This conclusion has been also drawn by past research using mixture preparation simulations [21, 22].
When the heaters are activated, the residual fuel increment seems to decrease by a factor of 10 to 20%. Comparing this result with that of Figure 5.6, where the heater arrangement of configuration A has been used, it is clear that the regions just upstream the intake valves are the most important port regions for mixture preparation. For the very early stages of the warm-up (1st minute), heater activation provides better mixture preparation
68

even compared to no-coolant tests because the thermal capacity of the heaters is much smaller than port and/or block thermal capacity. The slight increase of residual fuel increment in the second transient is very repeatable and it is attributed to some kind of interaction between the first two transients.
In order to acquire better understanding of the sensitivity of mixture preparation process to thermal environment for all above cases, the temperatures of some port regions and the relative AFR have been monitored. Figure 5.10 shows the temperature of region B1
(locally heated region) and Figure 5.11 the temperature of region TC1 (close to heated region) as the engine warms up for the first 5 minutes. Figure 5.10 shows that the heated region temperature is substantially higher when the heater is activated compared to both the baseline case and the case when the coolant is removed. The gradient of temperature when no coolant is in the engine is much steeper compared to all other cases. It is also interesting to observe that there is no impact on port thermal environment when the coolant is removed from the block. Figure 5.11 shows that the temperature of the region not affected by local heating is substantially higher when there is no coolant in the port compared to both baseline case and the case when the heaters are activated. This implies that the temperature of all non locally heated regions in the port are much higher when there is no coolant in the port, and therefore the shape of the curves in Figure 5.8 can be explained. As far as the very early stage of the warm-up is concerned, non locally heated region temperatures (as region TC1) seem to be the same for all cases due to port and/or block thermal capacity. However, the locally heated region temperature is significantly higher for the case when heaters are activated compared to all other cases explaining the low residual fuel increment at the very early stages of the warm-up when the heaters are on. Figure 5.11 shows that port temperatures are the same for the baseline case and the case when the coolant has been removed from the block, implying again that port thermal environment is not affected by block thermal environment. The above figure also proves once again that there is no heat diffusion from the heaters to the rest of the port and local heating is actually achieved. Finally, comparing Figures 5.10 and 5.11 it is observed that even if region B1 is closer to the combustion chamber than region TC1, it stays cooler since more fuel impinges and evaporates there.
69

70
60
50
U
40
30
0
20
10
0
0
No coolant in port
No coolant
No coolant in block
Baseline
1
Figure 5.10
2 3
Tine after firing [mim]
Temperature of region B 1
4 5
U
CL.,
60
50
80
70
40
U
30
E-
20
10
0
0
At
No coolant
No coolant in port-
21 J Heaters on
-Baseline
No coolant in block
1
Figure 5.11
2 3
Time after firing [min]
Temperature of region TC 1
4 5

1.20
1.15
1.10
1.05
1.00
0.95
0.90
0.85
0.80
0
-
-
-No coolant in head
1
1
Base'line
2 3
Time after firing [sec]
No coolant in block
4
Figure 5.12 Relative AFR
5
In Figure 5.12 the relative AFR variation for all above cases is depicted. Obviously the mixture preparation even for the quasi-static operating conditions during the warm-up is not the same for the above cases.
5.3 FUEL PERTURBATIONS
As already stated in Section 5.1 there are some sample baseline fuel transient tests. These tests are considered to be significantly easier to implement because air flow is constant during the transient and it is also constant during the warm-up since the flow is choked.
However, the residual fuel mass change is substantially smaller compared to throttle transient tests and as a result experimental uncertainties become important. Even though a similar engine calibration to the one described in Section 3.3 has been applied, test repeatability was not found sufficient enough to draw reliable conclusions. Figure 5.13 shows the results of the baseline case with both injectors. Tests have been also carried out with a locally heated port according to configuration A, but no effects have been recorded. Since no significant effects have been recorded with throttle transients when the
71

residual fuel changes are larger, it is reasonable fuel transient tests effects to be unobservable.
20
16
-+- Single spray
0- Dual spray
12
U
0
8
4
0
20 80 30
Figure 5.13
40 50
Coolant temperature [
0
C]
60
Fuel perturbation baseline results
70
72

6.1
INTRODUCTION
Summarizing the experimental results of Chapter 5, local heating effects do not seem to have any large impact on the mixture preparation process. Configuration A does not seem to improve mixture preparation at all (sometimes it deteriorates it), while as far as configuration B is concerned even though it is the right place for local heating, the improvement is substantial but it is still less than 20%. Studying complicated physical processes is a challenging task and finding a consistent explanation very often requires investigations on a theoretical basis. Working in this direction, it is suggested that a theoretical physical model characterizing the fuel transport mechanisms in the intake port should be applied. The model should be relatively simple in order to get semi-quantitative results but also sophisticated enough so that the physics can be sorted out.
6.2 X-TAU MODEL
6.2.1. Overview
In order to identify the fuel transport characteristics in the intake port, the best curve fit to the lean excursions of the mixture entering the cylinder that accompany throttle transients should be found. A very simple fuel transport model, which provides a reasonably good prediction of these excursions, is the x-tau model [6, 12, 19, 20]. This model is based on' the phenomenological description illustrated in Figure 6.1. The governing equations are the following:
73

11
= x e m
= (1 - x)m'j + Mr
(6.1)
(6.2)
In the above equations rihe is the mass fuel flow rate delivered into the cylinder and burned, tih is the mass fuel flow rate injected and rihi is the rate of change of the residual fuel mass. Even though Figure 6.1 shows the residual fuel in the port, it is not specified whether mr corresponds to port or cylinder residual fuel. Also, it is not specified whether the fuel leaving the residual fuel mass is in liquid or vapor phase. The parameter x represents the fraction of injected fuel that enters the residual fuel mass and C the time constant of release from this mass. These parameters are determined experimentally as described in Section 6.2.2. Therefore, the amount of fuel entering the cylinder and getting burned comes from two different paths: (a) directly from the injector: (1- x)rhin and (b) through the residual fuel mass: m, /T. The former is assumed to remain airborne, i.e. it evaporates before impinging on port walls, while the latter has already undergone fuel lag and therefore it is very likely that it comes from previous injections.
74
Figure 6.1 x-tau model

Setting one engine cycle as a time basis for the model, the discrete form of the ordinary differential equations (6.1) and (6.2) is derived when integrating over a cycle [23]:
. -1
m' = m' + xm -
i-1A t
A (6.3)
M =(1- x)m' + m
At
(6.4)
At is the time duration of a cycle, which is a constant at constant speeds.
Under steady state conditions the residual fuel mass is constant, i.e. m = m'-I so equation (6.3) yields: m- =xm lfin'At
(6.5)
Substituting back to equation (6.4): me =M (6.6)
Therefore, for steady state conditions the amount of fuel entering into the cylinder and getting burned equals to the amount of fuel injected. Even though fuel lag exists, there is no net effect because the amounts of fuel entering and leaving the residual fuel mass are the same.
However, under transient operating conditions this is not true due to the fuel lag effect.
The amount of fuel that remains airborne and enters into the cylinder directly from the injector responds to transients in one cycle, while the amount that passes through the residual fuel mass has a finite response time because the residual fuel mass increases during the transients. This increase is due to the fact that the mass of residual fuel scales with the mass of fuel injected as equation (6.5) implies (if x and T remain constant) and therefore, the residual fuel mass tries to reach a new steady state. Physically, the increase is attributed to residual fuel mass inertia as more fuel enters and less fuel can leave that mass during a transient. Since the residual fuel mass increases, equation (6.3) reads: m'-I r
< xm' in'
A
(6.7)
75

Substituting back to equation (6.4):
(6.8) inj
Therefore, during a transient and until the residual fuel mass reaches a new steady state condition, the amount of fuel burned is smaller than the amount of fuel injected due to the fuel lag effect.
In order to quantitatively estimate the amount of fuel lag and therefore the residual fuel increment during a step transient, the difference equation (6.3) is solved and then it is tMa substituted back to equation (6.4). If the fuel is increased in one step from min to i
(the superscripts b and a stand for before and after the transient respectively), then the solution to equation (6.3) is (Appendix B contains the detailed calculations): mi m b mr =Mra + a i -1 a xmi
(6.9)
The transient is initiated at i=1. In the above equation a =1- At and mb is the steady state residual fuel mass before the transient and according to equation (6.5) it is given by: mb = xm r njAt
(6.10)
Substituting equations (6.9) and (6.10) back to (6.4) and denoting Ami. = m' m (step increase in fuel): mi = ma - xAm.. in]j
At
T
(6.11)
Initial response (i=1): me = nin
+ Am. -(1 x)
New steady state (i->oo): mi = ma
The initial response implies that the mass of fuel burned jumps by Am (1- x) before starting to rise with the time constant c. This behavior has been found to agree well with the experimental data. Equation (6.11) implies that engine response to a step input is similar to a first order system. Even though this is an oversimplification of what happens
76

in the port during a transient, it turns out to be consistent with experimental data. Figure
6.2 shows a typical engine response to a step increase in load along with the x t model curve fit. The response reminds of a first order system and the model curve fit is very good if appropriate values of the parameters x and are chosen. Although there are models based on more fundamental description of the physical processes in the intake port [5], the x -r model has been found sufficiently accurate and relatively simple, therefore it is used in the current study.
35
30
C.)
C.) cj~
25
20
-Fuel injected
Fuel burned
------- Curve fit to x-tau model
15
-40 -20
Figure 6.2
0 20 40
Cycle number
60
Curve fit of x-tau model to engine transient
80 100
6.2.2 x and tau values
As already explained in Section 6.2.1, x and r values are determined experimentally.
Therefore, transients are initiated and the corresponding x and r values are determined so that the best curve fit to experimental data is obtained. Constructing look-up tables of x and r values at different operating conditions, the transient behavior of the engine at similar conditions can be predicted by interpolation. In practice such tables can be found in the ECU of a car engine, so that a lean excursion in the mixture delivered into the
77

cylinder can be predicted and fuel compensation is made accordingly in order to avoid this lean spike.
The current study deals with transients at fixed operating conditions: 1500 rpm and step increase in load from intake pressure of 0.7 bar to WOT. During the warm-up period, which is studied in particular, x and t do not stay constant. The dependence of x and T on engine coolant temperature is shown in Figures 6.3 and 6.4 respectively. These curves have been obtained so that the x-T model curve fit predicts engine transient behavior as accurately as possible. Figure 6.3 shows that x decreases as the engine warms up, since more fuel evaporates when being airborne, so the fraction of liquid fuel impinging on walls becomes smaller. Figure 6.4 shows a similar dependence of
T on engine coolant temperature. As the engine warms-up there is significant heat transfer from the walls
(port or cylinder) to the residual fuel mass. Therefore, vaporization is enhanced and the time constant of fuel leaving the residual fuel mass becomes smaller. Since the fuel leaving the residual mass is not specified whether it is in vapor or liquid phase, apart from enhancing vaporization, fuel viscosity becomes smaller. Therefore, the increasing coolant temperature causes both vapor and liquid fuel leaving the residual mass and many complicated phenomena occur in the intake port.
Figures 6.3 and 6.4 show a kind of linear dependence of x and
T to coolant temperature during the warm-up period. Even though the curve fit is not exactly linear, a linear function is used a lot in the subsequent calculations in order to simplify the analysis. The trends of the curves illustrated in Figures 6.3 and 6.4 are consistent with past research [6,
12, 19, 20].
78

1.6
1.2
CI
0 0.8
0.4
0
20
0.5
0.4
0.3
0.2
0.1
0
20 30 40 50
Coolant temperature [
0
C]
60
Figure 6.3 x value during warm-up
70 8 0
80 30
Figure 6.4
40 50
Coolant temperature [
0
C]
60 tau value during warm-up
70
79

6.2.3. Model modifications
The analysis described so far, assumes the existence of one residual fuel mass with certain properties fully described by the parameters x and T. These properties depend on operating conditions, whose impact is reflected on x and T values. Since one residual fuel mass is assumed, which is in a liquid form, it is also referred to as puddle. However, in reality the fuel not entering the cylinder may be randomly distributed in the intake port, due to splashing, exhaust gases back-flow and air flow entrainment effects. The fuel is not concentrated in one region and therefore, the term "puddle" does not seem to sufficiently characterize the fuel not entering the combustion chamber and the term
"residual fuel mass" is considered to be more appropriate.
Hence, since the residual fuel mass is distributed all over the intake port and cylinder walls, each mass element should have different properties according to the local conditions. These local conditions are reflected by the values of x and t as explained above. In other words, the temperature gradients in the intake port and cylinder should cause a gradient in x and t as well. Most favorable values are expected inside the cylinder and larger fuel lag is expected far upstream the intake valves. The wall temperature gradient mainly affects T and the air temperature gradient mainly affects x value. Such calculation becomes very complicated since both temperature and fuel distributions are required in order to determine a large set of x and T values that sufficiently describe the whole intake port and cylinder residual fuel mass. Therefore, the x- t model assumption of one residual fuel mass with common properties is not physically right. The results however are sufficiently accurate because the x and T parameters obtain mean spatial values without characterizing the behavior of the residual fuel in any particular location.
Since the x- t model picks mean spatial values for x and T parameters, determining the corresponding values when the port is locally heated should be looked at carefully. When the port is locally heated at specific regions, that part of the residual fuel lying in the vicinity of the heater has significantly different transport characteristics than the rest. As a result, local x and T values are different. Following the approach of the x- t model, which assumes mean spatial x and r values, two sets of these parameters should now be
80

used. The first set must refer to that part of the residual fuel affected by heater activation and the second to the rest of it, which remains unaffected. This modification of the x-t model requires the estimation of two things: (a) the fraction of fuel affected by heaters, which is referred to as fuel targeting and (b) the values of both sets of x and z parameters. The fuel targeting is determined according to the fuel deposition estimation described in Section 2.2 and it is different for the two heater arrangements (configuration
A and B). The values of x and t for the fraction of fuel that remains unaffected by heater activation are the same as the baseline case and they can be found directly from Figures
6.3 and 6.4. In order to estimate x and t for the rest of the residual fuel, Figures 6.3 and
6.4 can be used again since the local temperature rise is measured. For instance, if heater configuration B is used with the dual spray injector, which results in a temperature rise of
30K, and if the transient is initiated when the coolant temperature is 40 0
C, then the two sets of x and t values are determined from Figures 6.3 and 6.4 for coolant temperatures of 40
0
C and 40 + 30 = 70
0
C respectively. Therefore, the residual fuel not affected by local heating reads {x, t } = {0.38, 1.001 and the residual affected by local heating reads
{x, T } = {0.32, 0.33}. This method assumes that both coolant and port temperatures increase linearly as the engine warms-up, which is a relatively good assumption as
Figures 4.1 and 4.2 suggest. As the coolant temperature increases the linear relationship between x and T with temperature deteriorates, therefore this approach is used only for the very early stages of the warm-up (coolant temperature below 45 OC).
The discrete form of the governing equations (6.3) and (6.4) shows that the mass of fuel inducted into the cylinder and burned is a function of x and 'r. In particular equation (6.3) shows that more fuel is delivered into the cylinder as x and T decreases, i.e. the engine transient response is faster as the engine warms-up, which is consistent with the experimental data. However, the fuel delivered into the cylinder also depends on the mass of residual fuel at that particular cycle. Equation (6.3) shows that the smaller the values of x and T the smaller the amount of the residual fuel [21]. This is also clear under steady state conditions according to equation (6.5). The above suggests that as the engine warms up x and T decrease, but the residual fuel mass also decreases and therefore, the engine response is not as trivial as it initially appears. The solution of the difference equations
81

(6.3) and (6.4), which is given by equation (6.11) shows that when the residual fuel mass is eliminated, it is clear that the smaller the x and T values, the faster the engine response even though the residual fuel mass is smaller. Therefore, when the heaters are activated the engine transient response should be enhanced even when a small fraction of fuel impinges on the heated region. However, in order to obtain equation (6.11) from equations (6.3) and (6.4) it has been assumed that x and t remain constant during the transient. This seems to be a reasonable assumption because the transient is considered to be quasi-static and the coolant temperature barely changes in the time scale of a transient.
However, even though the coolant temperature is constant during a transient, port or cylinder wall temperature may locally change. As more fuel impinges on port and cylinder walls during a step increase in load, fuel vaporization may cool the walls. This is very likely to happen if the thermal capacity of the walls is sufficiently small. Even thougli the thermal capacity of port and cylinder walls is very large, the thermal capacity of the heaters is small and a local temperature drop when the heaters are on is very likely to occur. Figures 4.8 and 4.9 illustrate such behavior and it is clear that the large thermal capacity of the walls prevents any temperature drop for the baseline case, while the substantially smaller heater capacity cannot retain the local temperature at the same high levels during the transient.
Even though the local temperature when the heaters are activated is higher than the baseline case regardless of the subsequent drop during the transient, it is not clear if the engine transient response is faster. Equation (6.11) is not valid any more and there may be a combination of values of residual fuel mass, x and r that could result in slower dynamic response. As the heaters are activated the residual fuel mass may substantially decrease and the x and r values may not be as small as initially considered due to the temperature drop during the transient (even though they are always smaller compared to the baseline case). As a result the dynamic response may be slower for particular temperature ranges. The governing equations (6.3) and (6.4) now become:
82

m. m .
mn = m'r + x mj - m
-1 At mi = (1-x')m' + m-
-1At
(6.12)
(6.13)
In order to check whether there is a temperature range where the engine dynamic response can be slower even when the heaters are activated, equations (6.12) and (6.13) should be simultaneously solved using different x and t values for every cycle. The estimation of x and x during the transient is also done according to Figures 6.3 and 6.4 in the same way as described above, since the heated region temperature is continuously measured.
Therefore, the basic two modifications in the x- ' model are (a) the existence of two distinct residual fuel masses with different sets of x and c values and (b) the continuous change of x and r in the residual fuel mass affected by the local heating values during a transient due to the local temperature drop. The modified x-T model described in this section is applied to both heater configurations in Section 6.3.
6.3 X-TAU MODEL SIMULATION RESULTS
6.3.1 Configuration B
The modified x- r model is applied first to configuration B because the area of the port occupied by the heaters is significantly larger and the results are expected to be clearly studied.
Figure 6.5 shows the temperature history of heated region B 1 as measured with the corresponding sensor when the coolant temperature is 45 0
C. The graph shows both the baseline case (no heater activation) and the case when the local temperature is increased.
At cycle number around 30, the heaters are activated for the latter case and approximately
100 cycles later a transient is initiated for both cases. The temperature signal is almost flat for the baseline case since the thermal capacity of the port prevents local temperature drop during the transient. When the heaters are activated, there is a significant local temperature rise of more than 20K in 100 cycles (approximately 8 sec), which seems to be sufficient time for the local temperature to reach a steady state. When the transient is
83

initiated there is a local temperature drop of approximately 5K. This drop is approximately 20 to 25 % of the temperature rise due to heater activation and though small it is not negligible. In the same figure the corresponding values of x and 'r are shown at the points where the engine operation has reached local steady -state conditions, i.e. before heater activation, before throttle transient and after the engine stabilizes to new operating conditions. These values are obtained from Figures 6.3 and 6.4 according to temperature measurements. During temperature changes these figures are also used so that x and t values change on a cycle basis. For the baseline case, x and t correspond to the whole residual fuel mass, while when the heaters are activated two sets of x and t values are used accordingly. The residual fuel not affected by heater activation responds with the baseline x and t values. The fuel targeting (amount of fuel affected by local temperature increase) is set for this configuration to 70% after taking into account fuel deposition estimation results given in Section 2.2 and the possible effect of exhaust gas back-flow process.
U
45 x = 33.3% tau = 0.39 sec
40 --
Coolant temperature =45 C
CL
35
>ix
S30 30
_
Transient
= 33.9% tau = 0.46 sec
-
-Heaters on
Baseline 25 Heater activation
20x = 37.5% tau = 0.87 sec
0 40
Figure 6.5
80
Cycle number
120 160
Temperature of region B 1 during heater activation and transient
200
84

After estimating x and T values for every cycle and fuel targeting, the engine response to heater activation and throttle transient can be predicted by solving the governing equations (6.12) and (6.13). Figure 6.6 shows the relative AFR as predicted with the modified x- T model and as measured experimentally with the UEGO sensor. The figure shows that the agreement between the model and the experimental data is very good.
When the heater is activated part of the residual fuel is released from the residual mass resulting in a small rich excursion to the mixture delivered into the engine and when the transient is initiated a lean spike is observed as expected.
Figure 6.7 shows the residual fuel mass change when the heaters are activated and when the transient is initiated for both the baseline case and when the heaters are on. The figure clearly demonstrates that the initial condition of the residual fuel is much different for the two cases. Even though the residual fuel mass is significantly larger for the baseline case after the transient, the residual fuel increment, which is studied, is not significantly different for the two cases.
1.3
I .25 k
1.2
cr
IL-
1.15
CI
0)D
1.1
1 .05 -
1
0.95
0
UEGO sensor
.x tau model
40 80
Cycle number
120
J dA
160
Figure 6.6
Predicted and measured AFR with configuration B
200
85

140
120
Baseline
Heaters on
100
40
80 co
60
Heater activation
Transient
40 difference in initial conditions between the transients
20
0
Figure 6.7
40 80
Cycle number
120 160
Predicted residual fuel increment with configuration B
200
Figure 6.8 illustrates the predicted engine transient response and it is obvious that local heating has indeed a moderate positive impact. For this particular case (45 OC the residual fuel increment has been predicted to decrease from 63 to 54 mg resulting in a
14% drop in the residual fuel increment. Figure 5.8 showed that the experimental data at similar operating conditions (4 min after firing) agree well with the model: decrease from
52 to 46 mg resulting in a 12% drop in the residual fuel increment. The small discrepancy in the absolute levels of the residual fuel values is attributed to the rough approximation of the x and t absolute levels, but the change in residual fuel is sufficiently predicted by the model.
86

50
45
-40
Heaters on
Baseline
35
30
Fuel released due to heater activation
25
20
15
0
Figure 6.8
40 80
Cycle number
120 160
Predicted engine transient behavior with configuration B
200
Since fuel targeting (fraction of fuel affected by heater activation) and x and T variation during a transient (linear relationship has been assumed so far) are not accurately known,
Figure 6.9 summarizes all possible combinations and their corresponding impact on residual fuel increment in one chart. The fuel targeting is depicted on the horizontal axis and it ranges from 0 to 100%. Obviously in the former case heater activation has no impact at all, while in the latter case the whole residual fuel mass is affected by the heaters. The effect of the temperature drop is represented by the contour lines, which also range form 0 to 100%. This percentage refers to the temperature drop during a transient with respect to the temperature rise due to heater activation. Since a linear relationship has been assumed between x, t and temperature this percentage corresponds to x and t variation as well. If 0% is assumed, then the local temperature does not drop during the transient and the residual fuel mass affected by local heating behaves exactly like the rest with substantially lower but constant x and T values. If 100% is assumed then the local temperature drop during a transient equals the local temperature rise due to heater activation. On the vertical axis the residual fuel increment is depicted when the heaters are on divided by the baseline residual fuel increment, which is the same regardless of
87

88 fuel targeting and local temperature drop during a transient. If this ratio is smaller than unity, local heating effect is positive and if it is larger, local heating effect becomes negative.
2.0
5 1.6
1.2
14
. 0.8
0.4-
-
.
---4 relative x and tau increase
-o-100%
-- u--- 80%
-o--60%
-
.--
I
-20%
-4-0%
0.0
0
-
Figure 6.9
20 40 60
Fuel targeting on heated region [%]
80 100
Sensitivity analysis chart for engine transient behavior prediction
For the example studied above with a fuel targeting of 70% and a temperature drop of 20-
25%, local heating effects have an impact of the order of 15-20%, as found experimentally. If there were no temperature drop then local heating effects would result in a residual fuel increment drop of 40%. Therefore, this small temperature drop during a transient offsets more than half of the potential local heating gain. This chart also illustrates that even when the temperature drop due to transient is significantly smaller than the temperature rise due to heater activation, say just more than 40%, local heating effects can cause higher residual fuel increments than the baseline case. The combination of changes in residual fuel mass, x and t during a transient can justify such behavior.

6.3.2 Configuration A
The only assumption made to construct Figure 6.9 is that the engine coolant temperature is 45 0
C. However, since only temperature differences are shown in this figure, it is also valid for different coolant temperatures because linear relationships are assumed for the early stages of the warm-up. Therefore, Figure 6.9 can be also used to predict and explain the engine transient behavior when heater configuration A is used if appropriate values of fuel targeting and temperature drop during a transient are estimated.
Figure 6.10 shows the temperature of the heated region Al during a transient for the baseline case and when the heaters are on 5
. The coolant temperature in this case is 35
0
C and the corresponding values of x and t shown on the graph have been found from
Figures 6.3 and 6.4, as explained in Section 6.2.3. The temperature drop during the transient when the heaters are on is 7-8K, which is higher than in configuration B. The larger drop is attributed to the fact that the heaters occupy a smaller area in the port and hence their thermal capacity is even smaller. Therefore, x and T values increase approximately 30% relatively to their initial drop due to heater activation. Fuel deposition estimations suggest that fuel targeting is significantly smaller compared to configuration
B and it is in the order of 30%. However, the amount of fuel landing per heater may be larger for configuration A compared to B because the former uses just one heater while the latter uses six heaters. This also explains why local temperature during a transient is higher in that case.
5 In this example the single spray injector has been used because the results can be more clearly studied.
89

50
&
0
45
-Coolant
- - x= 33.3% tau = 0.39 sec temperature = 35 0 C x=.35.1% tau =0.58 sec
-
35
7930
30
225
20
15
0
x= 39.4% tau = 1.09 sec
20 40
Figure 6.10
Transient
Heaters on
-Baseline-
60 80
Cycle number
100 120
Temperature of region Al during heater activation and transient
140
According to the above assumptions the predicted engine transient behavior is depicted in
Figure 6.11. During the early stages of the transient local heating improves response, but it takes longer to reach the new steady state because of the continuously increasing x and xr. For this particular example there is almost no local heating gain since the residual fuel increment is 60 mg, while the corresponding value for the baseline case is 63 mg.
90

35
30
-~25
25__
_
-Heaters
-Baseline on
20_
15
0 20
Figure 6.11
40 60 80
Cycle number
100 120
Predicted engine transient behavior with configuration B
140
These results and their sensitivity to all the above assumptions can be also seen in Figure
6.9. For a fuel targeting of 30% and a relative temperature drop of 30%, there is almost no local heating net effect and therefore the experimental data are more prone to uncertainties concerning the measurements. This is believed to be the reason why almost no improvement (and sometimes deterioration) in engine transient behavior has been observed when local heating is applied to configuration A. The significantly smaller fuel targeting suggests that the wrong intake port region is heated and even the small improvement expected (not more than 15%) cannot be observed due to low heater thermal capacity effect as explained above.
91

92

7
SUMMARY OF RESEARCH COMPONENTS
The fuel targeting and the engine thermal environment effects on mixture preparation process of SI engines have been the main topics of this study. Fuel targeting has been studied by using two injectors with different spray characteristics. In order to study the thermal environment effects, mixture preparation has been observed during the warm-up period. On top of the continuously varying thermal state of the engine during the warmup, specific port regions have been locally heated and their impact on mixture preparation has been evaluated. There have been two different arrangements of locally heated regions: configuration A, where the regions of initial fuel impingement have been heated and configuration B, where the heated regions have been located just upstream the intake valves. Configuration A studies the coupling between fuel targeting and local thermal environment, while configuration B investigates the importance of the regions close to the intake valves for mixture preparation regardless of fuel targeting. Finally, the sensitivity of mixture preparation to the thermal state of the intake port and engine block individually have been derived by running the early stages of the warm-up process without cooling the port and the block of the engine respectively.
The mixture preparation process has been observed during throttle transients. When the steady state engine operation is perturbed, the accompanied fuel lag effect in the mixture preparation process is more prominent and its sensitivity to all the above factors can be more easily observed. During a throttle transient the engine behavior in terms of GIMEP response has been evaluated by processing the in-cylinder pressure data. Using the peak pressure location in order to account for the combustion phasing and carrying out tests at steady state recording AFR with the UEGO sensor, an appropriate engine calibration has been constructed. This calibration can be used to derive the mass of fuel burned during a transient from the corresponding pressure data. The parameter used to describe the
93

mixture preparation during transients is the residual fuel increment, which is the fuel injected but not burned due to the fuel lag effect.
Finally, a fuel transport model has been used to explain the mixture preparation process.
The so-called x- t model is widely used because of its simplicity (first order and linear) and sufficient accuracy. In the present study this model has been found sufficient enough to predict the fuel transport properties showing a relatively good agreement to the experimental data.
7.2 CONCLUSIONS
Based on both the experiments and the simulation of the fuel transport characteristics with the x- ' model the following conclusions can be drawn from this study:
* There is a substantial improvement of the mixture preparation process as the engine warms up. The residual fuel increment during tests with the same step increase in load drops by a factor of 4 or 5 as the engine reaches the fully warmed-up condition.
Obviously the time constant of fuel leaving the residual fuel mass (both port and cylinder) becomes smaller as the temperature rises and the fuel lag effect diminishes.
Also less fuel impacts on the walls and the fraction of fuel evaporating when being airborne increases due to higher intake air temperatures.
* No significant fuel targeting effects have been observed when switching between the two injectors. Even though the dual spray injector directs the fuel closer to the intake valves, which are the hotter regions in the port, than the single spray injector, the amount of fuel impinging directly onto the valves is not substantially different. Since there has been no recording of any substantial spatial temperature gradients in the port excluding the back of the intake valves, it is believed that the amount of fuel targeting the intake valves is very important. The dual spray injector has been found to be only slightly deflected by the airflow when OVI has been used as opposed to the hollowcone single spray injector. This deflection is believed to be responsible for the poorer mixture preparation of the single spray injector when OVI is used during the early stages of the warm-up.
94

* Local heating has little effect on the fuel delivery to the cylinder for the two fuel injector targeting configurations. Since only a fraction of fuel impinges on the heated regions, significant residual fuel increment changes are not expected. The fact that no local heating effects have been observed suggests that fuel splashing and flow away from the regions of initial deposition and also redistribution due to the back-flow process of the exhaust gases occur in the intake port. In many cases a small (less than
10%) but consistent negative impact of local heating has been recorded. That effect may be related to slight air flow modifications and different residual fuel initial condition arising from heater activation.
* Port regions very close to the intake valves have been found to play an important role to the mixture preparation process regardless of fuel targeting. A drop of 10-20% in the residual fuel increment has been observed, when the regions just upstream the intake valves are locally heated. These regions are important because liquid fuel flows through these regions on its way to the combustion chamber regardless of initial fuel deposition.
* Tests without coolant suggest that the mixture preparation is sensitive to the engine overall thermal environment. The engine responses in approximately 3 cycles, when running with out coolant for 4-5 minutes. By blocking the coolant passages only to the port, it has been found that port thermal environment plays a critical role in mixture preparation, while when blocking the coolant passages only to the block, the effects have not been that significant. These tests imply that mixture preparation is primarily dependent on port thermal environment since the fuel lag effect is most important there and studies should focus especially on it.
* The simple x-T model has been found to sufficiently predict the fuel transport phenomena in the intake port. The dependence of x and
T on coolant temperature has been found to agree well with past research. The model has been modified in order to account for the local heating effects: (a) there are two residual fuel masses with different properties (different x, r values), one for the fraction of fuel affected by local heating and one for the rest and (b) the x and
T of the first residual fuel mass
95

96 change during a transient when the heaters are activated according to their dependence on coolant temperature. The reason for x and T change during a transient is due to the fact that the local high temperature cannot remain constant because of the low thermal capacity of the heaters. The modified x- t model matches the experimental data and explains the relatively small (if any) impact of local heating on mixture preparation. The substantially different initial condition of the residual fuel between the baseline case and the case when the heaters are activated, and the low thermal capacity of the heating elements justify this behavior. Detailed calculations with the x- T model showed that even a small decrease of local temperature due to a transient (less than 8K) compared to the increase due to local heating (more than 25K) can almost offset the local heating gain as far as the residual fuel increment is concerned.

1. Heywood, J.B. Internal Combustion Engines Fundamentals. McGraw-Hill, New
York, 1988.
2. Bossert, J.C. "The Effects of Fuels on Engine Throttle Transients", Master's Thesis,
Dept. of Mechanical Eng., Massachusetts Institute of Technology, February 1994.
3. Fozo, R.S., Aquino, C.F. "Transient A/F Characteristics for Cold Operation of a 1.6
Liter Engine with Sequential Fuel Injection," SAE Paper 880691 (1988).
4. Shin, Y., Cheng, W.K., Heywood, J.B. "Liquid Gasoline Behavior in the Engine
Cylinder of a SI Engine," SAE Paper 941872 (1994).
5. Curtis, E.W., Aquino, C.F., Trumpy, D.K., Davis, G.C. "A New Port and Cylinder
Wall Wetting Model to Predict Transient Air/Fuel Excursions in a Port Fuel
Injected Engine," SAE Paper 961186 (1996).
6. Shayler, P.J., Davies, M.T., Colechin, M.J.F., "Intake Port Fuel Transport and
Emissions: The Influence of Injector Type and Fuel Composition," SAE Paper
961996 (1996).
7. Ely, J.F., Huber, M.L., "NIST Thermophysical Properties of Hydrocarbon Mixtures
Database (SUPERTRAPP)", National Institute of Standards and Technology,
Standard Reference Database 4-Version 1.0 Gaithersburg, MD (July 1992).
8. Chen, K.C., Witte, K.D., Cheng, W.K., "A Species Based Multi-Component
Volatility Model for Gasoline," SAE Paper 941877 (1994).
9.. Greiner, M., Romann, P., Steinbrenner, U., "BOSCH Fuel Injectors-New
Developments," SAE Paper 870124 (1987).
10. Bauer, W., Balun, P., Heywood, J.B., "Heat Transfer and Mixture Vaporization in
Intake Port of Spark-Ignition Engine," SAE Paper 972983 (1997).
11. Aquino, C.F., Plensdorf, W., Lavoie, G., Curtis, E.W., "The Occurrence of Flash
Boiling in a Port Injected Gasoline Engine," SAE Paper 982522 (1998).
97

12. Bossert, J.C., Shin, Y., Cheng, W.K., "Fuel Effects on Throttle Transients in PFI
Spark Ignition Engines," SAE Paper 971613 (1997).
13. Tseng, T.C., Cheng, W.K., "An Adaptive Air/Fuel Ratio Controller for SI Engine
Throttle Transients," SAE Paper 99010552 (1999).
14. Rublewski M. "Nitric Oxide Formation and Thermodynamic Modeling in Spark
Ignition Engines", Master's Thesis, Dept. of Mechanical. Eng., Massachusetts
Institute of Technology, February 2000.
15. Meyer R., Heywood, J.B., "Evaporation of In-Cylinder Liquid Fuel Droplets in an
SI Engine: A Diagnostic-Based Modeling Study," SAE Paper 99010567 (1999).
16. Curtis, E.W., Russ, S., Aquino, C.F., Lavoie, G., Trigui, N., "The Effects of Injector
Targeting and Fuel Volatility on Fuel Dynamics in a PFI Engine During Warm-Up:
Part 1-Modeling Results," SAE Paper 982519 (1998).
17. Martins, J.J.G., Finlay, I.C., "Fuel Preparation in Port-Injected Engines", SAE Paper
920518 (1992).
18. Shin, Y., Min, K., Cheng, W.K., "Visualization of Mixture Preparation in a Port
Fuel Injection Engine During Engine Warm-up," SAE Paper 952481 (1995).
19. Shayler, P.J., Teo, Y.C., Scarisbrick, A., "Fuel Transport Characteristics of Spark
Ignition Engines for Transient Fuel Compensation," SAE Paper 950067 (1995).
20. Russ, S., Stevens, J., Aquino, C.F., Curtis, E.W., Fry, J., "The Effects of Injector
Targeting and Fuel Volatility on Fuel Dynamics in PFI Engine During Warm-UP:
Part I-Experimental Results," SAE Paper 982518 (1998).
21. Saito, K., Sekiguchi, K., Imatake, N., Takeda, K., Yaegashi., T., "A New Method to
Analyze Fuel Behavior in a Spark Ignition Engine," SAE Paper 950044 (1995).
22. Imatake, N., Saito, K., Morishima, S., Kudo, S., Ohhata, A., "Quantitative Analysis of Fuel Behavior in Port-Injection Gasoline Engines," SAE Paper 971639 (1997).
23. Cowart, J.S., Cheng, W.K., "Throttle Movement Rate Effects on Transient Fuel
Compensation in a Port-Fuel-Injected SI Engine", SAE Paper 2000-01-1937 (2000).
98

Volumetric efficiency drop due to heater activation
Apart from affecting the fuel lag and as a result the fuel transport phenomena in the port, the heaters have also an impact on the air flow. When the temperature of some regions in the port is increased, the air gets hotter and therefore, the mass of air trapped into the cylinder gets smaller. The following is an estimation of the volumetric efficiency drop associated with the heater activation according to configuration B, where the heaters occupy a relatively large port surface area.
The energy balance in the control volume of the port region containing the heaters is:
Baseline: Nuaj~k (TWO -Ta )A raincp (Ta"f Ta) (Al)
Heaters on: Nudk (T "Ta A =ancP(Ta" Ta) (A2)
In the above relations Nud is the Nusselt number for a tube of diameter d with constant heat flux. A is the area occupied by the heaters, i.e. A = TdL where L is the heater length.
Twfo "is the temperature of the heated region, Ta the air temperature just before entering the control volume and Taf oron the air temperature leaving the control volume. For maximum heat flux from the walls to the air we have assumed that the temperature difference stays always maximum, i.e. (T "
-
Ta
)=
(T " -Ta ), which is the worst case assumption instead of using the more realistic logarithmic temperature difference.
Finally k, cp and kia,in are the thermal conductivity, the specific heat and the mass flow rate during the intake of air respectively.
Subtracting the equations (Al) and (A2) and modifying:
Ta
NudkirL onTff
off (T " -T w rha,inCpTa
(A3)
99

The air mass flow rate for the part load case has been found to be approximately 3.5 g/sec. However, this is the average value during the whole cycle. Since air is inducted only during the intake stroke the intake flow rate must be approximately 4 times higher.
Since the control volume concerns only one of the two runners of the intake port, (two intake valves per cylinder) the air mass flow rate should be then divided by 2. Therefore, the air mass flow rate for the control volume is 7 g/sec. The part load condition is used in these calculations because as equation (A3) shows the temperature rise is inversely proportional to the air flow rate. This effect more than offsets the higher heat transfer coefficient associated with higher flow rates because Nusselt number (non-dimensional heat transfer coefficient) is proportional to flow rate to the exponent 0.8 and therefore the worst case is investigated.
The mode of the flow is characterized by the Reynolds number:
Red _ puC~d p.
Ac lfaind d
_Iajn_
4
" (A4)
Red =7
-x18.43x10
4
6 x35
- Red =13817
The flow is highly turbulent (Red > 2300) and the Nusselt number yields:
Nud =0.023Re. 'Pr
04
=0.023x13817
08 x0.7
4
=>Nud =41 (A5)
In equations (A4) and (A5) [t and Pr are the dynamic viscosity and Prandtl number of air respectively.
Substituting (AS) and (A6) back to (A3), the relative temperature rise due to heater activation reads:
ATa
_
T o
41x 0.025 x r x 20 x 50 ATa 0.15%
7 x1000 x 300 T
(A7)
100

In the above relation it has been assumed that the temperature of the air just before entering the control volume is equal to the ambient temperature and the local temperature rise due to heater activation is 50K, which is the highest that can be achieved. Both assumptions lead to the highest possible air temperature rise.
Equation (A7) shows that the relative temperature rise, which is equal to the relative volumetric efficiency drop, is always less than 0.5%. The heater area turns out to be very small in order to affect the air flow and therefore, neglecting such effects is a very good approximation.
101

Solution of discrete form of x-r model
The discrete form of the x--L model is given by the following equations:
mi = m + xm - m - At r r ij r C i
M =(1- e x)mn + M ini r i-1At
At
(B1)
(B2)
If the transient is initiated at i=1 and the fuel is increased in one step from mb to m
(superscripts b and a stand for before and after the transient respectively) equation (B1) can also be written as:
In the above equation
mi =ami- + c (B3)
At a and c=xm a can be treated as constants, i.e.
independent of time.
i=1: mr =amr +c i=2: m2 =am + c a2m +c(l+ a) i=3: m. =am +c=am
0+c(1+a+a2) i=i: mr =am +c=a m +c(1+a+a
2+...+a'-)-
-
(B4) where mb =M= b' xm,-
At is the steady state residual fuel mass before the transient initiation. Note that i is an exponent for a and just a superscript for mr.
Substituting back to equation (B2) the mass of fuel burned becomes:
103

mi =M a
(
~At
' a
- xAm 1--- (B5) where Am j = a
m b is the se ic in fuel injection.
For i=1 (initial engine response), equation (B5) reads: mi n + Am (1 - x) (B6)
For i->oo (new steady state), equation (B5) reads: m
1 e
=:m. inj(M
(B7)
Therefore, the initial engine response jumps by Am..j (1- x) before starting to rise with a time constant -c and eventually it reaches the new steady state value.
The continuous form of the x--c model is given by the following equations: mx r = xrinj Mr
(B8) ri, =I x)rhij + Mr
The solution of the above differential equations (Laplace transform) yields:
Ti _ h e inj fi inj
1-xe
(B9)
(B10)
The transient is initiated at t=O. In order to derive the discrete solution from equation
(B10) it must be assumed that the finite integration time is much smaller than the time scale t. The integration time is the duration of one cycle and therefore this assumption is not very accurate. However, if the above is assumed then equation (B 10) can be written in discrete form: mi = mI. + AmI I
- xe
(i-I)At
I (B11) where the transient is initiated at i=1 and therefore the equations (B5) and (B11) are consistent.
104

Since it has been assumed that T=-<< 1, then e-T =1-T and therefore e
(i--1)At
= e
-At t
J
-
T
A
-t
Substituting back to equation (B 11) the resulting equation is identical to equation (B5). Therefore, the amount of fuel burned per cycle during a step transient can be predicted by equation (B5) accurately if the x and
T are determined experimentally, while equation (B 11) is not as accurate because it is assumed that the time constant for the transient is much larger than the duration of one cycle.
105