SIGNAL RESPONSE BASED ON SPEED AND REACTION OF OTTO CYCLE ENGINE

i

SIGNAL RESPONSE BASED ON SPEED AND REACTION OF OTTO CYCLE ENGINE

MUHAMMAD ZAIM BIN MOHAMED PAUZI

UNIVERSITI SAINS MALAYSIA

2016

i

SIGNAL RESPONSE BASED ON SPEED AND REACTION OF OTTO CYCLE ENGINE

by

MUHAMMAD ZAIM BIN MOHAMED PAUZI

Thesis submitted in fulfilment of the requirements for the degree of

Master of Science

November 2016

ii

ACKNOWLEDGEMENTS

First of all, thank God for given everything to me to achieve successful in my life. Next, I would like to appreciate my beloved parents who always give me support on my studies and teach me to be motivated while researching. Next, humbly I thank to my supervisor Dr. Elmi Abu Bakar and my co-supervisor Dr. Fauzi Ismail who supervised me on my master research study and support my finance by hiring me as research assistant. Many of opportunity has been given to gain my professionalism in research study such as involving in many of activities related to invention and innovations. From the opportunities, I develops my skill of research, paper write up and management skills.

Moreover, I would like to thank all the technicians who support my research especially Mr. Hisham by given physical help such bench installation and many more.

Furthermore, thanks to all of my friends in Innovative System and Instrumentation team who always give moral support and being my accompany.

Last but not least, my sincere thanks to School of Mechanical Engineering and School of Aerospace Engineering, Universiti Sains Malaysia for given opportunity to me to finish my study. Thanks to Ministry of Higher Education for providing funds for study fees.

iii

TABLE OF CONTENTS

Page

ACKNOWLEDGEMENT ii

TABLE OF CONTENTS iii

LIST OF TABLES vi

LIST OF FIGURES vii

LIST OF ABBREVIATIONS xii

LIST OF SYMBOLS xiv

ABSTRAK xvi

ABSTRACT xviii

CHAPTER ONE : INTRODUCTION

1.1 Background of research 1

1.2 Problem statement 4

1.3 Research objectives 5

1.4 Research scopes 6

1.5 Research approach 7

1.6 Research outline 8

CHAPTER TWO : LITERATURE REVIEW

2.1 Internal Combustion Engine Overview 10

2.2 Carburetion System 15

2.3 Electronic Systems 21

iv

2.4 Oxygen Sensor Principle and its Importance in Engine Diagnostic 25 2.5 Various Signal Pre-Processing Approaches for Engine Misfire 32

Detection (EMD) Algorithm

CHAPTER THREE : METHODOLOGY

3.1 Engine modelling using Simulink 47

3.1.1 Throttle 48

3.1.2 Intake Manifold 50

3.1.3 Intake and Compression Stroke Subsystem 51

3.1.4 Torque Generation and Acceleration for Engine Speed 53 Measurement

3.1.5 Misfire Actuation 54

3.2 Engine and Instrument Set Up 56

3.3 Artificial Engine Misfire Controller 65

3.4 Signal processing 76

3.4.1 Feature extraction 76

3.4.2 Signal Filtering 81

3.4.3 Discrete Wavelet Transform as Filter and Features Extractor 90

CHAPTER FOUR : RESULTS AND DISCUSSION 94

CHAPTER FIVE : CONCLUSION AND FUTURE RESEARCH

5.1 Conclusion 117

5.2 Contribution of the research 118

5.3 Recommendations for future research 119

REFERENCES 121

APPENDICES

Appendix A [Artifical misfire actuator program codes]

v

Appendix B [Signal filtering and misfire detection program codes]

LIST OF PUBLICATIONS

vi

LIST OF TABLES

Page Table 2.1 Vehicle Manufacturing Standard by the National

Automotive Policy (NAP, 2014)

23

Table 2.2 Types of Oxygen Sensors for Automotive Use (Ramamoorthy, Dutta, & Akbar, 2003)

28

Table 2.3 Electrical Conductivity of Solid Electrolytes at 600o (N.

M. Sammes, 1999; M. Mogensen, 2000; C. Xia, 2002;

N. M. Sammes, 1997)

29

Table 2.4 Testing data of exhaust emission (Ye, 2009) 44 Table 2.5 Relation indices and diagnosis results (Ye, 2009) 44 Table 3.1 Engine Specification (YZF-R15 OWNER’S MANUAL,

2010)

58

Table 3.2 Oscilloscope Specification (Technology, 2013) 64 Table 3.3 Arduino UNO controller specification

(https://www.arduino.cc)

68

Table 3.4 Frequency, k=0, 1, 2, and 3 for the waveform to correlate with sample distribution.

81

Table 3.5 Type of Low Pass Filter Behaviour [http://www.eetasia.com/]

84

Table 4.1 Signal to noise ratio of raw misfire signal 100 Table 4.2 Results obtained without heater activation on oxygen

sensor

106

Table 4.3 Results obtained with heater activation on oxygen sensor

107

Table 4.4 Results without heater activation on oxygen sensor 114 Table 4.5 Results with heater activation on oxygen sensor 115 Table 4.6 Reliability of EMD Comparison base on current and

previous studies.

116

vii

LIST OF FIGURES

Page Figure 1.1 Statistics of vehicles in Malaysia (Malaysia, 2015) 1 Figure 1.2 Statistics of emission of pollutants into the atmosphere

in Malaysia (Department of the Environment, 1997)

2

Figure 1.3 Types of compounds emitted by the exhaust at different conditions (To, 2012)

4

Figure 2.1 Four-Stroke Engine Configuration (Heywood, 1988) 11 Figure 2.2 Two-Stroke Engine Configuration (Heywood, 1988) 11 Figure 2.3 Four-Stroke Engine Mechanism Sequence (Heywood,

1988)

13

Figure 2.4 Sleeve valve system (Iliffe, 1919) 14

Figure 2.5 Poppet Valve System (Bissell, 1929) 14

Figure 2.6 Carburetor (Heywood, 1988) 16

Figure 2.7 Typical solenoid injector (Smith, D., 1980) 17 Figure 2.8 Main jet Structure

(http://www.s262612653.websitehome.co.uk)

18

Figure 2.9 Early Injection System by Rudolf Diesel (http://www.douglas-self.com/)

19

Figure 2.10 Electronic Control Injector (Akiyama, H., 2010) 21 Figure 2.11 Japan Emission Standards for Gasoline and LPG

Vehicles (Delphi, 2013)

22

Figure 2.12 Computer System Subsystem (http://www.edsim51.com/)

24

Figure 2.13 Typical Response of Zirconia Oxygen Sensor to AFR (A. D. Brailsford, 1998)

27

Figure 2.14 Wideband Oxygen Sensor Response to AFR (Hunt, 1993)

28

Figure 2.15 Unheated Oxygen Sensor (Bosch, 2012) 31

Figure 2.16 Heated Oxygen Sensor (Bosch, 2012) 31

Figure 2.17 Layer cross-sectional view planar (E. I. Tiffee, 2001) 32

viii

Figure 2.18 Light off time (J. Riegel, 2002) 32

Figure 2.19 Crankshaft Position Sensor Placement (Birnbaum R., 2000)

33

Figure 2.20 Misfire detection technique flowchart (Naik, 2004) 34 Figure 2.21 Power spectral density of engine without misfire

(Osburn, Kostek, & Franchek, 2006)

35

Figure 2.22 Signal of combustion interval acquired by using CPS and its moving average (Cavina et al., 2006)

37

Figure 2.23 Before Pre-Processing (Cavina et al., 2006) 38 Figure 2.24 After Pre-Processing (Cavina et al., 2006) 38 Figure 2.25 Simulated crankshaft position sensor signal (Tinaut et

al., 2007)

41

Figure 2.26 Misfire signal by temperature sensor (Tamura et al., 2011).

43

Figure 2.27 Misfire detection system block diagram (Johnson &

Rado, 1978)

43

Figure 2.28 Tooth machining error on flywheel (Naik, 2004) 45 Figure 3.1 Throttle manifold dynamics subsystem 49

Figure 3.2 Inside the throttle subsystem 49

Figure 3.3 Intake subsystem 51

Figure 3.4 Top level of the engine model system (The MathWorks, 1998)

52

Figure 3.5 Compression subsystem 53

Figure 3.6 Engine torque generation subsystem 53

Figure 3.7 Misfire actuator subsystem for CPS 55

Figure 3.8 Misfire actuator and oxygen sensor subsystem model 56

Figure 3.9 Experiment set up flowchart 57

Figure 3.10 Single cylinder spark ignition engine (http://etalkindia.com/)

59

Figure 3.11 Engine mounting model using Solidwork 59

ix

Figure 3.12 Engine mounting fabrication process 60 Figure 3.13 Experimental set up for data acquisition 60

Figure 3.14 Oxygen sensor installation process 61

Figure 3.15 Mathematical modelling for pressure distribution inside exhaust manifold

62

Figure 3.16 Pressure acquisition location along exhaust manifold 62 Figure 3.17 Pressure distribution along manifold 63

Figure 3.18 DAQ Interface 65

Figure 3.19 Wiring Diagram for Injection Signal Acquisition. 66

Figure 3.20 Injection Signal Without Noise 67

Figure 3.21 Injection Signal With Noise 67

Figure 3.22 Pull Up Resistor System 69

Figure 3.23 Injection Signal at High Engine Speed 70 Figure 3.24 Artificial Misfire Inducer Board Design 72

Figure 3.25 MOSFET Driver Circuit Diagram 73

Figure 3.26 Artificial Misfire Inducer Module 73

Figure 3.27 Program Flowchart 74

Figure 3.28 Injector Actuation Lagging 75

Figure 3.29 Injection Actuation Signal Without Extreme Lagging 76

Figure 3.30 Methodology Chart 77

Figure 3.31 Raw signal before transformation (Steven W. Smith, 1999)

78

Figure 3.32 Correlation waveform (Steven W. Smith, 1999) 79 Figure 3.33 Digital Butterworth filter design flowchart 83 Figure 3.34 Low pass filter diagram (Thede, 2004) 84 Figure 3.35 Stopband attenuation and Step responses of both

analogue and digital filters (Steven W. Smith, 1999)

86

Figure 3.36 Infinite impulse response of oxygen sensor 87

x

Figure 3.37 Phase response of FIR filter (Steven W. Smith, 1999) 88 Figure 3.38 Phase response of IIR filter (Steven W. Smith, 1999) 88 Figure 3.39 Correlation of signal with mother wavelet (Misiti &

Misiti, 1996)

90

Figure 3.40 Continuous Wavelet Transformation (Misiti & Misiti, 1996)

91

Figure 3.41 Haar wavelet function (Misiti & Misiti, 1996) 92 Figure 3.42 Multiple decompositions using DWT (Misiti & Misiti,

1996)

93

Figure 4.1 Summary of results 95

Figure 4.2 Injection signal had been missing to actuate misfire 95

Figure 4.3 Intake signal Response 96

Figure 4.4 Drop in speed during misfire 97

Figure 4.5 Oxygen sensor signal response with misfire 97 Figure 4.6 Real oxygen sensor response with misfire at 1000 to

1999 RPM with no heater

98

Figure 4.7 Real oxygen sensor response with misfire at 3000 to 3999 RPM with heater

99

Figure 4.8 Misfire signal frequency with respect to engine speed with and without heater

100

Figure 4.9 Unheated oxygen sensor frequency spectrum 101 Figure 4.10 Heated oxygen sensor frequency spectrum 102 Figure 4.11 Butterworth filter response at various order number 103

Figure 4.12 Filtered signal at 1st order 104

Figure 4.13 Filtered signal at 4th order 104

Figure 4.14 4th order signal filtering at 1000 to 1999 RPM without heater activation

108

Figure 4.15 4th order signal filtering at 1000 to 1999 RPM with heater

109

Figure 4.16 1st level decomposition of DWT 110

xi

Figure 4.17 3rd level decomposition of DWT 111

Figure 4.18 7th level decomposition of DWT 111

Figure 4.19 Average misfire detection accuracy using Butterworth filter

113

Figure 4.20 Average misfire detection accuracy using Haar DWT 113

xii

LIST OF ABBREVIATIONS

EMD Engine misfire detection OBD On board diagnostic AFR Air to fuel ratio ECU Engine control unit NOx Oxide of nitrogen

CO Carbon monoxide

HC Hydrocarbon

CO2 Carbon dioxide

CPS Crankshaft position sensor DWT Discrete wavelet transform DFT Discrete fourier transform IC Internal combustion TDC Top dead centre BDC Bottom dead centre

NAP National Automotive Policy CPU Central processing unit ADC Analog to digital converter DAC Digital to analog converter PPM Part per million

YSZ Yttria stabilized zirconia TPB Triple phase boundary SNR Signal to noise ratio IMM Interacting multiple model RPM Revolutions per minute CAD Computer aided design

xiii MIG Metal inert gas

CFD Computational fluid dynamics DAQ Data acquisition

USB Universal serial bus CSV Comma separated values EMF Electromagnetic force

FPGA Field programmable gate array

MOSFET Metal–oxide–semiconductor field-effect transistor PCB Printed circuit board

FIR Finite impulse response IIR Infinite impulse response CWT Continuous wavelet transform FAR False alarm rate

EGR Exhaust gas reciculation

xiv

LIST OF SYMBOLS

R Universal gas constant P’o2 Partial pressure of oxygen

µ Chemical potential

F Faraday constant

rm engine cycle frequency

fs Number of samples per engine cycle

m Mass

v Velocity

 Rotational velocity

I Moment of inertia

T Temperature

 Time constant

 Throttle angle

Pm Manifold pressure

Pamb Ambient pressure Mass flow rate

Vm Manifold volume

Rate of change of manifold pressure

N Engine speed

ma Mass of air in cylinder for combustion A/F Air fuel ratio

 Spark advance

J Engine rotational moment of inertia Engine acceleration

Ns Samples number

xv

No Order number

Nc Number of engine cylinders

ωp Passband frequency

ωs Stopband frequency

Ωp Passband pre-warping frequency Ωs Stopband pre-warping frequency

xvi

RESPON ISYARAT BERDASARKAN KELAJUAN DAN REAKSI ENJIN JENIS KITARAN OTTO

ABSTRAK

Sekian lama, enjin jenis kitaran Otto telah menjadi popular kerana teknologi moden telah menggantikan sistem enjin stim yang lama menyebabkan pencemaran akibat pembakaran arang batu untuk menjana gas wap Baru-baru ini, sebuah jurnal pengangkutan melaporkan bahawa statistik pengeluaran kenderaan telah meningkat setiap tahun kerana penduduk dan negara pembangunan yang semakin meningkat.

Oleh itu, ia telah meningkat lagi pelepasan gas yang dihasilkan oleh pembakaran enjin.

Pada masa ini. Enjin jenis kitaran Otto masih dalam siasatan untuk penambahbaikan kecekapan. Ekzos merupakan aspek utama yang perlu diberi perhatian kerana ia mengeluarkan gas berbahaya dan mempengaruhi persekitaran. Malah, pelbagai ciptaan telah dihasilkan untuk mengurangkan gas berbahaya daripada pelepasan ekzos enjin. Selain itu, kesan macet enjin telah didapati mengurangkan prestasi enjin dengan meningkatkan penggunaan bahan api, kuasa keluaran yang rendah, dan berisiko kepada penukar pemangkin dengan kerosakan yang disebabkan oleh pembakaran tidak cekap. Oleh itu, pengesanan macet pada enjin adalah penting untuk meningkatkan kecekapan enjin atau untuk mengurangkan masalah ini. Lebih-lebih lagi, dalam usaha untuk mematuhi peraturan kenderaan dan keselamatan, alat-alat diagnostik digunakan untuk mengesan secara automatik masalah kenderaan dengan menggunakan sistem komputer. Atas sebab ini, isyarat macet pada enjin perlu dikesan untuk sistem diagnostik supaya dapat mengenali masalah. Pelbagai deria boleh digunakan untuk pengesanan, tetapi persembahan dari segi tindak balas isyarat dan kelajuan adalah berbeza-beza. Dalam kajian ini, deria oksigen telah digunakan dan bukannya sensor lain kerana kelebihannya dari segi kos penyelenggaraan yang rendah. Di samping itu,

xvii

keupayaan deria oksigen telah diuji melalui pemodelan matematik dan eksperimen dengan menggunakan model silinder enjin empat lejang tunggal dengan sistem suntikan bahan api. Dari eksperimen, macet berlaku disebabkan oleh beberapa faktor;

sama ada masalah mekanikal atau keadaan alam sekitar yang menjejaskan pembakaran campuran bahan api. Secara umum, macet yang paling biasa yang sudah didapati berlaku kerana tidak keseimbangan udara kepada nisbah bahan api (AFR). Oleh itu, untuk menentukan parameter optimum dan keadaan, macet tiruan telah dibangunkan dengan mengenakan AFR yang tidak seimbang untuk pembakaran menggunakan komputer mikro. Selain itu, pelbagai penapisan isyarat digital telah digunakan untuk penyesuaian isyarat untuk menjelaskan isyarat keluar yang dihantar oleh sensor oksigen untuk pengesanan macet. Kemudian, diskret Fourier Transform telah dipilih untuk menganalisis isyarat AFR yang mempunyai macet. Untuk tujuan analisis, parameter penapis Digital Butterworth telah ditetapkan berdasarkan isyarat penganalisis itu. Di samping itu, untuk mengesahkan keputusan, Discrete Wavelet Transform telah dilaksanakan, dan juga untuk menilai isyarat macet. Di sini, pelbagai

“mother wavelet” telah digunakan untuk menjelaskan isyarat supaya isyarat macet dapat disahkan dengan menggunakan pelbagai teknik pra proses isyarat pada keadaan enjin yang berbeza. Melalui simulasi eksperimen, keputusan dipaparkan prestasi yang bagus.

xviii

SIGNAL RESPONSE BASED ON SPEED AND REACTION OF OTTO CYCLE ENGINE

ABSTRACT

Over the decade, the Otto Cycle engine has become popular since the modern technology has replaced the old steam engine system that caused more pollution due to coal burning to generate steam gas. Recently, a transport journal reported that the statistics of vehicles production has been increasing annually due to the increasing populations and country development. Therefore, it has further increased gas emission produced by the engine combustion. Currently, the Otto cycle type of engine is still being investigated for its efficiency improvement. Exhaust emission is a major aspect that needs to be given attention because it emits harmful gases and influences the surrounding environment. In fact, numerous inventions have been generated to reduce harmful gases from the exhaust emission of an engine. Moreover, the effect of engine misfire had been found to reduce the performances of engine by increasing fuel consumption, exerting low output power, and imposing risk to catalytic converter to damage due to inefficient combustion. Thus, engine misfire detection (EMD) is important to compensate or to reduce the problem. Moreover, in order to adhere to the vehicles regulation and safety, diagnostic tools or on board diagnostic (OBD) is used to automatically detect faulty of the vehicles by using computer system. For this reason, a signal of engine misfire needs to be detected for the OBD system to recognize the problem. Various sensors could be used for EMD, but the performances in terms of signal response and speed are varied. In this research, narrowband oxygen sensor was applied instead of wideband set due to its advantages in terms of low maintenance and cost which is 70% cheaper. In addition, the capability of the narrowband oxygen sensor was tested via mathematical modelling and experiments using a model of single

xix

cylinder four-stroke engine with fuel injection system. From the experiment, engine misfire occurred due to a number of factors; either mechanical problems or environmental condition that affected the fuel mixture combustion. In general, the most common misfire had been found to occur because of unbalanced air to fuel ratio (AFR). Therefore, in order to determine the optimal parameters and condition, an artificial engine misfire was developed by exerting unbalanced AFR for the combustion using a microcontroller. Moreover, various digital signal filtering had been employed for signal conditioning to clarify the output signal transmitted by the narrowband oxygen sensor for EMD. Later, Discrete Fourier Transform was chosen to analyse the AFR signal which include misfire. For analysis purpose, the Digital Butterworth filter parameters were set based on the analyst signal. In addition, in order to verify the results, Discrete Wavelet Transform was implemented, as well as to evaluate misfire signal. Here, various mother wavelets were used to clarify the signal in order to verify misfire signal using various method of signal pre-processing at different engine conditions. Through the experimental and simulations, the results displayed a good agreement of signal features pattern while average EMD accuracy are 91.67% using Butterworth filter and 86.67% using DWT on experimental signal output.

1

1. CHAPTER ONE…..

INTRODUCTION 1.1 Background of research

Engine is widely used for vehicle manufacturing around the world. In fact, there are many types of engines, such as steam, internal combustion, and electric/hybrid engine. The internal combustion engine, nevertheless, is the most popular and widely used. In Malaysia, the number of vehicles on the road has rapidly increased over a year. Based on the statistics provided by the Malaysian Road Transport Department, the chart shown in Figure 1.1 is evident to verify the growing number of vehicles in this country.

Figure 1.1 Statistics of vehicles in Malaysia (Malaysia, 2015)

Besides new vehicles, old vehicles are still in existence and are widely used by Malaysians. Thus, pollution potentially occurs since old engines have low efficiency in emission control. It leads to failure in the subsystem of some vehicles or limitations of control system to manage the aged engine. Furthermore, based on an investigation

2

done by the Malaysian Department of Environment, air pollution is mostly contributed by vehicles, instead of other industries, as shown in Figure 1.2. Thus, it is important to look into this issue towards improving the air quality in our country.

Figure 1.2 Statistics of emission of pollutants into the atmosphere in Malaysia (Department of the Environment, 1997)

Engine management system is indeed among the most important subsystems that have to be considered for enhancement as it controls the condition of the engine while it is running. In fact, some sensors are used to detect and record the parameters of an engine at various conditions. For instance, engine control unit (ECU) is a process for engine actuation based on the signals received by engine reactions. One common problem that could be controlled by using the electronic control system is engine misfire. Misfire is defined as the mixture that fails to properly combust in the combustion chamber at power stroke in an engine.

Therefore, it is important to control the mixture of air and fuel in the intake stroke to be fed into the combustion chamber. Enough mixture will burn the fuel completely; otherwise, unburnt fuel, due to mixtures that are too lean, increases the occurrence of misfire. Moreover, good fuel combustion avoids the engine running at

3

high temperature where it could break some engine parts, such as valve seal and piston ring, which are delicate to very high temperature. Besides, engine misfire reduces the power of engine since some power strokes are missed to combust the mixture appropriately. In this circumstance, the vehicles would experience jerking, while acceleration is uncomfortable when a driver drives the vehicle. On top of that, it is dangerous for motorcyclists while taking a corner that needs smooth engine running.

In terms of emission produced by the internal combustion engine, incomplete combustion produces harmful gases, such as oxide of nitrogen (NOx), while the engine is running on lean condition and high temperature. This gas is formed by the reaction of oxygen and nitrogen and it has more nitrogen atoms, as nitric oxide is produced by the engine. Furthermore, incomplete combustion emits carbon monoxide (CO) when the combustion has insufficient oxygen because the carbon atom is not fully oxidized into carbon dioxide (CO2). Other than that, emission of hydrocarbon (HC) could occur as well during this condition. HC is a raw fuel that is not burnt and it is produced by exhaust emission. The variation in emission product due to incomplete combustion can contribute to engine misfire. Figure 1.3 illustrates high NOx when the engine is running slightly lean, while HC and CO are high in a rich mixture condition. Hence, in order to avoid excessive incomplete combustion, the engine needs to run at a stoichiometric ratio, which is 14.7 of air to 1 part of fuel.

4

Figure 1.3 Types of compounds emitted by the exhaust at different conditions (To, 2012)

1.2 Problem statement

The accuracy of on board diagnostic (OBD) in detecting many problems in the engine system is important so that it complies with the standard of regulation. At present, most vehicles with internal combustion engine employ crankshaft position sensor (CPS) to detect the occurrence of misfires while the engine is running.

However, some limitations have been discovered that contribute to false misfire alarm or the inability to detect precisely. This has been due to the failure of the sensor to rightly detect the tooth of flywheel when the sensor is dirty or some teeth are misshaped and increase the tendency of false alarm rate (Naik, 2004). Moreover, misfire detection using CPS based on engine speed would share a similar signal with engine misfire when the vehicle has intermittent braking due to uneven road.