<%BANNER%>

Real-Time Adaptive Cutting Tool Flank Wear Prediction

Permanent Link: http://ufdc.ufl.edu/UFE0042405/00001

Material Information

Title: Real-Time Adaptive Cutting Tool Flank Wear Prediction
Physical Description: 1 online resource (176 p.)
Language: english
Creator: Lee, Gun
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: flank, power, wear
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: In modern manufacturing industry, automation is main stream to create products rapidly and economically. Many researches for automation are also going in the metal cutting industry. Therefore, computer Numerical Controlled (CNC) machines are widely used to achieve this goal while maintaining flexible production. Although the advent of CNC in the cutting industry has given many conveniences and benefits, CNC still has many limits. For example, contemporary CNC machines often cannot anticipate the problems caused by unexpected changes in the workpiece. Consequently, much research has been done to develop techniques to respond to these changes. For automation in metal cutting, it is very important to predict workpiece and tool condition. In turning operations, unexpected changes in the workpiece material properties can have negative effects on the efficiency of the operation and quality of the product. Variations in workpiece hardness and dimensions can cause variation in cutting forces, which can then lead to accelerated tool wear and even breakage. Such problems can be overcome during CNC operations by measuring the variation in hardness in the workpiece and adjusting the cutting conditions to account for increased forces. However, there are limitations to in-process measurements of material hardness. Conventional hardness measurement devices require contact with the material being measured, which can be time-consuming and may damage the workpiece. A method to detect variations in workpiece hardness that does not rely on contact could preserve tool life without costing additional time or creating damage in the workpiece. Theoretically, the spindle power required for turning operations in hard materials is higher than that required for soft materials. Therefore, a power sensor provides a novel means of detecting hardness changes in the work material without affecting the cutting process. Tool condition monitoring is important part for automation in metal cutting and many researches for tool condition monitoring have been done. Currently, many wear models are known. However, there is limitation because metal cutting process is very complex and has various conditions. In this research, flank wear was considered the main wear factor. Flank wear arises due to both adhesive and abrasive wear mechanisms from the intense rubbing action of the two surfaces in contact, i.e., the clearance face of the cutting tool and the newly formed surface of the workpiece. Its rate of increase at the beginning of the tool life is rapid, settling down to a steady state then accelerating rapidly again at the end of tool life. Flank wear leads to a deterioration of surface quality, increased contact area and, consequently, increased heat generation. Flank wear models have been developed for specific workpieces made of a single material in many studies. However, if workpiece material properties (such as hardness) are changed during operation, the existing flank wear models cannot be used, because general flank wear models cannot reflect the real-time workpiece material changes. Therefore, the flank wear model what is possible to use in cutting of workpiece jointed parts of different materials. First, the sensor what can detect the workpiece material change was decided. Three different types of sensors (power sensor, ultimate thermometer, and dynamometer) were tested for feasibility of detecting workpiece material changes. In the case of the ultimate thermometer, an infrared sensor was tested, and problems arose due to the difficulty of focusing on the tool edge. The dynamometer was found to be good for detecting the workpiece changes, but installation is difficult and also expensive. The dynamometer also adds unwanted vibrations by increasing the length of tool holder. The power sensor was installed to measure the spindle motor power. The power sensor was found to be the most practical choice for a sensor to detect the workpiece material changes. This is because installation of the sensor is easy compared to the dynamometer and it is also cheaper than the dynamometer. In this dissertation, the ultimate objective is to demonstrate the real-time flank wear model using a power sensor. This proposed model estimates the flank wear even for a workpiece with varying material properties. To validate proposed model, two different materials was used for cutting test. 8620 alloy steel was as soft metal and P20 tool steel was used as hard metal. Results show that proposed model can be used in cutting of workpiece combined with parts of different materials. However, consistency of tool is very important. In case of low quality tool, flank wear rate was not same when tool was changed. But, the flank wear rate was maintained in sing tool even thought cutting was performed with parts of different materials. Additionally, the study of tool should be performed. The hardness of tool or tool surface patterns may be one of reason that flank wear rate is not same even though same workpiece is cut with same condition. In the future, this technique can be implemented with adaptive cutting condition rules to make decisions that reduce cutting cost and maintain product quality.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Gun Lee.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Schueller, John K.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042405:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042405/00001

Material Information

Title: Real-Time Adaptive Cutting Tool Flank Wear Prediction
Physical Description: 1 online resource (176 p.)
Language: english
Creator: Lee, Gun
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: flank, power, wear
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: In modern manufacturing industry, automation is main stream to create products rapidly and economically. Many researches for automation are also going in the metal cutting industry. Therefore, computer Numerical Controlled (CNC) machines are widely used to achieve this goal while maintaining flexible production. Although the advent of CNC in the cutting industry has given many conveniences and benefits, CNC still has many limits. For example, contemporary CNC machines often cannot anticipate the problems caused by unexpected changes in the workpiece. Consequently, much research has been done to develop techniques to respond to these changes. For automation in metal cutting, it is very important to predict workpiece and tool condition. In turning operations, unexpected changes in the workpiece material properties can have negative effects on the efficiency of the operation and quality of the product. Variations in workpiece hardness and dimensions can cause variation in cutting forces, which can then lead to accelerated tool wear and even breakage. Such problems can be overcome during CNC operations by measuring the variation in hardness in the workpiece and adjusting the cutting conditions to account for increased forces. However, there are limitations to in-process measurements of material hardness. Conventional hardness measurement devices require contact with the material being measured, which can be time-consuming and may damage the workpiece. A method to detect variations in workpiece hardness that does not rely on contact could preserve tool life without costing additional time or creating damage in the workpiece. Theoretically, the spindle power required for turning operations in hard materials is higher than that required for soft materials. Therefore, a power sensor provides a novel means of detecting hardness changes in the work material without affecting the cutting process. Tool condition monitoring is important part for automation in metal cutting and many researches for tool condition monitoring have been done. Currently, many wear models are known. However, there is limitation because metal cutting process is very complex and has various conditions. In this research, flank wear was considered the main wear factor. Flank wear arises due to both adhesive and abrasive wear mechanisms from the intense rubbing action of the two surfaces in contact, i.e., the clearance face of the cutting tool and the newly formed surface of the workpiece. Its rate of increase at the beginning of the tool life is rapid, settling down to a steady state then accelerating rapidly again at the end of tool life. Flank wear leads to a deterioration of surface quality, increased contact area and, consequently, increased heat generation. Flank wear models have been developed for specific workpieces made of a single material in many studies. However, if workpiece material properties (such as hardness) are changed during operation, the existing flank wear models cannot be used, because general flank wear models cannot reflect the real-time workpiece material changes. Therefore, the flank wear model what is possible to use in cutting of workpiece jointed parts of different materials. First, the sensor what can detect the workpiece material change was decided. Three different types of sensors (power sensor, ultimate thermometer, and dynamometer) were tested for feasibility of detecting workpiece material changes. In the case of the ultimate thermometer, an infrared sensor was tested, and problems arose due to the difficulty of focusing on the tool edge. The dynamometer was found to be good for detecting the workpiece changes, but installation is difficult and also expensive. The dynamometer also adds unwanted vibrations by increasing the length of tool holder. The power sensor was installed to measure the spindle motor power. The power sensor was found to be the most practical choice for a sensor to detect the workpiece material changes. This is because installation of the sensor is easy compared to the dynamometer and it is also cheaper than the dynamometer. In this dissertation, the ultimate objective is to demonstrate the real-time flank wear model using a power sensor. This proposed model estimates the flank wear even for a workpiece with varying material properties. To validate proposed model, two different materials was used for cutting test. 8620 alloy steel was as soft metal and P20 tool steel was used as hard metal. Results show that proposed model can be used in cutting of workpiece combined with parts of different materials. However, consistency of tool is very important. In case of low quality tool, flank wear rate was not same when tool was changed. But, the flank wear rate was maintained in sing tool even thought cutting was performed with parts of different materials. Additionally, the study of tool should be performed. The hardness of tool or tool surface patterns may be one of reason that flank wear rate is not same even though same workpiece is cut with same condition. In the future, this technique can be implemented with adaptive cutting condition rules to make decisions that reduce cutting cost and maintain product quality.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Gun Lee.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Schueller, John K.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042405:00001


This item has the following downloads:


Full Text

PAGE 1

1 REAL TIME ADAPTIVE CUTTING TOOL FLANK WEAR PREDICTION By GUN LEE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSO PHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Gun Lee

PAGE 3

3 To my dear parents, whose continuous love, prayer, and support made my journey possible

PAGE 4

4 ACKNOWLEDGMENTS At first, I thank my parents for thei r support and love. I also thank my advisor, Dr. John K. Schueller, for his kind support and advice during my whole graduate education. I also thank my committee, Dr. Tony L. Schmitz, Dr. Hitomi Greenslet, Dr. Carl D. Crane, and Dr. Jacob Hammer for their support and guidance. Finally, I thank all my colleagues and friends at Machine Tool Research Center (MTRC).

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 LIST OF FIGURES .......................................................................................................... 8 ABSTRACT ................................................................................................................... 12 CHAPTER 1 INTRODUCTION .................................................................................................... 16 2 PREVIOUS RESEARCH IN TOOL CONDITION MONITORING SYSTEM (TCMS) ................................................................................................................... 20 2.1 Need of a Tool Condition Monitoring System .................................................... 20 2.1.1 Electric Motor Current/Power Measurement ............................................ 21 2.1.2 Cutting Forces (Dynamic and Static) ....................................................... 23 2.1.3 Acoustic Emission (AE) ........................................................................... 27 2.1.4 The Tool Tip/Cutting Edges Temperature ............................................... 30 2.1.5 Vibration Signatures ( Acceleration Signals) ............................................ 33 2.1.6 Miscellaneous Sensors and Methods ...................................................... 35 2.1.6.1 Optical methods ............................................................................. 35 2.1.6.2 Stress/strain measurements .......................................................... 36 2.1.6.3 Workpiece dimension ..................................................................... 37 2.1.6.4 Magnetism ..................................................................................... 37 2.1.6.5 Ultrasonic methods ........................................................................ 38 2.1.7 Sensor fusionsynergy of signal integration .......................................... 38 3 FORCE, POWER, FLANK WEAR, AND OPTIMIZED CUTTING CONDITION ....... 46 3.1 The Mechanistic Cutting Force for a Turning Operation ................................... 46 3.2 Conversion between Cutting Force and Power for a Turning Operation ........... 47 3.3 Wear Rate and Tool Life ................................................................................... 47 3.4 Cutting Tool Wear Prediction ............................................................................ 51 3.5 Derivation of the Wear Characteristic Equation ................................................ 53 3.6 Flank Wear Progress ........................................................................................ 56 3.7 Real time Flank Wear Model for a Workpiece Composed of Various Materials .............................................................................................................. 58 4 SENSOR FEASIBILITY TEST FOR REAL TIME WORKPIECE MATERIAL DETECTION ........................................................................................................... 64 4.1 Ultimate Thermometer (Infrared Sensor) .......................................................... 64

PAGE 6

6 4.2 Cutting Force Measurement on Hass SL10 Lathe ........................................... 65 4.3 Spindle Power Measurements on Hass SL10 Lathe........................................ 66 4.4 Power Measurement on Okuma LC 40 Lathe ................................................... 67 4.5 Power Measurement on SyiL CNC Lathe (C6B) ............................................... 68 5 RESULT OF FEASIBILITY TESTs FOR REAL TIME CUTTING POWER DETECTION ........................................................................................................... 79 5.1 Ultimate Thermometer (Infrared Sensor) .......................................................... 79 5.2 Power Meter and Dynamometer Test on Hass SL10 ....................................... 80 5.3 Signal Analysis of Okuma LC 40 Power Data ................................................... 83 5.4 Power Measurement Result on SyiL CNC Lathe (C6B) .................................... 86 6 FLANK WEAR MODEL FOR WORKPIECE COMBINED FROM DIFF ERENT HARDNESS MATERIALS ..................................................................................... 106 6.1 Adaptive Real Time Flank Wear Model for Workpiece Hardness Changes .... 106 6.2 Experimental Plan ........................................................................................... 107 7 RESULT OF FLANK WEAR MODEL FOR WORKPIECE COMBINED WITH TWO DIFFERENT HARDNESS MATERIALS ...................................................... 115 7.1 Hardness Measurement of 8620 Alloy Steel and P20 and Flank Wear Images ............................................................................................................... 115 7.2 Flank Wear Measurement ............................................................................... 116 7.2 Flank Wear Rate between 8620 Alloy Steel and P 20 ..................................... 118 7.3 Flank Wear Estimation Using Power ............................................................... 120 8 CONCULSIONS AND FUTURE WORK ............................................................... 141 APPENDIX A LINEAR POWER MODELING USING HASSSL10 TEST DATA ......................... 145 B ANALYSIS OF OKUMA LC40 POWER DATA (SECOND DAY SET) ................. 158 C SENSOR OUTPUT ............................................................................................... 164 LIST OF REFERENCES ............................................................................................. 167 BIOGRAPHICAL SKETCH .......................................................................................... 176

PAGE 7

7 LIST OF TABLES Table page 2 1 Previous research of tool condition monitoring system ....................................... 41 4 1 Ultimate Thermometer Specification ................................................................... 70 4 2 Specifications of the NI USB 6009 I/O board ...................................................... 70 4 3 Cutting conditions for power measurement on Okuma LC 40 ............................ 70 4 4 Workpiece materials specification (from McMaster CARR) ................................ 71 5 1 Mean cutting force values for aluminum steel aluminum steel workpiece on Hass SL 10 ......................................................................................................... 88 5 2 Results for a = 1 mm, fr = 0.1 mm/rev cutting tests using aluminum/steel workpiece ( Hass SL 10 ) ..................................................................................... 88 5 3 Results for a = 1 mm, fr = 0.2 mm/rev cutting tests us ing aluminum/steel workpiece ( Hass SL 10 ) ..................................................................................... 88 5 4 P ower during turning operations on Okuma LC 40 ............................................. 89 5 5 Predicted power due to radius change and measured power on Okuma LC 40 ....................................................................................................................... 90 5 6 Mean power change by spindle speed change on Okuma LC 40 ...................... 90 6 1 Cutting conditions in wear tests ........................................................................ 111 6 2 Digital microscope specification ........................................................................ 111 6 3 Cutting conditions in single insert tests ............................................................. 112 7 1 Specification of Wilson Rockwell hardness tester model 3JR .......................... 123 7 2 Chemistry data of P20 tool s teel and 8620 alloy steel ...................................... 123 7 3 Probability levels and the range of acceptable data of flank wear rates of case hard and soft ............................................................................................ 123 A 1 Measured cutting forces in the tangential direction ........................................... 148 A 2 Total error for all power models ........................................................................ 148 B 1 Maximum values of f acing operations, mean values and standard deviations of turning operations ......................................................................................... 159

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Flow of research ................................................................................................. 19 2 1 Force amplitude vs time (fixed cutting conditions). ............................................. 45 3 1 The cutting force component in turning operation ( Dimla, 2000) ........................ 59 3 2 Basic turning operation ....................................................................................... 59 3 3 Cutting tool wear forms in orthogonal metal cutting ( Dimla, 2000) ..................... 60 3 4 Conventional features of turning tool wear measurements ( Dimla, 2000) .......... 61 3 5 Development of flank wear width with time (redrawn from Tlusty, 2000) ............ 61 3 6 Threedimensional cutting model in turning ........................................................ 62 3 7 Stress distribution on tool ................................................................................... 62 3 8 Effect of cutting speed ( v) and feed ( fr) on the cost per part of components ....... 63 4 1 Experimental infrared sensor setup schematic ................................................... 72 4 2 Actual setup for infrared sensor test ................................................................... 72 4 3 Cutting force components measured by dynamometer during cutting tests ....... 73 4 4 Cutting force test setup on Hass SL10 .............................................................. 73 4 5 Aluminum steel aluminum steel workpiece for Hass SL10 power and force measurement ...................................................................................................... 74 4 6 Fast Response PH 13 Power Cell mounted in Haas SL10 power cabinet ........ 74 4 7 Fast Response PH 13 Power Cell mounted in Okuma LC 40 power cabinet ..... 75 4 8 Screen shot of data gathering program constructed by LabView for power measurement ...................................................................................................... 75 4 9 Finished pipe joint shape .................................................................................... 76 4 10 Power data while cutting an example part on Okuma LC 40 .............................. 76 4 11 SyiL CNC lathe, National Instruments USB DAQ board and Load Control Inc power sensor ...................................................................................................... 77

PAGE 9

9 4 12 Three different types of workpiece for power measurement test on SyiL CNC lathe .................................................................................................................... 78 5 1 Temperature readings from trials for infrared sensor test ................................... 91 5 2 Cooling profiles of infrared sensor test ............................................................... 92 5 3 Measured cutting force on Hass SL10 .............................................................. 93 5 4 Spindle power (no cutting) as a function of spindle speed on Hass SL10 ......... 94 5 5 Cutting force and spindle power on Hass SL10. ................................................ 95 5 6 Cutting force and spindle power on Hass SL10. ................................................ 96 5 7 Mean x direction cutting force and spindle power on Hass SL10 as a function of workpiece diameter ........................................................................... 96 5 8 Cutting force and spindle power on Hass SL10 for a =1mm and fr=0.1 mm/rev using steel/steel at 500rpm .................................................................... 97 5 9 Maximum values of facing operations, mean values of turning operations and Standard deviations ............................................................................................ 99 5 10 Max imum, mean, and standard variation LC 40 ............................................... 101 5 11 Power difference between worn and normal tool in second facing on Okuma LC 40 ................................................................................................................ 102 5 12 Mean and maximum value of measured power on Okuma LC 40 .................... 102 5 13 Measured power under air cutting on Okuma LC 40 ........................................ 103 5 14 Power measurement on Okuma LC 40 under different spindle speed ............. 103 5 15 Measured cutting power (sensor value in voltage) on SyiL C6B (workpiece A) 104 5 16 Measured cutting power (sensor value in voltage) on SyiL C6B (workpiece B) 104 5 17 Measured cutting power (sensor value in voltage) on SyiL C6B (workpiece C) 105 5 18 Measured cutting power (sensor value in voltage) on SyiL C6B (workpiece C) 105 6 1 Real time flank wear estimation model scheme ................................................ 113 6 2 Workpiece shapes for validation ....................................................................... 114 7 1 Wilson Rockwell hardness tester model 3JR .................................................... 124

PAGE 10

10 7 2 Rockwell hardness A of 8620 alloy s teel and P20 tool steel workpieces .......... 125 7 3 Box plots of Rockwell hardness A with whisker of 8620 alloy steel and P20 tool steel workpieces ........................................................................................ 125 7 4 Flank wear images of case soft second cutting test by microcope ................... 126 7 5 Flank wear of case hard ................................................................................... 127 7 6 Flank wear of case soft ..................................................................................... 127 7 7 Flank wear of case half half .............................................................................. 128 7 8 Flank wear of case mainly soft ......................................................................... 128 7 9 Flank wear of case single tool (TCMT32.52 carbide) ....................................... 129 7 10 Flank wear of case single tool (C 5 carbide) .................................................... 130 7 11 Flank wear rate versus Rockwell hardness A ................................................... 131 7 12 Flank wear surface of two different new inserts ................................................ 131 7 13 ANOVA screen shot case soft, case hard, case half half, and case mainly soft .................................................................................................................... 132 7 14 One way ANOVA screen shot for case half half (case half half 1) ................... 1 32 7 15 Flank wear rate versus Rockwell hardness A (case half half 1, Correlation coefficient r = 0.9149) ....................................................................................... 133 7 16 Flank wear rate versus workpiece materials (case half half 1, Correlation coefficient r = 0.8622) ....................................................................................... 133 7 17 Flank wear rate versus workpiece materials (case mainly soft 1, Correlation coefficient r = 0.7706) ....................................................................................... 134 7 18 Flank wear rate versus Rockwell hardness A (case mainly soft 1, Correlation coefficient r = 0.6249) ....................................................................................... 134 7 19 Flank wear rate versus workpiece materials (case single tool TCMT32.52 carbide) ............................................................................................................ 135 7 20 Flank wear rate versus workpiece materials (case single tool C 5 carbide) ..... 135 7 21 Power measurement of 8620 alloy steel cut ..................................................... 136 7 22 Hardness vs flank wear rate (case single tool C 5) .......................................... 136

PAGE 11

11 7 23 Hardness vs aver age cutting power (case single tool C 5) ............................... 137 7 24 Average cutting power vs flank wear rate (case single tool C 5) ...................... 137 7 25 Flank wear model from C 5 case soft and hard ................................................ 138 7 26 Real time flank wear estimation using power sensor ........................................ 139 7 27 Difference between flank wear and estimated flank wear ................................. 140 7 28 Flank wear rate (mm/m cutting length) during 50 mm cutting length ................ 140 A 1 Differenc e between experimental power and original power model .................. 149 A 2 Total error between experiment and model 1 according to changing a values 150 A 3 Model 1 and experimental power ...................................................................... 151 A 4 Total error between experiment and model 2 according to changes in a and b values ............................................................................................................... 152 A 5 Model 2 and experimental power ...................................................................... 153 A 6 Total error between experiment and model 3 according to changes in a and b values ............................................................................................................... 154 A 7 Model 3 and experimental power ...................................................................... 155 A 8 Total error between experiment and model 4 according to changes in a and b values ............................................................................................................... 156 A 9 Model 4 and experimental power ...................................................................... 157 B 1 Maximum values of facing operations, mean values of turning operations ....... 160 B 2 Standard deviations of turning operations ........................................................ 161 B 3 Figure Maximum, mean, and st andard variation on Okuma LC 40 (second day set) ............................................................................................................. 163 C 1 Connection of UPS and D.C motor, and turns .................................................. 165 C 2 Sensor output under 5 K Ohm and 1 turn of sensor set up .............................. 166 C 3 Sensor output under 20 K Ohm and 6 turn of sensor set up ............................ 166

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy REAL TIME ADAPTIVE CUTTING TOOL FLANK WEAR PREDICTION By Gun Lee December 2010 Chair: John K. Schueller Major: Mechanical Engineering In modern manufacturing industry, automation is main stream to create products rapidly and economically Many researches for automation are also going in the metal cutting industry. Therefore, c omputer Numerical Controlled (CNC) machines are widely used to achieve this goal while maintaining flexible production. Although the advent of CNC in t he cutting industry has given many conveniences and benefits, CNC still has many limits. For example, contemporary CNC machines often cannot anticipate the problems caused by unexpected changes in the workpiece. Consequently, much research has been done to develop techniques to respond to these changes. For automation in metal cutting, it is very important to predict workpiece and tool condition. In turning operations, unexpected changes in the workpiece material properties can have negative effects on the efficiency of the operation and quality of the product. Variations in workpiece hardness and dimensions can cause variation in cutting forces, which can then lead to accelerated tool wear and even breakage. Such problems can be overcome during CNC operations by measuring the variation in hardness in the workpiece and adjusting the cutting conditions to account for increased forces. However, there are limitations to inprocess measurements of material

PAGE 13

13 hardness. Conventional hardness measurement devices requi re contact with the material being measured, which can be timeconsuming and may damage the workpiece. A method to detect variations in workpiece hardness that does not rely on contact could preserve tool life without costing additional time or creating damage in the workpiece. Theoretically, the spindle power required for turning operations in hard materials is higher than that required for soft materials. Therefore, a power sensor provides a novel means of detecting hardness changes in the work material w ithout affecting the cutting process. Tool condition monitoring is important part for automation in metal cutting and many researches for tool condition monitoring have been done. Currently, many wear models are known. However, there is limitation because metal cutting process is very complex and has various conditions. In t his research, flank wear was considered the main wear factor. Flank wear arises due to both adhesive and abrasive wear mechanisms from the intense rubbing action of the two surfaces in contact, i.e., the clearance face of the cutting tool and the newly formed surface of the workpiece. Its rate of increase at the beginning of the tool life is rapid, settling down to a steady state then accelerating rapidly again at the end of tool life. Flank wear leads to a deterioration of surface quality, increased contact area and, consequently, increased heat generation. Flank wear models have been developed for specific workpieces made of a single material in many studies. However, if workpiece mater ial properties (such as hardness) are changed during operation, the existing flank wear models cannot be used, because general flank wear models cannot reflect the real time workpiece material changes.

PAGE 14

14 Therefore, the flank wear model what is possible to us e in cutting of workpiece jointed parts of different materials. First, the sensor what can detect the workpiece material change was decided. T hree different types of sensors (power sensor, ultimate thermometer, and dynamometer) were tested for feasibility of detecting workpiece material changes. In the case of the ultimate thermometer, an infrared sensor was tested, and problems arose due to the difficulty of focusing on the tool edge. The dynamometer was found to be good for detecting the workpiece changes, but installation is difficult and also expensive. The dynamometer also adds unwanted vibrations by increasing the length of tool holder. The power sensor was installed to measure the spindle motor power. The power sensor was found to be the most practi cal choice for a sensor to detect the workpiece material changes. This is because installation of the sensor is easy compared to the dynamometer and it is also cheaper than the dynamometer. In this dissertation, the ultimate objective is to demonstrate t he real time flank wear model using a power sensor. This proposed model estimates the flank wear even for a workpiece with varying material properties. To validate proposed model, two different materials was used for cutting test. 8620 alloy steel was as s oft metal and P20 tool steel was used as hard metal. Results show that proposed model can be used in cutting of workpiece combined with parts of different materials. However, consistency of tool is very important. In case of low quality tool, flank wear rate was not same when tool was changed. But, the flank wear rate was maintained in sing tool even thought cutting was performed with parts of different materials. Additionally, the study of tool should be performed. The hardness of tool or tool surface patt erns may be one of

PAGE 15

15 reason that flank wear rate is not same even though same workpiece is cut with same condition. In the future, this technique can be implemented with adaptive cutting condition rules to make decisions that reduce cutting cost and maintain product quality.

PAGE 16

16 CHAPTER 1 INTRODUCTION The final goal of manufacturing is to create products rapidly, economically, and with high quality. Computer Numerical Controlled (CNC) machines are widely used in the metal cutting industry to achieve this goal while maintaining flexible production. Although the advent of CNC in the cutting industry has given many conveniences and benefits, CNC still has many limits. For example, contemporary CNC machines often cannot anticipate the problems caused by unexpected changes in the workpiece. Consequently, much research has been done to develop techniques to respond to these changes. During metal cutting operations, if a tool fails it may damage the tool holder, the workpiece, or the machine elements. Also, as machining progresses and the tool wears out, the surface quality and the dimensional accuracy of the product degrade. Moreover, tool breakage may jeopardize operator safety, or may lead to problem in the manufacturing system. In turning operations, unexpected changes in the workpiece material properties can have negative effects on the efficiency of the operation and quality of the product. Variations in workpiece hardness and dimensions can cause variation in cutting forces, which can then lead to accelerated tool w ear and even breakage. Such problems can be overcome during CNC operations by measuring the variation in hardness in the workpiece and adjusting the cutting conditions to account for increased forces. However, there are limitations to inprocess measurements of material hardness. Conventional hardness measurement devices require contact with the material being measured, which can be timeconsuming and may damage the workpiece. A method to

PAGE 17

17 detect variations in workpiece hardness that does not rely on contact could preserve tool life without costing additional time or creating damage in the workpiece. T heoretically, the spindle power required for turning operations in hard materials is higher than that required for soft materials. Therefore, a power sensor provides a novel means of detecting hardness changes in the work material without affecting the cutting process. In this research, f lank wear will be considered the main wear factor. Flank wear arises due to both adhesive and abrasive wear mechanisms from the intense rubbing action of the two surfaces in contact, i.e., the clearance face of the cutting tool and the newly formed surface of the workpiece. Its rate of increase at the beginning of the tool life is rapid, settling down to a steady state then accel erating rapidly again at the end of tool life. Flank wear leads to a deterioration of surface quality, increased contact area and, consequently, increased heat generation. Flank wear models have been developed for specific workpieces made of a single mater ial in many studies. However, if workpiece material properties (such as hardness) are changed during operation, the existing flank wear models cannot be used, because general flank wear models cannot reflect the real time workpiece material changes. In thi s dissertation, to detect a workpiece material change, three different types of sensors (power sensor, ultimate thermometer, and dynamometer) were tested for feasibility of detecting workpiece material changes. In the case of the ultimate thermometer, an i nfrared sensor was tested, and problems arose due to the difficulty of focusing on the tool edge. T he dynamometer was found to be good for detecting the workpiece changes, but installation is difficult. The dynamometer also adds unwanted vibrations. The power

PAGE 18

18 sensor was found to be the most practical choice for a sensor to detect the workpiece material changes. This is because installation of the sensor is easy compared to the dynamometer and it is also cheaper than the dynamometer. In Chapter 5, the measured data will be presented, and the result will show that the power sensor data trend is the same as that of the dynamometer data. In this dissertation, the ultimate objective is to demonstrate the real time flank wear model to detect material change usin g a power sensor. This proposed model estimates the flank wear even for a workpiece with varying material properties. To validate the proposed model, two different materials which have different hardness were used for cutting test s Tests showed that tool variability made wear prediction difficult. However, when there was not tool variability, tool wear could be predicted. Therefore, it was concluded that proposed model can be used. In the future, this technique can be implemented with adaptive cutting condition rules to make decisions that reduce cutting cost and maintain product quality. However, the future study of tool and the sensor will be required. Figure 11 shows simplified flow of this research.

PAGE 19

19 Figure 11 Flow of research

PAGE 20

20 CHAPTER 2 PRE VIOUS RESEARCH IN TO OL CONDITION MONITOR ING SYSTEM (TCMS) Tool condition monitoring (TCM) is recognized to be important in CNC processes since the excessive wear or breakage of tool has to be noticed immediately in an automated manufacturing system to keep the qual ity and productivity. Much research has been carried out concerning the development of a reliable TCM system. H owever, none has yet found ubiquitous industrial use (Dimla, 2000 and Silva, 2007). One of the reasons for the lack of industrial applic ation of TCM systems is due to the fact that TCM systems have been developed based mainly on mathematical models, which require huge amounts of empirical data, reducing therefore their adaptation capacity. Another possible hindrance lies in the nature and characteristics of the utiliz ed sensor signals in general, which tend to be stochastic and nonstationary, and therefore difficult to model. There has been much research using acoustic emission (AE) tool dynamometer s, motor currents, etc. In this chapter, Tool Condition Monitoring Systems (TCMSs) will be introduced. TCMS s are classified by sensor type ( power, cutting force, acoustic emissi on, temperature, vibration, etc.). 2.1 Need of a Tool Condition Monitoring S ystem A tool failure such as wear, breakage, chipping, etc., could lead to poor product quality and even damage the machine tool and/or the fixture. M onitoring in a metal cutting process requires monitoring the machine dynamics, the cutting process dynamics, the cutting tools, and the workpiece t o ensure optim um performance of the system (Byrne, 1995) A tool condition monitoring system can therefore be viewed as serving as the advanced fault detection system for cutting and t ool, to check and safeguard machining process stability, a means by whic h machining tolerance is

PAGE 21

21 maintained on the workpiece to acceptable limits by providing a compensatory mechanism for tool wear offsets, and a machine tool damage avoidance system (Dilma, 2000) The TCMS may be also required detecting to including excessive power consumption, inaccurate tolerances, serrations, and uneven workpiece surface finish, which eventually could lead to machine tool and/or workpiece damage incurring unnecessary costs. Cook (1980) divided the methods of tool condition monitoring into di rect and indirect methods The direct methods are based on the direct measurement of the geometry of the tool using optical sensors or laser optical sensors (Fan, 1996 and Du, 1993) They can provide accurate measurements of tool conditions but also can be affected by chips, coolant, and various disturbances. In the indirect methods, tool condition is predicted based on various sensor signals such as cutting force, vibration, temperature, acoustic emission, and motor current or power In this chapter, reviews of power, cutting force, acoustic emission, tool temperature vibration signature (acceleration signals), and miscellaneous methods such as ultrasonic and optical measurements, workpiece surface finish quality, workpiece dimensions, stress/strain analys is and spindle motor current will be provided. 2.1.1 Electric Motor Current/P ower M easurement B hattacharyya et al. (2008) proposed a method for continuous online estimation of tool wear using spindle motor current and voltage measurement. They applied t he method to a face milling operation. They used a multi ple linear regression model to estimate tool wear in real time and obtained accurate predictions of tool wear in both laboratory and industrial experiments.

PAGE 22

22 Lee et al. (2002) developed a TCMS based on spindle motor current. They used the wavelet transform to measured spindle motor current and presented the energy level of tool conditions (new, worn, and bad). The proposed energy level method was measured the wear well. The change was very slight in ti me domain and it is noticeable in the specific energy level when processed in the wavelet analysis. The wavelet theory has been studied for the detection of tool wear conditions in various processes. Zhang et al. (1994) measured the spindle motor current of a vertical NC milling machine by using a Hall Effect sensor during cutting. The relationship between the meas ured motor current and the milling torque was modeled. Computer algorithms were developed to track the waveforms, rate of peak change and the relative eccentricities of the modeled relation. Cutter breakages were more reliably diagnosed with motor current measurements than force/torque measurements. Constantinides and Bennett (1987) measured the spindle motor power for estimation of wear and the detection of the end of effective tool life for a vertical milling machine. Spectral analysis was used and the effectiveness of using moving average, runningmeans and cumulative sum of the power spectrum values to estimate wear was evaluated. Between the cumulative sum of the power spectral energies and the total flank wear, there was a strong correlation. Spectral energy fluctuation of the spindle motor power was linearly related to the tool wear rate. Shaft power, cutting forces, torque and motor current are all related to each other, originating from, and depending entirely on one another. It suffices therefore to measure just one of these parameters as demonstrated by Rangwala and Dornfeld (1987 and 1990) They dropped the electric current in preference to the cutting forces.

PAGE 23

23 2.1.2 Cutting Forces (Dynamic and S tatic) The relationship between the cutting force and tool wear ( Lister, 1993; Ravindra, 1993; Dimla, 1999; Bayramoglu, 1998; Ko, 1994; Purushothaman, 1994; Tarng, 1994; Dornfeld, 1990; Lee, 1992; Marques, 1991; Kim, 1991; Oraby, 1991; Yao, 1990; Yao, 1992; Ramalingam, 1990; Lee, 1994) has been widely established. However, cutting force may not used in case of small breakage of tool. If small breakage has occurred, this may cause decreasing of the depth of cut. Generally, a smaller depth of cut decreases the cutting force. In practice, application and interpretation of this parameter has been diverse, but concentrated on studying the dynamic characteristic of the cutting force signal and interpreting its relation to tool wear levels. This can largely be attributed to the fact that force becomes important in worn tool conditions as a result of the variations produced due to friction between the cutting tool flank and the workpiece (Ko, 1994 and Dornfeld, 1990) Existing force based TCMSs typically operate indepe ndently of absolute force magnitude, measuring the relative change of the force that occurs as a new tool wears ( Elanayar, 1990 and Gould, 1988) or when it fractures (Choi, 1999). Experiments have shown that the three components of the cutting force (Fig. 31) respond differently to the various wear forms occurring on the tool. For example, the feed force is insensi tive to crater wear whereas the feed and radial forces may be influenced more than the main cutting force (Lister, 1993 and Gould, 1988) Dimla (1999 and 2000) devel oped an online tool wear monitoring system for a turning operation using cutting force measurements fused with vibration signatures. A tool post dynamometer was used to measure cutting force. Measurements of flank,

PAGE 24

24 nose and notch wear lengths were made immediately preceding recording the online data. Static and dynamic entities of the sampled cutting force were extracted as the mean and oscillatory components respectively (see Fig. 21) and analyzed in time and frequency domains from which features sensitive to tool wear were identified. Time domain established the nature and level of static force magnitude change while frequency analysis demonstrated the dynamic force s ignatures response to cutting conditions as well as accrued wear levels. The more succinct elements of tool wear as it gradually wore to catastrophic failure were observed better in the frequency domain with certain frequencies correlating very well to th e dynamic force changes. Overall, flank and nose wear were better indicators of tool wear than notch wear. Bayramoglu and Dngel (1998) present a methodical investigation on the use of cutting force ratios in TCM for turning operations. Tests using worn t ools were processed in order to examine how the force ratios could be used to monitor tool wear. Results showed that two of the force ratios to be particularly sensitive to the accrued flank wear, thus cutting force ratios could be applied in TCM operations. Ravindra et al. (1993) proposed a development of mathematical models to describe the wear time and wear force relationships for a turning operation. The wear progression was studied and the cutting forces modeled by a multiple regression analysis method. Experiments were designed to obtain data for new and worn tools and the obtained data were used to establish the effect of cutting condition on cutting forces and tool wear. Force models were constructed in terms of speed, feed, and depth of cut using m ultiple regression analysis. The r esult s showed that an increase in the magnitude of the component of the tri axial cutting force was related to the wear on the

PAGE 25

25 used insert. From the experiments, a good correlation between flank wear and feed and radial forces was provided. Lee et al. (1992) developed the TCMS to track the dynamic cutting force based on a personal computer with a Fast Fourier Transform (FFT) card installed inside The experiments were conducted on a Colchester Mascot 1600 lathe. The cutting forces and wear levels were measured and recorded and analyses of the obtained data showed that the feed and tangential dynamic force components had a good relationship to flank wear trend. Marques and Mesquita (1991) investigated the relationship betwe en the wear of sintered T15 highspeedsteel cutting tools and the associated cutting forces. The cutting force model was established taking into account the influence of both crater and flank wear. Experimental tests were conducted from during which the f orces were measured. The test types conducted consisted of short duration cuts to establish forcewear relationship, and longer duration cuts to observe the progressive influence of wear on the forces. There was good correlation between experi mental and t heoretical results. Kim and Lee (1991) presented an analytical model of dynamic cutting forces in orthogonal cutting and verified the proposed model. The dynamic cutting forces were expressed in terms of cutting coefficients that depend on the cutting var iable s. Good agreement between the theoretical limits of stability and experimental data was shown. Grabec (1988) and Khraisheh et al. (1995) have performed similar modeling of the dynamic cutting force for chatter prediction. Oraby and Hayhurst (1991) developed mathematical models to describe the wear time and the wear force relationships for steady centre lathe turning conditions. Cutting

PAGE 26

26 forces were found to correlate well with wear progress and with tool failure. To establish a universal rather than a case based example of individual characteristics, a mathematical method to quanti tatively formulate the wear cutting force relation to the cutting speed, feedrate and depth of cut to achieve repeatability and reliability for practical applications was pr oposed. Yao et al. (1990) investigated a comprehensive TCMS which included the measurement of major and minor flank, crater, and nose wear based on the analysis of dynamic cutting forces in oblique machining. Tool wear experiments were performed on a Colc hester lathe at varying cutting conditions using only one tool insert and workpiece type. To develop Autoregressive Moving Average Vector (ARMAV) time series models, the force, measured in terms of three orthogonal force components was used. Based on ARMAV, dispersion analysis (DA) was used to extract features sensitive to the rate of various types of wear. The results show that minor flank wear reaches a critical value first in finish machining, so that optimum cutting conditions or an appropriate tool change strategy must be determined on the basis of minor flank wear. The results also show that the method is a feasible means for online tool wear monitoring in finishmachining. Shi and Ramalingam (1990) performed artificial flank wear tests by measuring cutting force. That feed force to cutting force ratio can be a good indicator of flank wear was shown. This ratio and its derivative can be used to signal tool changes. Experimental results and slipline field analysis, a new real time tool condition sensi ng method using force ratio, was proposed.

PAGE 27

27 Lee et al. (1989) examined the nature and source of the dynamic force frequency spectrum and determined the extent of the correlations between the characteristic peak frequency and flank wear. They found good corr elation between the dynamic cutting force and flank wear. The dynamic force decreased rapidly before the entry into the tertiary zone as well as prior to the onset of catastrophic tool failure. 2.1. 3 Acoustic E mission (AE) Acoustic emissions ( AE) are comm only defined as transient elastic waves within a material caused by the release of localized stress energy. During metal cutting, the workpiece undergoes considerable plastic deformation associated with the generation of AE. AE is linked to the plastic def ormation process during chip formation, due to the interaction between the workpiece and cutting tool (Prickett, 1999) AE is generally found to be more sensitive to tool wear than cutting forces ( Blum, 1990) Other sources of AE include phase transformati ons, friction mechanisms (tool workpiece contact), and crack formation or extension fracture. Chen and Li (2007) proposed a technique based on AE signal wavelet analysis for tool condition monitoring. The local characterize of frequency band, which contai ns the main energy of AE signals, was presented by the wavelet multi resolution analysis, and the singularity of the signal is represented by wavelet resolution coefficient norm. They showed the advantages of the wavelet multi scale resolution method in co mparison to conventional data process. Choi et al. ( 1999) develop a real time tool breakage detection system for turning by the sensor fusion of an acoustic emission sensor and a built in force sensor. A built in piezoelectric force sensor was used to meas ure the cutting force without altering the

PAGE 28

28 characteristics of the machine tool dynamics. Two sets of experiments were done using carbide insert tips with one set slotted by wire EDM to accelerate fracture while the second was brazed to the workpiece to induce tool breakage. The recorded data was analyzed through a fast block averaging algorithm for features and patterns indicative of tool fracture. If tool breakage is occurred, a significant drop of cutting force follows an AE signal burst and this AE signal burst was used as a detecting the force change. In case of signal force drops below the preset threshold, it was considered to be tool breakage. Similar work conducted by Jemielniak and Otman ( 1998) used a statistical signal processing algorithm to ident ify the root mean square (RMS), skew, and kurtosis of the AE signal in the detection of catastrophic tool failure. Kakade et al. (1994) applied AE analysis on the effect of tool wear and corresponding change in chipform in face milling by selecting AE parameters namely ring down count. R ise time was recorded simultaneously with the corresponding flank wear land length measured at fixed intervals. The result s concluded that AE may be readily used for inprocess monitoring of chip status and consequently tool monitoring. Zheng et al. (1992) designed a novel intrinsic optical fibre sensor for detection of AE and designed a sensor used in monitoring tool wear. The sensor consisted of two distinct parts: the sensing element and an interferometer. The sensing element principally was used to produce a phase shift in the light transmitted through the optical fiber. The phase shift is generated by stress in an optical fibre due to coupling an AE wave into a backing material. Preliminary drilling and milling operation tests were performed. A c ommercial AE transducer was used for back to back tests and its response in 200 kHz ~ 1 MHz range compared to that from a commercial PZT sensor.

PAGE 29

29 The obtained results were compared and these showed a reasonable degree of agreem ent. Knig et al. (1992) applied AE to detect fracture and/or monitor the condition of small drills. Drilling operations were performed and a ceramic knock detector sensor designed for industrial application was used to measure the AE. Their result showed that at the closing phase of tool life (i.e., tertiary phase of tool wear), the RMS value increased dramatically. The rise, one could argue, was the case of tool breakage. Hence Knig et al. used this as a prescribed threshold which the RMS for normal operating drills should not exceed. This method however was found to be sensitive to tool chipping. Blum and Inasaki (1990) performed comprehensive experiments to determine the influence of cutting conditions on the generation of Acoustic Emission (AE) signal s during machining S45C steel. During the orthogonal cutting process, AE sensor signals and tool dynamometer signals provided extensive data and theoretical relationships between the energy content of the AE and the plastic work of deformation in the pri mary and secondary cutting zone was used. A relationship between the AE signals gener ated and the strain rate was estimated. The influence of flank wear on the generation of AE signals was emphasized. The feasibility of utilizing AE in tool wear sensing was confirmed with force measurement data. Moriwaki and Tobito (1990) analyzed and measured t he AE signals of turning of medium carbon steel with both coated and uncoated tools. They proposed a method to identify the conditions of the tool by AE signal s employing a pattern recognition technique. AE RMS values for the recorded AE signal and the wear values (initial,

PAGE 30

30 middle and tertiary stages of tool wear) were graphed on the same scale for comparison. The data was applied to a pattern recognition system and it performed reasonably well. Roget et al. (1988) investigated the possibility of detecting tool wear using AE experimentally. They used the custom made AE sensors for turning and milling test cuts. A comparison of the AE and flank wear was carried out and showed a remarkable similarity on both wear time and AE time plots. During metal cutting, there are many factors involved. Therefore, they concluded that AE is suited for determining tool wear only in specific and limited conditions. 2.1. 4 The Tool Tip/ Cutting Edges Temperature Metal cutting generates a significant amount of heat and the temperature in the cutting zone can change as the tool wear due to changes in the tool geometry and tools capability to cut. Therefore, the temperature can be used to monitor the tool condition. T he high temperatures around the cutting tool edges affects the rate and mode of cutting tool wear, the friction between chip and cutting tool, and also that between the cutting tool and the newly formed surface. Sarwar et al. (1996) developed a thermal imaging system consisted of a mathematical model of the energy partition during metal cutting for metal cutting applications. Thermal imaging data obtained showed that it was practicable and modifiable to fulfill the requirement s of orthogonal cutting. Lin (1995) developed an inverse approach to measure the cutting tool temperature during a milling process for real time tool/workpiece interface temperature. A least square inverse scheme was applied to solve the unknown boundary at the tool workpiece interface based on the

PAGE 31

31 surface temperatures measured outside the cutting zone by an infrared (IR) pyrometer. Lin concluded 1) an inverse technique was successfully developed to predict the milling temperature and heat flow by the measur ed temperature near the milled surface, 2) the mapping model was verified in a flame heating test, and 3) the average tool chip interface temperature and total heat through the milling area ware influenced by the thermal properties of the workpiece. Radule scu and Kapoor (1994) designed and tested an analytical tool temperature fields prediction model for use during continuous and interrupted cutting. Testing showed that the modeled tool chip interface temperature agreed well with experimental tests. Raman et al. (1992) proposed and developed a mathematical model for cutting tool temperature measurement based on the remote thermocouple sensing (RTS) principle. They described the application of the differential quadrature method for modeling the forward ther mal behavior of the cutting tool. Initial implementation of the quadrature method verified the suitability of using this technique for RTS. Stephenson and Ali (1990) summarized the result of studies on tool temperature effects on interrupted metal cutting theoretically and experimentally. Infrared and tool chip thermocouples were used to meas ure the cutting temperature during interrupted a nd turning tests on 2024 aluminum and gray cast iron at speeds up to 18 m/s. Temperatures were generally lower in interr upted cutting than in continuous cutting under the same conditions. Temperatures depended primarily on the length of cutting cycles and secondarily on the length of cooling intervals between cycles. The temperature measurements were found to be lower when the cutting was occasionally interrupted than for continuous cutting under the same cutting conditions. Generally, the

PAGE 32

32 instrument such as thermocouples, was not easy to install. The use of a noncontact measurement technique such as infrared thermal imagi ng was the best option. This technique was only capable of temperature measurements that might be considered at best averages rather than the true temperatures, and therefore tended to be dominated by chip images. Chow and Wright (1988) developed an onli ne temperature estimation technique in a turning operation. They used a standard thermocouple which was located at the bottom of the tool insert. The test cuts involved dry machin ing performed on plain steel tube (AISI 1020) with coated and uncoated cont rolled contact tool inserts. An increase in the tool wear resulted in an increase in the cutting temperature. The temperature increases were caused primarily by tool wear, which could be used to TCM during metal cutting. Shaw (1988) indicated that the mean tool face temperature was important role relative to the life of a cutting tool. Shaw proposed an approximate solution based on the principle of moving heat source and evaluated the solution. For online TCM by measuring temperature, remote thermocouple sensing appear s to be the only worthy way to measure the workpiecetool temperature since a direct measurement of the tool tip or rake face temperature distribution cannot be obtained. The temperature measurement of cutting edge is exceptionally difficult due to lack of direct access to the cutting zone. Boothroyd (1975) proposed the use of thermocouple techniques in the workpiecetool interface, and through such technique, the generated EMF at the junction is considered to be a measure of the mean temperat ure in that region. Temperature distribution is not taken by using this. Most currently available remote thermocouple sensor instruments can only allow either the

PAGE 33

33 cutting tool or workpiece interface or some other remote area temperature to be measured, and not the tool tip temperature. 2.1. 5 Vibration S ignatures ( A cceleration S ignals) Vibrations are produced by cyclic variations in the dynamic components of the cutting forces (Dilma, 2000) Usually, these vibration motions start as a small chatter respons ible for the serrations on the finished surface and chip thickness irregularities. Mechanical vibrations generally result from periodic wave motions. The nature of the vibration signal arising from the metal cutting process is such that it incorporates fac ets of free, forced, periodic and random types of vibration. Dimla (1998) presents a detailed investigation of progressive tool wear results obtained during a metal turning operation. The investigation showed that the amount of wear. I ts form was dependent more on the main cutting conditions. ElWardany et al. (1996) presented a study on monitoring tool wear and failure in drilling using vibration signature analysis techniques. They developed the features in both time and frequency domains. These vibration signature features were sensitive to drill wear and breakage but insensitive to cutting conditions, and sensor location. In the time domain, a monitoring feature was found to be effective for detection of drill breakage. In the frequency domain, a cepstr um ratio from the spectra of the vibrations monitored in both directions was also found effective for detection of breakage. By combin ing both techniques, it was possible to devise an effective drill monitoring system. Yao et al. (1991) investigated detection and estimation of groove wear at the minor cutting edge of the tool by using vibration signatures. During turning experiments, a

PAGE 34

34 miniature 3D accelerometer was used to measure the multivariate vibration signal produced by the turning process. A mult ivariate time series analysis was performed on the recorded vibration signals. The analysis showed that t he thrust cutting vibration was sensitive to the length of groove wear with two peaks: one at a very low frequency 200 Hz and the other at a high frequency 10 kHz. Rotberg et al. (1987) presented a tool wear monitoring method based on vibration in face milling. They focused on the milling tool entry and exit conditions. Face milling experiments were conducted and the ensuing flank and crater wear measured. The results indicated that the vibration signal was a suitable indicator of tool wear as it demonstrated considerable change during tool life. Rotberg et al. (1989) focused on tool wear monitoring usi ng vibration as the principal signal. The analysis was performed using two basic models characterized by low frequency and high frequency signal features. Experiments were conducted to validate these models utilizing measured acceleration signal features. The analysis showed the features from the vibration signals could be used in developing a TCMS as it correlated well with tool wear. Jiang et al. (1987) investigated the frequency composition of the signal and the changes of vibration pattern during tool wear process es and discovered a microbreakage stage. Then a frequency bandenergy method in which the problem in frequency domain was changed into that in time domain was proposed. Experimental investigations provided sufficient evidence that the vibrat ion signals were sensitive to tool wear states. The inter relationship between vibration signals and the cutting forces determines the dynamic nature of the cutting process, making the employment of these process

PAGE 35

35 parameters attractive in the development of TCMSs (Dimla 1998) The dynamic behavior on the other hand embodies vibration and certain aspects of the dynamic cutting force. In the real industry, the combination of cutting forces and vibration signals would be necessary. 2.1. 6 Miscellaneous Sensors and M ethods O ther various sensors are used for tool condition monitoring system. The optical methods, stress/strain measurement, methods based on measuring the workpiece dimension, surface finish quality measurement, and ultr asonic methods have been used for TCMS in metal cutting process es In this subchapter, a review for TCMS using other sensor s will be provided. 2.1. 6 .1 Optical methods Optical methods for tool condition monitoring system are suitable to evaluate tool wear in a laboratory as a direct measurement system. Most of the image skills are limited to 2D. However, Karthik et al. (1997) developed a 3D measurement system using a pair of stereo imaging cameras Shiraishi (1988) cited many studies have been made in the laboratory but commercially av ailable TCMSs based on optic were not available yet. For commercializing of an optic based TCMS, reliability in the actual environment will be required. C urrently, industrial application seems at an early stage because of high complexity and costs and low reliability and flexibility. However, vision based tool condition monitoring system will be one of major TCMS according to digital optic development. Wong et al. (1997) presented an optical method using the scatter pattern of reflected laser light for t he monitoring of tool condition in the roughing to near finishing

PAGE 36

36 range. The i mages were captured with a digital camera and the recorded images were processed and characterized using the mean and standard deviation of the scatter pattern and the intensity of the optical parameters, and their distribution correlated to the surface roughness. The scatter pattern was formed by a low power laser beam that was reflected from the surface of the workpiece. The deduced surface roughness in turning operation was then related to the ensuing state of the tool wear. Although to determine tool wear by observing machined surface roughness was very difficult, correlation between tool wear and intensity of the scattered light pattern was good. Recently, Atli et al. (2006) proposed a computer visionbased approach tool condition monitoring in drilling operation. They used a highspeed CCD camera for capture the images and a Canny edge detector was employed to extract tool features form the acquired images. Detection for the condition of all tests was performed well by using the proposed method. 2.1. 6 .2 Stress/strain measurements Noori Khajavi and Komanduri (1995) used f our sensors ( thrust, torque, and strains in two orthogonal directions of the machine table) for the study of the correlation of process parameters to drill wear. The measured strain signals were analyzed in both time and frequency domains. Meaningful correlation of the drill wear was obtained in the frequency domain. The area under the x axis PSD for the strai n sensor was found to correlate well to drill wear. Zhou et al. (1995) proposed a monitor ing system using the stresses acting in a cutting edge during a machining process. This provided a more reliable means to monitor and to predict tool spontaneous fai lure. A TCMS was based on a VME

PAGE 37

37 computer system and real time kernel was designed and implemented with a high speed data acquisition subsystem and a graphic presentation subsystem. They reported that prediction of spontaneous failures were possible by moni toring the risk factor defined as a ratio of the instantaneous stresses. Lee et al. (1994) presented a threedimensional loading and measured the loading on the rake face during cutting. From a finite element analysis (FEA), the principal and von Mises s tresses in the tool were examined to ascertain the points of highest stress. The principal and von Mises stresses could be used to predict the mode and location of tool failure. 2.1. 6 .3 Workpiece dimension ElGomayel and Bregger (1986) developed a sensing device to measure tool wear indirectly by monitoring the change of the workpiece diameter during turning operations. The workpiece diameter change was measured by electromagnetic sensors. Two electromagnetic sensors on opposite sides of the workpiece gav e a voltage output directly related to the gap between the sensor and the workpiece. The experimental data obtained for flank and nose wear were in agreement with the results obtained from the conventional method of measuring the wear by a toolmaker's micr oscope. 2.1. 6 4 Magnetism Jetley and Gollajesse (1994) proposed the magnetization of tool inserts and then, monitoring the magnetic field flux reduction as TCMS. Magnetized drills were used and implemented in order to validate their methodology. It was possible to accurately predict the end of tool life or fracture online by observing the magnetic flux, and they

PAGE 38

38 indicated that the system was cost effective with potential for implementation in most metal cutting environments. 2.1. 6 5 Ultrasonic methods Abu Zahra and Nayfeh (1997) developed a robust method for online gradual wear monitoring using normalized ultrasonic signals for tool wear monitoring for turning operations. A consistent calibration mark, cut in the lower corner of the tool nose, was us ed to generate a calibration echo. Experiments under various cutting conditions showed that the gradual wear measurements can be made tool independent by normalizing the measurements with the calibration mark. 2.1. 7 Sensor fusion synergy of signal integrat ion The single sensed signal can change with the cutting conditions such as machining parameters tool wear, etc. Therefore, more than one signal can be used in a complementary way to provide a more robust prediction of one or more machining features. The success of sensor fusion depends on which type of signals is suitable for a given machining outcome, which features are extracted, and in which way they must be complemented (Liang, 2004) In an early attempt, Lezansky and Rafalowicz ( 1993) developed the method using multi sensors. They used normal and tangential forces, vibration, acoustic emission, and both diameter and out of roundness to characterize the state of a grinding process. The methodology used can be seen as a multisensory rather than a sensor fusion, because the signals were used at the same time but were not complementary to each other.

PAGE 39

39 Bahr et al. ( 1997) developed a unique multisensory tool monitoring system using machine vision and vibration sensors for turning operations. They used a vibr ation signal as an online monitor to predict tool wear and detect breakage, and they also used machine vision between cuts to quantify the worn tool. In this way, a direct and an indirect technique were complemented and a more accurate tool monitoring system was developed because the machine vision can detect false signals from the vibration sensor. Azouzi and Guillot (1997) examined the feasibility for an intelligent sensor fusion technique. They presented an exhaustive analysis to determine the most sens itive process parameters (feed, depth of cut, cutting velocity) and signals (AE, forces, vibration) to predict the surface roughness and the final diameter error in machining. Based on experimental data and statistical tools, the feed, the depth of cut, and the radial and feed force components were selected as inputs to a neural network to predict the mentioned machining outcomes. Surface finish was assessed with an error varying from 2 to 25% under different process conditions, while errors ranging between 2 and 20 m were observed for the prediction of dimensional deviations. Etekin et al. ( 2003) studied the signals of force, AE, and spindle quill vibration to predict dimensional accuracy (bore size tolerance) and surface roughness in a CNC milling operati on. They used three different material types (6061T6 aluminum, 7075T6 aluminum, and ANSI 4140 steel). The RMS of the axial force component and the DC component of the AE signal were the candidate signals to be integrated and predict the quality character istics of the machined parts using three different workpiece materials.

PAGE 40

40 Recently, Cho et al. (2009) designed the effective multi sensor based TCM. The experimental was performed with 4340 steel and multilayer coated and multi flute carbide end mill cutter. They used the force, vibration, acoustic emission, and spindle power sensor for the time and frequency domain data. The experimental results showed that the design of TCM based on the feature level fusion can significantly improve the accuracy of the tool condition classification.

PAGE 41

41 Table 21 Previous research of tool condition monitoring system Main Sensor for TCMS Ref. Summary Electric motor current/power measurement (Bhattacharyya, 2008) A method for continuous on line estimation of tool wear using spindle motor current and voltage measurement was proposed. (Lee, 2003) Tool condition monitoring system with wavelet transform was proposed. (Zhang, 1994) A Hall effect sensor to measure the current supplied to the spindle motor drive of a vertical NC miller together with the cutting forces was used. (Constantinides, 1987) Spindle motor power was obtained and the PSD, the moving average, running mean and the accumulative sum power was used. (Rangwala, 1987 and 1990) The electric current in prefere nce to the cutting forces. Cutting forces (dynamic and static) (Dimla, 1999 Cutting force measurements fused with vibration signatures. and 2000) (Bayramoglu, 1998) A systematic investigation on the use of cutting force ratios in TCM for turnin g operations was presented. (Lister, 1993) The power spectrum of dynamic cutting forces was analyzed. (Ravindra, 1993) A mathematical model for tool wear estimation was developed. (Lee, 1992) Personal computer based fast Fourier transform software to track the dynamic cutting force signal was developed. (Marques, 1991) The relationship between wear of sintered high speed steel cutting tools and the associated cutting forces was investigated. (Kim, 1991) The modeling of dynamic cutting forces and experimental test data collected and compared with the theoretical model predictions was performed. (Oraby, 1991) A model for tool wear analysis in a turning operation by force characteristics within the different phases of tool wear was developed. (Yao, 1992) The measurement of major and minor flank, crater, and nose wear based on the analysis of dynamic cutting forces was performed. (Shi, 1990) The feed and cutting force components to flank wear length were correlated. (Dan, 1990) An inter rela tionship between the tangential, feed and normal components of the dynamic cut ting forces has been established. Acoustic emission (AE) (Chen, 2007) A technique based on AE signal wavelet analysis for tool condition monitoring was proposed. (No vak,1992) AE and cutting forces were fused. (Jemielniak, 1998) Statistical signal processing algorithm was used. (Kakade, 1994) AE parameters (ring down count, rise time, event duration, frequency and event rate) were selected.

PAGE 42

42 Table 21 Continued M ain Sensor for TCMS Ref. Summary Acoustic emission (AE) (Zheng, 1992) An intrinsic method for AE sensing based on an optic fibre sensor was presented. (Knig, 1992) Cutting test to detect fracture and/or monitor the condition of small drills using AE fe atures was performed. (Blum, 1990) Experimental test cuts to determine amongst other things, the influence of flank wear on the generation of AE signals was performed. (Moriwaki, 1990) A method based on AE measurement and analysis for coated tool life estimation was proposed. (Roget,1988) Machining tests from which the sensed AE signals from the cutting operation was carried out. (Tansel, 1991) Substantially little AE was thought to be generated compared to a larger AE accompanying tool breakage and fracture. (Lee, 1989) The tool tip/cutting edges temperature (Boothroyd, 1975) The thermocouple techniques in the workpiece tool interface and through such technique were proposed. (Sarwar, 1996) A thermal imaging system for metal cutting a pplications was developed. (Lin, 1995) Inverse approach for real time tool/workpiece interface temperature was devised. (Radulescu, 1994) An analytical tool temperature fields prediction model for use during continuous and interrupted cutting was desi gned and tested. (Raman, 1992) A mathematical model for cutting tool temperature measurement based on the remote thermocouple sensing (RTS) principle was proposed and developed. (Stephenson, 1990) Studies on tool temperature effects on interrupted meta l cutting were performed and theoretical and experimental results are reported. (Chow, 1988) An on line method for tool chip interface temperature measure ment in a turning process using a standard thermocouple inserted at the bottom of the tool insert w as devised. (Shaw, 1988) An approximate solution based on the principle of moving heat source was proposed and evaluated. Vibration signatures (acceleration signals) (Dimla, 1998) A detailed investigation of progressive tool wear results obtained during a metal turning operation was presented. (El Wardany, 1996) The use of vibration signature characteristics in on line drill wear monitoring and breakage was investigated. (Yao, 1991) Detection and estimation of groove wear at the minor cutting edge of the tool by monitoring vibration signatures are investigated. (Dan, 1990) A discrete modeling method called data dependent system to correlate vibration signals to cutting tool wear was employed. (Rotberg, 1987) The emphasis was on the milling tool entry and exit conditions. (Rotberg, 1989) Tool wear monitoring using vibration as the principal signal was focused.

PAGE 43

43 Table 21 Continued Main Sensor for TCMS Ref. Summary Vibration signatures (acceleration signals) (Jiang, 1987) A purposely buil t test rig was able to investigate the effects of vibration in the cutting and feed directions on the cutting tool. Optical methods (Atli, 2006) A computer vision based approach tool condition monitoring in drilling operation was proposed. (K arthik, 1997) a 3D measurement system using a pair of stereo imaging was developed. (Kurada, 1997) A review of the basic principle, instrumentation and various image processing schemes involved in the development of a vision based TCMS was presented. ( Oguamanam,1994) Implementation of optical methods was carried out. (Du, 1993) (Cuppini,1986) (Wong, 1997) A vision based TCMS using laser scatter pattern of reflected laser ray in the roughing to near finishing range was devised. Stress/stra in measurements (Noori Khajavi, 1995) Strain sensors of the correlation of process parameters to drill wear was used. (Zhou, 1995) Monitor the stresses acting in a cutting edge during a machining process in order to predict tool spontaneous failure was p roposed. (Lee, 1994) FEA and detailed stress analysis of the cutting edges were combined. Workpiece dimension (El Gomayel, 1986) A method for tool wear monitoring based on measure ments of workpiece deviation was proposed. Magnetism (Jetl ey, 1994) The magnetization of tool inserts and then, monitoring the magnetic field flux reduction as the tool wore was proposed. Ultrasonic methods (Abu Zahra, 1997) A normalized ultrasonic signal based method for in process tool wear monitoring f or turning operations was developed. Sensor fusion (Lezanski, 1993) Normal and tangential forces, vibration, acoustic emission, and both diameter and out of roundness to characterize the state of a grinding process were used. (Bahr, 1997) Vibrat ion signal as an on line technique to predict tool wear and detect breakage, and machine vision between cuts to quantify the worn tool were used. (Lou, 1996) That force is easily affected by the cutting conditions and acoustic emission is interfered by t he environmental noise was claimed. (Azouzi, 1997) AE, forces, vibration was used.

PAGE 44

44 Table 21 Continued Main Sensor for TCMS Ref. Summary Sensor fusion (Etekin, 2003) The signals of force, AE, and spindle quill vibration to predict dimensional accuracy and surface roughness in a CNC milling operation was used. (Cho, 2009) The effective multisensor based TCM was designed.

PAGE 45

45 Figure 21 Force amplitude vs time (fixed cutting conditions).

PAGE 46

46 CHAPTER 3 FORCE, POWER, FLANK WEAR, AND OPTIMIZED CUTTING CONDITION In this chapter, the cutting force mechanism of a turning operation, the relation between force and power, the tool wear rate and tool life, the flank wear progress, the optimal cutting conditions ( Usui, 1984 ; Dimla 2000; Tlusty, 2000; Matsumura, 2008) and the proposed real time flank wear model will be introduced. The following material in this chapter is extracted from the literature. Large portions are directly taken from Tlusty (2000), Dimla(2000), Usui(1984 ), and Matsumura (2008). 3.1 The M echanistic Cutting Force for a Turning O peration The cutting force acting on the tool is generated by actively engaging part of the cutting edge, which is drawn in Fig. 31. It includes the main cutting edge, the nose radius, and a small part of the secondary cutting edge. The direction of the force depends on the ratio of the components of the edge, and on the size of the radius with respect to feed rf The force F can be split into components: feed force fF which determines the direct load on the feed drive, the radial component rF which is decisive for the deflections affecting the accuracy of the machined surface, and the tangential force tF which has the direction of the cutting speed v and determines the cutting power tPFv which is tangential to the cut surface. The components tF and rF may be combined to give the force F which is normal to the cutting edge. The tangential cutting force is considered to be approximately directly proportional to the chip area A: tsssrFKAKbhKaf (3 1)

PAGE 47

47 where, as shown i n Fig. 32, b is chip width, h is chip thickness, a is depth of cut, rf is feed per revolution, and the subscript of the constant sK means specific, that is, per unit chip area. Conven tionally the dimensions to use are () tFN, () bmm () hmm 2() Amm and 2(/)sKNmm 3.2 Conversion between Cutting Force and Power for a Turning O peration Generally, the equation between power and torque is PT (3 2) where T The above equation can be rewritten as tsrPFrKafr (3 3) where tF is tangential cutting force and r is workpiece radius. Therefore, the equation between cutting force and power can be obtained. tP F r (3 4) If power can be measured, cutting force also can be estimated without a force sensor if r and are known. 3.3 W ear Rate and Tool Life Tool wear processes generally occur in combinations of the predominant wear modes, depending upon the cutting conditions, workpiece and tool materials, and the tool insert geometry. For a given cutting tool and workpiece material com bination, the tool wear form may depend on the cutting conditions, principally cutting speed v and the undeformed chip thickness t and a combination of the aforementioned wear mechanisms. Ranges of cutting speed where each type of wear is predominant can be

PAGE 48

48 identi fi ed by considering the product of these values as vt which is directly proportional to the cutting speed ( Shaw, 1984) Sometimes, the tool life can be consider ably reduced if the area of cut, the area swept by the cutting tool, is significantl y increased (i.e., mainly by increasing the depth of cut). At low cutting speeds, the tool wears predominantly by a roundingoff of the cutting point and subsequently looses sharpness. As the cutting speed increases, the wear land pattern changes to accomm odate the ensuing change with extremely high values leading to plastic fl ow at the tool point. Craters, on the other hand, depend largely on the cutting temperature, rather than on the cutting speed. The various forms of wear land pattern and prevailing cutting speed are shown in Fig. 33 for a turning operation. The more predominantly occurring forms of cutting tool wear often identified as the principal types of tool wear in metal turning using singlepoint tools are nose, flank notch and crater wear. Fi gure 34 shows how these wear features can be measured in a turning process through implementation of appropriate Inter national Standards Organi zation (ISO) criteria. Nose wear or edge rounding occurs predominantly through the abrasion wear mechanism on the cutting tools major edges resulting in an increase in effective negative rake angle. Nose wear can be dependent entirely on the implemented cutting conditions with tool sharpness lost through plastic or elastic deformation. At high cutting speeds, the edge deforms plastically and may result in the loss of the entire nose, as depicted in Figs 33(a) and 34(b). Edge chipping and cracking occurs during periodic breaks of the built up edge in interrupted cuts with a brittle tool and thermal fatigue. Catas trophic failure may also occur if the nose is considerably worn or as a result of the

PAGE 49

49 utilization of inappropriate machining conditions and br ittle tools such as ceramics ( Schey, 1987) Crater wear results from a combination of high cutting temperatures and high shear stresses creating a crater on the rake face some distance away from the tool edges, quantized by depth and cross sectional area (Fig. 33(c)). Crater wear also arises due to a combination of wear mechanisms: adhesion, abrasion, diffusion or t hermal softening, and plastic deformation. Severe depths of crater may trigger catastrophic collapse of the cutting point (see Fig. 33(d)). Flank wear arises due to both adhesive and abrasive wear mechanisms from the intense rubbing action of the two surf aces in contact, i.e., the clearance face of the cutting tool and the newly formed surface of the workpiece. Its rate of increase at the beginning of the tool life is rapid, settling down to a steady state, then accelerating rapidly again at the end of the tools life. Flank wear leads to a deterioration of surface quality, increased contact area and consequently to increased heat generation (Figs 33(b) and 34(c)). A wear notch can form at the depthof cut line as the tool rubs against the shoulder of the workpiece (Fig. 34(b) and (c)). A wear notch can lead to abrasion by the surface layers, and is accelerated by oxidation or chemical reactions, possibly leading to total tool failure. Flank wear typically increases with the time of cutting, as shown in Fig. 35. At the beginning, Phase I, there is initially a faster increase that is followed by a steady increase in proportion to cutting time, Phase II. When the wear depth reaches a certain size, it will accelerate and may lead to a sudden failure of the edge, Phase III. As an

PAGE 50

50 approximation to actual wear, expressed by the dashed line in the figure, wear can be assumed to be proportional to time according to: t r FWWw (3 11) where t FWW rw / is the rate of wear. A certain value of wear may be chosen as the permissible limit limFWW The time at which this limit is reached is called the tool life T The wear rate depends, for a given work/tool material combination, on cutting speed v and chip thickness h Within a practical range of v and h this relationship is usually expressed in algebraic form: q p r wh v C r (3 12) The exponent p on v is rather high, between 2 and 6, indicating the significant influence of v which is due to its effect on the temperatures of the tool. The exponent q on h is usually between 1.5 and 3 and is due to the influence of h on the load in the tool. Combining (311) and (312), t h v C FWWq p r (3 13) And for T t T h v C FWWq p rlim (3 14) or combining limFWW and rC *C T h vq p (3 15) Equation (315) is the tool life equation in a form very similar to that chosen by F. W. Taylor in 1905, except that his formulation involved a relationship between v and T

PAGE 51

51 only. Considering that in the turning operation, chip thickness is related, for a given side cutting edge angle to feed per revolution rf cosrf h (3 16) Equation (317) can be written C T f vq r p (3 17) where 2 *cos /C C Equation (317) is the usual form of the tool life equation expressin g the relationship between cutting speed v feed rf and tool life T and it is established considering a particular value of the angle However, Eq. (3 15) is more fundamental. It is common to obtain the parameters in Eq. (317) experimentally and express them graphically on loglog plot a relationship between two of the three variables involved. In this dissertation, the flank wear model which will be introduced in chapter 3. 6 will be used because the flank wear model in chapter 3.6 is more suitable for real time flank wear update. Equation 313 can only estimate the flank wear for one workpiece material. If the workpiece material is changed during the process, calculating the flank wear using Eq. 313 is impossible because time ( t ) depends on same workpiece material. 3.4 Cutting Tool Wear P rediction Since the tool life equation ( nVTC where T is the tool life, V is the cutting speed and n and C are constants) was discovered by F. W. Taylor, numerous research papers have been published throughout the world for better understanding of cutting tool wear and tool life characteristics. It should be noted that a goal of metal

PAGE 52

52 cutting research is to establish the theories or analytical m ethods which enable the cutting tool wear and other necessary parameters such as chip formation, cutting force, cutting temperature and surface finish to be predicted quantitatively without any cutting experiment. In order to predict cutting tool life anal ytically, it is necessary first of all to identify a simple wear characteristic equation for practical use which governs well the actual wear progress under practical cutting conditions. A function of wear rate, normal stress, and temperature, which includes only two material constants, is proposed as the characteristic equation for carbide tools and its appropriative is demonstrated experimentally. Considering a threedimensional turning operation, there is a need to predict analytically the chip formation and the three components of the cutting force for a given shape of tool edge and given cutting conditions, in order to calculate the normal stress and the temperature on the wear faces of a cutting tool. For this purpose, an energy method which utilizes orthogonal cutting data from machining is introduced and its outline is briefly explained. On the basis of predicted results from the energy method, it is possible to determine the distributions of stresses on the wear faces by introducing some appropriate assumptions. Since the frictional stress on the wear faces and the shear stress on the shear plane of cutting, which are predicted as above, determine the heat sources in the cutting zone, it is possible to calculate the temperature distribution on the wear faces through numerical analysis of the differential equation of heat transfer. The wear characteristic equation mentioned above is then utilized to predict analytically the wear progress (crater wear on the rake face and flank wear on the clearance face of the cutting tool) and the tool life by applying the normal stress and the temperature predicted already for the given cutting conditions.

PAGE 53

53 3.5 Derivation of the Wear Characteristic E quation Wear due to adhesion and abrasion appears to play the major role in the continuous dry cutting of steels with carbide tools without a built up edge. It is considered that the adhesion type of wear mechanism would be rate determining, while the abrasion due to hard particles in the matrix of steel such as carbide, sil ica ( 2SiO ) and corundum ( 23 AlO) may be complementary because temperatures and normal stress on the tool face are extremely high ( Shaw, 1954) and mutual diffusion of constituents between the steel and the tool is wellknown to take place ( Opitz, 1967) in the practical range of cutting conditions for carbide tools. Shaw and Dirke (1956) presented an Archard type of equation for adhesive wear: 'xc dWAZdL b (3 18) where dW is the wear volume for sliding distance dL xA the real area of contact, c the height of the postulated platelike wear particle, b the mean spacing of the asperities and Z the probability of producing a wear particle per asperity encounter (Holms probability). Regarding xA in Eq. (3 18) as the area for uni t apparent area of contact, Eq. (3 18) may be written as t c dWZdL Hb (3 19) where H is the asperity hardness and t is the normal stress on the contact surface. According to Shaw and Dirke (1956) / cb in the above equation may be regarded as being approximately constant owing to the existence of a size effect. Since the asperity hardness H depends more strongly on the bulk properties of the softer of the pair of

PAGE 54

54 mating surfaces than on those of the asperity itself, it may depend on the diffused layer, temperature, strain and strain rate on the chip surface in contact. Neglecting variation in the strain and strain rate in the practical range of cutting conditions for carbide tools, the following equation by analogy with the rate process may be assumed, since material strength and diffusion are similarly affected by temperature: 2 1exp A HA (3 20) where 1A and 2A are constants and is the temperature of the chip surface. The probability Z may be considered as that for yielding a weld which is strong enough to produce a wear particle when an asperity encounter takes place. Since such a weld formation is a kind of thermally activated rate process, the probability Z will be expressed by the following equation, if the interlocking time and the flash temperature rise during the encounter are regarded as being almost constant within a given range of cutting conditions: 1exp E ZB (3 21) where 1B is a constant, is Boltzmanns constant, E is the activation energy and is the temperature of the chip surface. Substituting Eqs. (320) and (321) in Eq. (319) and regarding / cb as constant, the following equation can be obtained. 2 1exptEA dW C dL (3 22) where 1 C is a constant. Although 2 EA in the above equation depends on the structure and the element concentration of the diffused layer on the contact surface, it

PAGE 55

55 may be regarded as being approximately constant if the variety of c utting conditions is limited. Then Eq. (323) can be obtained. 2 1exptC dW C dL (3 23) which is the same as the equation pr esented by Trigger and Chao (1956) If hard abrasive particles are involved at the asperities at various intervals, the average value of Z and H in Eq. (3 19) will be altered and hence the effect of abrasion is expected to appear in 1C and 2C of Eq. (3 23). This will be shown later experimentally. Equation (323) has been derived by introducing many crude approximations; however, it is simple enough and contains only two characteristic constants to be determined experimentally that it is convenient for practical use. The equation contains not only Boltzmanns canonical distribution as used in the analysis of the diffusion coefficient but also includes some mechanical effects (post diffusion process) represented by t and dL Although the real situation on the wear surface has not been clarified yet, it is still possible to depict another model of tool wear in which the abrasive action of the hard particles in the cut steel is taken as governing the process. Taking the crater wear in the rake face of a carbide tool as an example, it has been believed that the real area of tool chip contact is close to the apparent area ( Finnie, 1956) and viscous flow like deformation due to thermal soft ening takes place along the surface layer of the chip in contact with the rake face. In this situation, wear appears to be produced only as a result of plowing of the weak diffused layer on the rake face by the hard par ticles on the chip surface. Du et al. (1993) feel, however, that the observations of wear particles

PAGE 56

56 ( Uehara, 1976) appear to exclude the possibility of this wear model, although it may be worth noting that expression identical to Eq. (323) can be obtained if taking the Rabinowicz equation of abrasive wear ( Rabinowicz, 1961) and introducing thermal effects similar to those in Eq. (3 20). 3.6 Flank Wear P rogress The cutting force is predicted in the model based on the minimum cutting energy. Figure 36 shows the force model in the threedimensional cutting process. The process is interpreted as a piling up of the orthogonal cuts in the plane containing the cutting velocity and the chip flow velocity. The shear angle the friction angle the shear stress on the shear planes, and the tool chip contact length cl in the orthogonal cutting can be associated with the cutting velocity V the uncut chip thickness 1t, and the rake angle in the following equation: 1 1 1 1,, ,, ,, ,,cfVt gVt hVt lJVt (3 24) The orthogonal cutting data are prepared as a combination of the workpiece material and the cutting tool. If the chip flow direction is assumed, the cutting energy, which is the sum of the shear energy on the shear plane and the friction energy on the rake face, can be calculated wit h the orthogonal cutting model. The chip flow direction, which is specified by the angle c is determined such that the cutting energy is minimized. Then, the principal, feed, and radial components of the cutting force can be predicted in the chip flow model determined. The shear stress s on the shear plane

PAGE 57

57 and the friction stress t on the rake face can be distributed as shown in Fig. 37 ( Usui, 1978) The distributions of the normal stress f and the friction stress f are assumed on the flank wear land, where the friction stress is equal to the normal one ( Usui, 1984) The cutting temperature can be calculated via the finite element method ( Patankar, 1980) All mechanical work, which includes the plastic work on the shear plane and the friction work on the r ake face and the flank wear land, is converted into heat generation based on the stress distributions. The wear rate can be expressed as Eq. (3 23). 1C and 2C are the wear characteristic constants given in the combination of the workpiece and the tool. Then the flank wear rate ( BdV dt ) can be given by the following equation ( Rabinowicz, 1961) : 2 11 exp tan tanB f fdV C CV dt (3 25) where f and f are the normal stress and the temperature on the flank wear land, respectively, and V are the cutting speed, the rake angle, and the relief angle, respectively. The temperature distribution can be analyzed by assuming the friction stress f on the flank wear land. The wear rate is then calculat ed in Eq. (325). Because the wear rate is generally constant over the flank wear land, f and f are modified so that the wear rate is the same over the flank wear land. Finally, the stress distribution, the wear rate, and the temperature distribution can be determined. Then the flank wear BV at the cutting time T can be predicted in the following equation ( Matsumura, 1993) :

PAGE 58

58 0 0 t B BBdV VtV dt dt (3 26) where 0 BV is the initial wear offset, which is the width of fl ank wear land at the time T =0 in the calculation. 3.7 Real time Flank Wear M odel for a W orkpiece Composed of Various M aterials Equation (326) can be used in general turning operations only for workpieces with a single material. In other words, the flank wear model of Eq. (326) cannot reflect changes in the workpiece material. This is a kind of open loop estimation model and it is limited because all cutting conditions must be known. Therefore, if a workpiece has various materials, the flank wear model s hould be updated according to these changes. Therefore, this research proposes to extend the model to accumulate changes in material. It is proposed here that t he updated flank wear BV case of updated time updatedtt can be predicted in the following equation: 12 12 0 0 1 0 1 1...updated updated updated updated updated n updated updated tt BB B updatedB t material material nt B nt material n it B B it i materialdV dV VtntV dt dt dt dt dV dt dt dV V dt dt (3 27) At every update time, the flank wear rate /BdVdt should be changed according to changes of the workpiece properties. In the present research, workpiece material changes will be detected by a power sensor signal.

PAGE 59

59 Figure 31 The cutting force component in turning operation ( Dimla 2000) Figure 32 Basic turning operation

PAGE 60

60 Figure 33 Cutting tool wear forms in orthogonal metal cutting ( Dimla 2000)

PAGE 61

61 Figure 34 Conventional features of t urning tool wear measurements ( Dimla 2000) Figure 35 Development of flank wear width with time ( redrawn from Tlusty, 2000)

PAGE 62

62 Figure 36 Threedimensional cutting model in turning b, s, and c are back rake angle, side rake angle, and chip flow angle; and f and d are feed rate and depth of cut ( Matsumura, 2008) Figure 37 Stress distribution on tool face shear angle; t, friction stress on rake face; s, shear stress on shear plane; and f and f are friction stress and normal stress on flank wear land ( Matsumura, 2008)

PAGE 63

63 Figure 38 Effect of cutting speed ( v) and feed ( fr) on the cost per part of components due to machining time and due to tool cost ( Tlusty, 2000)

PAGE 64

64 CHAPT ER 4 SENSOR FEASIBILITY T EST FOR REAL TIME WO RKPIECE MATERIAL DET ECTION This chapter provides the description of feasibility test of sensors for real time workpiece detection. Three sensors were used for feasibility test. First, an infrared sensor was test ed by using heating plate. A dynamometer was tested on the Hass SL10. And a power meter was tested on Hass SL10 and Okuma LC40 CNC lathe s. Lastly, the mini CNC lathe (SyiL C6B) was assembled and the power sensor was tested with various workpieces. In thi s chapter, the setups, specifications, test conditions, etc. will be described and explained. The test result s will be shown in chapter 5. 4.1 Ultimate Thermometer (Infrared S ensor) In exploring the feasibility of a real time workpiece detection of t emperature for compensating tool wear, the ability to measure the temperature of small surface is required. The area for which infrared temperature sensors accurately detect temperature changes linearly with respect to the distance between the sensors lens and the object of interest. This dependency is characterized by the Distance to Spot diameter ratio or D/S ratio, acquired from the manufacturer. In this test, the Ultimate Thermometer 800043 D/S ratio of 7:1 is verified. Table 4 1 shows the specification of Ultimate thermometer. A schematic of the experimental setup c an be seen in Fig. 41. As show the infrared thermometer hangs with the sensor aimed at a black aluminum cylinder which is heated by a heating plate. Figure 42 is a picture of the actual setup. The aluminum cylinder is subjected by the heating plate to a heating and cooling process giving it a particular heating profile. A heating and cooling cycle is applied to the aluminum cylinder repeatedly as the

PAGE 65

65 thermometer is raised to varying heights. An embedded thermocouple records the temperature of the cylinder while another thermometer records the temperature of the hot plate. 4.2 Cutting Force Measurement on Hass SL10 Lathe The cutting force was measured while turning workpieces with step changes in the hardness on a Haas SL10 CNC lathe. Two workpieces were prepared, i.e., alternating sections of alloyed aluminum and low carbon steel and alternating sections of low carbon steel heat treated to different hardness levels (nominally 20 HRC and 30 HRC). These parts were test for turning operation under different depths of cut a and feed per revolution rf values and the three force components were measured using a cutting force dynamometer. Figure 43 shows a schematic of a turning operation and the resultant cutting force F components. These are the feed force fF and radial force rF within the plane of the insert cutting face (i.e., the rake face). These forces can be combined to give the normal force nF which acts perpendicular to the primary cutting edge. The tangential force, tF in the direction of the cutting speed can be combined with the normal force to give the resultant force. The setup for the heat treated steel workpiece is shown in Fig. 44. The cutting force dynamometer (Kistler 9257B) was mounted in the tool turret and the cutting tool holder and insert was clamped to the dynamometer. In thi s way, the cutting forces were directly recorded by the dynamometer The force components recorded were ()fFz ()tFx and ()rFy

PAGE 66

66 The insert was mounted on the bottom of the holder according to the photograph provided in Fig. 44. A photograph of the aluminum steel aluminum steel workpiece is provided in Fig. 4 5. The individual sections were approximately 20 mm long. 4.3 Spindle Power Measurements on Hass SL10 Lathe The tangential force component tF is directly proportional to spindle motor torque, and therefore power. From Eq. 41, T is torque, r is the workpiece radius, P is the only on the cutting force (which is a function of the workpiece hardness), but also on the instantaneous workpiece radius and selected spindle speed. (4 1) In these tests, a Fast Response PH 13 Power Cell from Load Controls, Inc. was used to monitor the (electrical) spindle power and compared the power levels for segmented workpieces as shown in Fig. 4 5 with cutting forces obtained using the dynamometer setup. The power sensor was wired directly into the Haas SL10 power cabinet as shown in Fig. 46. The addition of the sensor to the machine did not affect the cutting process or the machine functionality. As seen in Fig. 46, the threephase output from the spindle drive was fed through the three inductive sensors (three holes on the left side of the power cell). The same signals were wired to the top right of the unit to measure the reference voltage. Based on this data, an analog voltage is generated which is proportional to the load power consumption. This power signal was t tTFr PTFr

PAGE 67

67 then carried on a BNC cable to the data acquisition system (0 10 V corresponded to 015 Hp for this setup). 4.4 Power M easurement on Okuma LC 40 Lathe A Fast Response PH 13 Power Cell from Load Controls, Inc. was used to monitor the (electrical) machine input power on an Okuma LC 40 lathe. The power cell is designed to sense three phase power. It was used as a standalone transducer with an analog output (010 volts). With the installed current chip, this voltage range indicates to 0 to 107 hp. The power cell has three balanced Hall Effect devices, each with a flux concentrator. The Figure 47 show the sensor hooked up to the Okuma LC 40 lathe. A NI USB 6009 I/O board ( http://sine.ni.com/nips/cds/view/p/lang/en/nid/14605) in a laptop based data acquisition system and LabView software from National Instruments was used. Specifications of the NI USB 6009 I/O board are given in Table 4 2. Figure 48 shows a screen shot of the data measurement program written in LabView. The sampling resolution during gathering was 1 kS/sec. The spindle power measurement was performed on a Okuma LC 40. The lathe is machining specific parts and the end product shape is shown in Fig. 49. There are two facing operations, four roughing turning operations, and finishing facing and turning. For the first and second facing operations, the cutt ing conditions are 400 rpm spindle speed, 0.016 inch per revolution feed rate and 0.15 inch/pass depth of cut. The first roughing turning operation is simultaneous cut in the OD and ID with 350 rpm spindle speed and 0.02 inch per revolution feed rate. The second roughing turning operation is OD turning with 350 rpm spindle speed, 0.02 inch per revolution feed rate, and 0.2 inch depth of cut from 5.4750 inch diameter to 5.0750 inch diameter. It is followed by a third roughing

PAGE 68

68 turning operation that is OD tur ning with 350 rpm spindle speed, 0.02 inch per revolution feed rate, and 0.2 inch depth of cut from 5.0750 inch diameter to 4.6750 inch diameter. The last roughing OD turning has cutting conditions as 350 rpm spindle speed, 0.02 inch per revolution feed rate, and 0.1625 inch depth of cut from 4.6750 inch diameter to 4.3500 inch diameter. The finishing operation is the finishing facing and turning with 400 rpm spindle speed, 0.02 inch per revolution feed rate and 0.05 inch depth of cut. Table 4 3 shows the c utting conditions. Figure 410 shows a sample of data collection during machining a part with the operations labeled. The power measured from roughing facing operations is labeled 1 and 2. The data from the roughing OD and ID turning operation is 3 in Figu re 49. The data from roughing OD turning operations is 4, 5, and 6. Lastly, the data from finishing facing and OD ID turning operation is 7. 4.5 Power Measurement on SyiL CNC Lathe (C6B) Figure 411 shows a picture of the experimental setup, including the SyiL CNC lathe (C6B) and National Instruments USB DAQ board (USB 6009). The Load Control Inc.s power sensor (UPC) is installed in the input power line of the spindle motor and the data acquisition rate is set to 1 kHz in the LabView program. Three cyli ndrical workpieces of diameter of slightly less than 25mm were made for the cutting tests, shown in Fig. 412. All workpiece materials original diameter was 1 inch (25.4 mm) and was turned down for reducing the surface effects and exact same diameter. Table 4 4 shows the materials specifications. Workpiece A has four sections of 15mm length, each with a different material: aluminum alloy 6061, hot rolled 1035 carbon steel, P20 tool steel, and 1018 carbon steel. Workpiece B contains aluminum alloy 6061 and 1018 carbon steel. Workpiece C contains aluminum alloy 6061 and a

PAGE 69

69 stainless steel pin of 3.5 mm diameter to act as a hard spot. Before performing the cutting tests, the Rockwell A hardness values of all materials were measured by Wilson Rockwell hardness t ester model 3JR with three samples per section per workpiece. The average hardness values are shown in Figure 412. The cutting power should be sensitive to cutting conditions such as feedrate, depth of cut, hardness, and workpiece diameter. Spindle speed and feedrate were set to 1000 rpm and 0.1mm/rev, respectively. TCMT32.52 carbide inserts were used in this cutting test.

PAGE 70

70 Table 41 Ultimate Thermometer Specification Type Range Resolution Accuracy (meter only) Infrared -20 ~ 400C -4 ~ 752F 1 3% r dg or 3C or 5F (which ever is greater) Note: Accuracy is tested within 23 5C Data Output RS232 PC serial interface D/S Distance to Spot ratio is approximately 7:1. IR Wavelength Reg. 6 to 12 micrometer Table 42 Specifications of the NI USB 6 009 I/O board Feature USB 6009 AI Resolution 14 bits differential, 13 bits single-ended Maximum AI Sample Rate, Single Channel* 48 kS/s Maximum AI Sample Rate, Multiple Channels (Aggregate)* 42 kS/s DIO Configuration Open collector or active drive M ight be system dependent. Table 43 Cutting conditions for power measurement on Okuma LC 40 Operation Spindle speed (rpm) Depth of cut (inch) Feed rate (inch per revolution) Produced workpiece diameter (inch) Facing 400 0.016 Facing 400 0.15 0.016 OD ID turning 350 0.2 0.02 5.475 OD turning 350 0.2 0.02 5.075 OD turning 350 0.2 0.02 4.675 OD turning 350 0.1625 0.02 4.35 Facing and turning 400 0.05 0.02

PAGE 71

71 Table 44 Workpiece materials specification (from McMaster CARR) Material E asy to Machine Mold-Quality P20 Tool Steel Multipurpose Aluminum (Alloy 6061) General -Purpose Low -Carbon Steel General -Purpose Low -Carbon Steel Alloy/Low -Carbon Steel type P20 6 061 1018/1018 Carbon Steel 1035/Hot -Rolled 1035 Finish/Coating Ground Unpolished (Mill) Unpolished (Mill) Zinc -Galvanized Shape Rods and Discs Rods and Discs Rods and Discs Rods and Discs Diameter 1" 1" 1" 1" Diameter Tolerance +.045" .005" -.003" .009" Temper/Condition Hardened, Quenched, Tempered none none none Hardness Brinell 262-321 Brinell 95 Brinell 126-167 Rockwell B68 Yield Strength 110,000 PSI 35,000 psi 54,000 to 70,000 psi 30,000 to 44,000 psi

PAGE 72

72 Figure 41 Experimental infrared sensor setup schematic Figure 42 Actual setup for infrared sensor test

PAGE 73

73 Figure 43 Cutting force components measured by dynamometer during cutting tests Figure 44 Cutting force test setup on Hass SL10

PAGE 74

74 Figure 45 Aluminum steel aluminum steel workpiece for Hass SL10 power and force measurement Figure 46 Fast Response PH 1 3 Power Cell mounted in Haas SL10 power cabinet

PAGE 75

75 Figure 47 Fast Response PH 13 Power Cell mounted in Okuma LC 40 power cabinet Figure 48 Screen shot of data gathering program constructed by LabView for power measurement

PAGE 76

76 Figure 49 Finished pipe joint shape Figure 410 Power data while cutting an example part on Okuma LC 40

PAGE 77

77 Figure 411 SyiL CNC lathe (model C6B), National Instruments USB DAQ board (model USB 6009) and Load Control Inc. power sensor (model UPC)

PAGE 78

78 Figure 412 Three different types of workpiece for power measurement test on SyiL CNC lathe

PAGE 79

79 CHAPTER 5 RESULT OF FEASIBILIT Y TEST S FOR REAL TIME CUTTING POWER DETECTION 5.1 Ultimate Thermometer (Infrared S ensor) The infrared sensor test was repeated at distances of 50, 100, 150, 200, 250, 300, and 400mm between block and infrared sensor head. The heating plate was heated to 300C from room temperature, held at for 2 minutes then cooled to room temperature. The thermocouple and Ultimate thermometer readings were recorded. Figure 5 1 shows the result. Although the behavior of the infrared sensor closely follows that of the thermocouple, there are differences between the readings The infrared temperature appears to peak after thermocouple does. This delay occurs because the thermoc ouple is located lower on the aluminum cylinder. The thermocouple encounters the heat lump before the infrared sensor does. The infrareds highest readings are noticeably higher than the thermocouple readings at small distances. The spatial temperature distribution in the aluminum cylinder is such that the temperature at the center of the top surface reaches a higher peak than the upper sides of the cylinder. As the distance is increased, the infrared temperature readings tend to become less predictable. The large distance would lead to erroneous results; the intended reading would be affected by the surroundings in view of the infrared sensor. To better prove this inconsistency, a second test was conducted. The heating plate was held at room temperature while the aluminum cylinder was heated separately. The cooling profile from the thermocouple and the infrared sensor were recorded at only two distances, 30mm and 250mm. Figure 52 shows these results.

PAGE 80

80 The surrounding temperatures clearly affect the infr ared sensors readings at large distances. And the surface area is at least bigger than sensor head diameter. These results lead us to believe that the Ultimate Thermometer cannot provide good readings for its intended use. 5.2 Power Meter and Dynamometer T est on Hass SL 10 The mean cutting forces for the individual aluminum and steel sections for four different depths of cut/feed per revolution combinations are provided in Table 51. For each force component, the ratio is approximately 2:1 for steel (S) versus aluminum (A). The largest force, and highest signal to noise ratio, is seen for the x direction (tF). The same trend was observed for the steel steel steel (three different hardness steel) workpiece. However, the force differ ences were smaller due to the lower hardness difference between the individual sections. Based on these results, cutting force should provide a reasonable signal for real time workpiece detection. However, force measurement is not without its own difficult ies. Force measurements are, in general, actually based on displacement measurements. For example, strain gages are often used to estimate the load on structural members by providing a voltage that is proportional to displacement (within a Wheatstone bridge configuration). To obtain high signal to noise ratios, however, a local notch or other feature is typically added to increase the local displacement. In metal cutting, any reduction of the tool holder assembly stiffness leads to higher potential for chatter and reduced workpiece accuracy. Therefore, adding metrology (force measurement) to monitor workpiece hardness variations to the tool holder could lead to reduced performance overall. For example, the depth of cut may need to be decreased

PAGE 81

81 to obtain sat isfactory cutting conditions once the force sensor is added (i.e., the tool holder is modified). Measuring the spindle power provides an alternative to forcebased process interrogation. As a first step in the power sensor evaluation, the spindle power as a function of spindle speed (with no cutting load) was measured. Figure 54 shows the results of this testing. It is seen that the spindle power increases approximately linearly with spindle speed. Note that this power curve is associated with driving the spindle only and is not a function of the cutting process. To explore the combined effects of workpiece radius and cutting force on spindle power, multiple cutting tests were completed at three different depths of cut (0.5 mm, 1 mm, and 2 mm) and two feed rates (0.1 mm/rev and 0.2 mm/rev) over a range of workpiece radius values. The spindle speed was 500 rpm in all cases and two different workpieces (aluminum/steel and steel/steel) were used. Example results for the aluminum/steel workpiece are provided in Figs. 5 5 and 56. Figure 55 displays the variation in x, y, and z direction cutting forces, as well as the variation in spindle power with position of the cutting tool along the workpiec e. The depth of cut was 1 mm, the feed per revolution was 0.1 mm and the workpiece diameter was 67 mm for the external turning operation. The transition from aluminum to steel to aluminum and back to steel are seen as the step changes in force/power from low to high to low and back to high. The mean values are summariz ed in Table 52. In both the figure and table, it is seen that the power signal mimics the force results with reasonable signal to noise ratio (note that no filtering was applied to these signals, except an analog anti alias filter).

PAGE 82

82 Figure 56 shows the cutting forces and spindle power with a depth of cut of 1 mm and feed per revolution of 0.2 mm. The workpiece diameter was 66.5 mm (two workpieces were fabricated from the same bar stock). The same force/power transitions are again observed, but the level s are higher than the values seen in Fig. 55 due to the increase in feed with all other conditions remaining approximately constant. Again, the power signal successfully captures the hardness trend without influencing the cutting process. The influence of radius on power for the aluminum/steel workpiece is demonstrated in Fig. 57. For these tests, the depth of cut (0.5 mm) and feed per revolution (0.2 mm/rev) were maintained at constant values as the workpiece radius was reduced with each successive cut. The spindle speed was 500 rpm. The figure shows that the tangential (x direction) cutting force component remains essentially constant independent of radius (or diameter), while the power demonstrates an approximately linear increase with radius. A check of the Eq. (41) power expression can be performed using the Fig. 57 data. For a workpiece diameter of 70 mm (r = 35 mm) and aluminum Ft value of 150 N, the corresponding power should be 352 150500275 100060tPFr W This agrees reasonably well with the 294 W v alue shown in the figure (blue solid line). At a diameter of 60 mm (r = 30 mm), the power calculation is 302 150500236 100060tPFr W ; the experimental value was 217 W. The overall trend for both the aluminum and steel is a slightly higher slope than is predicted from Eq. (4 1). Additionally, the measured power values to be higher than the calculated values were expected due to the power

PAGE 83

83 consumed in rotating the spindle (Fig. 54). These issues could be explored further for the actual tool material lathe combinati on(s) used in practice. An example result for the steel/steel workpiece is shown in Fig. 58. The depth of cut was 1 mm and the feed per revolution was 0.1 mm; the spindle speed was 500 rpm. The change is cutting force (tangential or x direction) and power between the normalized sections (14.5 and 13.1 Rockwell C hardness values) and hardened section (28.9 Rockwell C) of 4130 steel is apparent. The local deviations in the force/power values in the hardened section is believed to be due to built up edge that occurred while cutting (no coolant was used in this study). 5.3 Signal Analysis of Okuma LC 40 Power D ata A power sensor was installed to measure machine input power in power cabinet of Okuma LC 40 lathe. Okuma LC 40 is horizontal machine type with an OS P controller. If cutting conditions such as a spindle speed, feed rate, workpiece material, and diameter are the same, a mean value of a power should be same. However, the power is expressed as tPFr the mean value cannot be used for facing operation because the radius is changed in the cutting progress. Therefore, a maximum value is used for data analysis for facing operation. A standard deviation was also calculated because tool wear and breakage causes more vibration. T o tal of 1000 samples were used to calculate standard deviation for each operation. Figure 59 and Table 54 show maximum values of facing operations, mean values of turning operations, and standard deviations of turning operations in one plot. The values for each cutting operation are Figure 510 for easy configuration.

PAGE 84

84 There should be a difference between normal and worn tools because tool wear or breakage causes large cutting forces. As shown in equation tsrPFrKafr where tF is tangential cutting force, r is workpiece radius, is spindle speed, sK is specific cutting force, a is depth of cut, and r f is feed rate, the power should be larger with a worn tool compared to a normal tool. As shown in Figure 511, after finishing part 5, roughing facing tool 1A was changed. As seen in figure 511, before changing the tools, maximum power values were higher than normal. Generally, tool wear and breakage causes more vibration in cutting. Therefore, the standard variation of power signal was calculated. However, no clear trend was found in this data. Perhaps no trend was found because the m achine operator changed tool before serious tool wear or breakage. If workpiece hardness is lower than normal, the cutting force should be smaller than normal. As shown Figure 512, the measured power of parts 6, 7 and 8 from the Aug. 18, 2008 data set (s ee appendix) is smaller than normal. If cutting conditions and cutting geometry are same and low hardness workpieces are machined, cutting force should be smaller than normal because ofsr PKafr Small cutting force should cause small spi ndle power because power is proportioned to cutting force. In this chapter, the lathe power was measured and analyzed for mean, maximum, standard deviation, and change due to radius change. The possibility to determine the force from the power was checked. If the spindle speed is fixed and the cutting geometry is known, feed rate can be changed to regulate the cutting force. To improve

PAGE 85

85 the productivity of the lathe operations, feed rate should be increased if hardness is low or DOC is smaller than for other parts. Based on the results from this test, the noninvasive, low cost power sensors offered by Load Controls, Inc., offer a feasible option for in process metrology of turning operations. Since the cutting power is caused by cutting force, the power under an air cutting was measured. This power isnt caused by the machining process. In this turning operation, noncutting power value was 0.7008 V (sensor value) When the estimated power was calculated, this value will be subtracted from measured power duri ng cutting. Figure 513 shows the measured power during air cutting. To estimate the power, the data from second, third, and fourth OD turning operations were used because all workpiece radiuses, depths of cut, and spindle speed are known in this period. The second and third OD have same feed rate and spindle speed. And if specific cutting force change due to hardness change is small, workpiece radius change will change power. Therefore, the power change relative to radius change was checked. Between sec ond and third OD turning operations, workpiece radius is decreased by about 7.6 percent. Estimated power of the third operation by the above equation should be 92.4 percent of second OD turning operation power. The measured power for the third OD turning operation is 90.5 percent of the measured for second OD turning operation. Another case is the second and forth OD turning operations. In this case, the radius is changed from 2.6375 to 2.2563 inch and depth of cut is changed from 0.2 to 0.1625 inch. Acc ording to above equations, the power estimates for the fourth OD

PAGE 86

86 turning should be 69.5 percent of the power for second OD turning operation. The measured power for the forth OD turning was 69.4 percent of the measured power for second OD turning operation. In turning operations, the power can be expressed using t noncuttingSr noncuttingPFrPKafrP (5 1) where tF is the tangential cutting force, is the spindle speed, r is the part radius, noncuttingP is the noncutting power, SK is the specific cutting force, rf is the feedrate, and a is the depth of cut. Therefore, power with 110% spindle speed override should be bigger than normal spindle speed (100%). Figure 514 show power difference of step 2 between normal (100%) and increased (110%) spindle speed. Non cutting power is around 0.68 in this machine. The experimental result is reasonable because cutting power ratio is 1.098. This value should be about 1.1. 5.4 Power Measurement Result on SyiL CNC Lathe (C6B) Cutting power data (sensor value in voltage) for all workpieces are provided in Figs 5 15~18. As expected, the average cutting power for harder materi als is higher than softer materials. In the case of workpiece B, which had a mixed aluminum and steel region, the cutting power is between that of the aluminum and steel sections separately (see Fig. 516). However, the region containing the relatively har der pin (stainless steel) cannot be distinguished by the power measurements in Fig. 517. This may be due to the small size of the pin, which only accounted for a maximum of approximately 5% of the workpiece perimeter. In Fig. 5 18, tool breakage occurred when the tool came in contact with the steel pin during a subsequent test. Close inspection reveals a peak corresponding to the instant when the tool was broken.

PAGE 87

87 These results show that the power sensor can detect changes in hardness that last for suffici ent duration. Accordingly, the sensor could be used as a sensor in an adaptive real time flank wear scheme. A simple method to detect variations in the hardness of a workpiece has been implemented in this research by measuring the power of a spindle motor for a turning operation. The power signal successfully captures the hardness trend without influencing the cutting process and requires no additional (external) hardness measurements. However, variations in hardness within a small section are difficult to detect, and may require increased sensitivity of this technique

PAGE 88

88 Table 51 Mean cutting force values for aluminum steel aluminum steel workpiece on Hass SL 10 Depth of cut(mm) Feed per revolution (mm/rev) Mean of F x (N) Mean of F y (N) Mean of F z (N) Al Steel Al Steel Al Steel Al Steel Al Steel Al Steel 1 0.1 160 318 167 341 123 241 109 258 79 166 76 177 2 0.1 191 376 148 426 133 220 74 217 97 190 78 256 1 0.2 236 537 258 576 152 329 148 330 120 240 105 275 2 0.2 443 1019 463 1079 1 66 341 128 339 278 591 283 623 Table 52 Results for a = 1 mm, fr = 0.1 mm/rev cutting tests using aluminum/steel workpiece at 500 rpm and 67 mm diameter on Hass SL10 Depth of cut (mm) Feed per revolution (mm/rev) Workpiece Diameter (mm) Mean value of force Power F x (N) F y (N) F z (N) Al (W) Steel( W) Al Steel Al Steel Al Steel 1 0.1 67 178 415 98 249 99 242 315 762 Table 53 Results for a = 1 mm, fr = 0.2 mm/rev cutting tests using aluminum/steel workpiece at 500 rpm and 66.5 mm diameter on Hass SL10 Depth of cut (mm) Feed per revolution (mm/rev) Workpiece Diameter (mm) Mean value of force Power F x (N) F y (N) F z (N) Al (W) Steel (W) Al Steel Al Steel Al Steel 1 0.2 66.5 257 567 138 337 104 246 452 1039

PAGE 89

89 Table 54 Maximum v alues of power during facing operations, mean values and standard deviations of power during turning operations on Okuma LC 40 Part # First Second First Second Third Fourth Finishing First Second Third Fourth Finishing facing facing OD ID OD OD OD OD ID OD ID OD turning OD OD OD ID Max. Max. turning turning turning turning turning turning SD turning turning turning mean mean mean mean mean SD SD SD SD 1 1.7140 4.2620 3.1697 2.7861 2.5750 2.1432 1.6784 0.0779 0.0273 0.0265 0.0233 0.0356 2 2.296 0 4.3670 3.0699 2.6882 2.4894 2.0666 1.6881 0.1476 0.0235 0.0249 0.0277 0.0267 3 1.4080 4.6420 3.0030 2.6371 2.4299 2.0081 1.7122 0.0317 0.0235 0.0191 0.0222 0.0289 4 3.6050 5.0000 3.0659 2.6970 2.4790 2.0653 1.6441 0.0775 0.0263 0.0235 0.0212 0.0239 5 1.9640 5.1050 3.0789 2.6783 2.4923 2.0815 1.5894 0.1072 0.0224 0.0223 0.0237 0.0278 6 2.0720 3.7160 3.0979 2.6239 2.4360 2.0270 1.5197 0.0875 0.0330 0.0243 0.0266 0.0255 7 2.0390 3.6960 3.0996 2.6763 2.4956 2.0791 1.5946 0.1972 0.0463 0.0253 0.0221 0.044 4 8 3.1340 3.6940 3.0861 2.6880 2.4997 2.0216 1.5341 0.1896 0.1602 0.0249 0.0238 0.1267 9 2.4880 3.6730 3.0635 2.7077 2.5018 2.0793 1.6410 0.2718 0.0787 0.0280 0.0244 0.0684 10 2.0610 3.7450 3.0976 2.6741 2.5051 2.0174 1.5749 0.1070 0.1956 0.0201 0.0224 0.0298 11 2.0120 3.7750 3.1071 2.6946 2.5133 2.0881 1.6333 0.0767 0.0334 0.0236 0.0231 0.0310 12 2.1530 3.7670 3.1036 2.6809 2.5053 2.0957 1.5889 0.1457 0.0437 0.0195 0.0215 0.0380 13 3.4390 3.7820 3.1177 2.7004 2.5072 2.0917 1.5930 0.0627 0.0248 0.024 0 0.0222 0.0227 14 2.9560 3.7870 3.0868 2.7128 2.5166 2.0882 1.6082 0.2383 0.0642 0.0185 0.0210 0.0553 15 1.8260 3.8050 3.1252 2.6907 2.5043 2.0961 1.5946 0.1096 0.0429 0.0215 0.0190 0.0363 16 2.3280 3.8880 3.1384 2.6945 2.5039 2.0970 1.5856 0.1040 0.01 90 0.0202 0.0199 0.0281 17 2.3050 3.9650 3.1035 2.7046 2.5130 2.0919 1.5907 0.0823 0.0174 0.0195 0.0215 0.0281 18 1.7980 3.6400 3.1142 2.7016 2.5170 2.1061 1.5863 0.1090 0.0291 0.0185 0.0180 0.0284 19 1.7880 3.5740 3.1464 2.6839 2.5100 2.0933 1.5674 0.0 844 0.0224 0.0241 0.0167 0.0296 20 2.2090 3.6940 3.1119 2.7099 2.5131 2.0931 1.5997 0.0673 0.0217 0.0177 0.0146 0.0207 21 1.9080 3.6650 3.0830 2.7003 2.4973 2.0990 1.5636 0.1897 0.1030 0.0274 0.0202 0.0849 22 2.1510 3.7660 3.0941 2.7151 2.5068 2.0958 1. 5964 0.1314 0.0303 0.0220 0.0230 0.0283 23 1.8720 3.7520 3.0409 2.7151 2.5130 2.0905 1.5866 0.1796 0.0390 0.0244 0.0212 0.0383 24 2.9970 3.7980 3.0674 2.6979 2.5126 2.0990 1.6017 0.0850 0.0457 0.0231 0.0198 0.0487 25 1.6940 3.8420 3.0879 2.6948 2.5174 2 .0970 1.5853 0.0899 0.0228 0.0219 0.0233 0.0241 26 2.0050 3.8020 3.0798 2.6963 2.5072 2.0884 1.5825 0.0957 0.0303 0.0216 0.0218 0.0345 27 1.8390 3.8010 3.0738 2.7063 2.5028 2.0884 1.6011 0.2479 0.0698 0.0251 0.0225 0.0761 28 1.5180 3.8310 3.0713 2.6954 2.5212 2.0835 1.5840 0.1184 0.0554 0.0229 0.0213 0.0473 29 1.3490 3.8720 3.0941 2.6884 2.5185 2.0934 1.5892 0.1262 0.0281 0.0224 0.0234 0.0298 30 1.5560 3.7960 3.0954 2.6998 2.5194 2.0820 1.5974 0.2003 0.0686 0.0331 0.0186 0.0580 31 1.2780 3.8620 3.1278 2.7123 2.5177 2.0965 1.5922 0.1092 0.0378 0.0198 0.0229 0.0299 32 3.1170 3.8450 3.1250 2.7193 2.5191 2.1001 1.5879 0.1606 0.0207 0.0200 0.0252 0.0192 33 2.8030 3.6990 3.0500 2.7143 2.5280 2.1074 1.5873 0.0406 0.0229 0.0229 0.0219 0.0239 34 2.2780 3.867 0 3.1340 2.7191 2.5322 2.1130 1.5890 0.0516 0.0307 0.0196 0.0207 0.0317 35 1.6430 3.9340 3.1379 2.7254 2.5155 2.1054 1.5932 0.0571 0.0213 0.0197 0.0184 0.0277 36 1.8360 3.8460 3.1406 2.6913 2.5278 2.1019 1.5592 0.0681 0.0559 0.0217 0.0172 0.0495 37 1.98 90 3.8990 3.1362 2.7099 2.5607 2.1067 1.5746 0.0796 0.0232 0.0692 0.0157 0.0294 Mean 2.1587 3.8942 3.0960 2.6957 2.5069 2.0843 1.5949 0.1202 0.0454 0.0238 0.0214 0.0389 S.D. 0.5775 0.3426 0.0314 0.0206 0.0229 0.0258 0.0350 0.0601 0.0382 0.0084 0.0028 0.0 216 95% CI 0.1861 0.1104 0.0101 0.0066 0.0074 0.0083 0.0113 0.0194 0.0123 0.0027 0.0009 0.0070

PAGE 90

90 Table 55 Predicted power due to radius change and measured power on Okuma LC 40 Operation Spindle speed (RPM) Feed rate (IPR) Depth of cut (inch) Workpiece radius Measured power (voltage) Ratio of workpiece radius Estimated ratio of power Experimental ratio of power Second OD turning 350 0.02 0.2 2.6375 2.6982 1 1 1 Third OD turning 350 0.02 0.2 2.4375 2.5088 0.9242 0.9242 0.9052 Fourth OD turning 350 0.02 0.1625 2.2563 2.0859 0.8555 0.6951 0.6935 Table 56 Mean power change by spindle speed change on Okuma LC 40 100% spindle speed 110% spindle speed Non cutting power Mean power 1.8989 2.0184 0.68 Estimated r atio (110%/100%) 1.1 Experimental result ( 110%/100%) 1.098

PAGE 91

91 0 200 400 600 800 1000 1200 0 50 100 150 200 250 300 350 offset=10mm time [sec]Temp. [deg C] heating plate temp.(thermo-couple) AL bolck(thermo-cople) AL Surface(infrared) 0 200 400 600 800 1000 1200 0 50 100 150 200 250 300 350 offset=50mm time [sec]Temp. [deg C] heating plate temp.(thermo-couple) AL bolck(thermo-cople) AL Surface(infrared) 0 200 400 600 800 1000 1200 0 50 100 150 200 250 300 350 offset=100mm time [sec]Temp. [deg C] heating plate temp.(thermo-couple) AL bolck(thermo-cople) AL Surface(infrared) 0 200 400 600 800 1000 1200 0 50 100 150 200 250 300 350 offset=150mm time [sec]Temp. [deg C] heating plate temp.(thermo-couple) AL bolck(thermo-cople) AL Surface(infrared) 0 200 400 600 800 1000 1200 0 50 100 150 200 250 300 350 offset=200mm time [sec]Temp. [deg C] heating plate temp.(thermo-couple) AL bolck(thermo-cople) AL Surface(infrared) 0 200 400 600 800 1000 1200 0 50 100 150 200 250 300 350 offset=250mm time [sec]Temp. [deg C] heating plate temp.(thermo-couple) AL bolck(thermo-cople) AL Surface(infrared) 0 200 400 600 800 1000 1200 0 50 100 150 200 250 300 350 offset=300mm time [sec]Temp. [deg C] heating plate temp.(thermo-couple) AL bolck(thermo-cople) AL Surface(infrared) 0 200 400 600 800 1000 1200 0 50 100 150 200 250 300 350 offset=400mm time [sec]Temp. [deg C] heating plate temp.(thermo-couple) AL bolck(thermo-cople) AL Surface(infrared) Figure 51 Temperature readings from trials for infrared sensor test

PAGE 92

92 Figure 52 Cooling profiles of infrared sensor test

PAGE 93

93 A) B) C) D) Figure 53 Measured cutting force on Hass SL10 for a = 1 mm and fr = 0.2 mm/rev(aluminum steel aluminum steel workpiece). A) in x direction, B) in y direction, C) in z direction and D) workpiece

PAGE 94

94 Figure 54 Spindle power (no cutting) as a function of spindle speed on Hass SL10

PAGE 95

95 0 10 20 30 40 50 60 70 80 -500 -450 -400 -350 -300 -250 -200 -150 -100 -50 0 Fx vs Feed feed (mm)force (N) 0 10 20 30 40 50 60 70 80 0 50 100 150 200 250 300 350 400 Fy vs Feed feed (mm)force (N) 0 10 20 30 40 50 60 70 80 0 50 100 150 200 250 300 350 400 Fz vs Feed feed (mm)force (N) 0 10 20 30 40 50 60 70 80 0 100 200 300 400 500 600 700 800 900 1000 Power feed (mm)Power(W) Figure 55 Cutting force and spindle power on Hass SL10 for a = 1 mm and fr = 0.1 mm/rev using aluminum/steel workpiece at 500 rpm and 67 mm diameter. (Top left) x direction force ( Ft). (Top right) y direction force ( Fr). (Bottom left) zdirection force ( Ff). (Bottom right) spindle power.

PAGE 96

96 0 10 20 30 40 50 60 70 80 -1000 -900 -800 -700 -600 -500 -400 -300 -200 -100 0 Fx vs Feed feed (mm)force (N) 0 10 20 30 40 50 60 70 80 0 100 200 300 400 500 600 700 800 Fy vs Feed feed (mm)force (N) 0 10 20 30 40 50 60 70 80 0 100 200 300 400 500 600 700 800 Fz vs Feed feed (mm)force (N) 0 10 20 30 40 50 60 70 80 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Power feed (mm)Power(W) Figure 56 Cutting force and spindle power on Hass SL10 for a = 1 mm and fr = 0.2 mm/rev using aluminum/steel workpiece at 500 rpm and 66.5 mm diameter. (Top left) x direction force ( Ft). (Top right) y direction force ( Fr). (Bottom left) zdirection force ( Ff). (Bottom right) spindle power. 55 60 65 70 -400 -350 -300 -250 -200 -150 -100 -50 0 Mean value of Fxworkpiece diameter (mm)force (N) Al Steel 55 60 65 70 100 200 300 400 500 600 700 Mean value of power workpiece diameter (mm)Power(W) Al Steel Figure 57 Mean x direction cutting force and spindle power on Hass SL10 as a function of workpiece diameter for a = 0.5 mm and fr = 0.2 mm/rev using aluminum/steel workpiece at 500 rpm. (Left) x direction force ( Ft). (Right) spindle power.

PAGE 97

97 Figure 58 Cutting force and spindle power on Hass SL10 for a =1mm and fr=0.1 mm/rev using steel/steel at 500rpm

PAGE 98

98 A)

PAGE 99

99 B) Figure 59 A) Maximum values of facing operations, mean values of turning operati ons B) Standard deviations of turning operations on Okuma LC 40

PAGE 100

100 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 Part numberVoltage (V)Max. power of first facing part A) 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 Part numberVoltage (V)Max. power of second facing part B) 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of first OD-ID C) 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of second OD D) 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of third OD E) 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of fourth OD F) 0 5 10 15 20 25 30 35 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of finishing OD-ID G) 0 5 10 15 20 25 30 35 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of fisrt OD-ID H)

PAGE 101

101 0 5 10 15 20 25 30 35 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of second OD I) 0 5 10 15 20 25 30 35 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of third OD J) 0 5 10 15 20 25 30 35 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of fourth OD K) 0 5 10 15 20 25 30 35 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of finishing OD-ID L) Figure 510 A) Max value of first facing, B) Max value of second facing, C) Mean value of first turning (OD and ID are simu ltaneous cut), D) Mean value of third turning (OD cut), E) Mean value of second turning (OD cut), F) Mean value of fourth turning (OD cut), G) Mean value of finishing turning, H) Standard variation of second turning (OD cut), I) Standard variation of first turning (OD and ID are simultaneous cut), J) Standard variation of third turning (OD cut), K) Standard variation of fourth turning (OD cut), and L) Standard variation of finishing turning on Okuma LC 40

PAGE 102

102 Figure 511 Power difference between worn and nor mal tool in second facing on Okuma LC 40 Figure 512 Mean and maximum value of measured power on Okuma LC 40

PAGE 103

103 0 50 100 150 200 250 300 0 1 2 3 4 5 6 7 8 time(sec)sensing value (voltage) Measured power for air cutting Figure 513 Measured power under air cutting on Okuma LC 40 140 150 160 170 180 190 200 210 220 230 240 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 time(sec)power(voltage) spindle speed 100% spindle speed 110% Figure 514 Power measurement on Okuma LC 40 under different spindle speed

PAGE 104

104 0 0.5 1 1.5 2 2.5 3 3.5 4 x 104 0.5 1 1.5 2 2.5 time [msec]power (sensor output in voltage) Aluminum alloy 6061 Rockwell A hardness : 42.3 P20 tool steel Rockwell A hardness : 68.3 Hot-rolled 1035 Low-carbon steel Rockwell A hardness : 55.8 1018 Low-carbon steel Rockwell A hardness : 58.8 Figure 515 Measured cutting power (sensor value in voltage) on SyiL C6B for spindle speed = 1000 rpm, a = 0.5 mm, fr = 0.1 mm/rev and workpiece diameter = 22 mm (workpiece A) 0 0.5 1 1.5 2 2.5 3 x 104 0.5 1 1.5 2 2.5 3 3.5 4 time [msec]power (sensor value in voltage) Aluminum alloy 6061 Rockwell A hardness : 43.2 Aluminum alloy 6061 and 1018 Low-carbon steel Rockwell A hardness : 43.2 and 60 1018 Low-carbon steel Rockwell A hardness : 60 Figure 516 Measured cutting power (sensor value in voltage) on SyiL C6B for spindle speed = 1000 rpm, a = 1 mm, fr = 0.1 mm/rev and workpiece diameter = 23 mm (workpiece B)

PAGE 105

105 0 0.5 1 1.5 2 2.5 3 x 104 0.5 1 1.5 2 2.5 3 3.5 time [msec]power (sensor value in voltage) Aluminum alloy 6061 + Stainless steel pin Rockwell A hardness : 42.3 + 74.8 Aluminum alloy 6061 Rockwell A hardness : 42.3 Aluminum alloy 6061 Rockwell A hardness : 42.3 Figure 517 Measured cutting power (sensor value in voltage) on SyiL C6B for spindle speed = 1000 rpm, a = 1 mm, fr = 0.1 mm/rev and workpiece diameter = 24 mm (workpiece C) 0 0.5 1 1.5 2 2.5 3 x 104 0.5 1 1.5 2 2.5 time [msec]power (sensor value in voltage) Tool breakage Aluminum alloy 6061 Rockwell A hardness : 42.3 Aluminum alloy 6061 Rockwell A hardness : 42.3 Aluminum alloy 6061 and stainless steel pin Rockwell A hardness : 42.3 and Figure 518 Measured cutting power (sensor value in voltage) on SyiL C6B for spindle speed = 1000 rpm, a =1 mm, fr = 0.1 mm/rev and workpiece diameter = 22 mm (workpiece C)

PAGE 106

106 CHAPTER 6 FLANK WEAR MODEL FOR WORKPIECE COMBINED FROM DIFFE RENT HARDNESS MATERIALS During metal cutting operations, changes in workpiece material properti es can cause problems for maintaining the operators safety, product quality, and machine and tool reliability. Therefore, detect ing workpiece material property changes in real time can be beneficial. In previous chapters, three different types of sensor s ( power sensor, ultimate thermometer, and dynamometer ) were tested for feasibility to detecting workpiece material changes. Based on the result s from chapter 5, a power sensor can be used for detect the workpiece material property changes with cost and installation advantage s compared to dynamometer In this research, the noninvasive, low cost power sensors offer a feasible option for inprocess metrology of turni ng operations. The final goal of this research is to develop the flank wear model based on power measurement to predict tool wear of tool even though workpiece hardness changes during the process. In this chapter, an adaptive real time flank wear model wi ll be proposed. H ow to validate the suggested model will be described. An e xperimental plan to find flank wear rates of different materials that can be used even if the workpiece has different materials will be described. 6.1 Adaptive Real Time Flank Wear M odel for Workpiece Hardness C hanges The goal is to design the real time flank wear model using a power sensor in turning operations to improve productivity and reduce tool wear. Generally, flank wear can be determined by using cutting conditions, but if w orkpiece hardness is variable, and the variation in the hardness cannot be measured, the current flank wear model cannot be predicted accurately without workpiece hardness est imation. There have

PAGE 107

107 been flank wear model research projects studying uniform sing le material workpieces. In this project, a power sensor will be used to measure the spindle power during cutting of materials with variable hardness High hardness material cutting should require more power than low hardness material cutting. A p ower model for workpiece material will be used to detect workpiece material property changes. Based on the power model, flank wear rate should be changed when workpiece material changes in hardness. The flank wear progression can be predicted accurately using the abrasive wear model with the identified wear characteristic constants and the identified initial wear. Currently, the flank wear model is limited to a workpiece with one material of uniform properties. An adaptive flank wear model will be proposed in this p roject Workpiece material properties such as hardness can be predicted by using power sensor. Workpiece material properties will be updated in real time, and the adaptive flank wear model will also be implemented in real time. The adaptive real time flan k wear scheme is shown in Fig. 61. 6.2 Experimental P lan Experiments were designed to obtain the wear rate for two different materials and validate that the proposed real time adaptive flank wear model is effective. The flank wear progress can be predicted by Eq. (325). In the case of a specific cutting condition, the terms 1 fC 2 fC and 1 tan tan V are constants. Therefore, the wear rate can be expressed a constant for fixed a cutting condition. The flank wear rate of Eq. (325) can be given by the following equation: (underspecificcuttingcondition)BdV D dt (6 1)

PAGE 108

108 where D is the constant for a specific cutting condition. The experimental plan for obtaining the wear rate was des igned in order to calibrate, by an inverse approach, the wear models based on abrasion mechanisms. Two different materials were used for workpieces, one is a soft steel (8620 alloy steel) and the other is hard (P20 tool steel), both with an initial diamete r equal to 18.8 mm. The Rockwell hardness A values of both materials were measured with Wilson Hardness tester and will be mentioned in the following chapter. A TCMT32.52 carbide tool was used for almost all tests as the default tool. A TCMT32.52 carbide t ool is low quality, low cost, not coated, and wear s fast In this experiment, four different sequence of workpiece materials (soft, hard, half half, and mainly soft) were used. The workpi ece shapes are shown in Fig. 6 2 In cases of soft, hard, half half, the flank wear was measured every 60 seconds cutting. In case of mainly soft, the flank wear was measured every 84 seconds of soft material cutting and 36 seconds of hard material cutting. The 60 seconds cutting is one pass of 50 mm length of workpiece, 84 seconds cutting is one pass 70 mm length workpiece, and 36 seconds cutting is 30 mm length workpiece cutting under 50 mm/min feedrate. All cutting were continued until total cutting length is 1000 mm or major tool breakage appeared. For this total cutting length, the wear rate can be considered stationary and, therefore, not dependent on the cutting time. Furthermore, in order to increase wear phenomena and temperature, no lubricant was used during the tests. The wear tests were performed according to the cutting conditions as shown in Table 61. The flank wear was measured by a USB type digital microscope (model: Dino Lite AM413T). The AM413T digital microscope is designed with a 1.3 megapixel image sensor. The possible m agnifications of the microscope ra nge from 10X to 230X.

PAGE 109

109 In this project, a 210X magnification was used for capturing the images. Microscope specifications are shown in Table 62 In this research, cases soft and hard were designed to obtain the wear rate Ds for each material. Cases half h alf and mainly soft were performed to validate the proposed real time adaptive flank wear model. Therefore, f our different hardness sequences of workpieces under same cutting conditions were implemented. The workpiece of case soft is 18.8mm diameter cylindrical shape of 8620 alloy steel. The workpiece of case hard is P20 tool steel of the same dimensions The workpiece of case half half has the sequence of the same lengths (50 mm for each) of soft and hard materials. The workpiece of case mainly soft is long, soft metal cutting and short, hard metal cutting. For this, a special workpiece was made which includes 70 mm of soft metal combined with 30 mm of hard metal. All cutting test were performed under 1000 rpm spindle speed, 50 mm/min feedrate, 1 mm depth of cut as shown in Table 61. To create the fast flank wear, a low quality non c oated carbide tool was used. For repeatability of the flank wear rate, the consistency of tool is very important. In order to reduce the impact of tool differences, additional cases of tests were performed where the three different cases are machined with one insert as single insert wear test s A TCMT32.52 carbide tool has three edges that can be used for cutting. The f ir st edge was used for cutting a soft workpiece. The s econ d edge was used for cutting a hard workpiece and the third edge was used for cutting the half half workpiece. A C 5 carbide insert from McMaster was used for single insert wear test for the progression of different tool wear. C 5 carbide insert is excellent in heat resistance and wear resistance. In C 5 carbide insert, three different edges of tool were used for cutting three different types of

PAGE 110

110 workpiece sequence (soft, hard, and half half). These two single insert cutting tests were performed as described table 63 The measured flank wear was compared to the estimated flank wear by proposed adaptive flank wear model (Eq. 327). All experiments were performed two or three times for verification of the repeatability.

PAGE 111

111 Table 61 Cutting conditions in wear t ests Case Tool Workpiece material sequence Depth of cut Feedrate Workpiece diameter Lubrication soft TCMT 32.52 carbide 8620 alloy steel 1 mm 50 mm/min 18.8 mm Dry hard P20 tool steel halt half 50 mm 8620 alloy steel and 50 mm P20 tool steel m ainly soft 70 mm 8620 alloy steel and 30 mm P20 tool steel Table 62 Digital microscope specification Model AM413T Dino Lite Pro Basic Specifications Interface USB2.0 Product Resolution 1.3M pixels. (SXGA) Magnification Rate 10x~50x 230x Senso r Color CMOS Frame rate up to 30fps Image Format jpg, bmp, avi Microtouch Yes LED Switchable (via DinoCapture) Yes Measurement Function (via DinoCapture) Yes Calibration Function (via DinoCapture) Yes Software Windows Vista, XP, 2000, or Windows Ser ver 2003 MAC OS (Microtouch function no supported) Unit Weight 90(g) Units Dimension 10cm (H) x 3.2cm (D)

PAGE 112

112 Table 63 Cutting conditions in single insert tests Tool Edge Workpiece material sequence Depth of cut Feedrate Workpiece diameter Lubricatio n TCMT32.52 carbide 1 8620 alloy steel 1 mm 50 mm/min 18.8 mm Dry 2 P20 tool steel 3 50 mm 8620 alloy steel and 50 mm P20 tool steel C 5 carbide 1 8620 alloy steel 2 P20 tool steel 3 50 mm 8620 alloy steel and 50 mm P20 tool steel

PAGE 113

113 Figure 61 Real time flank wear estimation model scheme

PAGE 114

114 Figure 62 Workpiece shapes for validation

PAGE 115

115 CHAPTER 7 RESULT OF FLANK WEAR MODEL FOR WORKPIECE COMBINED WITH TWO DIFFERENT HARDNESS M ATERIALS In previous chapter 6, the experimental plan to test the flank wear model for workpieces consisting of different hardness materials was described. In chapter 7, the experimental cutting test result will be presented t he possibilities of the proposed flank wear rate model discussed. The proposed flank wear rate model will be demonstrated with power sensor what was tested in chapter 5. 7.1 Hardness Measurement of 8620 Alloy S teel and P20 and Flank Wear I mages A Wilson Rockwell hardness tester model 3JR was used to measure the workpiece hardness using the Rockwell A scale Figure 71 shows the Rockwell hardness tester and specification s of the tester are shown in Table 71. Two different materials were used in the cutting test s One is the 8620 alloy steel as the soft metal and another is the P20 tool steel as the hard metal. Chemistry data of these two materials are shown in table 72 To confirm the material hardness difference, the Rockwell hardness A values of 8620 alloy steel parts (122 parts) and P20 tool steel parts (7 7 parts) were measured and these values are shown in Fig. 72. The mean value of Rockwell hardness A of 8620 alloy steel is 42.44 and the mean value of Rockw ell hardness A of P20 is 53.23. The standard deviations were 1.87 and 2.10 respectively. Figure 73 shows the box plots wi th whiskers of both m aterials. As shown in Fig. 73, the 8620 alloy steel and P20 tool steel materials have significantly different Rockwell hardness A value. To measure the flank wear, images were captured by using the U SB portable microscope with 210X magnification. The program Dino Capture provided with the Dino-

PAGE 116

116 Lite was used for capturing and measurement the length of flank wear. Figure 74 shows flank wear images of case soft. Each image was taken after 50 mm cutting of workpiece. 7.2 Flank Wear M easurement All cutting experiments were perfor med under cutting conditions of 50 mm/min feedrate, 1mm depth of cut, and 1000 rpm spindle speed (equivalent to a cutting speed of 55.92 m/min) All workpieces were 18.8 mm diameter cylindrical shape. During the cu tting, the power was measured by using the power sensor. The original diameter was recorded and the machined diameter was recorded to check the dimension of product. Case hard is P20 tool steel cutting. After every 50 mm cutting, an image of the flank wear surface was taken by USB portable microscope. Case hard was repeated 3 times. Figure 75 shows the flank wear versus cutting length. As shown in Fig. 75 after break in period, the flank wear has constant wear rate in case hard 1 and 2. But in case har d 3, the break in period is longer than the other cases. The cutting of the case soft was under same cutting conditions the as case hard. Afte r every 50 mm cutting, an image of the flank wear surface was taken by USB portable microscope. All cutting condit ions and measurement were same as case hard. The only difference was that workpiece material was not P20 tool steel but 8620 alloy steel as soft metal. Figure 76 shows the flank wear versus cutting length of three case soft cutting experiments. As shown in Fig. 76 after break in period, all flank wear has constant wear rate. In theory, the wear rate should be same if all cutting conditions are the same. As seen in Fig. 76 case soft 1 and 2 had almost the same slope, but the

PAGE 117

117 case soft 3 had a different slope. Tool differences may be one of the reasons why the wear rates are different. The cutting conditions of the case half half were the same as the p revious cases. But workpiece consisted of the two different materials a combination of 8620 alloy st eel and P20. After 50 mm of 8620 alloy steel cutting, 50 mm cutting of P20 was cut. This cutting sequence was repeated until cutting was stopped. Figure 77 shows the flank wear of case half half. The slope of flank wear for hard metal cutting should be higher than the slope of flank wear for soft metal cutting. As seen in Fig. 77 the slope of the period of P20 tool steel cutting (red color background) is higher than the slope of the period of 8620 alloy steel cutting (white color background) The cutting conditions of the case mainly soft were the same as the previous cases. But the workpiece s consisted of 70 mm of 8620 alloy steel and 30 mm P20 tool steel This cutting sequence was repeated. Figure 78 shows the flank wear of case mainly soft The slope of flank wear for hard metal cutting should be higher than the slope of flank wear for soft metal cutting. As seen in Fig. 78 the slope of the period of P20 tool steel cutting is higher (red color background) than the slope of the period of 8620 alloy steel cutting (white color background) In the case single tool, three different cutting test s were performed with only one insert All tool edges w ere used in three different types of cutting. The f irst edge was used for 8620 alloy steel cutting (case sof t), the second edge was used for P20 tool steel cutting (case hard), and the third edge was used for combi n e d workpiece of 8620 alloy steel and P20 tool steel cutting (case half half). Figure 79 shows the flank wear of the three different tool edges when using a TCMT32.52 carbide tool Bottom of Fig. 7 9

PAGE 118

118 shows the flank wear of case half half. Red color background is P20 tool steel cutting. Figure 710 shows the flank wear of the three different tool edges when cutting with C 5 carbide tool Bottom of Fig. 7 10 shows the flank wear of case half half. The r ed color background is P20 tool steel cutting. As shown in Figures 7 8 and 9, each flank wear rate if workpiece is combined of 8620 alloy steel and P20 tool steel is almost same as workpiece is one materi al. 7.2 Flank Wear Rate between 8620 Alloy S teel and P20 If cutting c onditions are the same, the flank wear rate should be constant in constant wear rate zone (phase II in Fig. 35) between break in period and rapid wear to failure Figure 71 1 shows that flank wear rate versus Rockwell A hardness from the case soft and hard data. Bad data was eliminated using Chauvenets criterion ( Huggins 1975). The critical boundary for rejection of data must be specified by the researcher prior to conducting the experi ment. The approach recommended herein is to accept as valid all data values,iX, which fall between the critical limits lX and uX such that: 1 1liuPXXX kN (7 1) where N is sample number and k is a constant specified by researcher. A common approach, known as Chauvenets criterion, is to let k equal 2. Table 73 shows the probability levels and the range of acceptable data. As mentioned previously, the tool difference may impact the flank wear rate. In the research reported in this dissertation, a TCMT32.52 carbide was used mainly. A TCMT32.52 is low c ost and uncoated tool. Figure 71 2 shows the flank wear surface of two different new inserts. As shown in Fig. 7 1 2 the consistency of tool is very poor in the pattern of surface. In statistics, analysis

PAGE 119

119 of variance (ANOVA) is a collection of statistical models, and their associated procedures, in which the observed variance in a particular variable is partitioned into components due to different sources of variation. To check the effects of tool and hardness, ANOVA wa s performed on the data of all tool edges hardness, and flank wear rate. Fact 1 is tool edge number and fact 2 is workpiece material s (8620 alloy steel or P20 tool steel) As seen in Fig. 71 3 fact 1 (tool edge number) has the probability of F value of 0.001%, enough to conclude statistical significance. But workpiece effect is not big ( the probability of Fvalue of 17.2%). Predictive Analytics Soft Ware (PASW) from SPSS Inc. was used for analyzing the data. To reduce the dominating effect of tool over workpiece the case single tool will be an alyzed next. When a workpiece combines different materials, the cutting period for each material should have specific own flank wear rate. The flank wear rate in interval s of 8620 alloy steel was compared with the flank wear rate in interval s of P 20. As to be predicted, the flank wear rate of hard material is higher than soft material. As shown in Fig. 71 5 ~1 8 flank wear rate of soft material cutting section is lower than flank wear rate o f hard material cutting section. Correlation c oefficient r values also show that flank wear rate is increased as the hardness increases even though workpiece materials are periodically changed. Figure 71 4 shows the one way ANOVA for case half half The fact i s workpiece material. As shown in Fig. 7 1 4 flank wear rates of 8620 alloy steel and P20 tool steel are different ( the probability of Fvalue of 0 %) if the tool is not changed. Theoretically, the flank wear rate should only depend upon the material hardness and the wear during case half half and mainly soft should be same as flank wear rate

PAGE 120

120 during the corresponding case hard and case soft. According to the results of the cutting test s, the flank wear rate was constant under same tool, but flank wear rate has different constant s with different tool s. This res ult may be caused by inconsistent tool quality. The c ase single tool was designed to reduce the impact caused by tool difference. Figure 71 9 and 20 show flank wear rates which are the empty circle symbol as the flank wear rate if only 8620 alloy steel is cut the empty triangle symbol as flank wear rate if only the P20 tool steel is cut, the solid circle symbol as flank wear rate of 8620 alloy steel section of case half half (combined 50 mm length 8620 alloy steel and 50 mm length P20 tool steel), and sol id triangle symbol is flank wear rate of P20 tool steel section of case half half. From the result of case single tool, the flank wear rates are maintained when workpiece was a combination of two different materials. 7.3 Flank W ear Estimation Using P ower I f tools are perfectly same with same cutting conditions, flank wear rate should be same in the stable wear increasing range. Previously, each flank wear rate of different materials has the same in same material with one insert. To minimize the effect of tool, the case single tool was performed. In case single tool, the flank wear rates in case soft and hard were maintained same as the rates case half half. A power sensor was used for detecting material changes. Figure 72 1 shows an example of cutting power (sensor value) during 8620 alloy steel cutting of case single tool cutting. The average power value of stable cutting range was recorded and compared to hardness and flank wear rate. Figure 72 2 shows the hardness versus flank wear rate (case single tool C 5). The relationship between hardness and flank wear rate is linear as shown in Fig. 7 2 2 As shown in Fig. 72 3 the relationship between hardness and average cutting power is

PAGE 121

121 also linear. Therefore, the average cutting power can be used for detecting t he material property changes instead of hardness measurement (see Fig. 72 4 ). In case of single tool C 5, estimated flank wear rates can be calculated by FWR0.39AP0.2 (7 2) where FWR/cuttinglength mmm is estimated flank wear rate and AP(sensor value) is average cutting power which is from case soft and hard of single tool C 5. Figure 725 shows the relationship between average power and flank wear rate. F lank wear estimation for fixed cutting conditions as everything is same except the workpiece materials will be following steps: Step 1: Power measurement and average power calculation for specific time (for example 1 sec) Step 2 : Flank wear rate decision from average power Step 3 : Flank wear estimation from multiple by cutting length and flank wear rate Step 4 : Update the flank wear and repeat the process step 1~ 4 again D uring cutting of two different materials as case single tool, flank wear can be estimated by summation of each workpiece section with flank wear rate decided using power value as shown in Fig. 72 6 Figure 727 shows the difference between flank wear captured by microscope and estimated flank wear from Eq. (72). Figure 728 shows the difference of flank wear rate (mm/m cutting length) during 50 mm cutting length between flank wear captured by microscope and estimated flank wear from Eq. (72). Initial length of estimated flank wear was the same as flank wear after first 50 mm length cut. Since flank wear is u pdate from initial point, the difference between flank wear and estimated flank wear

PAGE 122

122 can be increased after long cutting length. If initial flank wear can be update after specific cutting length, it is possible to estimate flank wear accurately without many times of flank wear measurement.

PAGE 123

123 Table 71 Specification of Wilson Rockwell hardness tester model 3JR Tests Rockwell Scales A,B,C,D,E,F,G,H,K,L,M,P,R,S & V Maximum Vertical Gap 8" Throat Depth 5 1/4" Overall Dimensions 13" x 20" x 28" Tall Table 7 2 Chemistry data of P20 tool steel and 8620 alloy steel P20 tool steel 8620 ALLOY STEEL Carbon 0.28 0.4 0.18 0.23 Chromium 1.4 2 0.4 0.6 Iron Balance Balance Manganese 0.6 1 0.7 0.9 Molybdenum 0.3 0.55 0.15 0.25 Phosphorus 0.03 max 0.035 max Silicon 0.2 0.8 0.15 0.35 Sulphur 0.03 max 0.04 max Nickel 0.4 0.7 Table 73 Probability levels and the range of acceptable data of flank wear rates of case hard and soft (Chauvenets criterion) N probability point 1 1 2 N t mean X Standard deviation range of acceptable data Xts case hard 1 9 0.94 2.1892 0.6953 0.1345 [ 0.4008 0.9898] case hard 2 3 0.83 2.1045 0.9573 0.1947 [ 0.5477 1.3670] case hard 3 8 0.94 2.2409 1.1415 1.5074 [ -2.2364 4.5194] case soft 1 12 0.96 2.3282 0.4482 0.2408 [ -0.1125 1.0088] case soft 2 18 0.97 2.3681 0.3192 0.1327 [ 0.0049 0.6335] case soft 3 22 0.98 2.5176 0.1373 0.0692 [ -0.0370 0.3116]

PAGE 124

124 Figure 7 1 Wilson Rockwell hardness tester model 3JR

PAGE 125

125 8620 alloy steel P20 30 35 40 45 50 55 60 65 Rockwell hardness A Figure 72 Rockwell hardness A of 8620 alloy steel and P20 tool steel workpieces 8620 alloy steel P20 30 35 40 45 50 55 60 65 Rockwell hardness A Figure 73 Box plots of Rockwell hardness A with whisker of 8620 alloy steel and P20 tool steel workpieces

PAGE 126

126 Figure 74 Flank wear images of case soft second cutting test by microcope

PAGE 127

127 Figure 75 Flank wear of case hard 0 200 400 600 800 1000 1200 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Cutting length (mm)Flank wear (mm) Case soft 1 Case soft 2 Case soft 3 Figure 76 Flank wear of case soft

PAGE 128

128 Figure 77 Flank wear of case half half Figure 7 8 Flank wear of case mainly soft

PAGE 129

129 Figure 79 Flank wear of case single tool (TCMT32.52 carbide)

PAGE 130

130 Figure 710 Flank wear of case single tool (C 5 carbide)

PAGE 131

131 35 40 45 50 55 60 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Rockwell A hardnessFlank wear rate (mm / m cutting length) case hard 1 case hard 2 case hard 3 case soft 1 case soft 2 case soft 3 linear R = 0.4772 Figure 71 1 Flank wear rate versus Rockwell hardness A Figure 71 2 Flank wear surface of two different new inserts

PAGE 132

132 Test s of Between Subjects Effects Dependent Variable:Flank wear rate Source Type III Sum of Squares df Mean Square F Sig. Corrected Model .032 a 21 .002 2.649 .000 Intercept .033 1 .033 56.168 .000 Tool .024 15 .002 2.766 .001 Workpiece .001 1 .001 1.889 .172 Tool Workpiece .002 5 .000 .566 .726 Error .075 129 .001 Total .154 151 Corrected Total .107 150 a. R Squared = .301 (Adjusted R Squared = .188) Figure 71 3 ANOVA screen shot case soft, case hard, case half half, and case mainly sof t Tests of Between Subjects Effects Dependent Variable:Flank wear rate Source Type III Sum of Squares df Mean Square F Sig. Corrected Model .000 a 1 .000 41.084 .000 Intercept .001 1 .001 130.391 .000 Workpiece .000 1 .000 41.084 .000 Error 4.412E 5 8 5.515E 6 Total .001 10 Corrected Total .000 9 a. R Squared = .837 (Adjusted R Squared = .817) Figure 71 4 One way ANOVA screen shot for case half half (case half half 1)

PAGE 133

133 8620 alloy steel P20 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Flank wear rate (mm / m cutting length) Figure 71 5 Flank wear rate versus workpiece materials (case half half 1, Correlation coefficient r = 0. 9149) 35 40 45 50 55 60 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Rockwell A hardnessFlank wear rate (mm / m cutting length) Figure 71 6 Flank wear rate versus Rockwell hardness A (case half half 1, Correlation coefficient r = 0. 8622)

PAGE 134

134 8620 alloy steel P20 0 0.2 0.4 0.6 0.8 1 Flank wear rate (mm / m cutting length) Figure 71 7 Flank wear rate versus workpiece materials (case mainly soft 1, Correlation coefficient r = 0.7706) 35 40 45 50 55 60 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Rockwell A hardnessFlank wear rate (mm / m cutting length) Figure 71 8 Flank wear rate versus Rockwell hardness A (case mainly soft 1, Correlation coefficient r = 0.6249)

PAGE 135

135 8620 alloy steel P20 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 flank wear rate (mm / m cutting length) Case soft Case hard Hard of case half-half (mean) Soft of case half-half (mean) Figure 71 9 Flank wear rate versus workpiece materials (case single tool TCMT32.52 carbide) 8620 alloy steel P20 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 flank wear rate (mm / m cutting length) Case soft Case hard Hard of case half-half (mean) Soft of case half-half (mean) Figure 720 Flank wear rate v ersus workpiece materials (case single tool C 5 carbide)

PAGE 136

136 0 2 4 6 8 10 12 x 104 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 time (msec)power cutting start cutting end spindle turn off spindle turn on (1000 rpm) Figure 72 1 Power measurement of 8620 alloy steel cut 40 42 44 46 48 50 52 54 56 58 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 hardness (Rockwell A)flank wear rate (mm / m cutting length) flank wear rate linear R = 0.6782 Figure 72 2 Hardness vs flank wear rate (case single tool C 5)

PAGE 137

137 40 42 44 46 48 50 52 54 56 58 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 hardness (Rockwell A)average power (V) average power linear R = 0.9270 Figure 72 3 Hardness vs average cutting power (case single tool C 5) 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 average power (V)flank wear rate (mm / m cutting length) flank wear rate linear R = 0.7652 Fi gure 72 4 A verage cutting power vs flank wear rate (case single tool C 5)

PAGE 138

138 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 average powerflank wear rate (mm / m cutting length) y = 0.39*x 0.2 flank wear rate (C-5 soft and hard) linear Figure 725 Flank wear model from C 5 case soft and hard

PAGE 139

139 Figure 726. Real time flank wear estimation using power sensor

PAGE 140

140 Figure 72 7 Difference between flank wear and estim ated flank wear 0 100 200 300 400 500 600 700 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 Cutting length (mm)Flank wear rate (mm/m cutting length) during 50 mm cutting length Flank wear rate Estimated flank wear rate Figure 728 F lank wear rate (mm/m cutting length) during 50 mm cutting length

PAGE 141

141 CHAPTER 8 CONCULSIONS AND FUTURE WORK Tool condition monitoring is recognized as impo rtant in CNC processes since excessive wear or tool breakage has to be noticed immediately in an automated manufacturing system to maintain the quality and productivity. During metal cutting operations, unacceptable tool wear or breakage can damage the tool holder, the workpiece, or the machine elements. Also, the surface quality and the dimensional accuracy of the product degrade can be caused by tool fail ure. Moreover, operator safety, or problem in the manufacturing system can also be caused by tool breakage. Therefore, tool s must be change d at the right time. The cutting c onditions in CNC machining such as cutting speed, feed per revolution, and depth of cut are conventionally based on prior experience. But, initial cutting conditions cannot anticipate inprocess variations such as depth of cut or material hardness change. Increases in hardness and depth of cut raise cutting force and may cause faster tool wear, and loss of dimensions and surface finish. Therefore, workpiece material detection would be beneficial. In chapter 4 and 5, three sensor s (power sensor, ultimate thermometer, and dynamometer) were tested for the detection of material changes. I t is confirmed that the power sensor is a cheap and non contact way for detecting material changes. Theoretically, the spindle power required for turning operations in hard materials is higher than that required for soft materials. Both theory and experiments documented in this work show that a power sensor provides a means of detecting hardness changes in the work material without affecting the cutting process.

PAGE 142

142 In this dissert ation, flank wear was considered the main wear factor. Flank wear arises due to both adhesive and abrasive wear mechanisms from the intense rubbing action of the two surfaces in contact, i.e., the clearance face of the cutting tool and the newly formed sur face of the workpiece. Its rate of increase at the beginning of the tool life is rapid, settling down to a steady state then accelerating rapidly again at the end of tool life. Flank wear leads to a deterioration of surface quality, increased contact area and, consequently, increased heat generation. Flank wear models have been developed for specific workpieces made of a single material in many studies. However, if workpiece material properties (such as hardness) are changed during operation, the existing f lank wear models cannot be used, because general flank wear models cannot reflect the real time workpiece material changes. The work demonstrate s a real time flank wear model using a power sensor. At first, flank wear rates for multiple material s are determined by cutting tests. These flank wear rates are used for modeling the flank wear of workpieces of co mbined different materials by summation of the flank wear for each section. Each section of workpiece with different materials can be known by average power of measured spindle power. This proposed model estimates the flank wear for workpieces with varying material properties. In chapter 6 and 7, the flank wear rat es for two materials were obtained and these two different rates can be used even if workpiece material is not one. However, case of low quality tool had some limitation. If tool and cutting conditions are exactly same, flank wear rate should be same. In case soft, the flank wear rate was different according to change of tool. Case hard also had different flank wear rates. The flank wear rates of soft material cutting and hard material cutting should be used for

PAGE 143

143 workpiece of soft hard materials combined. However, known flank wear rates of soft and hard could not use for combined workpiece if tool was changed. However, one cutting edge maintained the same flank wear rate in the same material cutting section. From this result, consistency of tool is very important factor for wear model. Therefore, the study of tool consistency should be required in t he future. The possible reason for different flank wear rates may be different hardness of tool or surface pattern. As shown in new tools surface, low quality tool flank surface patterns were different. In this dissertation, the possibility of real time fl ank wear estimation was studied when workpiece combined with two different materials. It was confirmed that the flank wear rate of specific workpiece material can be used for cutting workpiece combined with different materials. And power sensor was used for detecting workpiece materials changes. In the future, the advanced study about the sensor signal should be completed. The average of spindle power was used for detecting for workpiece change. To detect the tool breakage or other feature, different types of signal processing such as Wavelet transform should be required. As mentioned previous paragraph, the research about tool effect may be good topic. In high quality tool, consistency will be good. But low quality tool has no consistency The technique pr oposed in this dissertation can be implemented with adaptive cutting condition rules to make decisions that reduce cutting cost and maintain product quality. To apply this to metal cutting industry, suitable test plan for obtaining flank wear rates should be developed according to the specific plant. If a manufacturing facility has consistent tools, experiments can be conducted to

PAGE 144

144 establish the relationship between cutting power and tool wear rate. That relationship can be used in a real time model to predi ct tool wear based upon the power sensor.

PAGE 145

145 APPENDIX A LINEAR POWER MODELING USING HASSSL10 TEST DATA In turning operations, the power can be modeled using t noncuttingPFrP where tF is the tangential cutting force, is the spindle speed, r is the part radius, and noncuttingP is the noncutting power. The noncutting power is shown as a function of spindle speed in Fig. 54. The difference between the calculated power (based on the measured force) and experimental power for tests (using the aluminum steel hybrid workpiece) is shown in Fig. A 1. It is seen that the power is over predicted using the traditional model. The cutting conditions are provided below. The average forces for 20 workpiece diameters are shown in Table A 1. Condition 1: depth of cut = 0.5 mm, feed per revolution = 0.2 mm/rev, spindle speed = 500 rpm. The averages tangential forces (measured separately using a cutting force dynamometer) were 155 N (Al) and 307 N (steel). Condition 2: depth of cut = 1 mm, feed per revolution = 0.2 mm/rev, spindle speed = 500 rpm. The average tangential forces were 265 N (Al) and 611 N (steel). Four alternative linear models were tested for describing the experimental power. The noncutting power for driving the spindle at 500 rpm was taken to be 90 W (based on the Fig. 54 data). The model 1 fitting form wast noncuttingPaFrP. Figure A 2 shows the total error between the experimental power and model 1 according to changes in the a value. When a is 0.83, the total error is smallest. The total error (over the 40 experimental data points shown in Table 1) for model 1 ( 0.83t noncuttingPFrP ) was

PAGE 146

146 2135 W. This is smaller than the original model (t noncuttingPFrP ) error of 4673 W. The difference between model 1 and the experimental power is shown in Fig. A 3. The model 2 fitting form was t tnoncuttingPaFrbFP Figure A 4 shows the total error between the experimen tal power and model 2 according to changes in the a and b values. When a is 2.19 and b is 2.3, the total error is smallest. The total error (over 40 points) of model 2 ( 2.192.3t tnoncuttingPFrFP ) was 1798 W. This is smaller than the original model ( t noncuttingPFrP ) error of 4673 W. The difference between model 2 and the experimental power is shown in Fig. A 5. The model 3 fitting form was () t tnoncuttingPaFrbFP Figure A 6 shows the total error between the experimental power and model 3 power according to changes in the a and b values. When a is 1.26 and b is 0.43, the total error is smallest. The total error (over 40 points) of model 3 ( 1.260.43( )t tnoncuttingPFrFP ) is 891 W. This is smaller than the original model ( t noncuttingPFrP ) error of 4673 W. The difference between model 3 and the experimental power is shown in Fig. A 7. The model 4 fitting form was ttPaFrbF Figure A 8 shows the total error between the experimental power and model 4 according to changes in the a and b values. When a is 1.905 and b is 1.58, the total error is smallest. The total error (over 40 points) of model 4 (1.9051.58 tt PFrF ) is 563 W. This is smaller than the original model (t noncuttingPFrP) error of 4673 W. The difference between model 4 and the experimental power is shown in Fig. A 9.

PAGE 147

147 Table A 2 shows the error 40 2 ,exp, 11 40fittingierimentali iPP over all 40 experimental data points. It is seen that model 4 yields an 8.5 times reduction in the error compared to the original model. The linear model will be useful to estimate the cutting force.

PAGE 148

148 Table A 1 Measured cutting forces in the tangential direction Workpiece diameter Tangential force (N) (Condition 1) Workpiece diameter Tangential force (N) (Condition 2) Al Steel Al Steel 70 151 309 69.5 261 565 68.5 153 306 68 264 566 67 146 300 66.5 258 568 65.5 153 297 65 263 579 64 153 303 63.5 251 588 62.5 157 300 62 267 608 61 156 305 60.5 252 621 59.5 156 306 59 284 650 58 165 315 57.5 271 671 56.5 160 325 56 279 697 Average Tangential force 155 307 Average Tangential force 265 611 Table A 2 Total error for all power models Model # Model description a b Total error: 40 2 ,exp, 11 40fittingierimentali iPP cutting non tP r F P --117 W 1 cutting non tP r aF P 0.83 -53 W 2 cutting non t tP bF r aF P 2.19 2.3 45 W 3 ) (cutting non t tP F b r aF P 1.26 0.43 22 W 4 t tbF r aF P 1.905 1.58 14 W

PAGE 149

149 55 60 65 70 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 Expected Power and Experimental Power workpiece diameter (mm)Power(W) Condition 1(Al-experiment) Condition 1(Al-expected) Condition 1(Steel-experiment) Condition 1(Steel-expected) Condition 2(Al-experiment) Condition 2(Al-expected) Condition 2(Steel-experiment) Condition 2(Steel-expected) 56 58 60 62 64 66 68 70 150 200 250 300 350 400 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Expected power 56 58 60 62 64 66 68 70 350 400 450 500 550 600 650 700 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Expected power 56 58 60 62 64 66 68 70 300 350 400 450 500 550 600 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Expected power 56 58 60 62 64 66 68 70 800 850 900 950 1000 1050 1100 1150 1200 1250 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Expected power Figure A 1. Difference between experimental power and original po wer model

PAGE 150

150 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 Total error between experimental and fitting according to changes in a and b values total error (W) X: 0.83 Y: 0 Z: 2135 a Figure A 2 Total error between experiment and model 1 according to changing a values

PAGE 151

151 55 60 65 70 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 Model 1 Power and Experimental Power workpiece diameter (mm)Power(W) Condition 1(Al-experiment) Condition 1(Al-fitting) Condition 1(Steel-experiment) Condition 1(Steel-fitting) Condition 2(Al-experiment) Condition 2(Al-fitting) Condition 2(Steel-experiment) Condition 2(Steel-fitting) 56 58 60 62 64 66 68 70 160 180 200 220 240 260 280 300 320 340 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 350 400 450 500 550 600 650 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 300 350 400 450 500 550 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 800 850 900 950 1000 1050 1100 1150 1200 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Fitting model power Figure A 3 Model 1 and experimental power

PAGE 152

152 2 2.1 2.2 2.3 2.4 2.5 2 2.5 3 0 5000 10000 15000 total error (W)Total error between experimental and fitting according to changes in a and b values a b 2.05 2.1 2.15 2.2 2.25 2.3 1795 1800 1805 1810 Total error between experimental and fitting according to changes in a and b values total error (W) X: 2.19 Y: 2.3 Z: 1798a Figure A 4 Total error between experiment and model 2 according to changes in a and b values

PAGE 153

153 55 60 65 70 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 Model 2 and Experimental Power workpiece diameter (mm)Power (W) Condition 1(Al-experiment) Condition 1(Al-fitting) Condition 1(Steel-experiment) Condition 1(Steel-fitting) Condition 2(Al-experiment) Condition 2(Al-fitting) Condition 2(Steel-experiment) Condition 2(Steel-fitting) 56 58 60 62 64 66 68 70 160 180 200 220 240 260 280 300 320 340 360 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 350 400 450 500 550 600 650 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 300 350 400 450 500 550 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 600 700 800 900 1000 1100 1200 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Fitting model power Figure A 5 Model 2 and experimental power

PAGE 154

154 1 1.2 1.4 1.6 1.8 2 0.2 0.4 0.6 0.8 1 0 0.5 1 1.5 2 x 104 total error (W) Total error between experimental and fitting according to changes in a and b values a b 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 1.5 870 880 890 900 910 920 930 940 950 Total error between experimental and fitting according to changes in a and b values total error (W) X: 1.26 Y: 0.43 Z: 890.8a Figure A 6 Total error between experiment and model 3 according to changes in a and b values

PAGE 155

155 55 60 65 70 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 Model 3 and Experimental Power workpiece diameter (mm)Power(W) Condition 1(Al-experiment) Condition 1(Al-fitting) Condition 1(Steel-experiment) Condition 1(Steel-fitting) Condition 2(Al-experiment) Condition 2(Al-fitting) Condition 2(Steel-experiment) Condition 2(Steel-fitting) 56 58 60 62 64 66 68 70 160 180 200 220 240 260 280 300 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 350 400 450 500 550 600 650 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 300 350 400 450 500 550 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 800 850 900 950 1000 1050 1100 1150 1200 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Fitting model power Figure A 7 Model 3 and experimental power

PAGE 156

156 1.4 1.6 1.8 2 2.2 1 1.2 1.4 1.6 1.8 0 5000 10000 15000 total error (W)Total error between experimental and fitting according to changes in a and b values a b 1.78 1.8 1.82 1.84 1.86 1.88 1.9 1.92 1.94 1.96 562 563 564 565 566 567 568 569 570 Total error between experimental and fitting according to changes in a and b values total error (W) X: 1.905 Y: 1.58 Z: 563.2a Figure A 8 Total error between experiment and model 4 according to changes in a and b values

PAGE 157

157 55 60 65 70 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 Model 4 and Experimental Power workpiece diameter (mm)Power (W) Condition 1(Al-experiment) Condition 1(Al-fitting) Condition 1(Steel-experiment) Condition 1(Steel-fitting) Condition 2(Al-experiment) Condition 2(Al-fitting) Condition 2(Steel-experiment) Condition 2(Steel-fitting) 56 58 60 62 64 66 68 70 160 180 200 220 240 260 280 300 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 350 400 450 500 550 600 650 Condition 1:depth of cut=0.5mm,feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 300 350 400 450 500 550 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Al workpiece diameter (mm)Power(W) Experimental power Fitting model power 56 58 60 62 64 66 68 70 700 750 800 850 900 950 1000 1050 1100 1150 1200 Condition 2:depth of cut=1mm, feed per revolution=0.2mm/rev,spindle speed=500rpm,Steel workpiece diameter (mm)Power(W) Experimental power Fitting model power Figure A 9 Model 4 and experimental power

PAGE 158

158 APPENDIX B ANALYSIS OF OKUMA LC40 POWER DATA (SECOND DAY SET) The power measurement on Okuma LC 40 was performed during two days. The f irst day s measurement s are in chapter 5.3. In this appendix the second day s power measur ement data is provided. (Table B 1 and Figures B 1~3)

PAGE 159

159 Table B 1 Maximum values of faci ng operations, mean values and s tandard deviations of turning operations Part # facing Max. facing Max. OD ID mean OD mea n OD mea n OD mea n OD ID mean OD ID SD OD SD OD SD OD SD OD ID SD 1 2.954 3.776 3.0413 2.7194 2.5462 2.135 1.6379 0.2126 0.0255 0.021 0.02 0.0659 2 2.222 3.7398 3.0645 2.7288 3.0365 2.4559 1.6705 0.4232 0.3223 0.0282 0.024 0.192 3 3.176 3.786 3.0109 2.6946 2.515 2.0914 1.6172 0.1379 0.0561 0.0262 0.0276 0.049 4 2.388 3.785 2.9866 2.6994 2.5033 2.0892 1.6448 0.041 0.0244 0.0268 0.0245 0.0283 5 2.405 3.805 2.976 2.6858 2.5067 2.0745 1.6336 0.1614 0.0425 0.0225 0.0237 0.0411 6 2.365 3.222 2.5392 2.3414 2.18 08 1.8331 1.4797 0.2561 0.0575 0.0218 0.0202 0.066 7 2.275 3.415 2.5535 2.3226 2.1813 1.8285 1.4697 0.1588 0.0268 0.0183 0.0182 0.0253 8 3.635 3.405 2.5631 2.335 2.179 1.8429 1.4626 0.1425 0.0271 0.022 0.0203 0.0307 9 2.284 3.826 2.9746 2.6887 2.4935 2. 0718 1.625 0.0414 0.0303 0.0229 0.0235 0.0333 10 2.275 3.825 3.0042 2.6832 2.4985 2.0656 1.6195 0.1092 0.0235 0.0201 0.0212 0.026 11 1.735 3.866 3.0469 2.6989 2.5024 2.0796 1.6162 0.1898 0.0656 0.0291 0.0235 0.0617 12 2.163 3.857 3.0552 2.6937 2.5124 2. 0742 1.6316 0.0527 0.0241 0.0235 0.0199 0.0243 13 2.592 3.895 2.9907 2.6866 2.4943 2.073 1.6138 0.1399 0.0344 0.022 0.0212 0.0412 14 3.435 3.918 2.9901 2.7037 2.5103 2.0758 1.6371 0.0988 0.0219 0.0214 0.0234 0.0265 15 2.775 3.903 2.9756 2.6937 2.5043 2.0745 1.6273 0.1308 0.0205 0.0205 0.0194 0.0311 16 2.418 3.852 2.9579 2.6918 2.4996 2.077 1.5966 0.1567 0.1586 0.0218 0.0202 0.119 17 1.615 3.946 3.0011 2.6958 2.4987 2.0873 1.6163 0.1389 0.0254 0.0225 0.022 0.0285 18 2.257 4.007 2.9942 2.6909 2.4935 2.0 848 1.6012 0.0433 0.0334 0.0237 0.0219 0.0341 19 2.695 3.975 2.947 2.6787 2.4874 2.0938 1.6098 0.0839 0.0285 0.0215 0.0258 0.0324 20 2.161 4.008 2.9478 2.6877 2.4992 2.0829 1.6022 0.055 0.0178 0.0209 0.0196 0.0282 21 2.03 4.055 3.0238 2.6934 2.4999 2.07 84 1.6058 0.1291 0.0463 0.0272 0.0231 0.0585 22 2.528 4.04 2.9856 2.6755 2.4963 2.0793 1.5913 0.1433 0.0458 0.0526 0.0205 0.0647 23 3.385 4.138 2.9453 2.6939 2.4883 2.0734 1.6063 0.214 0.1022 0.0438 0.0189 0.1097 24 2.278 4.268 2.9191 2.6426 2.4643 2.04 27 1.6057 0.1884 0.0247 0.0203 0.0203 0.0318 25 4.099 4.356 2.9529 2.6389 2.4684 2.0407 1.5968 0.1639 0.0207 0.021 0.02 0.0346 26 2.555 4.097 2.9391 2.6308 2.4545 2.0396 1.6021 0.0627 0.0204 0.0207 0.0171 0.0272 27 3.345 4.298 2.9564 2.6414 2.4687 2.053 4 1.5651 0.2032 0.0231 0.0192 0.0182 0.045 28 2.528 3.882 2.9357 2.6577 2.4815 2.0541 1.6208 0.1065 0.0189 0.0155 0.0188 0.0228 29 2.301 3.865 2.914 2.6492 2.4719 2.0509 1.611 0.0454 0.0252 0.0184 0.0193 0.03

PAGE 160

160 0 5 10 15 20 25 30 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean, maximum value of measured power Max. of first facing Max. of second facing Mean of firsrt OD-ID Mean of second OD Mean of third OD Mean of fourth OD Mean of finishing OD-ID Figure B 1 Maximum values of facing operations, mean values of turning operations

PAGE 161

161 0 5 10 15 20 25 30 -0.2 -0.1 0 0.1 0.2 0.3 0.4 Part numberVoltage (V)Standard deviation of measured power Standard deviation of first OD-ID Standard deviation of second OD Standard deviation of third OD Standard deviation of fourth OD Standard deviation of finishing OD-ID Figure B 2 Standard deviations of turning operations

PAGE 162

162 0 5 10 15 20 25 30 0 1 2 3 4 5 6 7 Part numberVoltage (V)Max. power of first facing part A) 0 5 10 15 20 25 30 0 1 2 3 4 5 6 7 Part numberVoltage (V)Max. power of second facing part B) 0 5 10 15 20 25 30 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of first OD-ID C) 0 5 10 15 20 25 30 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of second OD D) 0 5 10 15 20 25 30 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of third OD E) 0 5 10 15 20 25 30 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of fourth OD F)

PAGE 163

163 0 5 10 15 20 25 30 0 1 2 3 4 5 6 7 Part numberVoltage (V)Mean power of finishing OD-ID G) 0 5 10 15 20 25 30 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of fisrt OD-ID H) 0 5 10 15 20 25 30 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of second OD I) 0 5 10 15 20 25 30 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of third OD J) 0 5 10 15 20 25 30 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of fourth OD K) 0 5 10 15 20 25 30 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Part numberVoltage (V)Standard deviation of finishing OD-ID L) Figure B 3 A) Max value of first facing, B) Max value of second facing, C) Mean value of first turning (OD and ID are simultaneous cut), D) Mean value of third turning (OD cut), E) Mean value of second turning (OD cut), F) Mean value of fourth turning (OD cut), G) Mean value of finishing turning, H) Standard variation of second turning (OD cut), I) Standard variation of first turning (OD and ID are simultaneous cut), J) Standard variation of third turning (OD cut), K) Standard variation of fourth turning (OD cut), and L) Standard variation of finishing turning

PAGE 164

164 APPENDIX C SENSOR OUTPUT To check the power sensor output, the current, voltage and sensor output were measured. The UPC power sensor from Load Controls Inc., was used to measure the DC power. Figure C 1 shows the connection of power sensor and DC motor, and also shows the turns (4 tur ns). Scaling for the analog output is the adjusted H.P. (1K ohm per H.P.) divided by the number of turns. For example, when H.P. scale is 5 K ohms and 4 turns, the full scale is 1.25 H.P. First test was perform ed under power sensor had 5 K Ohm and 1 turn a s full scale Figure C 2 shows the results. Second test was perform ed under power sensor had 20 K Ohm and 6 turns as full scale Figure C 3 shows the result. As shown in Fig. C 2 and C 3 power sensor was good to measure the real power.

PAGE 165

165 Figure C 1 Con nection of UPC and D.C. motor and turns (based on Load Controls Inc.)

PAGE 166

166 0 1 2 3 4 5 6 7 8 -100 0 100 200 300 400 500 600 700 800 900 1000 Test No.Voltage (V), current(A), and power(W) sensor output (V) current (A) voltage (V) power (W) = current x voltage power(sensor)(w) =sensor output x 746 x 5 K Ohm / 1 turn(S) / 10 Figure C 2 Sensor output under 5 K Ohm and 1 turn of sensor set up 0 1 2 3 4 5 6 7 8 9 10 11 -100 0 100 200 300 400 500 600 700 800 900 1000 Test No.Voltage (V), current(A), and power(W) sensor output (V) current (A) voltage (V) power (W) = current x voltage power(sensor)(w) =sensor output x 746 x 5 K Ohm / 1 turn(S) / 10 Figure C 3 Sensor output under 20 K Ohm and 6 turn of sensor set up

PAGE 167

167 LIST OF REFERENCES Abu Zahra, N.H., and Nay feh, T.H., 1997, Calibrated method for ultrasonic online monitoring of gradual wear during turning operations, International Journal of Machine Tools and Manufacture, Vol. 37, pp. 1475 1484. Atli, A. V., Urhan, O., Ertrk, S., and Snmez M., 2006, A computer visionbased fast approach to drilling tool condition monitoring, IMechE Vol. 220 Part B: J. Engineering Manufacture pp14091415. Azouzi, R., and Guillot, M., 1997, OnLine prediction of surface finish and dimensional deviation in turning using neural network based sensor fusion, Int. J. Mach. Tools Manuf., 37, pp. 1201 1217. Bahr, B., Motavalli, S., and Arfi, T., 1997, Sensor fusion for monitoring machine tool conditions, International Journal Computer Integrated Manufacturing, 10, pp. 314 323. Bayramoglu, M., and Dngel, ., 1998, A systematic investigation on the use of force ratios in tool condition monitoring for turning operations, Trans. Institute of Measurement and Control, Vol. 20 (2), pp. 92 97. Blum, T., and Inasaki, I., 1990, A study on acoustic emission from the orthogonal cutting process, ASME Trans. Journal of Engineering for Industry, Vol. 112 (3), pp. 203 211. Boothroyd, G., 1975, Fundamentals of Metal Machining and Machine Tools, McGraw Hill, New York. Burke, L.I., 1 989, Automated identi an application of self organising neural networks, Ph.D. thesis, Department of Industrial Engineering and Operations Research, UC at Berkeley, USA. Byrne, G., Dornfeld, D., Inasaki I., Ketteler, G., Knig, W., and Teti, R., 1995, Tool condition monitoring (TCM) the status of research and industrial application, Annals of CIRP, Vol. 44 (2), pp. 541 567. Chen X. and Li B., 2007, Acoustic emission method for tool condition monit oring based on wavelet analysis, Int J Adv Manuf Technol 33, pp968976. Cho, S., Binsaeid, S., and Asfour, S., 2009, D esign of multisensor fusionbased tool condition monitoring system in end milling, Int J Adv Manuf Technol. Choi, G.S., Wang, Z.X., Do rnfeld, D.A., and Tsujino, K., 1990, Development of an intelligent online tool wear monitoring system for turning operations,, Proc. USA Japan Symposium on Flexible Automation, A Paci ence ISCIE Kyoto, pp. 683 690.

PAGE 168

168 Choi, D., Kwon, W.T., an d Chu, C.N., 1999, Real time monitoring of tool fracture in turning using sensor fusion, International Journal of Advanced Manufacturing Technology, Vol. 15 (5), pp. 305 310. Chow, J.G., and Wright, P.K., 1988, Online estimation of tool/chip interface temperatures for a turning operation, ASME Trans. Journal of Engineering for Industry, Vol. 110, pp. 56 64. Constantinides, N., and Bennett, S., 1987, An investigation of methods for online estimation of tool wear, International Journal of Machine Tools and Manufacture, Vol. 27 (2), pp. 225 237. Cook, N. H., 1980, Tool Wear Sensors, Wear, 62, pp. 49 57. Cuppini, D., DEnrrico, G., and Rutelli, G., 1986, Tool image processing with application to unmanned metal cutting, a com puter vision system f or wear sensing and failure detection, SPIE, Vol. 701, pp. 416 422. Dan, L., and Mathew, J., 1990, Tool wear and failure monitoring techniques for turninga review, International Journal of Machine Tools and Manufacture, Vol. 30 (4), pp. 579 598. Dimla Snr. D.E., 1998, Multivariate tool condition monitoring in a metal cutting operation using neural networks, Ph.D. thesis, School of Engineering and the Built Environment, The University of Wolverhampton, UK. Dimla Snr. D.E., 1999, Tool wear monitoring using cutting force measurements, 15th NCMR: Advances in Manufac turing Technology XIII, University of Bath, 6 8 September, pp. 33 37. Dimla Snr. D.E., 2000, Sensor signals for tool wear monitoring in metal cutting operations a review of methods, Int ernational Journal of Machine Tools & Manufacture, Vol. 40, pp. 1073 1098 Dimla Snr. D.E., and Lister, P.M., 2000, On line metal cutting tool condition monitoring I: Force and vibration analyses, International Journal of Machine Tools and Manufacture, Vol. 40 (5), pp. 739 768. Dornfeld, D.A., 1990, Neural network sensor fusion for tool condition monitoring, Annals of the CIRP, Vol. 39 (1), pp. 101 105. Dornfeld, D.A., 1992, Application of acoustic emission techniques in manufacturing, NDT and E In ternational, Vol. 25 (6), pp. 259 269. Dornfeld, D.A., 1994, In process recognition of cutting states, JSME International Journal, Series C: Dynamics Control, Vol. 37 (4), pp. 638 650.

PAGE 169

169 Du, R., Zhang, B., Hungerford, W., and Pryor, T., 1993, Tool cond ition monitoring and compensation in finish turning using optical sensor, 1993 ASME WAM, Symposium of Mechatronics, DSC Vol. 50/PEDVol. 63, pp.245 253. El Gomayel, J.I., and Bregger, K.D., 1986, Online tool wear sensing for turning operations, ASME T rans. Journal of Engineering for Industry, Vol. 108, pp. 44 47. Elanayar, S.V.T., Shin, Y.C., and Kumara, S., 1990, Machining condition monitoring for automation using neural networks, Monitoring and Control of Manufacturing Processes (Winter Annual M eeting of the ASME), PED, 44, pp. 85 100 ElWardany, T.I., Gao, D., and Elbestawi M.A., 1996, Tool condition monitoring in drilling using vibration signature analysis, International Journal of Machine Tools and Manufacture, Vol. 36 (6), pp. 687 711. Er tekin, Y. M., Kown, Y., and Tseng, T L., 2003, Identification of common sensory features for the control of CNC Milling operations under varying cutting conditions, Int. J. Mach. Tools Manuf., 43, pp. 897 904. Fan, Y. Y., and Du, R., 1996, A laser di ffraction method for rotating tool condition monitoring, ASME J. Eng. Ind., 118, No. 4, pp. 664 667. Finnie, I., and Shaw, M. C., 1956, The friction process in metal cutting, Trans. ASME, Vol. 78, p. 1649. Gould, L., 1988, Sensing tool and drive element conditions in machine tools, Sensor, pp. 5 13. Grabec, I., 1988, Chaotic dynamics of the cutting process, International Journal of Machine Tools and Manufacture, Vol. 28 (1), pp. 9 32. Holmbom, P., Pedersen, O., Sandell, B., and Lauber, A., 1989, Fusing sensor systems: promises and problems, Sensor Reviews, Vol. 9 (3), pp. 143 152. Huggins, L. F., 1975, Analysis and interpretation, American Society of Agricultural Engineering, First Edition, pp. 1503~04 Jemielniak, K., and Otman, O., 1998, Tool failure detection based on analysis of accoustic emission signals, Journal of Material Processing Technology, Vol. 76, pp. 192 197. Jetley, S.K., and Gollajesse, A., 1994, Measuring tool wear using magnetism, Proceedings: Japan USA Symposium Flex ib le Automation, A Pacific Rim Conference, July 11 18, Kobe, Japan, pp. 345 347. Jiang, C.Y., Zhang, Y.Z., and Xu, H.J., 1987, Inprocess monitoring of tool wear stage by the frequency bandenergy method, Annals of the CIRP, Vol. 36 (1), pp. 45 48.

PAGE 170

170 Kakade, S., Vijayaraghavan, L., and Krishnamurthy, R., 1994, Inprocess tool wear and chip form monitoring in face milling operation using acoustic emission, Journal of Material Processing Technology, Vol. 44, pp. 207 214. Karthik, A., Chandra, S., Ramamoorthy, B., and Das, S., 1997, 3D tool wear measurement and visualisation using stereo imaging, Int. J. Mach. Tools Manuf., 37, pp. 1573 1581. Khraisheh, M.K., Pezeshki, C., and Bayoumi, A.E., 1995, Time series based analysis for primary chatter in metal cutting, Journal of Sound and Vibration, Vol. 180 (1), pp. 67 87. Kim, J.S., and Lee, B.H., 1991, An analytical model of dynamic cutting forces in chatter vibration, International Journal of Machine Tools and Manufacture, Vol. 31 (3), pp. 371 381. Kim J D., and Choi, I H., 1996, Development of a tool failure detection system using multi sensors, Int. J. Mach. Tools Manuf., 36, pp. 861 870. Kitagawa, T., Maekawa, K., Shirakashi, T., and Usui, E., 1988, Analytical prediction of flank wear of carbide tools in turning plain carbon steels (Part 1), Bull. Japan Soc. Precision Engineering, Vol. 22 (4), pp. 263 269. Ko, T.J., and Cho, D.W., 1994, Cutting state monitoring in milling by a neural network, International Journal of Machine Tools and Manufacture, Vol. 34 (5), pp. 659 676. Knig, W., Kutzner, K., and Schehl, U., 1992, Tool monitoring of small drills with acoustic emission, International Journal of Machine Tools and Manufacture, Vol. 32 (4), pp. 487 493. Kurada, S., and Bradley, C., 1997, A review of machine vision sensors for tool condition monitoring, Computers in Industry, Vol. 34 (1), pp. 55 72. Lee, G., Kim, K., and Jeon, D., 2002, Tool condition monitoring based on wavelet transform, International Conference on Control, Automation and System, Muju, Korea. Lee, K.S., Lee, L.C., and Teo, S.C., 1992, Online tool wear monitoring using a PC, Journal of Material Processing Technology, Vol. 29, pp. 3 13. Lee, L.C., Lam, K.Y., and Liu, X.D., 1994, Characterisation of tool wear and failure, Journal of Materials Processing Technology, Vol. 40, pp. 143 153. Lee, L.C., Lee, K.S., and Gan, C.S., 1989, On the correlation between dynamic cutting force and tool wear, International Journal of Machine Tools and Manufacture, Vol. 29 (3), pp 295 303. Lezanky, P., and Rafalowicz, J., 1993, An intelligent monitoring system for cylindrical grinding, CIRP Ann., 42, pp. 393 396.

PAGE 171

171 Lin, J., 1995, Inverse estimation of the tool work interface temperature in end milling, International Journal of Machine Tools and Manufacture, Vol. 35 (5), pp.751 760. Lister, P.M., 1993, Online measurement of tool wear. Ph.D. thesis, Manufacturing and Machine Tools Division, Depart ment of Mechanical Engineering, UMIST, Manchester, UK. Lister, P.M., and Barro w, G., 1986, Tool condition monitoring systems, Proceedings International 26th MATADOR Conference, Manchester, UK, pp. 271 288. Lou, K N., and Lin, C J., 1996, An intelligent sensor fusion system for tool monitoring on a machining center, Proceeding of the 1996 IEEE/SICE/RSJ,International Conference on Multisensor Fusion and Integration for Intelligent Systems, pp. 208 214. Marques, M.J.M., and Mesquita, R.M.D., 1991, Monitoring the wear of sintered highspeed steel tools, Journal of Material Proc essing Technology, Vol. 25, pp. 195 213. Martin, K.F., 1994, A review by discussion of condition monitoring and fault diagnosis in machine tools, International Journal of Machine Tools and Manufacture, Vol. 34 (4), pp. 527 551. Martin, K.F., Brandon, J .A., Grosvenor, R.I., and Owen, A., 1986, A comparison of inprocess tool wear measurement methods in turning, Proceedings International 26th MATADOR Conference, Manchester, UK, pp. 289 296. Matsumura, T., Obikawa, T., Shirakashi T., and Usui, E., 1993, Automonus turning operation planning with predicting tool wear and surface roughness, Transactions of the North American Manufacturing Institution of SME, Vol. 21, pp. 359366. Matsumura, T., Shirakashi, T., and Usui, E., 2008, Identification of wear c haracteristics in tool wear model of cutting process, International Journal of Material Forming, Vol. 1, pp. 555558. Moriwaki,T., and Tobito, M., 1990, A new approach to automatic detection of life of coated tool based on acoustic emission measurement, ASME Trans. Journal of Engineering for Industry, Vol. 112 (3), pp. 212 218. Noori Khajavi, A., and Komanduri, R., 1995, Frequency and time domain analyses of sensor signals in drilling I. Correlation with drill wear, International Journal of Machine Tools and Manufacture, Vol. 35 (6), pp. 775 793. Novak, A., and Wiklund, H., 1992, Reliability of tool wear monitoring, Annals of the CIRP, Vol. 42 (1), pp. 63 66. Oguamanam, D.C.D., Raafat, H., and Taboun, S.M., 1994, A machine vision system for wea r monitoring and breakage detec tion of singlepoint cutting tools, Computers in Industrial Engineering, Vol. 26 (3), pp. 575 598.

PAGE 172

172 Opitz, H. and Knig, W., 1967, On the wear of cutting tools, Adv. Mach. Tool Des. Res., Vol. 1, p. 173. Oraby, S.E., and Hayhurst D.R., 1991, Development of models for tool wear force relationships in metal cutting, International Journal of Machine Tools and Manufacture, Vol. 33 (2), pp. 125 138. P. Bhattacharyya, D. Sengupta, S. Mukhopadhyay and A. B. Chattopadhyay, 2008, On line tool condition monitoring in face milling using current and power signals, International Journal of Production Research, Vol. 46, No. 4, pp 1187 1201. Parra, L., Deco, G., and Miesbach, S., 1996, Statistical independence and novelty detection with information preserving non linear maps, Neural Computation, Vol. 8, pp. 260 269. Patankar S. V., 1980, Numerical Heat Transfer and Fluid Flow, Hemisphere Publishing Corporation, Washington, DC. Prickett, P.W., and Johns, C., 1999, An overview of approaches to end milling tool monitoring, International Journal of Machine Tools and Manufacture, Vol. 39 (1), pp. 105 122. Purushothaman, S., and Srinivasa, Y.G., 1994, A back propagation algorithm applied to tool wear monitoring, International Journal of Machine Tools and Manufacture, Vol. 34 (5), pp. 625 631. Rabinowicz, E., Dunn, L. A., and P. G. Russell, 1961, A study of abrasive wear under threebody conditions, Wear, Vol. 4, pp. 345. Radulescu, R., and Kapoor, S.G., 1994, An analytical model f or prediction of tool temperature fields during continuous and interrupted cutting, ASME Trans. Journal of Engineering for Industry, Vol. 116 (2), pp. 135 143. Raman, S., Shaikh, A., and Cohen, P.H., 1992, A mathematical model for tool temperature sensi ng, ASME Computational Methods in Materials Processing, PED 61, pp. 181 193. Rangwala, S., 1988, Machining process characterisation and intelligent tool condition monitoring using acoustic emission signal analysis, Ph.D. thesis, Mechanical Engineering Department, UC at Berkeley, USA. Rangwala, S., and Dornfeld, D., 1987, Integration of sensors via neural networks for detection of tool wear states, Winter Annual Meetings of the ASME, PED 25, pp. 109 120. Rangwala, S., and Dornfeld, D., 1990, Sensor int egration using neural networks for intelligent tool condition monitoring, ASME Trans. Journal of Engineering for Industry, Vol. 112 (3), pp. 219 228.

PAGE 173

173 Ravindra, H.V., Srinivasa, Y.G., and Krishnamurthy, R., 1993, Modelling of tool wear based on cutting f orces in turning, Wear, Vol. 169, pp. 25 32. Roget, J., Souquet, P., and Gsib, N., 1988, Application of acoustic emission to the automatic monitoring of tool condition during machining, ASNDT Materials Evaluation, Vol. 46, pp. 225 229. Rotberg, J., Br aun, S., and Lenz, E., 1987, Mechanical signature analysis in interrupted cutting, Annals of the CIRP, Vol. 36 (1), pp.249 252. Rotberg, J., Braun, S., and Lenz, E., 1989, Vibration generation models for cutting tool monitoring, Diagnostics, Vehicle D ynamics and Special Topics ASME Design Engineering Division, Vol. 18 (5), pp. 1 6. Sarwar, M., Li, J., Penlington, R., and Ahmed W., 1996, Development of thermal imaging systems for metal cutting appli cations, Proceedings Advanced Manufacturing Process es Systems and Technologies Conference (AMPST 96), Bradford, UK, pp. 361 369. Schey, J.A., 1987, Introduction to Manufacturing Processes (second Ed.), McGraw Hill Book Company, London, Chapter 8. Shaw, M. V., and Dirke, S. O., 1956, On the wear of cutti ng tools, Macrotecnic, Vol. 10(4), p. 187. Shaw, M.C., 1954, Metal cutting principles, Massachusetts Institute of Technology Press, Cambridge, MA, p. 111. Shaw, M.C., 1984, Metal Cutting Principles, Clarendon Press, Oxford, Chapter 3. Shaw, M.C., 1988, Temperature in cutting, ASME Proceedings Symposium on Thermal Aspects in Manufacturing, Vol. 30, pp. 133 143. Shi, T., and Ramalingam, S., 1990, Real time Meeting of the ASME, PED 43, pp. 157 170. Shiraishi, M., 1988, Scope of in process measurement, monitoring and control techniques in machining processes Part 1: In process techniques for tools, Precision Engineering, Vol. 10 (4), pp. 179 189. Silva, R. G., Wilcox, S., and Araujo, A. A.M.M., 2007, Multi sensor condition monitoring using spiking neuron networks, IADIS International Conference Applied Computing. Stephenson, D.A., and Ali A., 1990, Tool temperatures in interrupted metal cutting, Winter Annual Meeting of the ASME, PED 43, pp. 261 281. Tanner, R., and Loh, N.K., 1994, A taxonomy of multi sensor fusion, Journal of Manufacturing Systems, Vol. 11 (5), 314 325.

PAGE 174

174 Tansel, I.N., and McLaughlin, C., 1991, Online monitoring of tool breakage with unsupervised neural network, Trans. NAMRI/SME, pp. 364 370. Tarng, Y.S., Hseih, Y.W., and Hwang, S.T., 1994, Sensing tool breakage in face milling with a neural network, International Journal of Machine Tools and Manufacture, Vol. 34 (3), pp. 341 350. Teti, R., 1995, A review of tool condition monitoring liter ature database, Annals of the CIRP, Vol. 44 (2), pp. 659 666. Tlusty, G., 2000, Manufacturing Processes and Equipment, Prentice Hall, Upper Saddle River, NJ, USA, Chapter 8. Trent, E.M., 1977, Metal Cutting, Butterworths, London. Trigger, K. J., and Chao, B. T., 1956, The mechanism of crater wear of cemented carbide tools, Trans. ASME, Vol. 78 (5), p. 1119. Uehara, K., Takenshita, H., Nishina, K., and Sakurai, M., 1976, On the generating mechanism of wear particles of carbide cutting tools, Proc. 2nd Int. Conf. on Production Engineering, Tokyo, 1976, Japan Society of Precision Engineering, Tokyo, p. 203. Usui, E., Shirakashi, S., and T. Kitagawa, 1984, Analytical prediction of tool wear, Wear, Vol. 100, pp.129151. Usui, E., Shirakashi, T., and T. K itagawa, 1978, Analytical prediction of three dimensional cutting process (Part 3), Journal of Engineering for Industry, Transactions of ASME, Vol. 100, 236243. Warnecke, G., Jenewein, A., and Reinfelder, A., 1990, Tool monitoring based on process ide ntification, monitoring and control for manufacturing processes, Winter Annual Meeting of the ASME, Dallas, TX, USA, pp. 43 55. Weck, M., 1984, Metrological Analysis and Performance Tests, Handbook of Machine Tools (fourth ed.), Vol. 4, John Wiley and Sons, Chichester. Wong, Y.S., Nee, A.Y.C., Li, X.Q., and Reisdorf, C., 1997, Tool condition monitoring using laser scatter pattern, Journal of Materials Processing Technology, Vol. 63, pp. 205 210. Yan, D., El Wardany, T.I., and Elbestawi, M.A., 1995, A multi sensor strategy for tool failure detection in milling, Inter national Journal of Machine Tools and Manufacture, Vol. 35 (3), pp. 383 398. Yao, Y., and Fang, X.D., 1992, Modelling of multivariate time series for tool wear estimation in finishturn ing, International Journal of Machine Tools and Manufacture, Vol. 32 (4), pp. 495 508.

PAGE 175

175 Yao, Y., Fang, X.D., and Arndt, G., 1990, Comprehensive tool wear esti mation in machining via multivariate timeseries analysis of 3D cutting forces, Annals of the CIRP, Vol. 39 (1), pp. 57 60. Yao, Y., Fang, X.D., and Arndt, G., 1991, On line estimation of groove wear in the minor cutting edge for finish machining, Annals of the CIRP, Vol. 40 (1), pp. 41 44. Zhang, D., Dai, S., Han, Y., and Chen, D., 1994, Online monitoring of tool breakage using spindle current in milling, 1st Asia Pacific and 2nd Japan China International Conference Progress of Cutting and G rinding, Shanghai, China, pp. 270 276. Zheng, S.X., McBride, R., Barton, J.S., Jones, J.D.C., Hale, K.F., and Jones, B.E., 1992, Intrinsic optical fibre sensor for monitor ing acoustic emission, Sensors and Actuators,, Physical, Vol. 31, pp. 110 114. Z hou, J.M., Andersson, M., and Sthl, J.E., 1995, A system for monitoring cutting tool spontaneous failure based on stress estimation, Journal of Materials Processing Technology, Vol. 48, p. 231.

PAGE 176

176 BIOGRAPHICAL SKETCH Gun Lee was born in Pusan and raised in Seoul, Korea. He completed his Bachelor of Science degree in the m echanical e ngineering department and Bachelor of Art s in business administration at Sogang University in Seoul, Korea. He then finished Master of Science degree in the mechanical engineering department at Sogang University. After finishing his master s degree, Gun decided to continue his education towards a Ph.D. degree in the United States. He joined to his Ph.D. program at the Machine Tool Research Center (MTRC) in mechanical and aerospace engineering d epartment. Gun has worked on tool wear monitoring research. After graduation, Gun will continue h is career as an engineer in industry. His field s of interest include control s, image processing, and automation in manufacturing.