11
~ Pergamon Int, J. Mach. Tools Manufact. Vol. 35. No. 10, pp. 1385-1395, 1995 Copyright ~ 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0890-6955/9559.50 + .00 0890--6955( 94)00126-x AN INTEGRATED ULTRASONIC SENSOR FOR MONITORING GRADUAL WEAR ON-LINE DURING TURNING OPERATIONS TAYSIR H. NAYFEH,t~ OSAMA K. EYADA§ and JOHN C. DUKE, JR[[ (Received 1 September 1993; in final form 6 October 1994) Abstract--The condition of the tool and the cutting process are essential inputs to any productivity improvements through process optimization in conventional and unmanned machining. Tool replacement and tool wear compe.nsation strategies, which are based on prior experience and/or tool history are, in general, under performing. Currently, the methods of tool condition monitoring are either time consuming, as in the case of off-Line direct measurements of the tool, or are modestly successful, as in the case of the on-line indirect measurements, such as forces or acoustic emissions. This in part is due to the lack of suitable sensors and/or exact dynamic models, which relate the indirect measurements to the actual tool condition. This paper describe, s a promising ultrasonic method for on-line direct measurement of gradual wear in turning operations. All integrated (transmit and receive) single ultrasonic transducer operating at a frequency of 10 MHz is placed in contact with the tool. The change in the amount of the reflected energy from the nose and the flanks of the tool can be related to the level of gradual wear and the mechanical integrity of the tool. The experimental results show that under laboratory conditions, a correlation exists between the ultrasonic measurement and gradual wear and that it is tool dependent. INTRODUCTION Machining operations, conventionally or numerically controlled, present many opport- unities for productivity improvements including maximizing the metal removal rates, reducing tool failures, and improving the quality of the finished products. In either form of machining, the condition of the cutting tool and the dynamics of the overall maching process are the two dominant unknowns that must be determined in order to optimize the operation. The needs for on-line tool and process condition monitoring are becoming critical with the widespread use of CNC and the move toward unmanned machining centers. In these settings, it has been shown that 6.8% of all the down time is attributed exclusively to tool failures [1] and overall the processes are under per- forming by 20-80% depending on the part's geometry and complexity [3]. Tool condition monitoring can be performed on-line or off-line by either direct or indirect means. The on-line designation for a sensing method refers to the fact that it is performed while metal is being removed without the need for interrrupting the process. Off-line methods can be performed on the machine or away from the machine. In either case, these methods require either scheduling idle time for the measurements or actually interrupting the process. The indirect methods measure one or more of the parameters associated with the cutting operation, such as tool forces [4-16], acoustic emission [17-28], vibration [29-34], etc. The tool condition is then inferred from the values of the parameters. Due to the complexity of machining operations, there are no exact mathematical models that relate the cutting parameters to the actual tool condition. For example, tool forces, temperature, vibration, etc. may be changing due to tool wear, harder material, improper feed, speed or depth of cut. tDepartment of Industrial Engineering, Cleveland State University, Euclid Avenue at East 24th Street, Cleveland, OH 44115, U.S.A. (216) 687-6967. Author for correspondence. § Department of Industrial and Systems Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, U.S.A. [[ Department of Engineering Science and Mechanics/Materials Science and Engineering, Virginia Polytech- nic Institute and State University, Blacksburg, VA 24061, U.S.A. 35.~o-D 1385

An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

Embed Size (px)

Citation preview

Page 1: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

~ Pergamon Int, J. Mach. Tools Manufact. Vol. 35. No. 10, pp. 1385-1395, 1995 Copyright ~ 1995 Elsevier Science Ltd

Printed in Great Britain. All rights reserved 0890-6955/9559.50 + .00

0890--6955( 94)00126-x

A N I N T E G R A T E D U L T R A S O N I C S E N S O R F O R M O N I T O R I N G

G R A D U A L W E A R O N - L I N E D U R I N G T U R N I N G O P E R A T I O N S

TAYSIR H. NAYFEH,t~ OSAMA K. EYADA§ and JOHN C. DUKE, JR[[

(Received 1 September 1993; in final form 6 October 1994)

Abstract--The condition of the tool and the cutting process are essential inputs to any productivity improvements through process optimization in conventional and unmanned machining. Tool replacement and tool wear compe.nsation strategies, which are based on prior experience and/or tool history are, in general, under performing. Currently, the methods of tool condition monitoring are either time consuming, as in the case of off-Line direct measurements of the tool, or are modestly successful, as in the case of the on-line indirect measurements, such as forces or acoustic emissions. This in part is due to the lack of suitable sensors and/or exact dynamic models, which relate the indirect measurements to the actual tool condition.

This paper describe, s a promising ultrasonic method for on-line direct measurement of gradual wear in turning operations. All integrated (transmit and receive) single ultrasonic transducer operating at a frequency of 10 MHz is placed in contact with the tool. The change in the amount of the reflected energy from the nose and the flanks of the tool can be related to the level of gradual wear and the mechanical integrity of the tool. The experimental results show that under laboratory conditions, a correlation exists between the ultrasonic measurement and gradual wear and that it is tool dependent.

INTRODUCTION

Machining operations, conventionally or numerically controlled, present many opport- unities for productivity improvements including maximizing the metal removal rates, reducing tool failures, and improving the quality of the finished products. In either form of machining, the condition of the cutting tool and the dynamics of the overall maching process are the two dominant unknowns that must be determined in order to optimize the operation. The needs for on-line tool and process condition monitoring are becoming critical with the widespread use of CNC and the move toward unmanned machining centers. In these settings, it has been shown that 6.8% of all the down time is attributed exclusively to tool failures [1] and overall the processes are under per- forming by 20-80% depending on the part's geometry and complexity [3].

Tool condition monitoring can be performed on-line or off-line by either direct or indirect means. The on-line designation for a sensing method refers to the fact that it is performed while metal is being removed without the need for interrrupting the process. Off-line methods can be performed on the machine or away from the machine. In either case, these methods require either scheduling idle time for the measurements or actually interrupting the process.

The indirect methods measure one or more of the parameters associated with the cutting operation, such as tool forces [4-16], acoustic emission [17-28], vibration [29-34], etc. The tool condition is then inferred from the values of the parameters. Due to the complexity of machining operations, there are no exact mathematical models that relate the cutting parameters to the actual tool condition. For example, tool forces, temperature, vibration, etc. may be changing due to tool wear, harder material, improper feed, speed or depth of cut.

tDepartment of Industrial Engineering, Cleveland State University, Euclid Avenue at East 24th Street, Cleveland, OH 44115, U.S.A. (216) 687-6967.

Author for correspondence. § Department of Industrial and Systems Engineering, Virginia Polytechnic Institute and State University,

Blacksburg, VA 24061, U.S.A. [[ Department of Engineering Science and Mechanics/Materials Science and Engineering, Virginia Polytech-

nic Institute and State University, Blacksburg, VA 24061, U.S.A.

35.~o-D 1385

Page 2: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

1386 T.H. Nayfeh et al.

Direct on-line methods for tool and process condition monitoring have been difficult to develop and implement since the tool nose and flanks are obscured from vision during cutting. Furthermore, the debris and cutting fluids make it difficult to conduct measurements while the tool is idle. Several methodologies and systems have been explored for sensing the tool condition on-line directly [3]. Some of the methodologies explored are radioactivity tracers [35-40], optical scanning [41-44], electrical resistance, probes to measure the workpiece dimensions, and measurements of the distance to the tool post [4, 45-47], etc. However, none of these systems are practical for the shop floor.

Currently, there is no commercial direct on-line tool and process condition monitoring in application on the shop floor. However, there are several varieties of indirect measurement systems in operation [ 1-3] such as: tool presence sensing probes, vibration and acceleration sensors, load and power sensors, acoustic emissions, and force sensors. Most of these systems measure on-line a single property such as force, torque, current, etc. Although catastrophic tool failures can be predicted reasonably well by some of the methods, the level of reliability and accuracy of predicting gradual wear is certainly well below industrial expectation. An ultrasonic methodology for part gauging in wet turning was proposed in 1974 by Lemelson in a United States, Patent No. 4,118,139. Lemelson proposed that a fluid squirting system could be used to squirt cutting fluid on the part at the opposite end of the cutting tool. The ultrasonic transducer was to be placed at the back of the fluid squirter. Thus, the stream of fluid would provide a direct acoustical coupling mechanism of the ultrasonic waves and the part. This negates the effect of the free falling cutting fluid on the work. There is no evidence that Lemelson pursued the work further or developed a prototype of the system himself.

In 1990, Reed [48] built a commercial ultrasonic on-line part gauging system based on Lemelson's concepts. The two-way time of flight (TOF) of the ultrasonic waves and the amplitude of the reflected waves are used to measure the part diameter and surface finish. A 0.0002 of an inch accuracy in measuring part diameters was reported by Reed. No surface finish accuracy values were reported. Reed reported that the system is in successful operation at the shop floor of United Technology's Aircraft Engine Division (Pratt and Whitney). The main drawback of Reed's system is that it lags the cutting tool by 1/16th of an inch; so that, in the case of catastrophic failures, the part would have already been ruined.

EXPERIMENTAL SETUP

The experimental setup is depicted schematically in Fig. 1. The hardware used is available commercially, off the shelf and as such, certain limitations are imposed on the capabilities and performance of the experimental setup. The shape of the ultrasonic transducers selected for this research, dictated the cutting inserts' geometry. The transducer used is a Panametrics (V-203-RM) flat face contact transducer operating at frequency of 10 MHz. Its active element is 3.175 mm in diameter and has an overall cylindrical package geometry of 4.76 mm in diameter. An ultrasonic delay-line (isolator) is used in conjunction with the transducer to isolate it from direct contact with the hot cutting tool. The delay-line is a proprietary Panametrics (DLHT-3) 3.175 mm in diam- eter and 5.08 mm long. The combination of the transducer and the delay-line are cooled by injecting air inside the tool holder in order to achieve continuous operating contact with the tool.

An ultrasonic pulser-receiver operated in burst mode, i.e. short duration spikes, activates the transducer. The maximum repetition rate of the pulser is 2 kHz. In addition, the pulser can be triggered by an external source on command to provide synchronization with the data acquisition system. The receiver section provides a gain of 40 dB at 10 MHz and is capable of receiving pulses of up to 20 MHz.

The tool holder, depicted in Fig. 2, is specifically designed to accommodate the transducer. The tool clamp is shifted to the right side of the normal position, and a hole is bored from the back of the tool to the back-end of the holder. The transducer,

Page 3: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

An Integrated Ultrasonic Sensor 1387

Tool lmet't

~ & Qiroi C~pster

Fig. 1. Schematic representation of the system.

m //

Fig. 2. Tool holder and transducer assembly.

delay-line, and a solid handle are inserted into the tool holder, out of view and thus are protected from chips and debris. Part of the delay-line pierces the housing directly behind the cutting insert. A spring load keeps the transducer assembly in intimate contact with the back-end of the tool.

The tools used in the experiments are (TNG333) negative triangular carbide inserts, 4.76 mm thick and 1.587 mm nose radius. The tool holder geometry provides a 6 ° nose and main cutting edge tool clearance along with 6 ° negative rake angle. The flat geometry of the face of the transducer and the delay-line dictated the use of negative

Page 4: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

1388 T.H. Nayfeh et al.

150 ~ 1 1 O0 - Transducer Fire Pulse

i t ~T°olFlanks~°l 50 Tool Nose E c h o \ i

Fig. 3. Wave form of the tool's echo.

inserts. The cutting conditions were as follows: cutting speed 102 sm/m (335 sfpm), 0.203 mm/rev (0.008 in/rev), and at depth of cut of 1.016 mm (0.04 in).

THE ULTRASONIC METHODOLOGY

Gradual tool wear manifests itself in two locations which are the primary and secondary flanks, and the crater area. The present tool- transducer configuration can detect the first form. The second form, crater wear, can only be detected when it is very severe. The contents of the ultrasonic wave returning from the nose and the flanks can be broken down into two wave packets as illustrated by Fig. 3. The first is the direct reflection off the nose of the tool and the surrounding areas of the flanks. The extent of the contribution of the flanks is limited to the area encompened by a distance of 0.254 mm from the absolute nose. This area behaves as a flat reflector at the current sampling frequency. The second reflection is an internally reflected wave, which corresponds to the energy that strikes the flanks at the point of the start of the nose curvature. The wave, which strikes either flank, is internally reflected to the opposite flank, which is then reflected back, along a different path to the transducer. In the case of the second reflection, the travel path is longer than the path for the nose signal, which is manifested by the longer time of flight (TOF) exhibited in Fig. 4.

Fig. 4. Nose and flanks wave form.

Page 5: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

An Integrated Ultrasonic Sensor 1389

In the course of cutting and due to wear, a flat spot begins to develop at the tool nose and the flanks. This change in the geometry of the tool serves to change the total amount of the reflected ultrasonic energy. In this case, the flat area is a more favorable reflector. As such, the total amount of reflected ultrasonic energy increases with gradual wear for both components of the wave. In the ideal case, the increase in the amount of the reflected energy obeys the square law. In the case of turning, the principle does not strictly hold since the reflecting surfaces are marred and are at off angles from the normal to the transducer, thus resulting in complex wave intractions. In spite of the complex interactions, an overall increase in the energy content develops with wear.

The amount of energy transferred from the transducer to the tool is a function of the acoustic impedance mismatch between the transducer and the tool. Furthermore, the tool's acoustic impedance is a function of its temperature. Therefore, the amplitude or wear measurements, have to be conducted at either a nominal temperature, or after the wave form has been normalized to the effects of acoustic impedance changes due to temperature. The relationship between the reflected wave's amplitude and the tool's temperature are related through the TOF measurement. This is because an increase in the tool's temperature causes a decrease in the echo's amplitude and increases the two-way TOF simultaneously.

The two-way TOF was compared with the amplitude during actual machining while the tool was engaged and cutting fluid being applied; also, immediately after disengaging the tool and stopping the cutting fluid applications. The two types of test result are illustrated in Figs 5 and 6. It can be seen from Fig. 5 that the relationship between the amplitude and the TOF is not well behaved during actual tool engagement and forced application of cutting fluids. However, the relationship is well behaved and is repeatable immediately after the tool is disengaged as illustrated by Fig. 6. The relation between the amplitude and the A TOF was determined by a linear regression curve fit (see Fig. 7) which is:

A* = A/{1-I(800/M)*(O.O0362* A TOF) + 0.040381}

where: A* is the corrected amplitude, A is the current amplitude, M is the MHz multiplier of the trequency sampling rate.

The reason for the poor correlation during cutting is that the amplitude of the reflected wave is a function of the temperature and the degree of contact between the tool and the workpiece, i.e. the depth of cut, tool wear, chatter and contact with the

0.1

-0.1 •50

-8 -o.2

'~ ~ -0.3

q: -0.4

-0.5

-0.6

-0.7

• 1st Replication D 2nd replication * 3rdreplication

I I ~ [ I

50 100 150 200 250

Change in the Time of Flight in Microseconds X 800 ,

Fig;. 5. Temperature vs amplitude while the tool is engaged (wet).

Page 6: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

1390 T.H. Nayfeh et al.

0

-0.1

-0.2

-0.3

-0.4

- 0 . 5

-0.6

20

• 1st Replication

I

40

o 2nd Replication * 3rd Replication r I I I I

60 80 100 120

Change in the Time of Flight in Microseconds X 800

Fig. 6. Temperature vs amplitude tool disengaged (wet).

Amplitude-TOF Line Fit

I

140

- 0 1 [ T , - - , - , - , - - , - - , - - , . - , - - , - - , - - , - - , - - ~ ~ o~ , ~ " /

- 0 . 5

• Predicted Y - 0 . 6 a_

t~ y Delta TOF

Fig. 7. Linear regression fit of delta (TOF) vs delta amplitude.

metal chips. In addition, the flow of cutting fluid over the face of the tool is turbulent, which is due, in part, to the interaction with the cutting chips; this causes small erratic variations in the tool's temperature and thus the amplitude.

Gradual wear measurements can be made efficiently immediately after the tool is disengaged. The time required for making the measurement is on the order of 1-2 msec. The current TOF is recorded along with a full wave form. The change in the TOF (A TOF) from that of the nominal value is computed. The corrected wave form amplitude A* is then computed from the Amplitude-TOF Equation.

DATA ACQUISITION AND PROCESSING

During operations, the changes in the ultrasonic echo are small and rapid. Because of the signal's high frequency (10 MHz), a high speed data acquisition system is needed to provide the fine resolution. Thus, the major component of the data acquisition system is an AT style bus board that operates in the computer. The board is a SONIX- AD100 high speed analog to digital converter. The board's stand alone sampling rate is 100 MHz at 8 bit conversion resolution. The board can be operated in Equivalent

Page 7: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

An Integrated Ultrasonic Sensor 1391

Time Sampling (ETS) at rates ranging from 200 MHz to 3.2 GHz in doubling steps. This is achieved by taking repeated samples of the wave form at 100 MHz each and reconstructing the wave form to reflect the added resolution. Thus eight repeats of the wave form are acquired and synchronized to achieve an 800 MHz ETS sampling rate.

The basic acquiisition software uses Assembly language calls to the A/D board. The calls are integrated into a (proprietary SONIX, Inc.) C language program to provide input/output (l /C0 functions to the computer and data storage.

After each transducer firing, the wave form is full digitized and stored in a 64K buffer. Part of the wave form can be selected and displayed on the computer screen in a digital display oscilloscope format. The system provides the capability to determine and acquire the amplitude and the time of flight for up to eight peaks from the displayed wave form. The peaks are selected by positioning a series of gates on the screen. The gates' width and position in time, along with the threshold values for amplitude detection, are manually set by the user.

The relative positions in time of the peak detector gates are maintained by the use of a follower gate. This is an additional gate, which can be positioned on a peak in the wave form and activated. The relative positions of the remaining gates, with respect to the follower gate, are calculated and stored. Changes in the position of the peak in the follower ga~te trigger a correction to the position of the other gates. The positions are calculated and compensated after each firing of the transducer. Therefore, changes in the position of the wave form in time due to the increase in temperture are measured and are compensated.

The concern of this work is the feasibility and the mechanics of the actual on-line monitoring rather than the development of a fully operational system. Therefore, although the current system's hardward is capable of operating and delivering results on-line, the interface software has not been developed to the same stage. Thus the data processing and interpretation are performed off-line.

Gradual wear of the nose and flanks of the tool is a comparatively slow process on the order of minutes or perhaps hours in some cases. Determining gradual wear requires comparing the integral of the absolute value of the wave form to that of the new tool (base wave form),. All the captured wave forms are filtered by a 1X3 median and a 1X3 moving average filter to reduce the jitter caused by the 8 bit A/D converter. In addition, the amplitude of the wave form would have to be corrected if the measure- ments are taken at a different tool temperature from that of the base wave form.

EXPERIMENTAL RESULTS

Six tool wear tests were conducted to evaluate gradual wear measurement of the nose and the flanks. Wear was developed by cutting 4130 HR 52 steel at a cutting speed of 102 sm/m, a feed of 0.203 mm/rev and 1.016 mm depth of cut. Cutting fluid was applied during all machining operations. In all cases, the wear varied randomly at different measurement points from uniform to somewhat irregular. Maximum wear land height VB in the range of 0.381-0.432 mm was used as the criteria for tool life end. The ISO standards for uniform and irregular tool wear are, VB = 0.3 mm (0.0118") for uniform wear or VBmax = 0.6 mm (0.0236") for irregularly worn tools.

No effort was made to isolate the nose from flank wear since they are directly related to each other. In ;addition, isolating the individual wear contributions to the individual ultrasonic echos is not possible at the sampling frequency of 10 MHz. The wave length in carbide at this frequency is 0.5 mm (i.e. no area less than this can be individually isolated). However, the sum of both signals was believed to be representative of the overall nose and flank wear.

The measurement procedures were as follows:

(1) A wave form and the TOF were recorded for the new tool (base wave form); (2) the tool was then used to cut for 10-15 min; (3) after disengaging the tool, at random time intervals (20-100 sec), a new wave

form and the TOF were recorded;

Page 8: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

1392 T.H. Nayfeh et al.

(4) the tool holder was then removed and the wear land height and width was measured with a tool marker 's microscope while the tool remained in the holder and

(5) the cycle was repeated until the wear land height reached - 0.406 mm.

Since the measurements were taken at random intervals, the total number of measure- ment points per tool varied from eight to ten points. All wave forms were recorded at 800 KHz sampling rate and 500 mV full scale display. The Amplitude-temperature correction, in most cases, was small or negligible since the tools were cooled rapidly by the cutting fluid.

The objective of these tests was to determine if a correlation exists between the ultrasonic measurement and the gradual tool wear of the nose and the flanks. Several comparisons were made between the two types of data, which were:

(1) The change in the nose and flanks integrals vs wear land width; (2) the change in the nose and flanks integrals vs wear land height; (3) the sum of the changes in the nose and flanks integrals vs wear land width and

height; (4) the absolute value of the change in the nose and flanks integrals vs wear land

height and (5) the sum of the absolute value of the changes in the nose and flanks integrals vs

wear land height.

The change in the wave form integrals of the nose or the flanks, and their sum vs the wear land width followed a somewhat erratic behavior, in that the different in the integrals between two samples could increase or decrease. Although the overall trend increased with wear, the correlation with the width of the wear land was poor as illustrated in Fig. 8.

Comparing the change in the integrals of the nose or the flanks, and their sum to the wear land height resulted in similar erratic behavior as in the previous comparison. In essence, the ultrasonic data has not changed. In addition to the erratic behavior (see Fig. 9), the trend is considerably different for each tool.

The absolute value of the change in the integrals of the nose or the flanks yielded a much better correlation with the wear land height. The trends for both the nose and flanks were much more stable and followed the gradual wear trends. However, the sum of the nose and flanks measurements were much smoother and followed a much more stable trend as illustrated by Fig. 10.

All the tools exhibited excellent correlation between the combined nose and flanks

0.16

0.12

0.1

0.08

0.06

0.04

0.02

0

- - - = " " N o s e - - - c - - F l a n k e S u m

O

- . . . . . . . . . .

, ~ . . . o . - ° " ~=~=1¢-" " ' ~ ' l ' ~ I , I I I

0 5 10 15 20 25 30

Measured W e a r L a n d W i d t h ( i n 1 / 1 0 0 0 " )

q

35

Fig. 8. Differences in the wave form integrals vs wear land width.

Page 9: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

An Integrated Ultrasonic Sensor 1393

- . . m. . . N o s e - - - o - - F l a n k o S u m

0.18

0.16

0.14

0.12

0.1

0.08

0.06

0.04

0.02

0

-0.02

A f @ $

l _ , u . l m . . u . . . . . . . . n , . , . . . . . . -urn .-n . . . . . . 4 I l l - = f o . .

I I I " ";~ ° ° I I I I

2 4 6 8 10 12 14 16

M e a s u r e d F l a n k W e a r H e i g h t ( i n 1 1 1 0 0 0 " )

I 18

Fig. 9. Differences in the wave form integrals vs wear land height.

- . . nn- . . N o s e - ~ - F l a n k * S u m

0.35 == ,~ 0.3

~ .~ 0.25

~ 0.2

o -~ 0.15

0.05

0

~ . C ] - - -- B --.E3

I / P - ' c r y ' ' - " . . . . ,m . . . . . . . • . . . . . . " . . . . . . . "

, ~ g ~U m m o°° . °mm° ' " ..g ...... n..- -'D'°'°

~" I I ~ I I I I I

0 2 4 6 8 10 12 14 16

M e a s u r e d W e a r Height (in 1/1000")

I 18

Fig. 10. The absolute value of the differences in the wave form integrals vs wear land height.

measurements of the absolute change in the wave form integrals of the nose and flanks with the wear land height. However , the individual tool 's trends were different f rom each other.

The sum of the absolute values of the changes in the integrals of the nose and the flanks of six tools vs the wear land height is presented in Fig. 11. Although each tool exhibited its own individual trend, two main clusters appeared as in Fig. 11. The two clusters could not be attributed to the selection of the tools since all came from the same box, and only one cutting side of each tool was used in each of the experiments.

CONCLUSION

The system's response to gradual wear differed for individual tools; however, there was an excellent correlation between the ultrasonic measurements and gradual wear for each tool. Some engineering and wave mechanics issues remain to be resolved in order to utilize the gradual wear measurement . Differences in the trends between tools have to be resolwed and/or a calibration needs to be developed.

Application of the current system is limited to turning operations. In addition, the

Page 10: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

1394 T.H. Nayfeh et al.

• Tool 1 . ~ o Tool 2

= Tool 5 c Tool 6

* Tool 3 ~ Tool 4

0.45

~ .~ 0.4

gl 0.35 o.3

0.25

0.15

Ol

~' 0.05

0 0 2 4 6 8 10 12 14 16 18

Measured Wear Land Height( In 1/1000")

Fig. 11. Ultrasonic measurements of gradual wear vs actual wear land height.

cu r ren t sys tem ' s A / D convers ion accuracy , da t a acquis i t ion speeds , and the acquis i t ion sof tware res t r ic t its o p e r a t i o n to l a b o r a t o r y set t ings. In o r d e r to fully ut i l ize the sys tem on the shop f loor , a t r ansduce r capab le of coupl ing to the var ious shapes and styles of tu rn ing tools mus t be deve loped . A gener ic tool ho lde r is also n e e d e d in o r d e r to a c c o m m o d a t e the t r a n s d u c e r / t o o l assembly . In add i t ion , the m e a s u r e m e n t r e so lu t ion can be i m p r o v e d fu r the r by increas ing the A / D boa rds reso lu t ion to 12 bits.

REFERENCES

[1] Y. Koren, Adaptive control systems for machining, Mfg Review, 2 (1), 6-15 (1989). [2] J. V. Owen, Feedback from the cutting edge, Mfg Engng, 39-45 (January 1993). [3] Li Dan and J. Mathew, Tool wear and failure monitoring techniques for turning--a review, Int. J.

Mach. Tools Manufact. 30 (4), 579-598. [4] N. H. Cook, Tool wear sensors, Wear 62, 49-57 (1980). [5] H. Suzuki and K. J. Weinamann, An on-line tool wear sensor for straight turning operations, J. Eng.

Ind. 107, 397-399 (1985). [6] S. Jetty, Measuring cutting tool wear on-line: some practical consideration, Mfg Engng 55-60 (July

1984). [7] K. Langhammer, Cutting forces as parameter for determining wear on carbide lathe tools and as

machinability criterion for steel, Carbide J. Soc. Carbide Tool Engrs (1976). [8] K. Uehara, F. Kiyosawa and H. Takeshita, Automatic tool wear monitoring in NC Turning, Ann.

CIRP 28, 39-42 (1979). [9] S. Yie, Y. Zhang and L. Pan, On line tool wear monitoring for turning, Research paper of Nanjing

Aeronautical Institute, China, pp. 171-186 [in Chinese]. [10] J. Tlutsy and G. C. Andrews, A critical review of sensors for unmanned machining, Ann. CIRP 32,

563-572 (1983). [11] Y. Koren, K. Danai, A. Galip Ulsoy and Tsu-Ren Ko, Monitioring tool wear through force measure-

ment, Mfg Technol Review, Proc. 15th NAMRC pp. 463-468 (1987). [12] B. Lindstrom and B. Lindberg, Measurements of dynamic cutting forces in the cutting process, a new

sensor for in-process measurements, Proc. 24th Int. MTDR Conf. pp. 137-142 (1983). [13] P. M. Lister and G. Barrow, Tool condition monitoring systems, Proc. 26th Int. MTDR Conf.

pp. 271-288 (1986). [14] L. V. Colwell, Tracking tool deterioration by computer (during actual machining), Ann. CIRP 23,

29-30 (1974). [15] R. Mackinnon, G. E. Wilson and A. J. Wilkinson, Tool condition monitoring using multi-component

force measurements, Dept. of Electrical and Electronic Engineering, The Queen's University of Belfast. (1986).

[16] M. S. Lan and D. A. Dornfeld, In-process tool fracture detection, J. Engng Mater. Technol. 106, 111-118 (1984).

[17] E. Kannatey-Asibu, Jr and D. A. Dornfeld, A study of tool wear using statistical analysis for metal- cutting acoustic emission, Wear 76, pp. 106-118 (1984).

[18] E. Kannatey-Asibu, Jr and D. A. Dornfeld, Quantitative relationships for acoustic emission from orthogonal metal cutting, Trans. ASME J. Engng Ind. 103, 330--341 (1981).

[19] G. Chryssolouris and Michael Doniroese, Some aspects of acoustic emission modeling for machining control, Laboratory for Mfg and Productivity, MIT, U.S.A.

Page 11: An integrated ultrasonic sensor for monitoring gradual wear on-line during turning operations

An Integrated Ultrasonic Sensor 1395

[20] E. Emel and E. Kannatey-Asibu, Jr., Tool failure monitoring in turning by pattern recognition analysis of AE signals, J. Engng Ind. 110, 137-145 (1988).

[21] P. Citti, S. Bagnoli and C. Braccesi, A diagnostic method for assessing cutting tool condition, Condition- ing Monitoring '87, pp. 930-939. Pineridge Press, Swansea, U.K. (1987).

[22] K. Iwata and T. Moriwaki, An application of acoustic emission measurement to in-process sensing of tool wear, Ann. CIRP 25, 21-26 (1977).

[23] G. Dalpiaz and M. Remondi, Use of acoustic emission for cutting process monitoring in turning, J. Conditioning Monitoring 1, 1-26 (1988).

[24] G. Dalpiaz, Comparing different methodologies for cutting process monitoring through acoustic emis- sion, Pub. DIEM No. 2 University of Bologna (1988).

[25] M. S. Lan and D. A. Dornfield, In-process tool fracture detection, Tram. ASME. J. Engng Mater. Technol. 106, 106-118 (1984).

[26] I. Inasaki, S. Aida and S. Fukuoks, Monitoring systems for cutting tool failure using an acoustic emission sensor, JSME Int, 30, 523-528 (1987).

[27] S. Ramalingam and D. A. Frohrib, Intelligent inserts for real-time tool condition sensing, Mech. Engng Tram. 65-72 (lO89).

[28] A. B. Sadat and S. Raman, Detection of tool flank wear using acoustic signature analysis, Wear 115, 265-272 (1987).

[29] C. L. Lee, A study of noise emission for tool failure prediction, Int. J. Mach. Tool Design Res. 26, 205-215 (1986).

[30] S. M. Pandit, Stochastic linearization by data dependent systems, J. Engng Ind. 221-226 (1977). [31] S. M. Pandit, Data dependent systems approach to stochastic tool life and reliability, J. Engng Ind.

100, 318-322 (1978). [32] E. J. Weller, H. M. Schrier and B. Weichbrodt, What sound can be expected from a worn tool?, J.

Engng Ind. 525-534 (1969). [33] A. Del Taglia, S. Portunato and P. Toni, An approach to on-line measurement of tool wear by spectrum

analysis, Proc. 17th Int. MTDR Conf. pp. 141-148 (1976). [34] C. Jiang, Y. Zhang and H. Xu, In-process monitoring of tool wear stage by the frequency band energy

method, Ann. CIRP 36, 45-48 (1987). [35] K. Uehara, New attempts for short time tool-life testing, Ann. CIRP 22, 23-24 (1973). [36] K. Uehara, S. Kumagai, H. Mitsui and H. Takeyama, Relationship between the size of wear particles

and the mechanism of tool wear in metal cutting, Ann. C1RP 23, 13-14 (1974). [37] I. Ham, A. O. Schmidt and R. J. Babcock, Experimental evaluation of tool wear mechanism and rate

using electron miicroprobe analysis, Proc. 9th Int. MTDR Conf., pp. 973-987 (1968). [38] K. Uehara, On mechanism of crater wear of carbide cutting tool, Ann CIRP 21, 31-32 (1972). [39] N. H. Cook, Tool wear sensors, Wear 62, 49-57 (1980). [40] S. M. Arsovski, Wear sensors in the adaptive control systems of machine tools, Int. J. Prod. Res 21,

347-356 (1983). [41] R. Levi, A. Viii[a, G. Quaglia, R. Chiara and G. Rutelli, An expert control system for tool life

management in flexible manufacturing cells, Ann. C1RP 34, 87-90 (1985). [42] J. U. Jeon and S. W. Kim, Optical flank wear monitoring of cutting tools by image processing, Wear

107, 207-217 (1988). [43] K. B. Pedersen, "Wear measurement of cutting tools by computer vision, Int. J. Mach. Tools Manufact.

30, (4), 129-131 (1990). [44] F. Giusti and M Santochi, Development of a fibre optic sensor for in-process measurement of tool

flank wear, Proc. 20th Int. MTDR Conf., pp. 351-360 (1979). [45] H. Takeyama, Y. Doi, T. Mitsoka and H. Sekiguchi, Sensors of tool life for optimization of machining,

Proc. 8th Int. MTDR Conf., pp. 191-208 (1967). [46] M. Bath and R. Sharp, I, process control of lathes improves accuracy and productivity, Proc. 19th

Int. MTDR Con t~, pp. 1209-1221 (1968). [47] T. H. Stoferele and B. Bellmann, Continuous measuring of flank wear, Proc. 16th Int. MTDR Conf.,

pp. 573-578 (1975). [48] R. W. Reed, On-machine, in-process, ultrasonic gauging and flaw detection, Rev. Prog. in Quant.

Nondestructive Eval. 9, 1949-1956 (1990). [49] T. H. Nayfeh, A direct on-line ultrasonic sensing methodology for determining tool and process

condition during turning operations, Ph.D. Dissertation, Department of Industrial and Systems Engineer- ing, Virginia Polytechnic Institute and State University, Blacksburg, Va (1993).