26.3 Vibration Signal Recording

Back

The information from vibration sensors can be treated in numerous ways. The overall aim of a tool

condition monitoring system (TCMS) is to utilize the best processing techniques to extract the relevant

information from sensor signals. Generally, a TCMS consists of the steps depicted in Figure 26.8. Various

methods that could be used in each step will be discussed in more detail.

26.3.1 Direct and Indirect Systems

TCMSs can be divided in two categories, namely, direct and indirect. Direct methods are concerned with

a measurement of volumetric loss at the tool tip, while indirect methods use a pattern in sensor data to

detect a failure mode (Byrne et al., 1995). Direct methods do not utilize vibration and will not be

discussed here. In general, direct methods are sensitive to dirt and cutting chips, and consequently they

are not commonly accepted in industry. Indirect methods have found more acceptance in industry due to

the fact that most indirect methods are easily interpreted, cost-effective, and often more reliable than

direct methods. Also, for some applications, it might not be possible to use a direct monitoring method

due to the nature of the process.

26.3.2 Sensor Requirements for Tool Wear Monitoring

Machine tools represent very hostile environments for sensors. Sensors used for TCM (also see Chapter

15) must meet certain requirements, such as (Byrne et al., 1995) the following:

* Must measure as close as possible to the point of metal removal

* Must not cause a reduction in the stiffness of the machine tool

* Must not cause a restriction of the working space of the machine

* Should be wear and maintenance free, easy to replace, and of low cost

* Must have resistance to dirt, chips, and electromagnetic and thermal influences

* Should function independent of tool and workpiece

* Must provide reliable signal transmission, e.g., from rotating to fixed machine components

26.3.3 Force Measurement

Worn tools cause an increase in the cutting force components. It is also known that both the dynamic and

static components generally increase with tool wear due to frictional effects. The three components of the

cutting force each responds uniquely to varying machining parameters and the different wear modes.

Depending on the type of process that is investigated and the specific experimental setup, results among

researchers vary. This can be attributed to dynamic effects of the machine tool and the measurement

equipment. There are a number of different sensor configurations to collect forces from machining

operations and these are described below.

26.3.3.1 Direct Measurement Dynamometers

Tool holder dynamometers are by far the most popular method for collecting cutting forces. These

sensors utilize the piezoelectric effect and can measure quasistatic and dynamic cutting forces very

accurately. However, dynamometers are very expensive and bulky instruments and are not practical for a

sensor selection

and deployment

signal recording

and conditioning

generate signal

features

select wear

sensitive features

model features

and wear

relationship

FIGURE 26.8 TCMS steps.

Vibration-Based Tool Condition Monitoring Systems 26-7

© 2005 by Taylor & Francis Group, LLC

typical shop floor. Furthermore, their usable

frequency range is limited to approximately

1 kHz. An example of a tool holder dynamometer

is shown in Figure 26.9.

Tarmal and Opavsky (2000) investigated the

dynamics of a conventional force dynamometer

for machining operations. It was found that the

dynamometer has significant amplitude distortion

in the frequency range that is quoted as the

operating range by the manufacturer. The authors

suggest that the dynamic characteristics of the

dynamometer (while clamped as it would be

during measurements) be identified with a

modal test and the effect of dynamometer

dynamics be compensated for after measurements

are made to obtain the true cutting force.

26.3.3.2 Indirect Force Sensors

There are numerous small force sensors available

for the purpose of force measurement on machine

tools. These measure forces in load-carrying

components of the machine tool and are thus

not direct force measurement devices. The advantages

of these sensors are their size, low cost, and

significantly higher operational frequency range.

A disadvantage is that a suitable position for the

sensor can only be determined experimentally.

These sensors are suitable for tool breakage

monitoring in rough machining or detection of

other catastrophic events such as collisions. An

example of a three-component force sensor is

shown in Figure 26.10.

26.3.3.3 Piezoelectric Strain Sensors

The use of piezoelectric strain sensors for wear

monitoring of synthetic diamond tool inserts was

reported by Scheffer and Heyns (2000a). These

sensors are ultrasensitive to changes in cutting

forces if they are installed in an appropriate

location. The best location for the sensor must

once again be determined experimentally, but

generally it should be installed on a load-carrying

component of the machine as close as possible to

the tool tip, for example, on the tool holder itself

(Scheffer and Heyns, 2001b). An example of a

piezoelectric strain sensor that can be used on machine tools is shown in Figure 26.11.

26.3.3.4 Resistance Strain Gauges

A quite simple method to estimate both the static and dynamic components of cutting forces without any

distortion is to use resistance strain gauges (see Chapter 15). These comply with most of the requirements

for TCM sensors, and they can accurately follow the static and dynamic response of a system up to

FIGURE 26.9 KISTLER force dynamometer type 9121.

(Source: KISTLER Brochure 2002. Courtesy of Kistler

Instrumente AG.)

FIGURE 26.10 KISTLER three-component force sensor

type 9251A. (Source: KISTLER Brochure 2003.

Courtesy of Kistler Instrumente AG.)

FIGURE 26.11 KISTLER strain sensor type 9232A.

(Source: KISTLER Brochure 2004. Courtesy of Kistler

Instrumente AG.)

26-8 Vibration and Shock Handbook

© 2005 by Taylor & Francis Group, LLC

50 kHz. Scheffer and Heyns (2002a) developed a sensor-integrated tool holder using strain gauges. It was

shown that the system is robust, cost-effective, and fit for an industrial TCMS. The physical layout of the

strain gauges on a boring bar is shown in Figure 26.12. The system was calibrated with a special device to

directly obtain the three cutting forces from the strain gauge signals.

26.3.3.5 Customized Force Sensors

There are a number of customized force sensors available that can be used with specific machining

operations. These are:

* Force measuring plates, pins, and bearings

* Special force measuring bolts

* Force and torque measuring rings that fit on spindles

26.3.4 Acceleration Measurement

Piezoelectric accelerometers can measure the machine vibration caused by oscillations of cutting forces. It

is well known that high-frequency vibrations (higher than 1 kHz) yield large acceleration levels, giving

accelerometers an advantage over force-based monitoring. Accelerometers fulfill the environmental

requirements for tool wear monitoring because they are resistant to the aggressive media present during

machining. Accelerometers are also less expensive than force dynamometers and can measure vibration

levels within a very wide frequency range, typically 5 Hz to 10 kHz.

Various authors have shown that acceleration levels change with tool wear. Li et al. (1997) found that

the coherence function of two crossed accelerations can be used as an easy and effective way to identify

tool wear and chatter. They found that with progressive tool wear, the autospectra of the two

accelerations and their coherence function increase gradually in magnitude around the first natural

frequencies of the cross-bending vibration of the tool shank. As the tool approaches a severe wear stage,

the peaks of the coherence function increase to values close to unity. Scheffer et al. (2003) reported on the

use of an accelerometer for wear monitoring during hard turning. It was found that certain frequencies

show repeatable amplitude increase with increasing tool wear. These frequencies corresponded to the tool

holder natural frequencies. Some authors, for example, Bonifacio and Diniz (1994), also found that a

wear sensitive frequency will increase with increasing tool wear and then suddenly decrease near the end

of tool life. This can be attributed to an increased damping effect due to plastic deformation and

microbreakage of the cutting edge.

26.3.5 Acoustic Emission Measurement

Cutting processes produce elastic stress waves that propagate through the machine structure. Different

sources in the cutting process generate these stress waves known as acoustic emission (AE). Sources of AE

FIGURE 26.12 Application of resistance strain gauges. (Source: Scheffer, C. and Heyns, P. S., Mech. Syst. Signal

Process., Elsevier, 2004. With permission.)

Vibration-Based Tool Condition Monitoring Systems 26-9

© 2005 by Taylor & Francis Group, LLC

in metal cutting are:

* Friction on the tool face and flank

* Plastic deformation in the shear zone

* Crack formation and propagation

* Impact of the chip at the workpiece

* Chip breakage

A typical AE sensor for use on machine tools is

shown in Figure 26.13.

The fact that crack formation generates AE

makes AE ideal for tool breakage detection.

Collection of the AE requires special hardware

that can bandpass filter the signals to the AE range (between approximately 50 and 250 kHz).

Furthermore, amplification is required and an analogue root-mean-square (RMS) circuit with a short

time constant is generally also included to collect the AERMS. The different steps required to collect AE are

depicted in Figure 26.14 (adapted from Jemielniak, 2000).

Araujo et al. (2000) investigated sliding friction as a possible source of AE during metal cutting.

The AERMS values in different frequency ranges were collected for different widths of cut and also

with the tool rubbing against the workpiece without cutting. It was found that the level of AE

remains almost constant for all width of cut conditions, and hence it was concluded that the main

mechanism for AE during metal cutting is the sliding friction between the tool and workpiece.

Consequently, an increase or decrease of AE can be expected with tool wear depending on the effect

on the sliding friction due to that tool wear. Furthermore, it is believed that the cutting

temperatures will affect the AE due to thermal expansion effects. Chiou and Liang (2000)

investigated AE with tool wear and chatter effects in turning. A model is presented that can predict

the chatter AERMS amplitude with certain levels of flank wear. Good correlation was found between

the model and the experimental results. Kim et al. (1999) reported on the use of AE to monitor the

tool life during a gear shaping process. The AERMS is collected and used in a software program to

predict the remaining tool life.

Li (2002) presented an overview of using AE for TCM in turning operations. It is stated the AE is

heavily dependent on cutting conditions and, as a result, methods should be employed to handle this

problem effectively. Some methods are proposed that include advanced signal processing, sensor

fusion and modeling techniques. Many other AE-based tool wear and breakage monitoring systems

have been implemented successfully in research. One problem still lies with an appropriate

interpretation of the AE frequency spectrum. In most studies, an explanation for the choice of certain

frequencies and their advantages are not given or not investigated. In fact, Jemielniak (2000) found

that using the average value of AE (or AERMS) is the most suitable. A similar conclusion was made by

Scheffer et al. (2003), who compared different processing methods of the AE signal during hard

turning.

FIGURE 26.13 Kistler AE sensor type 8152B. (Source:

PCB Website 2002. Courtesy of Kistler Instrumente AG.)

FIGURE 26.14 Steps for collecting AE during turning. (Adapted from Jemielniak, K., Ultrasonics, 2000.)

26-10 Vibration and Shock Handbook

© 2005 by Taylor & Francis Group, LLC

26.3.6 Sensor Comparisons

Future research should be directed towards directly comparing different sensors for tool wear

monitoring. Choi et al. (1999) developed a single sensor for parallel measurement of force and AE. A

finite element analysis was carried out to determine the optimal position for the sensor away from the

tool holder because the sensor obstructed the working space of the machine. The approach was successful

for breakage detection but no wear estimations are reported. Barrios et al. (1993) compared AE,

vibration, and spindle current for TCM during milling. It was found that the spindle current is the most

sensitive sensor for detecting tool wear, with AE the least sensitive. However, contradictory results are

reported in other publications, and hence more research would be required to determine ultimately

which sensor is the best for which machining operation. Govekar et al. (2000) compared force and AE

methods for TCM, and concluded that the best result is achieved when sensory information is combined.

Dimla and Lister (2000a) compared the use of force and vibration signals for TCM and also combined the

information in a single decision-making technique (Dimla and Lister, 2000b). Similar comparative

studies were reported by Scheffer et al. (2003).

Навигация