Section III: Chromatic Monitoring of Mechanical Vibrations

This section describes the use of the chromatic approach for extracting information about mechanical vibrations and acoustic signals produced by various sources. Examples of such deployments are the monitoring of cutting forces during mechanical machining of metallic materials, the live monitoring of running railway tracks from a moving, in-service train and monitoring the condition of high- voltage electric power equipment via acoustic signals from a transformer.

Mechanical Machining Monitoring

C Garza

Introduction

Monitoring cutting forces during mechanical machining is needed for reducing geometric errors, improving efficiency and optimising and modelling the processing (Sukvittayawong and Inasak 1991). Methods which have been used include the deployment of strain gauges, piezoelectric devices and conductive polymer and magneto-elastic materials (Elbestaw'i 1999). Efforts have been made to integrate force sensors into machine tools (Byrne and O’Donell 2007). Piezoelectric force transducers are susceptible to electromagnetic interference and do not have the dynamic response needed for high-speed machining. Optical methods such as the use of optical fibres and photo-elasticity have been proposed (Jina et al. 1995).

The use of optical chromatic monitoring of a photo-elastic element with optical fibre transmission of polychromatic light for such machine cutting is described in this section. Both static and dynamic machine operation are addressed. The machining of a mechanically rotated cylindrical rod carrying a rectangular slot along its length as a defect under real machining cutting conditions is used to assess the transient response, and comparisons are made with results obtained with a piezoelectric load cell.

Test System

An experimental system for evaluating the use of a chromatic optical method for monitoring machine cutting consisted of two parts - a mechanical cutting system and the monitoring system (Figure 8.1) (Garza 2010, Garza et al. 2012).

The mechanical cutting system consisted of a tool holder attached to a tool fixture via two screws (A, B; Figure 8.1). Static tests could be implemented via various degrees of compression obtained by adjusting one of the screws. For dynamic tests, a lathe was used with cylindrical rods, each carrying a 2-mm-wide groove and made from different materials (aluminium, brass, stainless steel). The rods were rotated at speeds up to 850 rpm.

Chromatic system for monitoring the operation of a cutting tool

FIGURE 8.1 Chromatic system for monitoring the operation of a cutting tool.

The optical monitoring system was based upon a photo-elastic sensing element mounted between the tool fixture and cutting tool holder (Figure 8.1). The optical sensor consisted of a 3-mm-thick photo-elastic element with a reflective back plus polarisation filter addressed by white light from a light emitting diode (LED) source via a bundle of optical fibres, and the output was detected by three photo-detectors with overlapping wavelength responses. The optical fibre bundle consisted of seven, 400 mm to 400 pm fibres, with six fibres delivering the light to the sensor and one central fibre receiving the modulated light for transmission to the photo-detector. A piezoelectric load cell was attached to the tool cutter to provide comparative measurements of the stressed tool holder.

Chromatic Results

Both static and dynamic chromatic test results have been reported (Garza 2010, Garza et al. 2012). The static results were obtained by applying a series of mechanical stresses to the tool cutter, whilst the dynamic results were obtained by rotating the cylindrical rod containing grooves on a lathe.

Static Tests

A suitable location on a captured image of the photo-elastic element was chosen for addressing chromatically via the receiving optical fibre (Figure 8.1). The variation of each of the R, G, В output signals at this location with different mechanical loads (1-10 N) is given in Figure 8.2a, which shows how each of the three outputs varies differently with various loads. As a result, the values of the corresponding three chromatic parameters (H, L, S; Chapter 1) also tended to vary with the loads. An example of the variation of dominant wavelength H at the midpoint with a load is shown in Figure 8.2b. The variation is monotonic and so provides an unambiguous indication of the load value. A comparison of the susceptibility of the optical and piezoelectric orthogonal stresses indicated that the susceptibility of the optical sensor was <2% (Garza 2010, Garza et al. 2012).

Dynamic Tests

The response of the photo-elastic sensing system (Figure 8.1) has been shown to be independent of signal frequencies up to 7.5 kHz and within 2.2% for frequencies up to 15 kHz (Garza 2010,

Static chromatic addressing of the photo-elastic element,

FIGURE 8.2 Static chromatic addressing of the photo-elastic element, (a) Example of R. G. В outputs as a function of mechanical load, (b) Examples of chromatic H at three locations as a function of mechanical load.

Garza et al. 2012). The time response of the entire system (Figure 8.1) with optimised photo-elastic sensing and using the chromatic H parameter are shown in Figure 8.3a and b. The variation of the H parameter with time for turning of an aluminium rod at 850 rpm with major pulses occurring every 0.07 s is shown in Figure 8.3a. A time-expanded view of both chromatic H and piezoelectric output (Lc) during a groove transit with aluminium at 850 rpm is shown in Figure 8.3b.

Chromatic Results Implications

The test results indicated that the chromatic approach can address the complex structure of the optical pattern produced by a photo-elastic element subjected to machine operating conditions. Consequently, chromatic parameters H. L, S can in principle be used for monitoring such machine cutting operations. The H parameter provides a first-order quantification with high optical efficiency compared with monochromatic detection and with less susceptibility to spurious monochromatic disturbances. Also, chromatic H and S are independent of optical intensity and so provide some immunity to extraneous light intensity variations.

Static Performance

Under static conditions (Figure 8.2), the chromatic parameter H varies monotonically over a range of 55° for a load change of 0-Ю N with a coefficient of regression R2 = 0.99. A sensitivity of 6°/N with a noise level of only 1° and a resolution of 0.164 N have been demonstrated. A sensitivity to orthogonal stresses was <2%, with a repeatability of <2°.

Dynamic test results for a cylindrical rod with groove, (a) Time variation of chromatic parameter H. (b) Fine time scale variation of chromatic H and piezoelectric output (Lc)

FIGURE 8.3 Dynamic test results for a cylindrical rod with groove, (a) Time variation of chromatic parameter H. (b) Fine time scale variation of chromatic H and piezoelectric output (Lc).

Dynamic Performance

The dynamic performance of the chromatic photo-elastic unit was evaluated via two graphs (Garza 2010, Garza et al. 2012). The first was the use of the chromatic parameter H values to determine the rotational period of the machined rods (i.e., time between pulses in Figure 8.3a) versus the nominal rotation rate (Figure 8.4a). The second was the use of the chromatic parameter FI values to determine the pulse width versus the calculated value from the nominal rotation speed and known groove width (Figure 8.4b).

The chromatic photo-detection unit had no change in attenuation for frequencies below 7.5 kHz but had a change in attenuation of 2.2% for frequencies above 7.5 kHz up to 15 kHz. The angular speeds of rotation of the grooved rod deduced from the time separation between consecutive

Measured versus calculated time scales for rotating workpieces of different materials and rotation rates

FIGURE 8.4 Measured versus calculated time scales for rotating workpieces of different materials and rotation rates: (a) rotation period from chromatic H measurements; (b) groove transit time from chromatic H and piezoelectric load cell output.

chromatic H pulses (Figure 8.3a) showed a linear variation with the nominal rotation rate of the rod for the rotational speed range of 165-850 rpm (Figure 8.4a). The linear variation was similar for the different rod materials (aluminium, steel, brass), and the sensitivity was 9.47N.

The values of the pulse w idth associated with the rotating groove determined chromatically showed good agreement with the calculated values (Figure 8.4b) for the same rotation rates (165-850 rpm) and the different rod materials (aluminium, steel, brass). Comparison of the piezoelectric sensor results for the pulse width showed a greater scatter than the chromatic sensor results (Figure 8.4b).

The coefficient of determination (R2) for the chromatically measured and calculated periods of rotation (Figure 8.4a) was 0.999, whilst a comparison of the groove transit time (Figure 8.4b) gave R2 = 0.996. The piezoelectric measurements of groove width (Figure 8.4b) gave R2 = 0.977.

Overview and Summary

The feasibility of using a chromatically based sensing system for addressing a photo-elastic element stressed by the mechanical rotation of a mechanically machined cylindrical rod has been demonstrated. The system did not use mono-mode optical fibres nor monochromatic light. Calibration graphs of the chromatic parameter H versus static load, turning period and groove transit time had coefficients of determination of 0.99. The repeatability of static load results from repeated tests is claimed to be at least 2%, whilst machining different materials agreed with each other to within +/-1.5%, and susceptibility to orthogonal stresses was <2%.

Further investigations are needed to assess long-term drift effects and susceptibility of the system to ambient temperature variations. The possibility of chromatically extending the force measurement range also warrants investigation.

References

Byrne, G. and O'Donell, G. E. (2007) An integrated force sensor solution for process monitoring of drilling operations. Ann. CIRP 56, 89-92.

Elbestawi, M. A. (1999) Force measurement in The Measurement, Instrumentation and Sensors Handbook Chapter 23 (J. G. Webster Ed.) CRC Press Boca Raton. FL, USA, 23-7-23-9.

Garza, C. (2010) Investigation of an optical sensor for cutting force measurement through chromatic modulation, PhD Thesis, University of Liverpool.

Garza, C„ Jones. G. R.. Hon, К. К. B., Deakin. A. G. and Spencer. J. W. (2012) Feasibility of monitoring mechanical machining with a chromatically addressed optical fibre photo-elastic sensor. Strain, 49, 68-74.

Jina. W.L., Venuvinod. P. K. and Wang. X. (1995) An optical fibre sensor based cutting force measuring device. Int. J. Mach. Tools Manufact, 35, 877-883.

Sukvittayawong, S. and Inasak. I. (1991) optimisation of turning process by cutting force measurement. ISME Int. J. Ser. C, 34. 546-555.

Q Chromasonics

Chromasonics: Monitoring Rail-Track Faults

R. K. Todd

Introduction

Railway operators have a need to ensure that their running track is in good condition and safe for freight and passenger traffic. Current methods for doing so include video scanning of the rails, eddy current generation, ultrasonics and walking the tracks.

The first two methods are unable to detect many rail faults, especially in the early stages. The latter is slow and can only detect obvious effects. Ultrasonic methods have been shown to perform well (Hesse, 2007) in detecting nearly all rail defects but are expensive in requiring an ultrasonic test car and involve slow operations (about 40 mph) along the tested rail track. This means that they can only be used w'hen main line traffic is not running, Thus, a method based upon the use of a high-speed main-line locomotive which provided fault detection during normal operation would be advantageous. Such a possibility arises from the use of chromatic techniques in the acoustical domain called Chromasonics. Such a system fitted on a locomotive would be cost effective compared to ultrasonics, could be used daily for monitoring during normal running schedules and could relay information to a control centre whence remedial action could be taken. Chromasonics has been tested and run in the United Kingdom and United States and shown to be capable of detecting many important rail defects. An example is shown in Figure 9.1.

A description of the Chromasonics approach is given along with examples of defects detected in real tests w'ith the method.

Chromasonics

Chromasonics is the adaptation of chromatic techniques in the acoustical domain for monitoring acoustic signals from railway tracks on board a locomotive during its passage along the railway track. The basis of the Chromasonics approach is described, followed by a description of instrumentation used for its deployment on board operating locomotives.

Vertical crack on a rail track

FIGURE 9.1 Vertical crack on a rail track.

Principles of Operation

The adaptation of chromatic techniques for monitoring acoustic signals involves deploying three acoustic filters (R, G, B) with non-orthogonal frequency responses to address a Fourier-transformed acoustic signal in the frequency domain. This compares with deployment of three non-orthogonal optical filters for processing optical signals as described in Chapter 1.

Figure 9.2 show's three frequency domain acoustical signals at each of three different times, the first corresponding to a normal operating condition and the other two to two different fault conditions at later times. Also shown in Figure 9.2 are the responses of three non-orthogonal acoustic filters (R. G, B) superimposed upon the three acoustic signals in the frequency domain.

The outputs from the three acoustic filters are chromatically processed to yield three chromatic parameters H. L, S (Chapter 1) which represent the dominant frequency, effective signal strength and signal spread and which can be represented on polar diagrams of H:S and H:L (Chapter 1). Figure 9.3a and b show typical representations of the H, L, S signature of an acoustic signal. Thus, different acoustic signals may be graphically represented by their coordinates on such H:S and H:L polar maps.

Chromatic processors R. G. В addressing acoustical signals at different times (t,, t, t,)

FIGURE 9.2 Chromatic processors R. G. В addressing acoustical signals at different times (t,, t2, t,).

Schematic chromatic polar diagrams for a signal О at a given time

FIGURE 9.3 Schematic chromatic polar diagrams for a signal О at a given time: (a) H:S, (b) H:L.

Monitoring System

The basic structure of a Chromasonics system is shown in Figure 9.4. This consists of a standard microphone in association with a personal computer (PC) and data storage means plus data transfer capability.

A microphone with a frequency range of 10 Hz to 20 kHz was mounted on the locomotive in a suitable position for receiving acoustic signals from the rail track produced by the rotating wheels of the locomotive. The output from the microphone was fed into a PC for chromatic processing. The outputs were Fourier-transformed into frequency domain signals for chromatic monitoring. The resulting information was stored on the unit with a provision for wireless transmission to a central control hub via general purpose radio system.

A chromatic system has been tested in the United Kingdom and United States. Extensive testing with Union Pacific Railways USA enabled comparisons to be made with results from an ultrasonic

Chromasonics data collection and processing system

FIGURE 9.4 Chromasonics data collection and processing system.

test vehicle which indicated that the Chromasonics system could detect most defects indicated by the ultrasonic system.

RAIL-DEFECT Detection

Tests performed with Chromasonics both in the United Kingdom and United States have provided results which demonstrate the capabilities of the approach for detecting rail defects with H:S and H:L chromatic maps on running tracks and with operating locomotives.

Normal Track Signal Plus Squat

Squats are an indentation in the rail head which tend to fill up with dirt and grit and are black in appearance. They are frequently caused by track ballast on the rail track run over by a locomotive wheel. They are of variable size, ranging from a few millimetres to a few centimetres. A squat can lead to the formation of defects such as vertical cracks.

Figure 9.5a and b show H:S and H:L chromatic maps which display signals from both a normal length of railway track plus a single squat. For the normal track, the dominant frequency (H) varied continuously with mid-range dominant frequencies (H ~ 180°) and a relatively low level of effective magnitude (L). However, a squat is clearly distinguishable with a high value of effective magnitude (L) and a well-defined frequency (H) of dominant value less than 90°.

Normal Track Signal Plus Horizontal Crack

Figures 9.6a and b show H:S and H:L chromatic maps which display signals from both a normal length of railway track plus a horizontal crack. The normal track has a different signature from the

Chromatic polar diagrams of a normal railway track and squat signals

FIGURE 9.5 Chromatic polar diagrams of a normal railway track and squat signals: (a) H:S, (b) H:L.

Chromatic polar diagrams of a normal railway track and crack signals

FIGURE 9.6 Chromatic polar diagrams of a normal railway track and crack signals: (a) H:S. (b) H:L.

Chromatic polar diagrams of a normal railway track and defective weld signals

FIGURE 9.7 Chromatic polar diagrams of a normal railway track and defective weld signals: (a) H:S, (b) H:L.

track signal of the railway track shown in Figure 9.5 but with some similar features. The signal shows similar variability but has a lower spread (S) and a lower dominant frequency (H), with a relatively low effective strength (L).

The presence of a horizontal crack of about 15 cm length produced a narrow' frequency band (S) signal with a much higher dominant frequency (H) than the background rail signal. Two points are apparent which may give an indication of the length of the crack.

Normal Track Signal Plus Defective Weld

Figures 9.7a and b show' H:S and H:L chromatic maps which display signals from a normal length of railway track plus a defective weld between two lengths of the track. The normal track signal in this case showed substantial variation in signal frequency spread (S) and the dominant frequency (H) at relatively low levels of effective signal magnitude (L).

The defective weld produced a distinctive signal with a much higher dominant frequency (H) and narrower spread (S) plus a higher effective magnitude (L).

Worn Track Signatures

Figures 9.5-9.7 show that the chromatic signatures of lengths of different tracks can vary with regard to the dominant frequency (H) and spread (S) but less so for the effective magnitude (L). However, extensive tests on different tracks have show n that the acoustic signature of a railway track tended to vary as it became older (Figure 9.8). Nonetheless, defects were observed to be located in the same frequency domain on both H:S and H:L chromatic maps. The H:L map of Figure 9.8 shows how the presence of squats was detectable regardless of the track being new or old. Figure 9.8 also shows the occurrence of wheel burn (i.e., the locomotive drive w'heels spinning without the locomotive moving due to a heavy load) and the track becoming excessively heated and deformed.

Overview and Summary

Real track tests with Chromasonics test units on board travelling locomotives indicate that the Chromasonics unit is capable of detecting defects on railway running tracks during normal locomotive operation (Todd, 2005, 2007).

The tests have shown that different tracks can have different H, L, S coordinate values which for each track form clusters together. The occurrence of various faults (squats, cracks, faulty welds) are identifiable by changes in the H, L, S signatures being distinctly different from the normal track signatures. Different faults may also be distinguishable from each other. However, further

Chromatic H:L polar diagram of new and old railway tracks

FIGURE 9.8 Chromatic H:L polar diagram of new and old railway tracks.

research and evaluation are needed to realise the full potential of the technique over a wide variety of conditions.

References

Hesse. D. (2007) Rail Inspection Using Ultrasonic Surface Waves. Department of Mechanical Engineering Imperial College London SW7 2AZ. (pdf Section 3 P. 43)

Todd. R. K. (2005) Rail Flaw Detection using Acoustic Techniques for Network Rail (File reference PRJ- DF-LD 79565: Page 46; Section 7.5; Chromasonic Data analysis and appendix A and B)

Todd, R. K. (2007) Chromatic processing technology: Defect detection of railway tracks, doi: 10.1049/ etr.2016.0090; ISSN 2096-4007; www.ietdl.org

 
Source
< Prev   CONTENTS   Source   Next >