Monitoring acoustic emission levels
with changes in rock cutting conditions

E Williams and PC Hagan

The University of New South Wales (UNSW), Sydney

The emission of acoustic signals or micro seismic activity in rock subjected to stress is a well established phenomenon that has been exploited in geomechanics to understand for example changes in stress levels around excavations in active mine areas. Another potential application is in the area of rock cutting.

The development of a passive, on-line system capable of detecting changes in rock type for horizon control or deterioration in cutting performance with bit wear would greatly reduce cutting machine downtime and improve cutting efficiency, production and safety.

This paper outlines the findings of a study to investigate whether changes in rock cutting conditions are reflected in the nature of acoustic signals generated in rock. A test facility was established comprising a linear rock cutting machine, acoustic transducer and data acquisition system. The study examined changes between new and worn cutter picks, depth of cut and attenuation of the acoustic signal with distance.

The results show that there were measurable changes in the acoustic signal. Further work is suggested to expand on the range of variables considered, for example changes in rock mass type and structure.

Introduction

It has long been known that most solids including rock, concrete, glass, wood, metals, ceramics, plastics and ice, emit acoustic emissions or micro seismic activity (AE/MS) when subjected to stress or some form of deformation (Hardy, 1981). This phenomenon underpins the passive, indirect techniques that are currently used in industry to continuously monitor AE/MS signals in ground surrounding active mining areas.

Machine rock cutting is a form of excavation often used in mining soft rock with machines such as longwall shearers, continuous miners and roadheaders. Central to the design of many of these machines is a rotary cutter head around which is deployed an array of picks as shown in Figure 1. With rotation of the cutter head, each of the picks in turn impact against the rock surface as shown in Figure 2. This impact causes a rise in stress levels within the rock that will eventually lead to given sufficient energy crack initiation, crack growth and the formation of discrete rock chips. As the tool continues to move though its arc of cutting there is a continual rise and fall in stress levels within the rock as chips are broken away from the surface. This cyclic pattern of loading and unloading in stress levels during cutting is reflected in the variation in force as measured by the pick and illustrated in Figure 3. Both the initial impact and subsequent changes in stress levels are analogous to a series of micro-seismic events.

Figure 1. Longwall shearer cutting coal.

Monitoring of AE/MS associated with the rock cutting process was first identified in the 1960’s as a possible technique that could be used as part of an automated system to control the operation of rock cutting machines. Limitations in technology at the time however prevented further investigation.

Figure 2. Schematic arrangement of a pick during cutting (top insert)
mounted on the rotating cutter head of a continuous miner (bottom).
(after Roxborough and Pedroncelli, 1982)

Figure 3. The variation in cutting and normal forces with time during rock cutting.

A project was undertaken using resources within the University of New South Wales (UNSW) to assess the potential of monitoring AE/MS during the rock cutting process given the advances in sensor technology and data collection. Specifically, the project attempted to determine whether AE/MS could be detected during rock cutting and whether changes in cutting conditions would translate to a change in the pattern or “signature” of the signal. This paper contains the results of this project.

Test Arrangement

The following equipment was used in the project.

1. Modified Invicta 6M linear rock cutting machine with a triaxial force dynamometer.

2. Standard 12.5 mm wide tungsten carbide cutting bits as used in rock cuttability tests.

3. Brüel & Kjær accelerometer, model 4370.

4. Brüel & Kjær charge amplifier, model 2635.

5. Two 8 channel signal conditioning boards, National Instruments model SC-2043-SG.

6. Two analogue to digital (A/D) cards, National Instruments model 6032E and 6034E.

7. DASYLab data acquisition software system, version 7.

An experimental test facility was assembled that comprised a linear rock cutting machine and data collection system in the School of Mining Engineering and acoustic signal detection equipment from the School of Mechanical Engineering at UNSW. The cutting machine shown in Figure 4 is a modified shaping machine capable of cutting a series of grooves in sandstone up to 10 mm deep.

Figure 4. Linear rock cutting machine with triaxial dynamometer (left)
and data acquisition system (right)

The triaxial dynamometer attached to the cutter head measures the force on the cutting tool and resolves it into three orthogonal vectors; that is cutting, normal and lateral force. A tungsten carbide bit was fitted to a tool post holder attached to the triaxial dynamometer on the cutter head. A schematic of the cutting arrangement is shown in Figure 5.

A sandstone block having dimensions of 395 mm (l) x 275 mm (w) x 270 mm (h) and a uniaxial compressive strength of approximately 45 MPa was secured at the front of the rock cutting machine.

Figure 5. Schematic of rock cutting arrangement.

The accelerometer or acoustic transducer was attached to the block of sandstone using softened bees wax. The wax which has good signal transmissions characteristics allowed for easy placement at different locations around the block. The acoustic transducer was connected to the charge amplifier by a miniature coaxial cable.

Signals from the charge amplifier and dynamometer were digitally recorded via a signal conditioning module and two A/D cards on board a PC computer. DASYLab software was used to control data acquisition during each test. Figure 6 shows a schematic arrangement of the various components in the test facility while Figure 7 shows the design of the data acquisition system (DAQ).

The DAQ was configured to record two force channels (cutting and normal forces) and the acoustic signal at a data sampling rate of 500 Hz per channel.

Figure 6. Schematic of various components that comprised the test facility.

Figure 7. The arrangement of the data acquisition system.

Following a series of calibration tests, the charge amplifier was preset to the following settings in the rock cutting tests.

· Integrator amplifier: 316 mV/unit output

· Operating mode: acceleration

· Lower frequency limit: 0.2 Hz

· Upper frequency limit: 10 kHz

Figure 8 shows the cutting tool in action during a test. The accelerometer can be seen mounted on the lower front surface of the sandstone block.

Figure 8. Rock cutting in action with the acoustic transducer
 located at base of sandstone block.

The project involved a series of 12 tests that examined:

· cutting depth;

· state of wear of the cutting tool;

· distance between transducer and cutting groove;

· location of transducer with respect to the cutting direction.

Results

The results of the test program are summarised in Table 1.

The variation in force as measured during the first test in the program is shown in a series of three graphs in Figure 9. The top two graphs are of the cutting and normal forces respectively. The duration of rock cutting was 1.67 s and the length of the groove cut in sandstone was 233 mm. Considering the sampling rate of 500 readings per second, 835 readings were recorded during the test with an average of 3.5 readings per millimetre.

Figure 9. Variation in force levels during cutting Test No. 1, Cutting Force (top),
Normal Force (middle) and Trend/Moving Average of Cutting Force (lower).

The third graph in Figure 9 shows the trend in cutting forces as represented by the moving average with a period of 20 data points. This third graph more clearly indicates the loading and unloading cycles of which there were approximately 10. This corresponds to the loading cycle having a period of approximately 0.17 s acting over an average distance of 23 mm.

Figure 10. Variation in acceleration levels during cutting.

A graph of the variation in AE levels with time in Test No. 1 as measured by the acceleration transducer is shown in Figure 10. The graph indicates four different states in the acoustic signal could be detected during a test (indicated as 1, 2, 3, and 5 in the graph) these being:

· State 1: background noise

· State 3: electric drive motor started (State 2) and drive gear engaged, cutter head begins to move

· State 4: cutter bit impacts the rock surface and rock cutting takes place

· State 5: cutting finished, drive disengaged and power to electric motor turned off

Figure 11 is a graph where the calculated moving average for cutting force is superimposed over the measured level for acceleration in Test No.1. The graph indicates rapid changes in both parameters during the test.

A closer examination over a shorter duration of just 0.3 s as shown in Figure 12 more clearly indicates some correlation in terms of the rise in cutting force corresponding to that observed for acceleration.

Figure 11. Superposition of cutting force
and AE acceleration levels in Test No.1.

Table 1

Summary of test results

			Location of accelerometer (mm)				Mean force (kN)
Test	Depth of Cut (mm)	Distance cut (mm)	X	Y	Z	State of pick	cutting	normal	RMS accel. (m/s^2)	Notes
1	5	233	233	113	135	new	1.027	0.771	2.142	new pick in middle of rock, transducer at end
2	5	191	191	113	135	worn	2.114	2.617	2.022	worn pick in middle of rock, transducer at end
3	5	261	261	113	135	worn	2.051	2.544	1.052	worn pick in middle of rock, transducer at end
4	5	244	344	113	135	new	1.294	0.949	2.433	transducer at end
5	5	300	300	113	135	new	1.199	0.935	2.673	transducer at end
6	5	136	155	93	0	new	1.177	0.948	3.419	transducer front mounted high
7	5	184	155	93	0	new	1.211	0.983	3.641	transducer front mounted high
8	5	240	245	221	0	new	1.293	1.026	2.428	transducer front mounted low
9	5	294	245	221	0	new	1.155	0.933	2.100	transducer front mounted low
10	8	350	245	221	0	new	1.682	1.086	3.192	deep cut, transducer front mounted low
11	10	354	254	113	140	new	2.097	1.225	2.472	deep cut, transducer at end
12	10	281	281	113	140	new	1.740	1.097	2.443	deep cut, transducer at end

Figure 12. Superposition of cutting force and AE acceleration levels
over the first 0.3 s of cutting.

Signal Analysis

Several approaches to the analysis of acceleration levels were examined to quantify the effects of changes in rock cutting conditions on the measured acoustic signal. These analysis techniques included Fast Fourier Transform (FFT) and root mean square (RMS) analysis.

FFT is particularly useful when analysing irregular signals as they provide a way of isolating characteristic frequencies and quantifying the signals. A limitation of the technique is the Nyquist criterion that states a reliable frequency spectra can only be produced for frequencies less than half of the sampling frequency (ME 82, 2003). As the sampling rate in the test program was 500 readings per second, analysis was therefore limited to a folding frequency of 250 Hz.

Figure 13. FFT Analysis of acceleration levels in Test No. 1.

Figure 13 shows the frequency spectra resulting from the FFT analysis for Test 1. The rise in frequency spectra at frequencies approaching the folding frequency of 250 Hz indicates the likelihood of frequency content at higher frequencies; higher than the analysis could identify given the data sampling rate. Therefore the frequency spectrum is incomplete as the frequencies featured are probably the result of aliasing. Hence due to the relatively low sampling rate, analysis of the test results in this program using this technique could not produce meaningful results.

The RMS analysis of data was able to provide a measure of the magnitude of the acoustic signal. A calibration factor was applied to the raw results based on an earlier calibration test using a standard signal generator.

Changes in cutting conditions

The values for mean cutting force and mean normal force were calculated for each test and these values are shown in Figure 14.

Figure 14. Variation in the mean force levels
between each test.

The graph indicates that in Tests 2 and 3 when a worn cutter bit was used, the forces were significantly higher especially the normal force compared to the forces with a new bit.

Conversely, the calculated RMS acoustic level in Test 2 was slightly less than the RMS level measured in tests with a new bit but it was even less so in Test 3 as shown in Figure 15. 

Figure 15. Variation in the RMS acoustic level between tests.

The effect of state of wear on cutting force and accelerometer is summarised in Figure 16 with a doubling in cutting force and a near halving in the RMS acoustic level. A possible explanation for this difference in reaction between force and acoustic signal is that although the cutting forces are larger with a worn bit, the reduction in forces on rock fracture is much lower than with the new cutting bit. The worn bit may have leads to a more irregular fracture pattern in the rock with generally smaller sized rock fragments. It is possible that the more frequent fracturing when using the worn bit produced more acoustic activity but with an overall lower amplitude. This would translate into a lower RMS acoustic level with the worn bit. 

Figure 16. Effect of cutter tool wear on cutting forc
 and the RMS acoustic level.

In terms of the effect of a change in cutting depth, Figure 14 shows a doubling in depth with Tests 11 and 12 resulted in much higher cutting forces. For the same two tests, Figure 15 indicates a less significant change in the RMS acoustic level. The effect of cutting depth on force and RMS acoustic level is summarised in Figure 17 with an 87% increase in cutting force and a smaller 15% increase in the RMS acoustic level.

Figure 17. Effect of depth on cutting force (CF) and
the RMS acoustic level.

The increase in the RMS acoustic level could be due to the larger forces required to fracture the rock resulting in greater amounts of elastic strain energy being released on failure. It is this elastic strain energy that is responsible for the majority of the signals detected by the accelerometer.

In terms of the location of the acoustic transducer with respect to the line of cutting:

· when the accelerometer was placed at the front of the sandstone block, the RMS acoustic level decreased with distance;

· when the accelerometer was placed at the end of the block, there was no obvious link between distance and the RMS acoustic level.

As Figure 18 tends to indicate, location of the transducer did not appear to have any significant affect on the acoustic signal.

Figure 18. Variation in the RMS acoustic level with distance.

Conclusions

1. The test program was successful in that it showed that an acoustic emission is generated during the process of rock cutting and that it is technically feasible to measure this acoustic emission.

2. The acoustic emission was found to vary with time during rock cutting and there seemed to be a correlation between the rise in cutting force and the level of acoustic emission.

3. Changes in rock cutting conditions appeared to have some measurable impact on the nature of the acoustic emission.

4. Several methods of analysis of the measured acoustic emission were investigated. It was found that the low sampling rate of 500 Hz, limited the use of Fast Fourier Transform (FFT) analysis. Calculation of the Root Mean Square (RMS) value did provide a means of characterising the level of the AE signal for each test.

5. The project has shown that signal analysis may be a key factor to its implementation of a measurement system in the field. Better analysis methods that can be undertaken in real-time will be required if this is to be developed into a system that will be integrated into a machine control system.

6. While the project was successful in detecting an acoustic signal, the test program was limited to using an available transducer in the Faculty of Engineering. Other types of transducer are available with greater sensitivity and capable of achieving higher sampling rates than that used in this program. A study of AE during coal cutting by Hardy and Shen (1996) indicated a monitoring system that can handle frequencies in the range of 100 kHz to 450 kHz may be required.

7. With better instrumentation and signal analysis there is scope to investigate additional variables associated with field conditions, including differences in rock mass properties and structural boundaries; wave attenuation; changing wave velocities; changes in operating conditions, including cutting speed and force; additional acoustic background activity (including electrical interference, traffic, blasting and low level seismic activity) and wave complexities due to boundaries and rock structures.

Acknowledgements

The authors wish to thank the support and contributions made by Prof. R Randall and Mr R Overhall, School of Mechanical and Manufacturing, UNSW; Prof. J Wolfe, School of Physics, UNSW; and, Prof. H R Hardy, Pennsylvania State University, USA.

References

Hardy, H R Jr, 1981. Application of acoustic emission techniques to rock and rock structures: a state-of-the-art review, in Proceedings ASTM Symposium on Acoustic Emissions in Geotechnical Engineering, Detroit, June, pp 4-92 (American Society for Testing and Materials)

Hardy, H R Jr and Shen, H W, 1996. Laboratory study of acoustic emission and particle size distributions during linear cutting of coal, in Rock Mechanics Tools and Techniques: 2nd North American Rock Mechanics Symposium, pp 835-841. (Balkema: Rotterdam)

ME 82, 2003. Fourier Transforms, DFT’s and FFT’s – Mechanical Engineering Measurements, Department of Mechanical and Nuclear Engineering, accessed online http://www.me.psu.edu/me82/Learning/FFT/FFT.html August 20 2005 (Pennsylvania State University, USA)

Roxborough, F F and Pedroncelli, E J, 1982. A practical evaluation of some coal-cutting theories using a continuous miner, The Mining Engineer (London), September, pp 145-156.