采矿工程 毕业外文翻译

附件D：原文姓名：王小丹学号：20087281 Efficient mine microseismic monitoring

Maochen Ge

Pennsylvania State University, University Park, PA 16802, USA

Received 6 May 2004;

revised 24 August 2004;

accepted 7 March 2005.

Available online 12 April 2005.

Abstract：

During the past 20 years, the microseismic technique has grown from a pure research means for rockburst study to a basic industrial tool for daily safety monitoring at rockburst-prone mines. This article examines the important issues for efficient mine microseismic monitoring programs. The key technical issues for such a program are discussed from three aspects: monitoring planning, data processing, and microseismic event location. An efficient monitoring program would be impossible without a firm commitment of the mine management. Issues related to the management and mine operations are discussed, including monitoring program integration, efficient use of microseismic data, and the benefit of monitoring programs for mine safety and productivity.

Keywords: Microseismic; Rockburst; Source location; Ground control; Mining

1. Introduction

Rockbursts and coal bumps are sudden and violent releases of energy stored in rock masses and geological structures. They have been a persistent threat to mine safety, causing catastrophic failures of mine openings, paralyzing mining operations, damaging mining equipment, and posing a severe safety threat to underground workers. In 1958, a rockburst at the Springhill Coal mine in Nova Scotia claimed 75 lives. In the U.S., a total of 100 rockburst-caused fatalities were reported in the last 60 years (Blake and Hedley, 2003).

The energy released by a rockburst can be staggering. In 1995, a rockburst with a local magnitude of 5.2 M L was recorded at the Solvay trona mine, Wyoming, when an entire 1000 m × 2000 m panel collapsed. The US coal mining industry has experienced bumps since the 1920s, with magnitudes up to 4.5 (Blake and Hedley, 2003).

The technique that is widely used for studying rockburst activities is the microseismic monitoring technique. The technique utilizes signals generated by the material to study fracture/failure processes. The real time monitoring capability of the microseismic technique, in terms of event source location, magnitude and source mechanisms, makes it an ideal tool for studying mine seismicity and related ground control problems.

The phenomenon of the emission of micro-level sounds by stressed rocks was first

discovered in the late 1930s by two U.S. Bureau of Mines (USBM) researchers, Obert and Duvall, when they carried out sonic studies in a deep hard rock mine (Obert, 1975). In the early 1960s, South African researchers began to utilize this phenomenon to study the rockburst problem associated with deep gold mines (Cook, 1963). This early study convincingly demonstrated the feasibility of the rockburst location by the microseismic technique, the central element of mine microseismic monitoring.

In the middle of the 1960s, the USBM started a major research program in order to make the microseismic technique an efficient tool for mine safety monitoring. The hardware and software developed from this program, as well as the research and field tests carried out during this period, laid the foundation for the industrial use of the microseismic technique (Leighton and Blake, 1970 and Leighton and Duvall, 1972).

From the middle of the 1980s to the early 1990s, severe rockburst problems occurred spontaneously in Canadian mines. Over 20 rockburst-prone mines installed microseismic systems for daily monitoring purpose. From the late 1980s to the 1990s, large-scale rockburst research was carried out in Canada, sponsored by the Canadian federal government, the Ontario provincial government, and major mining companies. This research fundamentally changed the role of the microseismic technique in the Canadian mining industry. It is no longer a pure research tool, but the basic monitoring means for mine safety and ground control.

This article examines the important issues for efficient mine microseismic monitoring programs. The discussion is carried out from three aspects: monitoring planning, data processing, and microseismic event location. Although the focus of this paper is the technical issues, it is important to note that an efficient monitoring program would be impossible without a firm commitment from the mine management. For this reason, we will also discuss issues related to the management and mine operations, including monitoring program integration, efficient use of microseismic data, and the benefit of an efficient monitoring program for mine safety and productivity.

2. Planning and optimization of monitoring systems

Careful planning is the foundation for establishing an efficient monitoring program and has a profound impact on the system's long-term performance. There are three important issues to be resolved at this stage: engineering assessment of monitoring objective and monitoring condition; determination of the monitoring system size (number of channels); and optimization of the sensor array layout. Also, the harsh mining environment requires a rigorous maintenance program because monitoring systems degrade rapidly.

附件D：原文姓名：王小丹学号：20087281

2.1. Engineering assessment of monitoring objective and monitoring condition

The first task at the planning stage is a thorough assessment of monitoring objectives, including target areas, monitoring accuracy, and associated monitoring conditions. Since mining is a dynamic process, this assessment should take into account both short-term and long-term monitoring needs.

In order to achieve this goal, a comprehensive analysis should be carried out on potential rockburst hazards in relation with the mining conditions, such as mining method, mine layout, ground control practice, mine development operations, geological materials and structures, and stress conditions at the mine site. As a result of this engineering assessment, the size of the monitoring system can be determined. This analysis should also yield useful information on feasible locations for sensor installation.

2.2. Using a large channel system

The number of channels needed depends on several factors. The most important ones are the size of the area to be covered, the location accuracy required, the signal level expected, and rock formations. An initial estimation may be made with the reference of the mines with the similar conditions.

It is always good practice to use a relatively large channel system. Why is a large channel system critical for daily monitoring programs? A simple answer to this question is that the efficiency of a microseismic system is first measured by its capability of catching enough signals. If a system has difficulty detecting the expected signals, the value of the system diminishes. This has been a major problem faced by the microseismic technique prior to the use of large channel systems.

The efficiency of large channel systems for signal detection is due to two mechanisms. First, with a large channel system, we effectively shorten the distances between a potential source and sensors. If we consider the fact that the energy of a microseismic event decays rapidly with the distance because of both attenuation and geometric spreading effects, shortening the signal travel distances is the only solution to the problem. Second, the emissions of microseismic signals are generally directional with significant variations in signal strength tied to direction. The only solution to this problem is to have sufficient sensors surrounding potential sources.

Before the early 1980s, microseismic systems were small, typically four or five channels. At that time, the most serious problem was the difficulty of collecting data. It hampered early applications and made daily microseismic monitoring a difficult task. The single most important factor for the successful rockburst monitoring in Canada is the use of large channel systems. These systems ranged from 32 to 128 channels. The other important function of large channel systems is that they make it possible for sensor array optimization.

2.3. Sensor array design and optimization

Sensor array geometry refers to the configuration of the sensors to be used for event location. From a technical point of view, it is probably the most important factor affecting the monitoring accuracy and reliability. The fundamental importance of the sensor array geometry lies in the fact that it determines the stability of the source location system, or in other words, it determines the impact of initial errors on the location result. A good array effectively minimizes the impact of initial errors on the location result. The relative location accuracy associated with a three-sensor array is demonstrated in Fig. 1.

Fig. 1. The hyperbolic field associated with a triangular sensor array, where circulars denote sensor locations (after Ge, 1988).

The array effect is shown by the density of the hyperbolic field associated with the array, which is an indication of the relative location accuracy for the array (Ge, 1988). It is clear from the figure that the location accuracy is best at the center of the array and rapidly decreases away from the array. The worst areas are those behind the sensors.

Since errors are inevitable for input data (such as arrival times, velocity and sensor coordinate), the source location accuracy depends greatly on the efficiency of reduction of the impact of these initial errors. Good array geometry is essential for the reliable and accurate source location. The particular importance of the sensor array is its long-term effect on daily monitoring programs, as the achievable monitoring accuracy at a mine site largely depends on the array used. As such, the sensor array design is the central task at the planning stage.

There are a number of important aspects which have to be carefully considered in the design process. The following is a brief discussion of these aspects.

2.3.1. Long- and short-term monitoring needs

Installation of an underground monitoring system is a very time-consuming and costly operation. In order to minimize the changes to be made at a later point, both the current and long-term monitoring needs have to be thoroughly assessed.

2.3.2. Field investigation

附件D：原文姓名：王小丹学号：20087281 During the planning stage, the physical conditions at the potential sensor locations should be assessed. In addition to their accessibility, the sites should not be shielded by large openings or major discontinuities. Rocks at the mounting site should be competent and a good coupling effect can be achieved.

2.3.3. General planning

Mining is a dynamic process. The production and development activities are often carried out at several different locations. In order to design an array that is efficient not only for the mine as whole but also for those specific areas, one has to understand those basic array effects. The following are several basic rules for general planning.

? Two-dimensional arrays should be avoided. This type of array gives very poor accuracy in its perpendicular direction.

? Special sensor pairs may be designed for reinforcement of coverage in certain directi ons at particular locations ( 2.3.4. Simulation analysis

After a general plan is made, its effect may be further studied through a simulation analysis. The emphasis of the study should be the pattern of the location accuracy, not the inpidual numbers. The array can be fine-tuned as the result of this study.

2.3.5. Calibration study

Calibration studies should be regularly scheduled after the array is in place as they will provide the most reliable information on the monitoring accuracy as well as the effect of the sensor array. Both rockburst and blast data can be utilized for the purpose.

2.4. System maintenance

Practical experiences have shown that one of the most critical factors to keep a mine monitoring system at its best performance level is regular maintenance. Mines represent an extremely harsh environment for microseismic monitoring. Sensors and wires can be easily damaged by mining activities and falling rocks. Water and excessive moisture may cause the problem of malfunction for sensors. Fractured ground may significantly reduce signal strength at sensors. Local disturbances from mining, transportation, and ventilation may create high background noise. Any of these problems could severely affect the performance of monitoring systems.

A simple and effective means to detect these problems is a regular check of sensors' signal intensity. If there are no signals, sensors or their wires may be damaged and need to be repaired. If signals are weak for a prolonged period, the site or sensor installation may not

be appropriate and sensors should be reinstalled or moved to new sites. If there is a constant and high level of background noise, we may have to temporarily suspend the affected sensors to prevent their interference to the system.

3. Microseismic data processing

The microseismic data recorded at a mine site can be extremely complicated. This is usually due to the excessive background noise presented at the mine site. Microseismic signals are often partially or even completely swamped by noises, making it difficult to identify the actual arrival times of incoming signals.

“Clean” signals can also be very complicated. Some complications are due to other activities unrelated with the event under consideration. Furthermore, a good portion of these signals may be caused by S-wave arrivals instead of P-wave arrival as we would normally assume. If these signals are used without discrimination, it will result in significant contamination of the database. The Canadian experience of daily monitoring has shown that efficient monitoring is dependent on the ability to process microseismic data. In this section, we will discuss two important aspects of microseismic data processing: noise filtering and identification of arrival types.

3.1. Frequency analysis and data filtering

A primary task in data processing is to filter background noise. This requires a detailed study of the frequency distributions for both signals and noise. If the dominant frequency range for signals is different from that of noise, we may separate signals from background noise by using a set of required filters. The following is an example from a recent study by the author at a limestone mine, where the monitoring efficiency of the microseismic system was severely affected by background noise.

A detailed study was carried out on the characteristics of microseismic signals and noise, including a manual inspection of all waveforms from the database, frequency analysis of representative waveforms, and case testing. It was determined that the microseismic signals at the mine site were primarily confined in the range of 10–200 Hz, with a dominant frequency spectrum of 10–130 Hz. Three typical noise types were identified, which were high-frequency noises (>200 Hz) due to various onsite mining activities, low-frequency and cyclic noises (<10 Hz) caused by distance machinery activities, and electrical activity noise at 60 Hz. Three types of filters were therefore used to eliminate the noises: a 10-Hz high-pass filter, 100–200-Hz low-pass filters and a 60-Hz notch filter. By using these filters, most of background noise can be filtered out. Fig. 2shows such an example.

附件D：原文姓名：王小丹学号：20087281

Fig. 2. A microseismic signal with a high noise level filtered by a 100-Hz low-pass filter [(a) original waveform, (b) filtered signal by a 100-Hz low-pass filter, (c) frequency spectrum of the original signal, (d) frequency spectrum of the filtered signal].

In this figure, Part a is the original waveform, where the microseismic signal is present, but its arrival is buried by the background noise and cannot be determined. The dominant frequency of the background noise for this particular case, as shown by Part c, is somewhere near 270 Hz. Because the dominant frequencies for microseismic signals at this mine site are typically less than 200 Hz, with a concentration in the range of 10–130 Hz, a low-pass filter (<100 Hz) was applied. The filtered signal is shown in Part b. It is clean and its arrival is ready to be detected.

A more drastic effect of this filtering technique is shown in Fig. 3. In this case, no recognizable microseismic signals can be observed from Part a. However, a very clear signal, as shown in Part b, emerged from this highly noised channel after a 100-Hz low-pass filter is applied.

Full-size image (67K)

Fig. 3. Recovering a microseismic signal buried in the background noise [(a) recorded signal, (b) filtered signal by a 100-Hz low-pass filter, (c) frequency spectrum of the original signal, (d) frequency spectrum of the filtered signal].

3.2. Identify physical status of arrivals

In addition to noise elimination, another important task in data processing is the identification of arrival types. The first arrival detected by a sensor is not necessarily due to a P-wave as it has been assumed for most microseismic studies. It is a much more complicated phenomenon. In addition to P-waves, first arrivals may be due to S-waves, or even outliers. Outliers are those arrivals which are not due to the physical source triggering the majority of the stations during an event time window.

The importance of being able to identify these arrivals relies on two facts. First, it would introduce significant and systematical errors into our database if these arrivals are mixed and assumed to be P-wave arrivals. Second, S-wave and outlier triggering are not rare or isolated

events. The analysis of actual daily monitoring data showed that a significant portion of recorded signals could be due to these arrivals (Ge and Kaiser, 1990). Fig. 4 is such an example.

Full-size image (16K)

Fig. 4. Distribution of arrival types from a survey of 434 events.

This figure shows the distribution of arrival type picks as a function of triggering sequence for 434 events from a mine's database. S-wave arrivals account for 41% of the total picks and outliers are about 10%. If the P-wave arrival assumption were used for these S-wave arrivals and noises, the input data for event location would be severely contaminated. In fact, this is the single most important problem responsible for the poor performance of many daily monitoring systems in the early days.

Identifying the physical status of signal arrivals from an automatic picking system is an inherently difficult task. It is often difficult even when manual analysis of waveforms is performed. In searching for the solution to the problem, Ge and Kaiser (1990) developed a phase association theory for the interpretation of the physical status of arrival picks without waveforms. It consists of two parts, arrival time difference analysis and residual analysis. The theory is based on the idea that the physical status of an arrival pick, although unknown when judged by the arrival time alone, will manifest in its connections with other related parameters. Such connections may be found from the observed arrival time differences relative to the distances of the associated sensors and to the array size and density. Connections may also be found from the size and sign of station residuals relative to the assumed arrival type and to the observed arrival time differences.

This technique provides a unique means of identifying arrival types and has been adopted by many mines, resulting in substantial improvement of the monitoring accuracy and efficiency. The following two examples demonstrate the importance of identifying outliers and S-wave arrivals for the daily microseismic monitoring programs.

3.2.1. Example showing the effect of outliers

Outliers are extremely harmful for a meaningful source location. The presence of even one outlier in an event, regardless of the size of the event, is likely to ruin the solution. While outliers have complex causes, the most common ones are interference of seismic activities and culture noises. Human errors can also be a factor from wiring problems to errors in recording station coordinates.

To demonstrate the frequent occurrence of outliers and their severe consequence, four

附件D：原文姓名：王小丹学号：20087281 Event number Arrival status b Outlier excluded (m) Outliers included (m)

x y z x y z

39 DPPPPPPPPPPP 5720 5651 2080 –––

40 DPPPPPPPPPPP 5717 5669 2084 –––

41 PPPPPPPPPPPD 5727 5663 2092 –––

43 PPPPPPPPPPPP 5736 5675 2088 5736 5675 2088 recorded events are presented in Table 1.

Table 1. Effect of outliers on source location accuracy a

Full-size table

a Denotes the coordinates which are out of the mine boundary.

b P represents P-wave arrivals and D outliers.

The four events presented in the table were caused by a nearby blast and recorded within 6 s. By the analysis of the triggering sequence, it is known that their locations should be very close. The automatic analysis indicated that three out of four events recorded during this very short time period were contaminated by outliers, denoted by D in Table 1. A further manual analysis confirmed that the outliers for the first two events (first triggered sensor) were due to the interference of the other local events, and the one with the third event (last triggered sensor) was due to remote noise because the sensor was located in a different mining area, not belonging to this local array.

The devastating effect of these outliers can be seen by comparing the two sets of location results. When the outliers are included in the calculation, the locations are all out of the mine boundary, and the errors are in the order of more than 10 km. However, when the outliers are excluded, all four events have very similar locations. If the blast site (5725, 5660, 2088) is used as an approximation of the origin of these events, the errors for these three events are only 13 m, 13 m, and 5 m, respectively.

3.2.2. Example showing the effect of an S-wave arrival

P- and S-waves travel with very different velocities. The velocity for S-waves is typically 60% of that for P-waves. If S-wave arrivals are simply assumed as P-wave arrivals, it will introduce large errors into the input data and cause severe location problems. The example given

in Table 2 shows the consequence if an S-wave arrival is assumed to be a P-wave arrival or simply ignored.

Table 2. Effect of an S-wave arrival on source location accuracy

Solution Velocity model a Coordinate (m) Error (m)

x y z

Solution Velocity model a Coordinate (m) Error (m)

x y z

1 PPPPS 5647 5573 2656 29

2 PPPPP 5374 555

3 266

4 290

3 PPPPD 5476 5568 2656 187

Blast site 5663 5552 2643

Full-size table

a P and S represent P- and S-wave arrivals, and D denotes the sensor not used.

The event presented in this table was related to a blast which coordinates were known. The triggering pattern for the event based on the automatic analysis is PPPPS, which means the first four were triggered by P-wave arrivals and the last one was by an S-wave arrival. The location error for this velocity model is 29 m (solution 1), which is very close to the blasting site (Fig. 5).

Fig. 5. The effect of the sensor array geometry on the location of a blasting event (PPPPS denotes the location with a right arrival time picking, PPPPP represents the location with a wrong assumption of P-wave arrival for the fifth sensor, and PPPPD is the location without using the fifth sensor).

If the S-wave arrival at the last sensor was mistakenly assumed as a P-wave arrival, the location error would be 290 m (solution 2), which is 10 times higher than the one with Solution 1. Solution 2 was located on the other side of the sensor array (Fig. 5).

For Solution 3, the S-wave arrival was excluded from the calculation and only the first four P-wave arrivals were used. The error for this solution is 187 m. Why does the solution without significant input errors have such a large location error?

If we examine the array geometry formed by these five triggered sensors relative to the blasting site, we can see that this is a fairly weak array which lacks good control in the horizontal direction. This explains the large error associated with Solution 2 in the horizontal direction. If sensor 5 is removed, the array is almost on a straight line. This location system is highly unstable in the horizontal direction. Even a small error in the input data may cause significant location error in that direction. This is shown in Fig. 5by the mirror image locations of solution 3 and the blasting site about the array.

The example above demonstrates the importance of using S-wave arrivals for preserving the information and maintaining the integrity of the sensor array geometry. Both of these

附件D：原文姓名：王小丹学号：20087281 factors are critical for an efficient and accurate microseismic monitoring.

4. Criteria on selection of source location code

The basic function of a daily mine microseismic system is to delineate the locations of rockbursts and the associated microseismic activity. Its efficiency is largely measured by the accuracy and reliability of event locations. In this regard, choosing a suitable source location code is critical for an efficient monitoring program.

Source location is a very broad subject. There are many different approaches and methods. The discussion here is limited to how to choose a suitable method for the daily monitoring purpose at mines. For more information, readers may refer to the author's two recent articles (Ge, 2003a and Ge, 2003b), which provide the detailed discussion of various major methods used in seismology, microseismic monitoring, and acoustic emission, including the triaxial sensor approach.

When selecting a code for the mine microseismic monitoring, particular attention should be paid to the following four aspects:

4.1. Convergence character of searching algorithms

One of the major considerations in selecting the source location method is the convergence character of searching algorithms. The convergence character here refers to the stability of the solution searching process carried out iteratively. If a searching algorithm has a poor convergence character, it is usually prone to the pergence problem. When this happens, the location searching process is in effect stopped, either in the form of oscillation or causing a system breakdown. For a daily monitoring program, this would be an intolerant problem as it is impossible for a manual analysis of hundreds of daily recorded events.

Currently, there are three typical algorithms used for mine microseismic monitoring: the USBM method, the Geiger's method and the Simplex method. The USBM method is a widely used mine-oriented source location method, developed by the U.S. Bureau of Mines' researchers in the early 1970s (Leighton and Blake, 1970 and Leighton and Duvall, 1972). The method is simple and easy to use, and, because of its non-iterative algorithm, has no pergence problem. However, the method is severely limited for the daily monitoring purpose because it cannot simultaneously handle P- and S-wave arrivals.

Geiger's method, developed at the beginning of the last century (Geiger, 1910 and Geiger, 1912), is the best known and most widely used source location method. In seismology, it is used almost universally for local earthquake locations. The algorithm is efficient and flexible, but prone to the pergence problem and, therefore, not suitable for the daily monitoring purpose.

The Simplex method developed by Nelder and Mead (1965) searches the minimums of mathematical functions through function comparison. The method was introduced for source location purpose in late 1980s by Prugger and Gendzwill (Prugger and Gendzwill,

1989 and Gendzwill and Prugger, 1989). The mathematical procedures and related concepts in error estimation for this method were further discussed by Ge (1995). The most important advantage of the Simplex method over the other popularly used iterative algorithms is its robust convergence character. Divergence is essentially not an issue for this method. This advantage, together with its efficiency and flexibility, makes the method the top choice for daily microseismic monitoring purposes.

4.2. Using P- and S-wave arrivals simultaneously

The ability to use P- and S-wave arrivals simultaneously has a number of important implications for accurate event location. First, it allows an efficient use of the available data. An important phenomenon frequently observed in both seismology and microseismic monitoring is the higher amplitude for S-wave arrivals. In many cases, we may only see

S-wave arrivals, instead of P-wave arrivals. Fig. 4 is a clear demonstration of this phenomenon.

Second, most microseismic events are very small, with only five or six arrivals. If the location code can only use P-wave velocity and there are S-wave arrivals, we then have to discard the S-wave data or use the P-wave arrival assumption for all arrivals. As it was discussed earlier, neither of these approaches is acceptable. Using the P-wave assumption will incur large location errors, and discarding S-wave arrivals may severely limit the sensor array and make the solution vulnerable. The other important advantage for using P- and S-wave arrivals simultaneously is that it introduces a new error control mechanism for improving source location accuracy.

4.3. Optimization method

Accurate source location depends greatly on our ability to limit the impact of initial errors. There are two principal approaches, array and data optimization. The importance of array optimization, as discussed earlier, is to create a stable mathematical system which will not be overly sensitive to initial errors. The objective of data optimization is to find the best fit solution for the observed data through a statistical analysis. The efficiency of the analysis depends on how well the error that is used to define the best fit characterizes the data to be analyzed.

For the microseismic source location, there are two commonly used methods: the least-squares method and the absolute-value method. The least-squares method is by far the best known method. One problem with this method in our case is that it assumes the Gaussian distribution for errors, which is difficult to fulfill if events are small. The consequence is that the method may be overly sensitive to large errors and loses its function.

Unlike the least-squares method which defines the total error as the sum of squares of mismatches, the absolute-value method defines the total error as the sum of the absolute values of mismatches. This method is not overly sensitive to those relatively large errors. The

附件D：原文姓名：王小丹学号：20087281 following is an example that demonstrates this point.

The event to be studied here is an actual microseismic event recorded at a limestone mine. The event triggered five sensors. The arrivals for the first four sensors are clear and can be precisely determined. The arrival at the fifth sensor, however, is very weak, which led to a wrong pick of 67.25 ms at the initial stage. A late analysis showed that the actual arrival should be 32.47 ms. Therefore, the error for our initial pick is 34.78 ms.

The source location results for this erroneous pick by the least-squares method and the absolute value method are given in Table 3. It is seen from the table that the solution given by the absolute value method is actually very accurate, with a location error of only 2.7 m (9 ft). In contrast, the location error for the least-squares method is 58.5 m (195 ft).

Table 3. Comparison of location results by the absolute-value method and the least-squares method with an error pick at channel 5

Optimization method Coordinate (m) Location error (m)

X Y

Absolute value method 1640 462 2.7

Least squares method 1672 511 58.5

The cause of the performance difference between the least-squares method and the absolute-value method is shown in Table 4 where the mismatches or errors for five sensors are listed for each solution. For the absolute-value method, the error is concentrated at sensor 5, which is 34.36 ms. If we compare this calculated error and the actual error of 34.78 ms, it is known that the error occurred at sensor 5 was quarantined at the sensor. With the least-squares method, however, a significant amount of the error occurred at sensor 5 spread to others.

Table 4. Comparison of channel residuals (ms) for the solutions by the absolute value method and the least squares method with an error pick at channel 5

Sensor number 1 2 3 4 5

Absolute value method 0.08 0.00 ?0.85 0.00 34.36

Least squares method ?3.22 ?8.01 ?12.98 1.98 22.22

4.4. Reliability analysis

Microseismic data is complex in nature. A considerable amount of the events initially collected by the mine monitoring systems may not be located or accurately located. If these event locations were used, they could severely contaminate the monitoring result and diminish their usefulness. For a reliable mine monitoring program, a vigorous reliability analysis of the source location result is needed.

The assessment of the reliability of the source location result is conventionally based on the

size of the event residual. The event residual, a measurement of the total effect of the mismatches between the observed and calculated data, is an important and basic indication of the reliability of event locations. However, it is important to note that residual provides only a partial characterization. The other important part is sensitivity, which measures the vulnerability of source location solutions. For instance, the actual location error for an event outside of the array may be significantly larger than an inside-array event with the same residual. A comprehensive analysis of the reliability of event location results may include three aspects: residual, sensitivity, and hit sequence.

4.4.1. Residual analysis

Residual analysis should include both the size of the residual and the distribution of the channel residuals. The size of the residual gives an idea of how accurate the observed data is matched. The distribution of channel residuals may reveal important information on the origin of the error and how it spreads.

4.4.2. Sensitivity analysis

Sensitivity analysis is used to assess the stability of the event location system. A sensitivity index may be defined in many different ways. The simplest one is the distance between the event location and the location obtained with a slightly varied velocity, say 10%. Sensitivity analysis is a strong indication of the potential influence of initial errors. An important usage of this parameter is to determine the direction that is most sensitive to initial errors. Since the impact of the initial errors is largely governed by the sensor array geometry, it is also an indication of the suitability of the sensor array.

4.4.3. Hit sequence analysis

Hit sequence analysis studies the patterns of observed hit sequence and compares it with the calculated sequence. Since the observed hit sequence delineates the feasible region for the recorded events, the comparison of two sequences provides the intuitive information on the reliability and feasibility of the calculated result.

All three analyses can be programmed and carried out automatically. A procedure developed by the author has been used by many mine monitoring systems for the final screening of the location results before they are used for the mine safety and ground control purposes.

5. Microseismic monitoring and mining operations

A daily microseismic monitoring program is not just a technical issue. To be truly an industrial tool, it also requires both leadership and commitment from the mine management.

附件D：原文姓名：王小丹学号：20087281 This section discusses how to integrate a daily monitoring program as a part of the basic mining operations and its benefit to mining companies.

5.1. Benefit of microseismic monitoring

An effective mine microseismic monitoring program can contribute significantly to safe and productive mining operations. The following is a brief discussion of several main benefits. 5.1.1. Mine rescue operations

When a rockburst occurs, the first question asked by mine management is “where is the burst and how big is it.” If the mine has a daily mo nitoring system, management can determine almost on a real-time basis whether the burst is an immediate threat to underground workers. With this information, a rescue plan can be quickly made.

5.1.2. Risk management

One of the most important functions of the daily monitoring program is risk management. At a mine without such a system, a rockburst or anomalous seismic activity could easily cause panic and result in a significant production delay. When a rockburst or anomalous seismic activity can be located and its hazard potential can be evaluated, the problem often becomes more manageable. As a result, production disruptions are often eliminated, avoiding costly loss of revenue.

5.1.3. Cause of rockburst and better mine design

When the locations of rockbursts and associated activities are known, the cause of rockbursts may be objectively assessed in terms of geology, mine layout, mining method and stress field. The mine may be re-planned and the mining method may be altered to alleviate the rockburst potential. The effect of remedial measures can be assessed immediately by the intensity of the associated microseismic activity.

5.1.4. Ground control

The microseismic monitoring technique has been used to study a variety of ground control problems, such as pillar stability, roof conditions, effect of various support techniques, and mechanism of backfill. At many rockburst-prone mines, this technique has become a primary tool for assessing overall ground conditions. Maps of microseismic activity are used to evaluate potential ground control problems as well as the further monitoring needs by other methods.

5.1.5. Rockburst prediction

Rockbursts are a complex phenomenon. Predicting the precise timing of a rockburst, like an earthquake prediction, is inherently difficult. In the author's opinion, the degree of success for a timing predication will largely depend on local conditions. A recent study carried out at a limestone mine showed a consistent trend of microseismic activity proceeding convergence in the roof fall areas, a strong indication of the feasibility of using the microseismic technique as an early warning system at this mine site (Iannacchione et al., 2004). In any case, we have to emphasize that timing is only part of rockburst prediction; the more important part is potential locations and the measures to alleviate or, in some cases, to eliminate the rockburst potential in the areas. As has been discussed, major progress has been made in this area. 5.2. Microseismic monitoring in Canadian mines

In the 1980s, many rockburst-prone mines in Canada adopted the daily microseismic monitoring programs in response to a sudden surge of rockburst activities. Although the monitoring efficiency was low at the beginning stage, the effort of improving the monitoring efficiency was steady. From the early 1990s, daily microseismic monitoring gradually emerged as a mainstream mine safety and ground control tool and became the primary monitoring means at rockburst-prone mines. A survey at those mines shows that

“they would have serious reservations about their ability to mine in such ground without a microseismic monitoring system (Bharti Engineering Associates Inc., 1993)”.

The success of daily microseismic monitoring programs in Canadian mines is not simply a result of technical advance, such as the adoption of the techniques discussed in the article. A strong support from the mine management has played an important role. This can be easily understood by glancing at a typical daily operation.

Each morning, the system operator prepares a brief, daily report on the microseismic activity during the past 24 h. This report, along with other information, is forwarded to the engineering department for the detection of any new concerns as well as for the general assessment of ground condition. The findings are then passed to the underground supervisor or ground control staff so that the identified areas can be checked out at the beginning of each shift.

The most encouraging development of daily microseismic monitoring programs used in Canadian mines is the culture change. Microseismic data is no longer the “property” of the management and engineers. It is encouraged to be used by everyone. Various communication channels are used for this purpose. At the Kidd Creek mine, an online service was developed for the real-time access of microseismic data. In 1993, the mine changed its safety regulation after high-level location accuracy was achieved through a major system updating. Under this new safety regulation, it is mandatory that all shift bosses have to check maps of microseismic activity for the past 8- and 24-h periods before they go underground. A computer terminal was specifically set at the mine entry for this purpose.

附件D：原文姓名：王小丹学号：20087281 6. Conclusions

During the past two decades, the microseismic technique has emerged from a pure research means to a mainstream industrial tool. The daily monitoring program has greatly enhanced mine safety and productivity at rockburst-prone mines. The success of daily microseismic monitoring is due to both technical advances and tremendous efforts made by the mining industry itself.

The technical issues discussed in the paper are basic and essential elements for efficient daily monitoring programs. Among these issues, four are of particular importance, which are large channel system, sensor array geometry, identification of arrival types, and location code selection. The capability of a location code for daily monitoring purpose may be evaluated from four aspects, namely, convergence characteristics, ability of using P- and S-wave arrivals simultaneously, optimization method(s), and reliability analysis.

Acknowledgements

I thank J. Hower, A. Iannacchione, and X. Luo for reviewing the manuscript. Their comments greatly improved the manuscript.

References：

Bharti Engineering Associates Inc., 1993 Bharti Engineering Associates Inc., 1993. A technological review of the seismic and microseismic monitoring systems in Ontario mines. Blake and Hedley, 2003 W. Blake and D.G.F. Hedley, Rockbursts: Case Studies from North American Hard-Rock Mines, Society for Mining Metallurgy and Exploration, Inc., Littleton, CO (2003).

Cook, 1963 N.G.W. Cook, The seismic location of rockbursts, Proc. 5th Rock Mechanics Symposium, Pergamon Press, Oxford (1963), pp. 493–518.

Ge, 1995M. Ge, Comment on “Microearthquake location: a non-linear approach that makes use of a Simplex stepping procedure” by A. Prugger and D. Gendzwell, Bull. Seismol. Soc. Am.85 (1995), pp. 375–377.

Ge, 1988 Ge, M., 1988. Optimization of Transducer array geometry for acoustic emission/microseismic source location. Ph.D. Thesis, The Pennsylvania State University Department of Mineral Engineering, 237 pp.

Ge, 2003a M. Ge, Analysis of source location algorithms: Part I. Overview and non-iterative methods, J. Acoust. Emiss.21 (2003), pp. 14–28.

Ge, 2003b M. Ge, Analysis of source location algorithms: Part II. Iterative methods, J. Acoust. Emiss.21 (2003), pp. 29–51.

Ge and Kaiser, 1990 M. Ge and P.K. Kaiser, Interpretation of physical status of arrival picks for microseismic source location, Bull. Seismol. Soc. Am.80 (1990), pp. 1643–1660. View Record in Scopus | Cited By in Scopus (9)

Geiger, 1910 L. Geiger, Herbsetimmung bei Erdbeben aus den Ankunfzeiten, K. Gessell. Will. Goett.4 (1910), pp. 331–349.

Geiger, 1912 L. Geiger, Probability method for the determination of earthquake epicentres from the arrival time only, Bull. St. Louis Univ.8 (1912), pp. 60–71.

Gendzwill and Prugger, 1989 D. Gendzwill and A. Prugger, Algorithms for micro-earthquake location, Proc. 4th Conference of Acoustic Emission/Microseismic Activity in Geological Structures and Materials, Trans Tech. Publications, Clausthal-Zellerfeld (1989), pp. 601–605. Iannacchione et al., 2004 A.T. Iannacchione, P.R. Coyle, L.J. Prosser, T.E. Marshall and J. Litsenberger, The relationship of roof movement and strata-induced microseismic emissions to roof falls SME Annual Meeting, February 23–25, 2004, Denver, CO (2004) Preprint 04-58, 8 pp..

Leighton and Blake, 1970 Leighton, F., Blake, W., 1970. Rock noise source location techniques. USBM RI 7432.

Leighton and Duvall, 1972 Leighton, F., Duvall, W.I., 1972. A least squares method for improving rock noise source location techniques. USBM RI 7626.

Nelder and Mead, 1965 J.A. Nelder and R. Mead, A simplex method for function minimization, Comput. J.7 (1965), pp. 308–313.

Obert, 1975 L. Obert, The microseismic method: discovery and early history, Proc. 1st Conference of Acoustic Emission/Microseismic Activity in Geological Structures and Materials, Trans. Tech. Publications, Clausthal-Zellerfeld (1975), pp. 11–12.

Prugger and Gendzwill, 1989 A. Prugger and D. Gendzwill, Microearthquake location: a non-linear approach that makes use of a Simplex stepping procedure, Bull. Seismol. Soc. Am.78 (1989), pp. 799–815.

本文来源：https://www.bwwdw.com/article/gvmq.html