PART IV DATA ANALYSIS & RISK ASSESSMENT (Chapters 9-11)

Transcription

1 PART IV DATA ANALYSIS & RISK ASSESSMENT (Chapters 9-11) Australian Centre for Geomechanics 50

2 9. SEISMIC MONITORING SYSTEM DATA ANALYSIS This chapter describes the analysis, interpretation and presentation of quantitative SMS data. The objective is to specify the seismic fingerprint or personality of each mine sector. This can be done using MS-RAP features. A considerable effort should go into analysing the data after the SMS database is validated. Obviously analysis will be limited by inadequate attention to foundational factors such as SMS coverage or data processing. It is necessary to ensure a robust seismicity database. AS/NZS 4360:2004 Risk Management Framework 3.3 Identify risks What can happen, where and when? Tools and techniques 3.4 Analyse risks 9.1 Analysis Principles Routine analysis is mandatory, to allow personnel to become adept at analysing and interpreting, to become proactive about hazards, and to relate anomalous seismic response to any other indicators. There are numerous techniques for the analysis and interpretation of event data from a seismic monitoring system. Some are direct and practical and can be done in limited time on mine sites. Some are more the domain of consultants or researchers. Some suit certain types of seismic behaviour. Regular analysis will establish which techniques are best for the Mine. The engineer performing this task must have a powerful computer with dual screens, and familiarity with the mine s visualisation software package. It is important to note that MS-RAP analysis depends on the source of the energy causing failure being due to the mining itself (e.g. pillar failure, abutment failure, which are directly induced by mining) rather than being regional (e.g. slip on major faults, which may be triggered by mining in a rockmass that is already loaded by stress and close to a state of instability). 9.2 Frequency of Analysis and Reporting Hudyma and Brummer (2007) note that future seismicity in a mine is strongly related to past seismicity in a mine. Our greatest opportunity to become proactive about mine seismicity hazards and risk is through routine back-analysis and interpretation of seismic data. Despite this, the impetus to back-analyse seismic data is often a very large event or a major incident. There is no question that the more seismic analysis we do, the better we get at doing the analysis and interpreting the results. This will undoubtedly lead to a better understanding of the local seismic response to mining, and to the capabilities and limitations of the seismic monitoring system. Alternatively, if seismic data is not routinely analysed, it is possible that a data collection problem will arise in the seismic monitoring and go undetected for an extended period of time. Only when it is suddenly important to analyse the seismic data, is it realised that there is a problem, or that the data is inadequate. The outcome of analysis and interpretation is to be documented. Ensure good communications of monitored data - updates in pre-shift meetings, regular geotechnical presentations on seismicity, display seismic monitoring results. Analysis is necessary on several time levels: a) daily and weekly verify system functionality; identify larger events and recent trends affecting production; look for anomalous rockmass response; activate any required re-entry restrictions. b) after-blast review look for anomalous rockmass response; activate any required re-entry restrictions. c) short-period review, e.g. monthly update and communicate seismic hazard status. d) long-period review, e.g. yearly, or an external review or audit knowledge synthesis. Australian Centre for Geomechanics 51

3 Periodic formal reports, say monthly, are suggested to summarise seismicity, and disseminate feedback information sufficiently frequently and regularly. Mercier-Langevin & Hudyma (2007) and Hudyma & Brummer (2007) suggest that such reports may: Show plan and section of the seismically active areas List and discuss all large events and their suspected causes List all damaging events and the support system in use that was damaged Present facts per mine sector (emergence of new clusters of events, spikes in activity, inferred causes) Identify short term (1 month) abnormal trends in seismicity often associated with blasts Identify long term (1 year) abnormal trends in seismicity (often associated with large scale structures or geometric configurations) Present a detailed seismic risk analysis from MS-RAP including ramifications for production Identify seismicity management failures and successes To make these periodic reports possible, it is clearly necessary for the Mine to keep very good historical records, and have a robust system for records and information management. MSRRM Major Sponsor mines have the benefit of an analysis of their seismicity data by MSRRM staff. State if this has been done, and if so when, where the records are, and the key findings. 9.3 Clustering and Cluster Tracking Create and update seismic clusters and groups in MS-RAP. Analysis is much more successful if logical groups of seismic events are analysed together. This is discussed further in Appendix A. Create clusters and allocate to groups. Set up groups of events defined by location, faults, structures, dykes, other mining or geological features. The premise is that SMS clusters give advance notice of where significant activity may be during stoping, and give early identification of seismically active structures. Groups need 100 or more events over some months to provide a degree of confidence in the analysis. Avoid analysis that is highly dependent on one or a few large events. Update MS-RAP frequently (e.g. monthly) with surveys, faults/structures, geological refinements, mining development and stopes The site Operation Manual needs to specify the location of these files and how to perform the updates. The opening screen of MS-RAP is the Cluster Tracking Monitor (or CTM), an example of which is shown in Figure 22. The CTM contains the listing of seismic cluster groups at the mine, and their activity over time periods selected by the pull down menus at the top of each column. Also shown for each time period selected are the total number of events in each group (#), the number of significant events (#S) and the number of large events (#L). The Seismic Hazard Scale rating, the confidence in the rating and mechanism for each group is shown on the far right. It is important to remember that the data which is ticked (note the check boxes next to the cluster group s names) and the time setting for the far right column (set to All in Figure 22) determines which seismic events will be included in all subsequent analysis. The tabs above the CTM also allow seismic data filtering by time, space or quality. New clusters will appear at the bottom of the CTM as unnamed. It is important to group these clusters and give them names according to nearby locations, mining geometries or geological features. CTM cells shown as orange indicate that the frequency of events for that cluster group and for that time interval is 2 to 5 times the long term average. Red cells indicate the frequency is above 5 times the long term average. This is simply to highlight which clusters or areas of the mine are currently seismically active and whether or not this activity may be anomalous. You can also sort the columns in the CTM by clicking the #, #S or #L headings. This is a good way to get a quick indication of which areas of the mine are seismically active. Australian Centre for Geomechanics 52

4 Figure 22 - Cluster tracking monitor from a MSRRM sponsor s mine. 9.4 How to Perform Daily Seismic Data Analysis At most mines, there is a desire or need to get a quick overview of the recent seismicity in the mine, particularly the events in the last shift or last day. This information is often required at an operations meeting at the start of the morning shift. Typical information required may be: Where have the events been over the last day? Has there been an unusual pickup of activity in the mine in the last 24 hours? Have there been any large magnitude events in the mine in the last 24 hours? What was the seismic response to recent mine blasting? These questions generally relate to short-term seismic hazard in the mine. Specifically: Where is it active in the mine? Is there an increased likelihood of a large magnitude event? While there is no good seismological technique for measuring short-term seismic hazard, with time and experience people develop a feeling for short-term seismic hazard by looking at recent event rates and the location of recent significant events. This section presents a standard procedure for analysing and interpreting seismic data on a daily basis with MS-RAP. Ideally, by following standard procedure, rock mechanics engineers and technicians use the available tools to develop a better feeling for what is normal and unusual at the mine. An important feature of MS-RAP is to generate fast reports to allow quick interpretation of seismic data. In a few minutes, a user should be able to get a very good idea of the location, size and number of events that have occurred in the last 1 to 3 days, and generate a graphical report containing this information. The flowchart in Figure 23 is a model for integrating MS-RAP into a daily seismic data interpretation routine. There are five main steps: Australian Centre for Geomechanics 53

5 1. Import unprocessed seismic data into MS-RAP (typical duration: 1 minute), 2. Generate a daily report in.pdf format (typical duration: less than 2 minutes), 3. Short-term seismic hazard assessment (typical duration: 5 to 15 minutes), 4. Manually process seismic data on seismic system (typical duration: 0.5 to 4 hours), and 5. Update MS-RAP with manually processed events and make seismic hazard interpretation (typical duration: minutes). After the first 3 steps, the user should be able to make an assessment of the recent events in the last day or few days. Note that most Australian mines with ISSI seismic monitoring systems now make use of the ISSI remote processing service. When the geotechnical engineer arrives on site each morning, much of the previous 24 hours worth of seismic data has been processed. At these sites, step 4 is likely to be far less time consuming than what is shown in Figure 23. Each of the steps is described in detail below in the following sections. Figure 23 - Flowchart for performing daily seismic data analysis. Typical Duration 1 minute Import Data into MS-RAP Use MS-RAP Scheduler to load all new events into MS-RAP Trigger short-term update 2 minutes Generate Daily Report Make list of events Plot mine plans with significant and large events with report generator minutes Seismic Hazard Assessment Look for areas with: High frequency of events Abnormal number of significant events Any occurrence of large events Watch for mine blasts that have slipped past the blast filters Print key charts Prepare to distribute or show the key pages of the report or discuss recent seismicity in a morning/operations 0.5 to 4 hours Process data manually Focus on: Location of significant and large events Elimination of significant and large outliers Process data manually on seismic system minutes Update MS-RAP: Update events Update mine plans, if necessary Input blast details for Omori analysis Final Data Interpretation Load manually processed events into MS-RAP Events changed or deleted in manual processing will be automatically changed or deleted in when MS-RAP is updated Identify significant/large outlier events blasts tag as outliers and blasts in MS-RAP Look for new clusters Look for unusual activity rates, particularly significant size events Australian Centre for Geomechanics 54

6 Import Data into MS-RAP Key Steps: 1. Start the MS-RAP Scheduler. 2. Click on Trigger button for Import Recent Data. This will update MS-RAP with new events or changes in events in the last 14 days. 3. You can shut down the MS-RAP Scheduler, if you chose. It is no longer required until more data is to be loaded into MS-RAP. Note that MS-RAP is designed to replace or update previously imported events. In addition, if an automatically processed event is later deleted during manual processing with WaveVis TM or JMTS TM, MS-RAP will purge the automatically processed event from the MS-RAP database in the next data import. Basically, you should have no concerns about importing automatically processed events into MS-RAP. Generate a Daily Report Key Steps: 1. Start MS-RAP. 2. Open the Report Manager in MS-RAP (Figure 24). 3. Ensure that the desired report settings are selected. Recommended report settings should include: o List of events, o Summary of the Cluster Tracking Monitor o Long section of the mine, showing all event locations, o Plan view of all large events or if there are 3 or more significant events on a plan, and o A Magnitude vs Time chart for all of the events. Ensure that the Backdate Report button IS NOT ticked (Figure 24). Figure 24 - Report Manager in MS-RAP. A reasonably good seismic sensor array should give sufficiently accurate automatic source locations to identify unusual seismic event behaviour. If your automatic source locations are not reliable enough for a first pass data analysis, re-calibration or improvements to the seismic system should be considered. Australian Centre for Geomechanics 55

7 Daily Seismic Hazard Assessment The Cluster Tracking Monitor and charts generated in the daily report will show whether there have been unusual levels of activity recently. If there is little activity, or the level of activity is normal a short-term seismic hazard assessment may not be required. If there is unusual activity, steps should be undertaken to investigate the unusual seismic activity. What is unusual recent seismic activity? A substantial increase in event frequency may be expected following a mine blast, but blastinginduced stress-change events are mostly small events. Unusual recent seismic activity may be: o An substantial increase in event frequency not related to mine blasting, particularly in one area of the mine, or o A substantial increase in the number of significant events, particularly in one area (the CTM should identify this), or o The occurrence of any large events. Key Steps: If unusual recent seismic activity is found or suspected, interactive use of MS-RAP should be used to investigate the unusual behaviour. 1. Look for areas in the mine with a high frequency of events. You can do this by looking at the mine plan plot, or with the summary of the Cluster Tracking Monitor. o Are there unusual concentrations of seismic events? Has there been a mine blast to cause these events? o Is there a recent concentration of significant events? o Are there any large events? 2. Magnitude Vs Time Chart. This is the most important tool. o What is a normal event frequency? The cumulative number of events line in the Magnitude vs Time chart will show whether the number of events is above normal. A change in slope of the cumulative line represents an increase in event frequency. o Has there been a mine blast that caused unusual event increases? Mine blasts typical cause a Step in the cumulative number of events line in the Magnitude vs Time chart. The number of events after blasts may increase dramatically for a few hours. o In general, what is the shape of the cumulative number of events line in the Magnitude vs Time chart? Step-like behaviour is seismicity induced by mine blasting. A gradual slope is seismicity independent of blasting. o For this part of the mine or cluster group, using a Magnitude vs Time chart over the last 12 months, when are the large events occurring? Are the large events occurring primarily with blasting (in conjunction with steps in the chart) or are the events occurring unrelated to blasting. The Diurnal chart will also show the time of day of the significant seismic events. 3. Are there large events? o Have large events occurred in this area in the recent past? Recent large events in this area should be considered a very strong warning sign of high short-term seismic hazard. o Have large events ever been recorded in this area. Large events that have ever occurred in an area imply a relatively high seismic hazard. It is rare for the ground to fail and become destressed" causing a decrease in seismic hazard. It is necessary for a user to become very comfortable with using MS-RAP to investigate individual cluster groups using the Magnitude Vs Time, b-value, diurnal and mine plan plotting analyses in MS-RAP. This will enable you to conduct these analyses quickly during a daily seismic hazard assessment. Watch for mine blasts that may have been missed by the MS-RAP blast filters. These will be significant or large events that occur at blast time. Australian Centre for Geomechanics 56

8 Make note of significant and large events that occur with no small events preceding or following the significant and large events, especially if these events do not follow directly after mine blasts. This is unexpected behaviour, and a strong sign of geologically controlled (particularly fault-related) seismicity. Manual Data Processing Manual data processing is the most time consuming task in operating a seismic system, typically taking 30 minutes to several hours to complete. It is important that the manual processing is done routinely so that the best possible data is included in MS-RAP for future analyses. From a seismic data processing perspective, it is far more important to eliminate blasts from the seismic system, than it is to get good source location on small seismic events (local magnitude -2). Below is a recommended seismic data processing list of priorities: 1. Remove or tag all blasts on the seismic system. 2. Locate large events (Richter magnitude +1) as accurately as possible. 3. Locate significant events (Richter magnitude 0) as accurately as possible. 4. Locate smaller sized events (-1 Richter magnitude < 0). 5. Locate small events (-2 Richter magnitude < -1). 6. Locate very small events (-2 Richter magnitude). Ideally, on a daily basis, steps should be done as soon as possible at the start of each day. Steps 4 and 5 should be kept up to date as much as possible, preferably keeping up within a day. Ideally, step 6 is kept up to date when possible, but if something has to be left undone, it is the processing of the small events (-2 local magnitude). Typically the very small events comprise more than half of the events on the seismic system and may not be easy to manually process or locate accurately. It is imperative that all mine blasts are tagged as blasts with the seismic system software (WaveVis TM, JMTS TM ). Blasts pose a serious problem for seismic data interpretation, as they occur frequently (a few to several per day) and in MS-RAP have the appearance of large seismic events. Including only a few blasts per month as real events will severely compromise our ability to understand seismic hazard. Final Data Interpretation Key Steps: 1. Start the MS-RAP Scheduler. 2. Click on Trigger Button for Import Recent Data. This will update MS-RAP with the changes in events made during the manual processing. 3. You can shutdown the MS-RAP Scheduler, if you chose. It is no longer required until more data is to be loaded into MS-RAP. 4. Allocate ungrouped clusters to cluster groups. As new seismicity is recorded and new clusters are generated by MS-RAP, these clusters may potentially be part of existing cluster groups. 5. Allocate significant and large un-clustered events to clusters. Very large seismic events may not be well located and are frequently not clustered. Often the events that occur after large events identify which clusters these large events should be associated with. 6. Add mine blasts to the blast database. The blast database is very useful for building cause-effect relationships between mining and seismicity. Information in the blast database can be displayed in most of the MS-RAP data analysis techniques. If significant or large events are not clustered, check to see if these events can be manually allocated to a cluster (see the user s manual). Ideally, the vast majority of significant and almost all large events are associated with clusters and cluster groups. Chapter 11 describes some approaches for conducting short term, day-to-day seismic hazard assessment. Australian Centre for Geomechanics 57

9 9.5 Other Points Also note the following miscellaneous comments and cautions: Ensure that any seismically quiet periods are correctly handled and understood: If due to SMS downtime or non-operation of certain sensors, the effect of those periods must be removed from or accounted for in charts, graphs and calculations. If due to a temporary lack of mining, be aware that this quiescence does not reflect the seismic potential of an area when mining is active. If due to a theorised physical change in the rock mass (such as an area becoming destressed, or certain pillars fractured, or punching into the footwall) then the truth and extent of the change needs to be demonstrated. Beck (2003) cautioned that some source parameters (source radius, apparent stress, corner frequency, stress drop) produced by seismic monitoring systems are very heavily model dependent, also that the stress drop inferred from SMS will not match the real stress change in the rock. Errors arise due to assumptions, sensor distribution, and asperities. Notes on the Gutenberg-Richter plot (b-value plot): The larger events in a Gutenberg-Richter (GR) plot are few in number by definition, so statistically they appear more variable. It is not appropriate to use these few larger events alone to attempt to forecast a maximum size event. The GR plot should always be accompanied by a statement of the time period (weeks, months etc) of the data comprising the plot. Comparisons of such plots for different clusters/groups should be on the basis of similar time periods. The GR plot should show a fairly linear slope. If not linear, and the number of events is statistically sufficient, review the cluster for: o Appropriate cluster definition does it comprise two (or more) mechanisms superimposed; o Effects resulting from seismic system changes e.g. if a nearby sensor becomes nonfunctional, lower-end events may stop being recorded, warping the GR plot. Australian Centre for Geomechanics 58

10 10. THE SEISMICITY KNOWLEDGE BASE (SKB) This chapter deals with explaining the seismicity in light of the mine environment and mining practices. The repetitive nature of mining provides the opportunity to apply, to future design, experience gained in every prior mining step (Kaiser, 1996). This implies procedures and methods be developed to integrate seismic data with important nonseismic information and considerations (such as mine sequencing, presence of major geological features, stress, numerical modelling). Damage is highly variable according to local site factors, so the latter need to be collected, filed and interrogated to determine their relevance and influence. It is not necessarily easy to develop an engineering appreciation of all the factors that have a part in directing seismicity. The Mine may have some of this material in current and archived documents and papers, which should be consulted accordingly. The outcome of the SKB process will ideally: Define how seismic character is related to the various physical aspects of the Mine. Provide sufficient objective information to allow the rigorous evaluation of the merits and usefulness of the SKB topics (based on Durrheim et al, 2007). Provide material for research, continuous improvement and back analysis. AS/NZS 4360:2004 Risk Management Framework 3.3 Identify risks Why and how can it happen? 3.4 Analyse risks Evaluate existing controls 10.1 Definition of the Seismicity Knowledge Base (SKB) The Seismicity Knowledge Base (SKB) is a database that integrates everything that is known about the seismicity at the Mine. Rockburst risk questions are very complex and often subjective to answer. A large amount of information, both seismic and non-seismic in nature, needs to be synthesised or integrated to arrive at knowledge conclusions. It is necessary to formalise data in an objective and robust manner. It is proposed that for a robust SRMP, the synthesis method is to concisely summarise what is known in a database called the SKB as follows: The SKB contains facts. It summarises knowledge of the specific aspects of the Mine that are considered to be related to seismic damage. The SKB is a layered database, and may have a dozen or more layers. It is a major document requiring time and effort to compile. The SKB has similarities to the Ground Conditions Model (GCM) 2 launched by Western Mining Corporation (WMC) a decade ago. The term Knowledge Base is adopted in the SRMP instead of alternatives such as Model, as the latter can be confused with numerical modelling, with results or interpretations instead of facts, and with geological models. 2 The GCM (Singh et al, 2002) presented data as a series of overlays of parameters such as mine history, structural geology, Q-rating, blast quality, installed ground support, bursts, falls, adverse ground conditions, and stability problems. The GCM was regularly updated with information from the various data gathering methods. The primary function of the GCM was to synthesise the relevant information from the geological /geotechnical databases into an appropriate format for application to mine design, planning and ground support requirements. See also Newmont Australia Ltd (2003). Australian Centre for Geomechanics 59

11 The SKB also represents a feedback loop. It is the knowledge repository and a place where improvements are decided. Entries made to the SKB may trigger further entries to the SRMP, GCMP and related procedures. The SKB is a live database new data is added as it becomes available. The SKB is part of the continuous improvement process. It facilitates the trending of and repair of systematic failures and deviations. As noted by Beck (2003) every failure... should be treated as a failure of the rock mechanics hazard management system... must investigate, and implement into the system the fundamental requirements to avoid or manage the event. It is usually difficult to visually discover relationships among a large quantity of disparate data. Many aspects and uncertainties can muddy the waters. There are tradeoffs between many parameters. It takes time and a comprehensive study of these features to identify the fundamental controls on seismicity. A powerful technique known as CSIRO Self Organising Maps or CSOM (Fraser et al, 2006) is available and is recommended to better understand relationships between the SKB parameters The SKB Structure It is likely that the SKB will comprise layers in more than one form. Possibilities are: A written summary a statement of the factual links between seismicity and a range of parameters. Links can sometimes be expressed as a series of rules. The written summary clarifies thinking. A pictorial SKB (e.g. sets of Level plans) can be viewed and interrogated when assessing seismic clusters and trends. Map on Level plans the extent of and degree of damage due to events, including notes on the installed support, whether that support was adequate, and proximity to stoping, and estimated distance from the source. Via the MS-RAP Minodes function. A future capability will be to attach files to Minodes e.g. incident reports, photos, monitoring data. A database this could contain data relevant to specific occurrences, such as significant seismic events. A set of charts when data lends itself to plotting, do so. Construct charts combining several of the measured or observed parameters. These charts will become tools to develop knowledge and understanding of seismicity at the Mine. Repeat for each mining area if more than one. Most importantly, SKB knowledge is a synthesis of data per location, not per topic (although topical knowledge can and should be extracted from it). The structure of the SKB must be based on location. Whatever the particular structure used physical or electronic it must be designed to collate and view the full range of data available per location. In practical terms, the unit of location will most likely be the stope or the drive The SKB Contents Practitioners are urged to consult Singh et al (2002) for illustrations of the types of information that belong in the SKB. The SKB needs to address the following topics, and perhaps others as relevant, and the interactions between these topics: Location Numerical modelling results Geology Rock Types Australian Centre for Geomechanics 60

12 Rock Properties Significant Geologic Structure Structure Mechanical Properties Rockmass Characterisation Groundwater Stress Regime Ground Conditions Ground Control Measures Excavation Span, Geometry o Development headings o Creation of stress attractors Mining Sequence o Stope sequencing o Stope retreat direction Stope blasts Production rate - rate of extraction of rock Workplace activities Backfill Previous Mining Refer to Appendix E for more information on each of these topics. Australian Centre for Geomechanics 61

13 11. SEISMIC HAZARD AND RISK ASSESSMENT This chapter is about how to define the level of seismic hazard and risk in The Mine. It describes techniques for assessing seismic hazard in the short term, as well as a procedure for the demarcation of high seismic hazard areas and assigning a risk level, based on the seismic system data analysis plus key mining / environment parameters. A quantitative framework for risk assessment is used in MS-RAP. It stands in contrast to the qualitative approaches. Qualitative methods use words to describe consequences and likelihoods. They are characterised by being based on experience, taking time and skill to develop at a new Mine, relating hazards to similar conditions (which can over-estimate some hazards and miss others) and relying on correct interpretation of seismic mechanisms. The MS-RAP quantitative approach uses numerical values for likelihood, consequences and risk. This permits the objective measurement of the success or failure of the particular risk assessment. This approach has the benefit of being based on real time data, but it is important to note that: Estimates are completely dependent on data quality and an adequate seismic system Robust data management systems are necessary A wildcard with both qualitative and quantitative approaches is that future seismic hazard is not guaranteed to be related to past seismic hazard. If for example tectonic stresses are locked into the rockmass, future seismicity can be much larger than could be inferred from past seismicity. Both these methods rely on input from past seismicity, and cannot make provision for such as these. Therefore it is critical to refer to the SKB for guidance in the risk assessment process. AS/NZS 4360:2004 Risk Management Framework 3.4 Analyse risk Consequences and likelihood Types of analysis Sensitivity analysis 3.5 Evaluate risks 11.1 Short-Term Seismic Hazard Assessment There are no consistently reliable short-term seismic hazard assessment techniques. However, at some mines, rock mechanics personnel have developed a local intuition or understanding of seismic hazard by reviewing and analysing seismicity on a daily and weekly basis. On a regular basis, rock mechanics personnel are required to use their best judgement or intuition with regard to the likelihood of hazard seismicity. MS-RAP can aid that judgement, by: Providing real-time data analysis tools, The means of back-analysis of past seismicity, Development of intuition is still in the hands of the operator. MS-RAP is simply an application to aid that development. One of the first applications of seismic monitoring systems was to analyse recent activity rates, particularly following mine blasts and large events. If the seismic event occurrence rate was unusually high, a management decision was made to exclude or keep out of the active areas. The activity rate to cause exclusion is based on local site experience, allowing significant potential for bias and subjectivity. The nature of failure in a rock mass is that it is rarely instantaneous. It is a gradual process that is caused by a set of conditions related to stress, geological features, and the influence of mining. Australian Centre for Geomechanics 62

14 The process evolves over time gradually growing and accelerating. This process generates significant seismicity that often gives strong indication of the seismic hazard (maximum potential event size) and the seismic source mechanism. The objective of proactive use of MS-RAP is to identify high seismic hazard situations and to identify patterns in seismicity. Objectives There are three primary considerations in short-term seismic hazard assessment: Where is seismic hazard elevated, When is a high hazard event most likely to occur, and How big could the event be. There are different MS-RAP applications for each of these considerations. Tools and Techniques In MS-RAP, the main tools for short-term seismic hazard assessment are: Spatial plotting of event magnitude, Frequency-magnitude analysis, Magnitude-Time History analysis, Apparent Stress Time History analysis, Event frequency and the Cluster Tracking Monitor, Instability analysis (EICAV), Daily histograms, and Diurnal analysis. For clustered seismic events in MS-RAP, each of these techniques has been subjectively rated for their value in determining where, when and how big an event may be (Table 8). The success rate of each technique has also been subjectively rated. Table 8 - Success in short term seismic hazard assessment. Short Term Seismic Hazard Assessment Techniques Where? When? How Big? Success Rate Spatial Plotting of Magnitude Frequency-Magnitude Analysis Magnitude Time History Analysis Apparent Stress Time History Cluster Tracking Monitor Seismic Hazard Map Instability Analysis Daily Histogram of Event Frequency Diurnal Analysis Notes: =Not useful =Rarely useful =More successful =Most successful No single technique is universally applicable or infallible. Seismic hazard is best evaluated using a combination of techniques. The following pages describe some of the techniques available in MS-RAP for short-term seismic hazard assessment. A more detailed discussion of the techniques can be found in Appendix A and in the MS-RAP electronic help file or manual. Australian Centre for Geomechanics 63

15 Spatial Plotting of Event Magnitude Where? When? How Big? Success Rate Spatial Plotting of Magnitude Description: Plotting of seismic events in 2D and 3D can be very revealing about future seismicity. When possible, plotting should be done on up-to-date mine plans and include all past stoping, and geological features. Seismic Hazard Considerations: The largest local event is an indicator of the local worst case event. Where are the significant and large events? Future significant and large events are likely to be close to past events. How many significant and large events have occurred? If there are numerous significant and large events, there is a significantly greater likelihood of future events. If there are few or no significant or large events, the likelihood of such an event in the near future is reduced. Frequency-Magnitude Analysis Where? When? How Big? Success Rate Frequency-Magnitude Analysis Description: Frequency-magnitude analysis is the fundamental technique for seismic hazard assessment in mines. Seismic Hazard Considerations: A lower b-value generally implies that there is a greater proportion of large events in a particular population of events. This fact alone suggests that there is an increased likelihood of larger events. The x-axis intercept of the theoretical frequency-magnitude relation is often a good indicator of the largest potential event in the group. This is particularly true when the frequencymagnitude relation is well-behaved and linear. Groups of events that have few or no significant and large events have a lower seismic hazard. Groups of events with numerous significant and large events have a higher relative seismic hazard. Magnitude-Time History Where? When? How Big? Success Rate Magnitude Time History Analysis Description: Magnitude-time history analysis is perhaps the most valuable tool for investigating seismic hazard. Australian Centre for Geomechanics 64

16 Seismic Hazard Considerations: Look for an increase in the number of events (change in slope in the blue Cumulative Number of Events line). Larger events often occur during periods of increasing event frequency. Kijko and Funk (1994) give a very crude rule of thumb of the largest expected event: where: m max = X max + (X max X n-1 ) m max is the largest expected event, X max is the largest event that has occurred, and X n-1 is the second largest event that has occurred. So if the largest event is local magnitude +1.5 and the second largest event is +1.1, the largest expected event would be Are step increases in event frequency due to mine blasting? Do the significant and large events tend to occur during periods of high event frequency? This suggests that large events are primarily caused by stress change due to mine blasting. Do significant and large events occur at times of low event frequency? This suggests that the significant and large events occur largely unrelated to mine blasting. This is typical of fault-slip seismicity. There is a setting to allow SHS to be shown for a cluster. This is helpful for identifying situations in which unexpected large events could occur. In the Blast Panel of MS-RAP, there is a setting to allow mine blasts to be shown on the magnitude-time history chart. This is helpful for building a relationship between mine blasts and mine seismicity. When events are occurring, is the maximum event magnitude increasing and are there changes in event frequency? This is analogous to the local deformation rate. Apparent Stress Time History Where? When? How Big? Success Rate Apparent Stress Time History Description: Apparent Stress is a seismic source parameter measuring the relative amount of energy in a particular seismic event. It has been shown that when stress is increasing, Apparent Stress is higher than expected. Apparent Stress Time History is a means of quantifying changes in Apparent Stress over time. Seismic Hazard Considerations: In many cases, large seismic events occur shortly after stope blasting. Typically the seismic events following mine blasting have unusually high Apparent Stress. Apparent Stress Time History can be used to show when stress change is occurring and the likelihood of large seismic events is higher (elevated seismic hazard). To successful apply Apparent Stress Time History, it needs to be used in back-analysis to best determine analysis dependent parameters. Apparent Stress Time History is less likely to be an indicator of high seismic hazard for seismicity that is not related to stress change, in particular to seismicity on mine faults. Large events on mine faults often do not have any significant precursory seismicity, so the Apparent Stress Frequency is likely to be low. In the Blast Panel of MS-RAP, there is a setting to allow the date of mine blasts to be shown on the Apparent Stress Time History chart. Australian Centre for Geomechanics 65

17 Event Frequency and the Cluster Tracking Monitor Where? When? How Big? Success Rate Cluster Tracking Monitor Description: The Cluster Tracking Monitor (CTM) is used to evaluate seismic event frequency for clustered seismic data. The CTM allows the user to view the number of seismic events in mine clusters for four different time periods. Typically one time period is selected for: Short term (1 to 3 days), Medium term (1 to 3 months) Longer-term (in the range of 3 months to 1 year), and Very long term (a few years or more). For each cluster and group in MS-RAP, the long term frequency is calculated for the number of events, the number of significant events and the number of large events. For each time frame, the number of events recorded is compared to the long term frequency. If the frequency of events in a particular time frame is more than two times the long-term frequency, the time frame is coloured orange. If the frequency of events in a particular time frame is more than five times the long-term frequency, the time frame is coloured red. Seismic Hazard Considerations: The CTM alone is the single most important feature in MS-RAP. It shows where and when events are occurring, in particular the significant and large events. If a group of events has had few or no significant and large events, the seismic hazard is low. For example, in Figure 25, group G20 has had more than 4727 seismic events, but only 8 significant and no large seismic events. Despite the high number of events, seismic hazard is relatively low If a group of events has had many significant and large events, the seismic hazard is high. For example, in Figure 25, group G17 has had 1525 events, which is relatively low, but it has 62 significant and 9 large events. This group has the second highest number of significant and large events (after group G13). The CTM also shows the number of un-clustered seismic events. If there are a considerable number of un-clustered seismic events (particularly significant and large seismic events), the level of confidence in the MS-RAP clustering is reduced. In Figure 25, only 29 of the significant and only 1 of the large events have not been clustered. This is very good clustering, with almost all of the significant and large events in clusters. Figure 25 - Example of a Cluster Tracking Monitor. Australian Centre for Geomechanics 66

18 Each Time-Magnitude cell in the CTM shows if the event frequency is 2 times or 5 times greater than the long-term frequency. When a particular group of events shows high event frequency for multiple cells, the seismicity can be considered anomalous. The Backdate feature can be used to investigate historical event frequencies. This is a useful exercise prior to large events or during the blasting of particular stopes to investigate past seismicity rates, and to develop local rules of thumb on what are normal and abnormal levels of seismicity. Seismic Hazard Mapping Where? When? How Big? Success Rate Seismic Hazard Map Description: Seismic hazard mapping is a combination of three MS-RAP seismic hazard techniques: spatial event plotting, the Cluster Tracking Monitor, and frequency-magnitude analysis. MS-RAP generates Minodes, which are automatically generated nodes spaced on about 5 metre spacing for all mine development. There are typically several thousand Minodes calculated for an average size mine. Seismic hazard is calculated for each Minode and is updated with the occurrence of each new seismic event loaded into MS-RAP. Seismic Hazard Considerations: The greatest benefit of a seismic hazard map is that it is overlain onto mine development allowing the worst-case seismic hazard to be viewed for the mine development. This information can be compared to the location of past events and used in determining ground support requirements and exclusion zones following mine blasts. Instability Analysis (EICAV) Where? When? How Big? Success Rate Instability Analysis Description: Use of Instability Analysis (EICAV) is relatively common in South Africa. These analyses are rarely successfully used at Australian or Canadian mine sites. A local level of confidence in the application of Instability Analysis should be achieved before it is used as a short-term seismic hazard tool. This is done through back-analysis of past seismicity, focussing on tuning the Energy Index averaging parameters. Seismic Hazard Considerations: An increasing or variable Energy Index (EI) is purported to be an indicator of increasing stress. A decrease in EI is potentially related to rockmass destressing. This is often seen as the key characteristic of this analysis. A sharply decreasing EI suggests that significant rockmass failure has occurred and may be a precursory indicator of a large event. The Cumulative Apparent Volume (CAV) is related to the rockmass cumulative deformation. A flat CAV suggests that no significant rockmass deformation has occurred. A steadily increasing CAV suggests accumulating rockmass deformation. Large steps in CAV suggests that seismic energy is being released in large seismic events. There is a setting to allow the date of mine blasts to be shown on the EICAV chart. Australian Centre for Geomechanics 67

19 Limitations: Successful use of Instability Analysis requires a high degree of local experience in applying the technique. Confidence in this technique can only be gained through successful back-analysis of local seismic data. The technique also relies on a high level of system sensitivity in the zone it is applied (possibly as far down as Richter Magnitude -2.0). Daily Histogram of Event Frequency Where? When? How Big? Success Rate Daily Histogram of Event Frequency Description Amongst the first applications of seismic monitoring was identifying anomalous event frequency levels. If there are an unusually high number of events, the ground is perceived to be working and is more likely to have a significant ground failure. This is still the most widely used approach, particularly after mine blasts. Seismic Hazard Considerations: A daily histogram allows the user to compare recent daily event rates to past daily event rates. There is a setting to allow the user to calculate a past moving average. The default is 30 days. This allows the user to look for long term trends, such as a steadily increasing event rate. In the Blast Panel of MS-RAP, there is a setting to allow the date of mine blasts to be shown on the daily histogram chart. Diurnal Analysis Where? When? How Big? Success Rate Diurnal Analysis Description: Diurnal (or time of day analyses) can be used to determine whether the majority of events are occurring directly as a result of blasting, or whether they are independent of blasting. Diurnal charts are very simple. They show the hour of the day in which the seismic events occur. Seismic Hazard Considerations: A diurnal chart allows the user to determine whether or not seismic events (particularly significant events) occur directly following blasting or with no obvious relationship to blasting. Commonly seismicity associated with faults or other major structures exhibit little or no relationship to blasting. Whilst the diurnal chart shows little information about the level of seismic hazard, it provides some indication as to the time when this hazard may occur. Other Considerations Most Probable versus Worse Case Seismic Event The user has to choose whether seismic hazard should consider the most probable large seismic event, or the worst-case seismic event. Selecting the worst-case seismic event will provide the most conservative seismic hazard evaluation, but this may be unduly pessimistic. The most probable seismic event may be more appropriate: Australian Centre for Geomechanics 68

20 In cases of relatively low mining extraction, During development mining activities, When there has been no recent stope mining in proximity of a workplace (in the last few months). The worst-case seismic event may be more appropriate: In areas of high extraction, i.e. near large underground voids, or in mine pillars, When production blasting is in progress in the proximity of a workplace, Near geological features, with a past adverse seismic history, which show signs of movement. In MS-RAP, the entire seismic record should be used when assessing the worst case seismic event. A most probable event may be more realistic for areas that have not experienced a high event rate for a long period of time. In this case, the user could consider using only a more recent time period for estimating the largest potential event. For instance, if an area of the mine has not had a significant event rate for five years, a most probable event could be calculated using frequencymagnitude relation for the last 3 or 4 years of seismic data. The Role of Clustering in Understanding Seismicity The seismic hazard methodology proposed for MS-RAP is founded upon the concept that the seismic related failure in mines occurs as a result of some combination stress, geological structure, and the influence of mining. Clusters of seismic events represents rockmass failure processes, or sources of seismicity. If an adequate seismic record has been collected, the seismic hazard (or worst case seismic event) associated with each seismic source can often be estimated. There are numerous assumptions and limitations with this regard, which are detailed in the MS-RAP manual or electronic help file. Exceptions There will always be a percentage of seismic sources for which the seismic hazard potential is much greater than could be inferred by past seismicity. In some of these cases, tectonic stresses are "locked" into the rockmass, and the influence of nearby mining triggers seismic events much larger than would be expected. MS-RAP cannot make provision for such cases. However, based on several years of experience of detailed analysis of seismicity in mines, the relative frequency of occurrence of unexpected, large, "triggered" seismic events is very low. There are some similarities in some of the cases of these events: They occur associated with faults that are significant in spatial extent, and They tend to occur during development mining or at relatively low levels of extraction. It is assumed that a seismic hazard methodology can be developed without the consideration of the occurrence of unexpected, large, "triggered" seismic events. Australian Centre for Geomechanics 69