SIM040207 - Bushveld Closure Guidebook - 2006

Document Sample
SIM040207 - Bushveld Closure Guidebook - 2006 Powered By Docstoc
					Guidelines for measuring and analysing
continuous stope closure measurements
   in platinum mines of the Bushveld

                D.P. Roberts
                 D.F. Malan
          A.L. Janse van Rensburg
                 M. Grodner
                 R. Handley

         Mine Health and Safety Council
                   June 2006
                                           TABLE OF CONTENTS

INTRODUCTION....................................................................................................................... 3
TERMINOLOGY........................................................................................................................ 4
GENERAL CHARACTERISTICS OF CLOSURE IN THE BUSHVELD.................................... 8
USE OF CLOSURE TO WARN OF IMPENDING COLLAPSES ............................................ 12
USE OF STEADY-STATE CLOSURE RATE DISTRIBUTIONS ............................................ 18
USE OF THE STOPE CLOSURE DECAY PARAMETER...................................................... 23
GUIDE ..................................................................................................................................... 25
CLOSURE SURVEYS............................................................................................................. 28
GUIDELINES FOR CLOSURE MEASUREMENTS................................................................ 31
INSTRUMENTATION TO MEASURE CONTINUOUS CLOSURE......................................... 32
REFERENCES........................................................................................................................ 36


          A: Elastic convergence solution for the incremental                                                               37
               enlargement of a stope
          B: Correction procedure for clockwork closure data                                                                39


This booklet was produced at the behest of the Mine Health and Safety Council (MHSC) to
provide a general guide for measuring and analyzing stope closure in the platinum mines of
the Bushveld Complex.

This booklet is the culmination of MHSC project SIM040207 - Closure profiles of the UG2
and Merensky reef horizons at various depths, using different pillar types, in the Bushveld
Complex (Roberts, et al., 2006) – conducted from 2004 to 2006. An extensive monitoring
programme was undertaken involving continuous and long-period closure measurements on
various sites on the Merensky and UG2 reef horizons. Sites with different pillar
configurations, in different areas and at various depths were monitored. Geotechnical
evaluations and numerical modelling were also conducted to aid comparison of data between
the sites.

Analysis of the data revealed that steady-state closure rate distributions were useful in
identifying changing geotechnical conditions. A methodology was suggested for regular
systematic evaluation of closure profiles. An exponential function characterised by the stope
decay parameter (SCD) was also shown to indicate the relative stability of panels. A closure
survey technique was also used to quickly estimate the distribution of closure within panels.

This booklet provides basic characteristics of closure in the Bushveld Complex platinum
mines and presents a number of methods and parameters which are useful in detecting
changes in closure behaviour.

This is a first step towards using closure data as leading indicators of hazardous or
deteriorating rockmass conditions. At some point in the future, it is envisaged that each
stope will contain multiple closure stations which will be monitored in real-time, providing
alarms to indicated increase fall-of-ground (FOG) potential or imminent panel collapse. The
nature of these stations and the criteria used to generate alarms are not known at this point.
It is left to practitioners, consultants and researchers to conduct meaningful analysis of such
data and provide these critical parameters.

The section is repeated (with some amendments) from Malan (2003). Note that the
definitions of closure and convergence below do not agree with the original terminology
proposed by the ISRM in 1975. As the meaning of these words in reports and publications in
the South African industry has changed with time, it is considered more appropriate to use
the definitions given below.

•   Closure: Relative movement of the hangingwall and footwall normal to the plane of the
    excavation (Figure 1).
•   Ride: Relative movement of the hangingwall and footwall parallel to the plane of the
    excavation (Figure 1). It should be noted that for a dipping tabular excavation in three
    dimensions, the ride consists of two components namely ride in the strike direction and
    ride in the dip direction. In some cases an apparent ride component is also observed if
    the fractured rock of the footwall moves into the strike and dip gullies. This is very
    common on the Vaal Reef (Roberts, 2003)
•   Convergence: Elastic component of closure. For deep tabular stopes, the closure is the
    sum of the convergence and inelastic movements caused by fracturing and the slip and
    opening of discontinuities such as bedding planes.



                                     DIP ANGLE

                             Figure 1. Definition of closure and ride.

•   Long period closure measurements: Discrete closure measurements with a typical
    interval of 24 hours or longer between successive data points. These measurements are
    commonly plotted as a function of distance to face or time (usually in days or months).

    Figures 2 and 3 illustrate some examples. Note that if closure is plotted as a function of
    distance to face, the graph does not go through the origin as the closure instruments are
    always installed some distance behind the face. The data in Figures 2 and 3 was
    recorded by closure stations at a fixed location in a panel and, with time, the face moved
    away from the measurement point due to regular blasting. These results can be
    misleading as the curves are a complex aggregate of face position changes and the
    time-dependent behaviour of the rock.




                                     1                                       Graph of closure versus time
                                                                             at Blyvooruitzicht
                                                         Graph of closure versus time
                                                         at West Driefontein

                                          0   50   100    150    200   250    300       350   400   450     500   550

                                                                    TIME IN DAYS

    Figure 2. Typical closure curves obtained from long period measurements (after
       Wiggill 1965). In this example the closure is plotted as a function of time.

Figure 3. A typical closure curve obtained from long period measurements where the
 closure is plotted as a function of distance to face (after Gürtunca & Adams 1991).

•   Continuous closure measurements: Closure recorded in a continuous fashion with
    suitable instrumentation such as clockwork closure meters. Closure collected with
    electronic data loggers with a sample frequency of greater than 1 sample/15 minutes will
    also be referred to as continuous. These measurements are always plotted as a function
    of time (usually in minutes or hours). An example is given in Figure 4. Note that the

      effects of changes in geometry, the time-dependent behaviour and seismic events are
      clearly distinguishable.
•     Time-dependent closure: Slow ongoing closure observed between successive blasts
      when there is no change in the mining geometry. This consists of a primary and steady-
      state phase as indicated in Figure 5.
•     Primary closure phase: This is the component of time-dependent closure following a
      blast and is characterised by a period (≈ 3 to 5 hours) of decelerating rate of closure. It is
      also observed after large seismic events.
•     Steady-state closure: The component of time-dependent closure following the primary
      closure phase (see Figure 5). The rate of steady-state closure appears to be constant in
      the short term but it gradually decreases when there is no blasting or seismic activity.



          Closure (mm)

                      30         SEISMIC EVENT

                      20                                                        BLAST

                             0   1000   2000     3000   4000    5000     6000    7000   8000    9000
                                                        Time (minutes)

    Figure 4. An example of continuous closure data collected in a VCR stope where the
    hangingwall consists of hard lava. Note that the unit of time in this figure is minutes,
                                while it is days in Figure 2.

•     Instantaneous blast closure: The instantaneous closure component occurring at
      blasting time (Figure 5). Due to the delays in the detonation sequence between adjacent
      blast holes in the face, this closure phase is not really instantaneous but can last for
      several minutes.

•     Instantaneous seismic closure: The instantaneous closure component occurring during
      a seismic event. Similar to the blast closure, the instantaneous seismic closure is followed
      by a primary and steady-state closure phase.

                                                   Steady-state closure

                           Primary closure phase


                                                               ∆t                 ∆ST


Figure 5. Typical continuous stope closure after blasting and the definition of closure

•   Rate of steady-state closure (RSSC): Defined as

                                                        SSS =                                               (1)

    where   ∆Sss      is the increase in steady-state closure over the period ∆t (see Figure 5). As

    the rate of steady-state closure gradually decreases in the absence of blasting activity, it
    is important to mention the period used for the calculation. The period                 ∆t   is often taken
    from six hours after the blast (to avoid the effect of the primary phase) to 24 hours after
    the blast (or until the next blast occurs, whichever comes first).

•   Closure ratio: The ratio of the instantaneous blast closure to total closure following a
    blast. To avoid to risk of making inappropriate conclusions about sudden unexpected
    increases in closure ratio, this parameter is only defined for the closure following a blast
    and not for a seismic event.
•   Geometric closure rate (GCR): The rate of closure as a function of face advance,
    expressed in mm/m. Geometric closure rates incorporate the primary, instantaneous and
    steady-state closure phases. The rate should be calculated after every blast and, like the
    steady-state closure rate, will tend to decrease with increasing face advance. The face
    advance at each blast must therefore be known. Geometric closure rates are obviously
    meaningless in the absence of mining.

Similar to measurements in the gold mines, it was found that Bushveld closure profiles
contain a significant time-dependent component, even in the absence of mining. The
characteristics of the closure profiles varied considerably from site to site, and even between
adjacent stopes in the same area. A typical closure profile from a crush pillar site is
presented in Figure 6. Note changes in the rate of mining and the distinctive time-dependent
component of closure.

                             Frank II Shaft
                                                                                                                Rate of steady-state
                   35                                                                                           closure = 1.5 mm/day

                   30                                                                         Rate of steady-state                                Blast
                                                                Rate of steady-state          closure = 0.14 mm/day                       Blast
                                Rate of steady-state            closure = 1.65 mm/day
 C losure (m m )

                   25                                                                                                                  Blast
                                closure = 2.27 mm/day           CR = 0.27                                                         Blast
                                CR = 0.3
                   20                                                                                                          Blast
                   15       11h33                                                               Blast
                            7.4 m to face                               Blast      Blast
                   10                                           Blast

                        0                     5                 10              15               20             25             30                 35           40
                                                                                           Time (days)

Figure 6. Typical closure profile (with variations in the mining rate) from a crush pillar
                              site on the Merensky Reef.

The rate of steady-state closure (RSSC – expressed in mm/day) captures the time-
dependent component of behaviour and provides a consistent parameter for comparing site
responses and tracking changes in closure behaviour. The geometric closure rate (GCR -
expressed in mm/m) is also useful, but shows greater variability than the analysis of the
RSSC. The closure ratio used to characterize closure in gold mines was difficult to isolate
and quantify and was therefore not used as a criterion for comparison.

A detailed analysis of closure profiles in terms of the mining parameters was conducted as
part of SIM040207. A discrete averaging process was suggested for analyzing closure data.
This allows the data to be condensed to a single value representative of the average. The

use of discrete averages tends to reduce the influence of outliers and also provides a
consistent means of comparing site behaviour. This technique is discussed in more detail in
the relevant section.

The main findings of the analysis are outlined below:

   •   Geometric closure rates (in mm/m of face advance) are generally greater than the
       steady-state closure rates (in mm/day).
   •   The initial spans and relative positions of instruments seem to have had little effect on
       closure rate, or these effects are obscured by other site-specific factors.
   •   Numerical modelling indicated that Young’s moduli of between 2.2 GPa to 8.6 GPa
       were required to match the measured closure rates. This level of material property
       manipulation is highly questionable and indicates that elastic boundary element
       models are not suitable for estimating closure in the Bushveld.
   •   No clear correlation between closure rates and depth could be established, though
       there was a weak trend towards increased rates with increasing depth.
   •   Examination of the ratios between geometric closure rates and steady-state closure
       rates indicated that the geometric closure rates are notably higher in the crush pillar
   •   Rock mass ratings did not explicitly correlate with the closure rates, however there is a
       correlation between the observed panel conditions and closure rates. This is
       especially true for the steady-state closure rates.
   •   No distinction between closure behaviour on the different reefs could be discerned,
       but there is limited data from the UG2 reef at this time.
   •   In at least one site, nearly all of the closure was associated with footwall movement. It
       is not appropriate to make any conclusions based on a single survey, but the result
       indicates that more effort must be concentrated on determining relative footwall and
       hangingwall movements.
   •   Ride values more than two times greater than the vertical closure were measured. No
       correlation between ride and other parameters could be established.

The following conclusions are drawn with respect to the usage and value of geometric and
steady-state closure rates:
   •   It appears that geometric closure rates may be affected by the pillar type. Geometric
       closure rates are generally higher (in relation to steady-state closure rates) in crush
       pillar areas than in yield or stable pillar panels.

   •   Rates of steady-state closure generally provide a good indication of rockmass
       conditions within the panel, and therefore provide a measure of the potential for

The strongest correlations were found between observed panel conditions and rates of
steady-state closure. The sites are grouped according to RSSCs and observed conditions in
Table 1. It must be emphasized that this table is compiled on limited data and should be
updated and emended where more data becomes available.

    Table 1. Site grouping according to rates of steady-state closure and observed
                                rockmass conditions
      Site            GCR      RSSC      Conditions           Characterisation
                       ave.     ave.                          and criterion
                     [mm/m] [mm/day]
       Impala 10 #
                       3.65       0.23     Stable conditions
       Rowland #       1.30       0.33     Stable conditions
       Impala 10 #
                       1.73       0.36     Stable conditions        Low closure
                                                                    RSSC < 0.75
       Impala 1 # -
                       1.47       0.44     Stable conditions
       17c13 1S
       Impala 9 # -
                       1.02       0.48     Stable conditions
       17c45 3W
       Impala 9 # -                        Poor conditions in
       17c45 4W                            face due to
                       1.90       1.02
                                           presence of nearby
                                           pothole                 Moderate closure
       Impala 9 # -                        Moderate conditions    0.75 < RSSC < 1.5
                       1.19       0.86
       17c45 8W                            – FOGs reported
       Frank II # -
                       4.06       1.08     Stable conditions
       Impala 10 # -                       Generally good
                       2.94       1.68
       1969 3S                             conditions
       Impala 9 # -                        Poor conditions –
       17c45 7W        3.54       2.38     many FOGs                 High closure
                                           reported               1.5 < RSSC < 4.0
       Site B                              Poor conditions –
                       7.41       2.61     large FOG closed
       Site A (UG2
       under                                                      Very high closure
                       10.18      5.89     Poor conditions
       Merensky                                                     RSSC > 4.0

Arrays of closure-ride stations showed the distribution of closure across panel spans.
Generally, the highest values were recorded at centre-span, but in crush pillars sites the
highest closure was recorded at the furthest distance from the abutment, near the crush
pillars. This indicated that the effective span extends across the entire panel. Distributed

closure readings such as this can provide an indication of whether pillars are indeed

Some very encouraging results were obtained regarding the use of closure data to warn of
large collapses. In some cases, warning of up to a week prior to the collapse could be
obtained due to accelerating rates of steady-state closure. However, the closure rates did not
always indicate a hazard when a FOG occurred, nor did high closure rates (or changes in
closure rates) always result in FOGs. Determining criteria for generating alarms based on
closure readings should make up a major part of future research in this area.

The successful use of closure data to indicate impending stope collapses would be a
significant advance in safety and risk management. If criteria could be established to indicate
an increased potential for collapse using real-time monitoring equipment, the stope could be
evacuated before the collapse happens. During the current study, the authors were fortunate
that further examples of large panel collapses could be studied. Some examples are
described below.

Collapse at Site E
At this site, closure instrumentation was installed in several panels as shown in Figure 7. A
large collapse eventually occurred in Panel 4S as shown in Figure 8. A noticeable
acceleration in closure was observed before the collapse (see Figure 9 and Figure 10).
Calculated values of the rate of steady-state closure for this area are given in Table 2.

A significant degree of variability is observed up to 03/04/05: values vary from 0.5 to
2.7 mm/day. The last two values of 6.3 and 12.5 mm/day, however, indicate very large
changes and very large values relative to the average. Even with a limited knowledge of
“typical” Bushveld closure behaviour, this change served as an indication of a greatly
increased hazard and appropriate action (which in this case was limited to evacuating and
abandoning the panel) could be taken in good time.

                                                         Panel S2

                                                       Panel S3

                                       closure meter

                                                              Panel S4
                                   Face position

                     Figure 7. Area in which the fall of ground occurred.
Figure 8. Photograph showing the panel collapse (left) and a wedge-shaped fallout in
                     the face caused by a curved joint (right).


            Closure (mm)


                                                                                     08/03/2005 9h57
                                                      Rate of steady-state closure
                           10                         1.7 mm/day
                                    01/03/2005 9h15


                                0        2000         4000       6000        8000        10000 12000
                                                         Time (minutes)
   Figure 9. Time-dependent closure in panel S4 for the period from 01/03/2005 to

                              45                            Rate of steady-state closure
                                                            12.5 mm/day
                                                                                     Closure exceeded

               Closure (mm)
                                                                                     instrument range
                              30                                                     29/03/2005 02h54

                              25                                                     FOG followed after this
                                                                                     and occurred some
                              20       23/03/2005 11h33                              time before
                                                                                     30/03/2005 8h00
                                                     Rate of steady-state closure
                               5                     6.3 mm/day
                                   0           2000          4000           6000           8000         10000
                                                            Time (minutes)
 Figure 10. Time-dependent closure in panel S4 for the period immediately before the
                     collapse. Note the acceleration in closure.

    Table 2. Rate of steady-state closure for different panels in this area before the
                          Panel Date          Rate of steady-
                                              state closure
                            S2     27/11/04          0.6
                            S2     08/01/05          0.5
                            S3     05/02/05          1.1
                            S3     07/02/05          1.5
                            S3     18/02/05          0.7
                            S4     23/02/05          2.4
                            S4     24/02/05          2.7
                            S4     04/03/05          1.7
                            S4     23/03/05          6.3
                            S4     26/03/05          12.5
                            S4     30/03/05        Collapse

Collapse at Site B
The following sequence of four closure curves illustrated the rapid increase in closure
experienced in Panel S3 at this site during late November 2005. This data was initially used
to motivate the stopping of blasting activity in the panel and then to completely evacuate the
panel after the accelerating rates of closure shown in Figure 14 were observed. The changes
in closure in this case are not as significant as the actual values of the closure rates, which
are greater than 28.6 mm/day – greater than any other steady-state closure rate recorded in
the project.

A collapse in the face of the panel occurred some time after 30 November. The full extent of
the fall is not known, however, as the area was abandoned and access could not be gained
to do an investigation.

                                        Spud Shaft
                                        26/17 3S


                Closure (mm)   15

                                                                                                                                22/11/05 09h36
                               10        17/11/05 10:11                                                                         14.2 m to face
                                         12.1 m to face


                                    0          1000             2000            3000            4000            5000     6000    7000            8000
                                                                                       Time (minutes)

         Figure 11. Closure in Panel S3 for the period 17/11/2005 to 22/11/2005.

                                        Spud Shaft
                               50       26/17 3S                                                     25/11/05 08h31
                                                                                                     17.1 m to face

                                                                                 Rate of steady-state closure
                                                                                 13.7 mm/day
                                                 Rate of steady-state closure

                                                 24.3 mm/day


                               20       22/11/05 10h03
                                        14.2 m to face


                                    0                500            1000              1500                2000         2500     3000             3500
                                                                                       Time (minutes)

   Figure 12. Closure in Panel S3 for the period 22/11/2005 to 25/11/2005. The rate of
        steady-state closure increased significantly after the blast on the 22nd.

                                             Spud Shaft
                               45            26/17 3S

                               40                                                                                             28/11/05
                                                                                                                              17.1 m to face
                               35                                              19.1 mm/day

                Closure (mm)




                                                              25/11/05 08h47
                                    5                         17.1 m to face

                                        0            500           1000          1500         2000       2500       3000      3500             4000
                                                                                        Time (minutes)

Figure 13. Closure in Panel S3 for the period 25/11/2005 to 28/11/2005. The high rate of
            steady-state closure persisted in the absence of any blasting.

                                              Spud Shaft
                                    40        26/17 3S

                                    35                                                         28.6 mm/day                  30/11/05 09h03
                                                                                                                            17.1 m to face

                     Closure (mm)


                                                                 18.9 mm/day
                                             28/11/05 10h07
                                    10       17.1 m to face


                                         0                 500                 1000           1500           2000          2500              3000
                                                                                        Time (minutes)

   Figure 14. Closure in Panel S3 for the period 28/11/2005 to 30/11/2005. The rate of
        steady-state closure accelerated further in the absence of any blasting.

Clearly, accelerating closure rates generally indicate that a collapse is imminent, however,
the closure rates did not always indicate a hazard when a FOG occurred, nor did high
closure rates (or changes in closure rates) always result in FOGs. It may be that mining in
nearby panels could result in unusually high closure rates, especially where crush pillars are
being used. This will tend to artificially raise the baseline values for typical closure, so
activities in adjacent panels should be considered in future closure research. Determining
criteria for generating alarms based on closure readings should make up a major part of
future research in this area.

Some cases of collapses were also observed where no prior warning was indicated by the
closure data. In these cases, the closure meters were installed a considerable distance from
the fall and it therefore appears that positioning of the meters is critical. Ideally, the entire
panel should be monitored, though this may prove economically and practically infeasible.


A motivation and methodology are suggested in the final report for SIM040207 for quantifying
closure behaviour. As mentioned above, the steady-state closure rates were found to provide
an indication of rockmass stability. In particular, changes in rockmass conditions due to
changes in steady-state closure rates are revealed by this technique. The method is
described below and examples are provided.

The first step in the analysis is to select a consistent period of face advance. For the data
considered here, closure meters were installed at 7 m to 12 m face advance, and remained
in position for up to 30 m face advance. A period of 10 m to 30 m face advance is therefore

The steady-state closure rate is calculated after each blast. It is essential that the distance-
to-face be recorded after each blast for this technique to be applicable. Recall that the
steady-state closure rate is simply the change in closure (as a function of time) from 6 hours
after the blast to 24 hours after the blast, or until the next blast if it occurs before 24 hours
have passed. This provides a parameter which is calculated consistently and can, in general,
be expected to decrease with face advance.

The calculated steady-state closure rates can be expressed as a function of distance-to-face
after each blast. This provides a closure rate distribution which is processed further to
provide a single value for comparing different data sets. Examples are now presented to
better demonstrate the technique. Distributions have been generated for adjacent panels
from site 12 (see Roberts, et al., 2006) in Figure 15 and Figure 16. Note that the RSSCs are
plotted against distance-to-face. The sets are separated in time (i.e. set 1 was recorded, then
set 2, then set 3, etc.). It is clear even from these distributions that different behaviours are
evident in the two panels. It is also evident that conditions in 7W are changing significantly.


             Rate of steady state closure [mm/day]   5





                                                         5                   10                     15                 20                  25
                                                                                       Distance to face [m]

                                                                 SET 1              SET 2                 SET 3               SET 4
                                                                 SET 5              Linear (SET 2)        Linear (SET 1)      Linear (SET 3)
                                                                 Linear (SET 4)     Linear (SET 5)

                                                         Figure 15. Distribution of RSSCs for Site 12 panel 7W

             Rate of steady state closure


                                                                 5            10             15              20               25           30
                                                                                            Distance to face [m]

                                                                            SET 1                 SET 2               SET 3
                                                                            Linear (SET 2)        Linear (SET 1)      Linear (SET 3)

                                                         Figure 16. Distribution of RSSCs for Site 12 panel 8W

Though straight-line fits have been applied, it is clear that the confidence of these fits are low
due to significant scatter in the distributions. To obtain a more representative distribution and
to limit the impact of outliers, a discrete averaging process is employed. The face advance
range in divided into segments of 5 m length each. These begin with the segment from 7.5 m
to 12.5 m and end with the segment from 27.5 m to 32.5 m. Readings from each set are
averaged over these segments. As an example, the averages from panel 7W are tabulated in
Table 3.

    Table 3. Central-difference averaging procedure for all data sets from panel 7W
         Segment       Average     Average rates of steady-state closure [mm/day]
          [m – m]        face
                       advance      SET 1      SET 2 SET 3 SET 4 SET 5
         7.5 - 12.5        10        1.97       2.23      3.5               2.35
        12.5 - 17.5        15        0.7        2.68     4.75      3.44     0.44
        17.5 - 22.5        20        0.5        1.79     4.15
        22.5 - 27.5        25                   1.6       4.8
        27.5 - 32.5        30
        AVERAGES                     1.06       2.08      4.3      3.44     1.40

The discrete averages are then averaged for each set (as shown in the last row of Table 3) to
provide a single value for comparison. These values can now be plotted as a function of time
(or data set) to show changes in panel conditions. The resulting graphs for panels 7W and
8W are presented in Figure 17 and Figure 18, respectively. Maxima and minima are also
provided to indicate the ranges involved.

                  Average rate of steady state closure



                                                         3                                       Maximum


                                                             0   1   2      3        4   5   6

             Figure 17. Variation in steady-state closure rates for panel 7W


                   Average rate of steady state closure


                                                          3                               Maximum


                                                              0   1      2        3   4

              Figure 18. Variation in steady-state closure rates for panel 8W

The graphs above indicate that panel 7W is subject to significantly higher closure rates than
panel 8W. Underground observations also showed greater instability in panel 7W, though
FOGs were also recorded in 8W. The increase in RSSCs in Figure 17 corresponds with
deteriorating rockmass conditions in the panel. The subsequent decrease in rates may be
associated with the introduction of a pillar in the middle of the panel.

A general methodology is proposed for the systematic analysis of closure. The steps are
outlined below:
   1.     Capture site details including a geotechnical survey.
   2.     Install continuous closure instrument 5 m to 10 m from the face, recording current
   3.     Monitor for a fixed face advance (up to 30 m is usually feasible), recording
          distance-to-face and time of each blast. Any changes in panel conditions and
          deviations from planned mining or support rules should be documented.
   4.     Generate distributions of steady-state closure rates with face advance. This can be
          achieved automatically using the CIVAT programme provided with this booklet.
   5.     Calculate 5 m discrete averages and average these for the data set.
   6.     Insert the data into a spreadsheet or graphing program and record the changes in
          average RSSC.
   7.     Assess closure in terms of support design. Calculating the geometric closure rate
          (mm/m) will allow for the support design to be evaluated and amended if
   8.     Repeat from step 2 for the life of the panel.

By monitoring over the same face advance for each set, the use of average values becomes
more meaningful. It should be noted that it is to be expected that the closure rates will
change as the overall span of the excavation increases, however, for the ranges studied in
this project, initial span did not appear to have a very strong influence on closure rates.

At this point, it is not appropriate to suggest which values of closure would represent a
dangerous situation. The data has shown that changes indicative of an increased potential
for collapse or FOG are also not easy to quantify, and clearly vary from area-to-area, and
even from site-to-site. The application of the suggested methodology will allow for these
critical values and changes to be identified for specific areas. The creation of a national or
industry database for this data will allow for areas to be characterised according to closure
behaviour. Individual site behaviour can then be quantified in terms of potential for FOG or

The closure rate analysis above concentrated on determining “average” and representative
values of closure for regular systematic analysis on a scale of days and weeks. The SCD
(Stope Closure Decay) parameter may also assist in evaluating geotechnical conditions.

The experimental sites in the platinum and gold mines have shown that the rate of closure
decreases with time in the absence of blasting (Malan, 2003). To simulate this decay in the
rate closure, the following empirical equation is used (Malan, 2003):

∆ Sss = a 1 − e − ( SCD ) t   )                 (2)

In this equation t represents time and a and SCD are empirical closure parameters. This
analytical model can be fitted to measured data. A smaller value of SCD indicates that the
rock takes a longer time to reach a quasi-static equilibrium after blasting, which might
possibly indicate more unstable conditions.

The values of SCD for different areas are given in Table 4, showing significant promise to
use this parameter for geotechnical classification. Care should, however, be taken when
calculating this parameter as the transition phase from instantaneous closure during the blast
to the steady-state closure phase might take from several hours up to a day. This transition
phase should not be included when calculating the SCD parameter. This complication arises
as the decrease in rate of closure after the blast is more complex than simple exponential
decay and the duration of the transition phase is not immediately obvious.

When using this equation to calculate the SCD parameter, a data set with a period of at least
5 days of no blasting which is preceded by at least one day of no blasting is recommended.
Due to these complications, further work is required to investigate the possible use of this
parameter as an indicator of poor hangingwall conditions and the practicality of calculating
this parameter.

                Table 4. Calibrated values of SCD for different mining areas
     Area                  SCD-value Qualitative description of conditions
     Site 5                   0.012     Stable hangingwall. Good conditions in panel.
     Site 2                   0.006     Large collapses in adjacent panels. Open joints in
                                        hangingwall. Panel stopped.
     Site C                   0.0036    Blocky hangingwall conditions with the possible risk
                                        of large collapses. Panel stopped.

Mponeng Mine   0.015   Stable hangingwall. Hangingwall discontinuities
                       tightly clamped due to dilation ahead of face.



                                                                        Blast Time

                              Figure 19. CIVAT user interface.

CIVAT is a tool for analyzing stope closure over time. Closure profiles can be plotted, and
any available blast time data is displayed. If there is no explicit blast time data, blast times
can be generated at specified intervals, and can be moved interactively by the user. Closure
rates can be calculated interactively, by specifying two points on the closure profile, or
automatically from the blast times. A report will then be generated suitable for pasting into

Program installation
The CIVAT folder on the attached CD can be copied directly to the user’s hard disk. Double-
click the “closure.exe” file to run this program. Should any problems arise, run the
“dotnetfx11.exe” file. This file makes up part of Service Pack 2 for Windows XP and should
already be installed on most systems.

Toolbar commands
   1.     “Open File”: Use this to open a closure profile data file (either with .ccp or .txt
          extensions.) In the simplest case, these files consist of two columns of numbers;
          the time (in days) and the closure (in mm). There may be corresponding face-time
          data, which is contained in files with the extension .ftd, and have two columns
          consisting of time (also in days) and face advance (in metres.) If a corresponding
          face-time dataset is found, it will be loaded automatically. That is, if you load

          ‘test.ccp’, the program will look for ‘test.ftd’ and load it if it exists. All times are in
          days from the first date of measurement.
   2.     “Copy to Clipboard”: This copies the graph to the clipboard, in a form suitable for
          pasting into a report.
   3.     “Copy Data”: This copies the actual numbers to the clipboard.
   4.     “Smooth Data”: This applies a moving-average to the data. The parameter is the
          averaging period, in days.
   5.     “Zoom In”: You can drag the mouse to expand any area of the plot.
   6.     “Zoom Out”: This restores the original display.
   7.     “Blast Time Setup”: These are the parameters used to automatically generate face
          times and analyze the data.
   8.     “Shift All Blast Times”: If a blast time has been manually moved, then this
          command will move all other blasts by the same amount. This is useful for getting
          a good alignment of blasts and closure profile events.
   9.     “Generate Steady-State Closure Report”: This will give you a report with the
          following columns: Blast Time, Face Advance, Closure Rate, Closure. It will be
          placed on the clipboard.

Interactively Calculating Closure Rates
Click on any point on the curve; the message on the status bar will read ‘Please click on
another point on curve’. Click again and the line will be drawn. The reported slope (in units of
mm/day) will be reported in the curve legend (bottom right of plot.) A detailed view, showing a
least-squares fit of the data between the points, will be shown. This will show the best fitted
line slope for comparison, as shown in Figure 20.

              Figure 20. Example of interactively calculating closure rates.

How Blast Times are Automatically Calculated
If explicit blast data is not available, blast times can be calculated from the initial blast time (in
days from the start of measurement, as usual) and the blast frequency (in hours.) By default,
blasts are assumed to start immediately and occur every 24 hours. To change these defaults,
use the toolbar command “Blast Time Setup” (7 above). The resulting window is shown in
Figure 21. The blast frequency and initial blast time are entered in the first two fields. The
initial and final face positions (in metres) are also required. The face advance rate is
assumed to be constant over the period.

                          Figure 21. Blast time adjustment window.

How Steady-State Closure Rates are Calculated
The definition used in CIVAT is this: steady-state closure rate is calculated from the time 6
hours after the blast, to 24 hours after the blast, or, if a blast occurs less than 24 hours later,
that blast time will be used instead. These defaults can be modified by the “Blast Time Setup”

Interactively Modifying Blast Times
If the mouse is held down near a blast line and dragged, then that blast time can be moved.
Releasing the mouse will cause the data to be updated and replotted. The “Shift All Blast
Times” command (8 above) is useful in the case where the blast intervals are mostly correct,
but the initial time needs to be adjusted. After any blast time has been altered, this command
will adjust the other blast times accordingly. Blast times can be saved in the .ftd file.

                              CLOSURE SURVEYS
Significant differences in closure can be experienced in stopes in the Bushveld Complex,
even in adjacent panels. Closure surveys are based on estimating stope closure from the
deformation of support units and serve as a quick and simple way of revealing differences in
closure behaviour and identifying problematic areas where more extensive and detailed
closure monitoring is necessary. This simplicity and efficiency of the method allows for easy
data collection from a large number of different sites. The utility of this method is
demonstrated below.

The deformation on support units is used to estimate closure. Since estimating the
deformation on individual units is a quick operation, a map of closure across the panel (or
across multiple panels) can be generated within a short period of time.
For example, in an area supported by pencil props, the following procedure is followed. For
pencil props, a ring of wire is bound around the unit some distance below the end of the pod
(Figure 22). A number of pencils are measured to obtain the average position of the wire
from the end of the pod. The difference between this reference measurement and the
measured distance between the wire and the hangingwall gives the approximate amount of
closure to which each pencil is subjected. Care should be taken not to include elongates in
the survey that have been installed as additional support at a later date.

                                          Distance between
                                          wire and the end of
                                          the pod

           Figure 22. Measurement between the end of the pod and the wire.

A closure survey was conducted at an un-named site on the UG2 reef horizon. The
Merensky Reef was extensively mined in this area and the middling between the two reefs
was approximately 18 m. The mining layout in this area is shown in Figure 23. Two lines
were surveyed in panel 2N and one line in panel 3N. The resulting distribution of closure is
presented in Figure 24.

               Figure 23. Area where the closure survey was conducted.


                                                                          Updip side : Panel 2N
                                                                          Survey line 1
                   Closure estimate (mm)



                                                                                     Downdip side : Panel 2N
                                                                                     Survey line 2

                                                                              Less than 20 mm closure in Panel 3N
                                                                              Survey line 3

                                                 0   5    10             15                       20                25
                                                         Distance to face (m)

                  Figure 24. Closure as a function of distance to face.

The closure in panel 2N varies considerably from the up-dip side of the panel to the down-dip
side of the panel. The closure in panel 3N is also significantly lower than panel 2N. The total
amount of closure experienced in panel 2N ranged from 100 mm in the up-dip part of the
panel to 50 mm in the down-dip part. Closure in panel 3N was less than 20 mm.

The increase in closure towards the up-dip side of panel 2N may be related to the proximity
of the overlying Merensky pillar. Conditions in the panel also indicate deterioration of the
rockmass towards the pillar. Based on this survey, the area most likely to suffer adverse
rockmass conditions and associated FOGs has been identified. Armed with this information,
the rock engineering practitioner on the mine can take remedial action before conditions
deteriorate even further. In areas where the cause of the variable deformation is not as
obvious, such a survey can indicate where more detailed instrumentation is necessary.

To conduct the kind of analyses suggested here requires that reliable data is recorded
consistently. The following should always be considered when undertaking closure

   •   It is important to record the exact distance to face and measurement position in the
       stope when installing closure instruments. This information should be included in any
       report referring to the rate of closure.
   •   Blasting times and face advance following each blast should be recorded.
   •   When calculating rate of closure from long period measurements, it is important to
       record the exact time interval (for rates in mm/day) or distance interval (for rates in
       mm/m) used in the calculation.

The aim of closure monitoring is primarily to indicate increased FOG potential or imminent
panel collapse. This can be achieved by examining the closure profiles in real-time and
identifying sustained acceleration in the profile. Systematic analysis using the methodologies
suggested above of the data is also valuable in identifying changing conditions.

There is tremendous scope for further work in Bushveld closure monitoring and
interpretation. The closure rates and ranges quoted here should be updated and this booklet
amended as more data becomes available.

As this guide focuses on continuous closure measurements, instrumentation commonly used
to collect long term closure measurements such as spring vernier closure meters and four-
peg closure ride stations will not be discussed in this guide. The reader is referred to Ryder
and Jager (2002) for a description of these instruments and their use. The following is a list of
the instruments used to collect continuous closure data in the MHSC project. It should be
noted that various companies are continually involved in the design of new closure meters
and the following list contains only the instruments relevant to the MHSC project.

Clockwork closure meters
Although the clockwork closure meter is an old mechanical design, it is a reliable device and
is still occasionally used to collect closure data. In fact, clockwork closure meters were used
exclusively in the MHSC project. A benefit of this instrument is that it can be used in areas
where intrinsically safe instruments are required. A drawback is that the data is collected on
graph paper and significant effort is required to digitise and analyse the data. The
mechanism of this instrument is shown in Figure 25.
                               To hangingwall

                                                      needle pivot

                            needle guide
                            fixed to upper
                            pipe                      rotating
                                                                     needle on
                                spring                               graph paper


                                                     base fixed
                                                     to lower pipe

                                  To footwall

                  Figure 25. Mechanism of the clockwork closure meter.

The clockwork closure meter comprises two telescopic tubes with one of the tubes sliding
against the pressure of a spring. The base of the instrument is fixed to the lower tube with the
needle being moved by a guide fixed to the upper tube. The instrument is installed with the
needle point close to the bottom of the graph paper on the drum. Stope closure moves the
upper tube further into the lower tube with the needle point moving upwards on the paper.
The clock typically rotates the drum once every seven days, resulting in a continuous record

of closure over this period. The needle arm can be designed with various possible pivot
points to allow for different amplification factors of the closure magnitude. Unfortunately this
design is inherently flawed as instantaneous closure is recorded as an arc, thereby distorting
the time axis. The data needs to be corrected before use as indicated in Appendix B. An
example of the raw data recorded on the graph paper is given in Figure 26. Figure 27
illustrates the same data after it was digitised and corrected according to the procedure in
Appendix B.

   Figure 26. An example of continuous closure data recorded using the clockwork
                                   closure meter.

Figure 20 only shows a small section of the graph paper sheet and is not printed to scale.
The horizontal grid lines are separated by a vertical distance of 5 mm.          Note that the
horizontal axis represents time, which increases to the left. The days printed on the paper
indicate an increase of time to the right, but this is incorrect as the paper was mounted
upside down on the drum. The vertical axis represents closure magnitude which increases
upwards. The pivoting needle arm resulted in the actual closure magnitude being multiplied
by a factor of 1.86 on the graph paper.




                 Closure (mm)



                                     0   500   1000   1500        2000   2500   3000   3500
                                                      Time (minutes)

Figure 27. The corrected closure data for the measurements depicted in Figure 20. The
            data points give an indication of the digitisation interval used.

When installing the instruments underground, it is important that the two tubes make contact
with solid rock in the hangingwall and footwall respectively. The tubes are typically secured to
the hangingwall using quick-setting putty. When installed close to the face, the instrument
should be protected from blast damage by installing it behind support units. Figure 28
illustrates a typical installation underground.

  Figure 28. Installation of a clockwork closure meter in a stope. It is important when
    installing these meters that the lower tube makes contact with solid rock in the

In dipping stopes, the instruments should be installed normal to the plane of the reef to
measure the closure component. It should be noted that with time, a significant ride
component will tilt the instrument, resulting in a composite measurement.

1.   Roberts, D.P., Malan, D.F., Janse van Rensburg, A.L., Grodner, M. and Handley,
     R. 2006. Closure profiles of the UG2 and Merensky reef horizons at various
     depths, using different pillar types, in the Bushveld Complex. SIM040207 Final
     Report. Mine Health and Safety Council.
2.   ISRM. 1975.
3.   Wiggill, R.B. 1965. The effects of different support methods on strata behaviour
     around stoping excavations. Symposium on Rock Mechanics and Strata Control in
     Mines: J. S. Afr. Inst. Min. Metall.: 1-35.
4.   Gürtunca R.G. & Adams D.J. 1991. Determination of the in situ modulus of the
     rockmass by the use of backfill measurements. J. S. Afr. Inst. Min. Metall. 91(3):
5.   Salamon, M.D.G. 1968. Two-dimensional treatment of problems arising from
     mining tabular deposits in isotropic or transversely isotropic ground. Int. J. Rock
     Mech. Min. Sci., 5: 159-185.
6.   Malan, D.F. 1998. An investigation into the identification and modelling of time-
     dependent behaviour of deep level excavations in hard rock. PhD Thesis,
     University of the Witwatersrand, Johannesburg.
7.   Malan, D.F. 2003. Guidelines for measuring and analysing continuous stope
     closure behaviour in deep tabular excavations. SIMRAC publication, ISBN 1-
8.   Ryder, J.A. and Jager, A.J. 2002. A textbook on rock mechanics for tabular hard
     rock mines. SIMRAC.

Elastic convergence solution for the incremental enlargement of a

The elastic convergence (Sz ) of a horizontal (dip = 0º) parallel-sided panel in isotropic

ground without contact between the hangingwall and footwall is given as (Salamon, 1968)

                                                (0 )         4(1 − ν 2 )Wz 2
                                           Sz          (x) =               L − x2                       (B1)

for   x ≤ L where

                                                               W z = ρgH                                (B2)

and 2L = span of the stope,           ρ   = density of the rock,    g   = gravitational acceleration,   H   =

depth below surface, ν = Poisson' ratio, E = Young' modulus and
                                s                 s                                 x is the distance from
the centre of the stope.

Consider the incremental enlargement (both sides of the stope are simultaneously mined) of
a stope as indicated in Figure 8. The convergence of the original stope of half-span L is given
by equation (B1). After mining an increment ∆ on both sides, the convergence is given by

                     (1)         4(1 − ν 2 )Wz
                Sz         (x) =                            (L + ∆ )2 − x 2     for     x ≤L+∆          (B3)

The increase in closure after one increment is therefore

                ∆S1 = S z             (x ) − Sz (0 ) (x )
                  4 1 − ν 2 Wz   )              (L + ∆ )2 − x 2 −       L2 − x 2       for    x≤L       (B4)

When mining a second increment, the total closure is given by

                                   (2 )     4(1 − ν 2 )Wz
                                 Sz ( x ) =                              (L + 2∆ )2 − x 2
                                                  E                                              (B5)
                                 for x ≤ L + 2∆

The new increment in closure caused by this second mining step is given by

                                    (2 )
                      ∆S2 = S z             (x ) − Sz (1) (x )
                        4 1 − ν 2 Wz)                (L + 2∆ )2 − x 2 − (L + ∆ )2 − x 2
                              E                                                                  (B6)
                      for x ≤ L + ∆

The total increment in closure caused by these two mining steps is

                       ∆S T = ∆S1 + ∆S 2

                         4 1 − ν 2 Wz   )       [ (L + 2∆ ) − x  2   2
                                                                         − L −x 2   2
                                                                                        ]        (B7)


Similarly, for any number n of mining increments, it can be shown that the increment in
closure is given by

                       ∆S =
                             4 1 − ν 2 Wz        )     [ (L + n∆ ) − x
                                                                     2      2
                                                                                − L2 − x 2   ]
                                   E                                                             (B8)
                         for     x ≤L

Correction procedure for the clockwork closure meter data

As the design of the instrument is such that instantaneous closure is recorded as an arc
(Figure C1), the time scale is distorted and the data needs to be corrected.



   Figure C1. Recording mechanism of the clockwork closure meter. The particular
  setting of the needle arm for the meters results in the closure being amplified by a
                                    factor of m on the drum.

   Figure C2. Sample of the calibration curves of a clockwork closure meter (not to

For the correction procedure, the radius r needs to be determined. Although this can be
directly measured at the instrument, it can also be calculated from the calibration curves in
Figure C2.



                                   o      r-b               b

             Figure C3. Calculation of the radius r from the calibration curves.

From Figure C3

                                                r 2 = a2 + (r − b )

                                             r 2 = a2 + r 2 − 2rb + b2                  (C2)

                                                            a2 + b2
                                                         r=                             (C3)

For correction of the data, it is assumed that the x-values along the horizontal line in the
centre of the graph paper (marked 60, Figure C2) is correct. Any deviation from this line
gives an error in x-value, Cx(y), which becomes progressively worse as the distance from
this line increases (Figure C4).

                                  y   Cx(y)            x = (r2 - y2)½

                                                              r            y


  Figure C4. Illustration of the error in x-value as the curve deviates from the central
  horizontal line. The construction of the meter and the rotation of the drum are such
            that time (x-axis) increases away from the concave part of the arc.

The equation of the circle is given by        x2 + y2 = r 2        or   x = r 2 − y 2 and   therefore the

error in x-value as a function of y is given as

                                                   C x (y ) = r − r 2 − y 2                          (C4)

For a list of faulty data points ( x old , y old ) each data point should be corrected according to

                                 xnew = x old − Cx (y ) = x old − r − r 2 − y old
                                                                                                 )   (C5)

                                                                               y old
                                                                  y new =                            (C6)

where m is the closure amplification factor (see Figure C1).

As a first step in the correction procedure, the data must be digitised and any commercial
package can be used. The data can then be corrected according to equations (C5) and (C6).


Shared By: