You are on page 1of 18

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/334054290

Stope performance assessment at the Goldcorp Eleonore mine using bivariate


analysis

Conference Paper · October 2017

CITATIONS READS

4 894

3 authors, including:

Sebastien Guido Martin Grenon


Laval University Laval University
4 PUBLICATIONS 6 CITATIONS 54 PUBLICATIONS 670 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Ground Support Guidelines View project

Stope design and reconciliation View project

All content following this page was uploaded by Sebastien Guido on 21 January 2021.

The user has requested enhancement of the downloaded file.


Stope performance assessment at the
Goldcorp Eleonore mine using bivariate analysis

S. Guido1, 2, M. Grenon2, and P. Germain1


1 Goldcorp – Eleonore mine, Québec, Canada
2 Université Laval, Québec, Canada

Comparing stope performance relative to the engineering design is of paramount


importance for any mining operation. A database was created of 105 stopes mined between
170 m and 800 m below surface from July 2014 to November 2016 at the Goldcorp Eleonore
mine. The database consists of more than 20 parameters defining spatial, mining,
geometrical, geological, drill-and-blast, and temporal stope properties. Univariate and
bivariate statistical analyses were conducted using the database to better understand the
root causes of stope performance. The study has allowed parameters controlling stope
performance to be identified. These are the spatial localization of the stope (mining
horizon, mining lenses, and central location within the orebody), the mining method, and
the orientation of the major structures relative to stope orientation, the global stope drilling
pattern and the stand-up time.

INTRODUCTION

Open stope mining is one of the most commonly used mining methods in Canada, allowing high levels
of mechanization and thus increased productivity. Maximization of stope value is closely linked to
achieving the planned or engineered stope shape. Stope performance is therefore computed by
comparing the planned stope shape with the actual or mined shape. The latter is obtained using a cavity
monitoring survey system (CMS) (Miller, Potvin, and Jacob, 1992).

Stope overbreak (OB) is defined as the volume of rock by which the actual stope shape exceeds the
planned stope shape. Stope underbreak (UB) is the inverse. In narrow vein mining, stope performance
is usually assessed by measuring the equivalent linear overbreak/slough (ELOS), as proposed by Clark.
This is calculated by measuring the ratio between OB volume and the area of a given stope surface:
hangingwall, footwall, crown, or endwalls.

Although conceptually defined in 3D, the common computing methodology used in the mining
industry to evaluate ELOS relies on 2D calculations on various sections or slices cut along the stope
(Cepuritis and Villaescusa, 2006). In large bulk-mining operations, stope performance is generally
calculated by dividing the total OB volume by the planned stope volume (Potvin et al., 2016; Suorineni,
2010). The latter measurement is generally expressed in percentages. This computation, although to a
lesser extent, is also regularly computed in 2D along stope sections or slices.

The study of stope parameters controlling stability, via OB quantification, has been the focus of several
papers in the scientific literature. The pioneering work of Mathews et al. proposed the stability graph
method, linking stope geometry (hydraulic radius) and geomechanical properties (Q’; A, a stress factor;
B, a joint orientation factor; and C, a gravity factor) to stope stability. Over the last 30 years, several
authors have worked on improving this initial work (Potvin, 1988; Clark, 1998; Diederichs and Kaiser,
1999; Bewick and Kaiser, 2009; Vallejos et al., 2016).

It is recognized that the stope stability graph method, despite its most recent developments, has many

2–7 October 2017· Cape Town

1
limitations, as it does not account for many important factors such as mining method, complex
geometries, geological structures, blasting effects, stand-up time, etc. (Suorineni, 2010; (Cepuritis and
Villaescusa, 2006). For these reasons, recent studies focus on relating some of these factors with OB
measurements (ELOS), generally by modifying existing factors or by adding additional factors
(Suorineni et al., 2001). Other authors have focused on multiple regression models aiming at predicting
stope OB (Wang, 2004; Hughes, 2011), with limited success. Recently, Potvin et al. presented a site-
specific investigation where geomechanical and operational stope parameters were studied with regard
to defining OB and UB for mining stopes.

This paper presents a case study where stope investigations were carried out at Goldcorp’s Eleonore
mine to assess factors controlling stope performance. A 105-stope database was created, including stope
design and field parameters (input parameters) and stope performance indicator parameters (output
parameters). A tridimensional assessment of stope performance (OB, UB, and ELOS) was fully adapted
and implemented at the site. A bivariate analysis is presented to identify the controlling parameters on
stope performance at Eleonore. The proposed methodology can easily be applied to any other mining
operation.

ELEONORE MINE

Location and General Layout


Goldcorp’s Eleonore mine is located approximately 1000 km northwest of Montreal and 650 km north
of the Abitibi mining district, in the James Bay area in Northern Quebec, Canada (Figure 1). It is one of
the most important gold-producing operations in Canada. The mine is accessible by road for material
transport and by plane for workers.

Figure 1. Location of Eleonore mine.

The orebody is located beneath Hydro-Quebec’s Caniapiscau Reservoir. Underground access is via two
shafts—the initial exploration shaft has been converted to a ventilation intake—and one ramp, located
on the hangingwall of the orebody. The exploration shaft reaches 720 m while the main production shaft
reaches 1140 m. The ramp currently extends 1040 m below surface and will reach the production shaft
depth later in 2017.

2
Local Geology
Eleonore’s Roberto deposit strikes northwest and dips steeply towards the east at 75°. The deposit
consists of multiple discrete lenses a few metres wide that are contained within a sedimentary basin of
thinly bedded greywacke with important intrusions of pegmatite and diabase dykes. Quartz veins and
breccia are also common . The strike lengths of the different ore lenses are 1 km and they reach a depth
of more than 1.5 km.

Figure 2 presents a typical level plan view. Three principal ore lenses (5010, 5050, and 6000) are
subparallel to the haulage drift. Access drifts to longitudinal stopes are located every 100 m or so. Stope
names are based on the mucking level, ore lens, and mine grid ID (e.g. 0800-5050-358 indicates that the
mucking level is 800 m below depth, lens is ’5050’, and mine grid ID is ’358’).

Figure 2. Typical plan level showing mining drifts, ore lenses, and mines stopes.

Production Rate and Mining Methods


Stope production began in July 2014 and commercial production began during the second quarter of
2015. In 2016, the average daily throughput was approximately 4600 t. Ramp-up is advancing in order
to reach full production capacity, and mining pyramids have commenced in four different horizons (H1
to H4). A fifth horizon is currently under construction (Figure 3). The principal extraction method is
open stoping with longitudinal retreat (Figure 4). Some transverse primary-secondary stopes are mined
where the orebody is thicker. The mine uses paste backfill and unconsolidated rockfill (UCR) in the
mining sequence. Typically, 30% of the voids are filled using UCR to optimize costs.

3
Figure 3. Longitudinal view of Eleonore orebody and mining horizons. Horizons 1–4 are
currently active and horizons 5–6 are upcoming.

Figure 4. Left and center: isometric and transversal section view of typical longitudinal stope with fanned drilling
pattern and drill-holes parallel to the hangingwall. Right: surface area of stope for ELOS and hydraulic radius
calculations.

STOPE DATABASE

The stope database covers 105 stopes, mined between July 2014 and November 2017, at depths ranging
between 170 m and 800 m below surface. The database was subdivided in two broad sections: the stope
performance quantification parameters (output parameters) and the stope design and field parameters
(input parameters).

Output Parameters: Stope Performance Quantification


To assess stope performance, various output parameters were evaluated in full 3D for all stope surfaces
(hangingwall, footwall, sidewalls, and crown), OB and UB volume, surface area, surface perimeter, and
ELOS. Table I summarizes the final stope performance indicators (output parameters) used in this
paper.

4
Table I. Output parameters: stope performance indicators (calculated for all surfaces).

Parameter Unit Details


Overbreak (OB) volume m³
Underbreak (UB) volume m³ Rejected due to possible CMS blind areas
Overbreak percentage (OB %) % Defined as overbreak w.r.t. planned volume
ELOS m Defined as overbreak w.r.t. surface area

OB and UB volumes as well as surface areas were processed using Deswik . The ’reconciliation wizard’
enables calculations of OB and UB for all surfaces (hangingwall, footwall, endwalls, and crown) of a
given stope, comparing the design and final stope volumes. Merging multiple CMS results is sometimes
necessary to survey the entire final stope. This technique allows 3D volume calculations of OB and UB
for each compared surface. Hangingwall and footwall surfaces were determined by adding both
overcut and undercut drift walls, as shown in Figure 4 (right). UB calculations were excluded from the
following performance analysis, as various shadow zones (blind spots) occurred when using the CMS,
resulting in large overestimates of the UB volume for several stopes.

Input Parameters: Stope Design and Field Parameters


The second portion of the database consists of more than 20 parameters defining spatial, mining,
geometrical, geological, operational, and temporal stope properties (stope input parameters). The
parameters contained in this database are shown in Table II under their associated categories.

Table II. Summary of the input field parameters for the stope performance database.

Parameter Unit Details


Spatial
X, Y, Z m Stope center of gravity (COG)
Horizon - [1, 2, 3, 4]. Mining horizon
Level - Mucking level

Mining
Method - [Transverse, longitudinal or intersection]. Intersection is a longitudinal
stope with a drawpoint /access drift breaking through the hangingwall
surface.
Blasting void - [Drift, Second lift, adjacent stope]

Geometrical
Planned volume m³ Planned volume of stope
Hydraulic radius m Surface area divided by surface perimeter. See Right: (bold outline)
Width m Perpendicular dimension between HW and FW. See Figure 5.
Dip ° Angular dimension between horizontal and surface.
Strike variation - [Straight, undulating, irregular]. See Figure 6.
Dip variation - [Regular, irregular, switch]. See Figure 7. Switch indicates that
hangingwall and footwall sides are changing along the strike length.
Undercutting – average m Average width at which undercut drift is not aligned with stope surface.
See Figure 5.
Undercutting – portion % Percentage of stope affected by the undercutting. See Figure 5.
Undercutting – equivalent m Average width (m) at which undercut drift is not aligned with stope
surface multiplied by undercutting portion (%). See Figure 5.

Geological
Distance to structure m Distance between stope COG and nearest structure
Orientation of structure - [‘-‘ or NA (indicating no structure), perpendicular, parallel to HW, parallel to
FW].

5
Drill and blast
Slotraise type - [Dropraise, raisebore, V30].
Drill factor t/m Tonnage of stope per metre drilled.
General drilling pattern - [Dice 5, fan, 50-50]. See
Figure 5.
HW drilling pattern - [Parallel, fanned, 50–50]. See
Figure 5.

Temporal
Final blast date - Date of principal/final blast
Stand-up time days Time between end of mucking and final blast dates

Figure 5. (a) Dice 5 drilling pattern with parallel hangingwall drilling, (b) fanned drilling
pattern with fanned hangingwall drilling, (c) undercutting indicators.

Figure 6. Strike variations: (a) straight, (b) undulating, and (c) irregular.

6
Figure 7. Dip variations: (a) straight, (b) irregular, and (c) switch.

DATA PROCESSING AND ANALYSIS

The stope data acquired during this investigation is presented in the next section using univariate
analysis for both output and input stope parameters.

Univariate Analysis
The objective of univariate analysis is to describe the various parameters of the database. It also enables
conclusions to be made on the distributive nature of the parameter.

Output Parameters: Stope Performance Quantification


The cumulative distribution of total stope OB, hangingwall (HW) OB, and footwall (FW) OB is
presented in Figure 8. Total OB is closely linked to HW OB above a cumulative frequency of 10%. Below
that threshold, it appears that, both FW and stope ends play an important role in defining stope total
OB. Furthermore, Figure 8 shows that hangingwall stability is of greater concern at Eleonore. Indeed,
50% of hangingwall surfaces have 13% OB and 0.5 m ELOS, compared to 4% OB and 0.2 m ELOS for
the footwall surfaces. For this reason, the upcoming data processing and analyses will focus on the
hangingwall surface. The last row of Figure 8 compares the empirical cumulative distribution with the
theoretical lognormal distribution for footwall and hangingwall data. This theoretical distribution is the
statistical best-fit for all empirical distributions (OB% and ELOS).

7
Figure 8. Univariate analysis of stope performance indicators.

Input Parameters: Stope Design and Field Parameters


A univariate analysis, comprising histograms and cumulative frequencies of all the different parameters
contained in the input parameter database, is presented in Figure 9, enabling a visual and quantitative
appreciation of the collected data.

Bivariate Analysis
To assess the effects of the different parameters on HW OB performance (OB and ELOS), different
bivariate analyses were conducted. Only the statistically significant analyses are presented. Bivariate
analyses are presented according to HW OB and ELOS, as both stope performance indicators differ
depending on stope geometry (Figure 10). Indeed, the majority of outlying points above the general
trend line are associated with stopes that were mined transversally. These high-volume stopes have
relatively low hangingwall surface areas, resulting in high ELOS values when compared to the OB%.

Bivariate analyses were conducted on both qualitative and quantitative data from the database.
Qualitative parameters are presented using box-and-whisker diagrams as well as cumulative frequency
plots. Box-and-whisker diagrams enable numerical data to be graphically displayed through their 25%,
50%, and 75% quartiles. Similar to boxplots, notches (narrowing of the box around the second quartile)
allow statistical comparison of medians (second quartile) with a 95% confidence level. Cumulative
frequencies allow visual comparison of performance under various thresholds of HW OB and ELOS.
As a reminder, +OB and ELOS need to be log-transformed (base 10) as these parameters follow a
lognormal distribution. Quantitative parameters are presented using scatter plots with superimposed
bar charts. Scatters allow direct trends to be observed, while bar charts show the relative frequency of
stopes exceeding a given threshold, corresponding to acceptable OB. The current study uses thresholds
of 15% HW OB and 0.5 m ELOS.

8
Figure 9. Univariate analysis (histograms and cumulative frequencies) of input parameters.

9
Figure 10. Comparison between ELOS and OB to assess stope performance: raw data (left)
and log-transformed data (right).

Spatial
Stope performance varies spatially according to three parameters in the stope database: the stope COG,
the mining horizon, and the ore lens. Figure 11 presents a colour map of the 105 stopes based on their
associated OB% and ELOS. Lower performance is observed in the center of the orebody and in the
easternmost ore lens. The ELOS box-and-whisker plot (Figure 12) confirms that stopes in horizon 1 (H1)
have a median OB% and ELOS smaller than stopes located in H2 and H4. Setting a 0.5 m ELOS
threshold, the cumulative frequency plot (Figure 12) indicates that 90% of stopes within horizon 1 are
below this threshold, compared to only 30% of H2 and H4 stopes.

Figure 11. Colour map of stopes according to overbreak (left) and ELOS (right).

Mining
Stope performance is controlled by the mining method and the blast void configuration. As shown in
Figure 13, stopes mined longitudinally and not intersected by an access drift (intersection stope) have
less overbreak. Longitudinal stopes have a wider spread compared to both intersection and transverse
stopes. As previously mentioned, transverse stopes are generally associated with larger volumes and
smaller hangingwall surface areas. Since the notches in the box plot do not overlap, one can conclude
with 95% confidence that the true ELOS median of the longitudinal stopes differs from that of the other
mining methods. Looking at a 15% OB threshold, the cumulative frequency plot indicates that 70% of
longitudinally-mined stopes are below this value, compared to only 35% for transverse stopes. A similar
result is obtained when using the 0.5 m ELOS threshold (longitudinal: 65%; transverse: 20%). Very few
cases exist where a stope is being blasted within a horizontally adjacent open stope (double length).
This is shown by the strong step-like line in Figure 14. Double lift stopes have a different median ELOS

10
than longitudinal stopes, but the same cannot be concluded for OB%. This can arguably be explained
by the inadequacy of ELOS in defining stope performance for stope geometry associated with double
lift.

Figure 12. Box-and-whisker diagrams of HW OB (top left) and ELOS (top right) according
to the mining horizon. Cumulative frequencies of OB (bottom left) and ELOS
(bottom right) for each mining horizon.

11
Figure 13. Box-and-whisker diagrams of HW OB and HW ELOS according to the mining
method (top) and blasting void (bottom).

Figure 14. Cumulative frequencies of OB and ELOS for each mining method (top) and each
blasting void type (bottom).

12
Geometrical
From the investigated stope database, geometrical properties appear to have a somewhat limited impact
on stope perfomance. Stope HW dip is the most important parameter but its impact remains modest.
Figure 15 presents the scatter plot of HW dip versus stope perfomance indicators. A superimposed bar
chart indicating relative frequency of stopes exceeding the 15% HW OB and 0.5 m ELOS thresholds
allows one to quantify stope performance for a given criterion.

Geological
The orientation of major structures relative to the stope strike length is the most important geological
parameter controlling stope performance. The box-and-whisker diagrams presented in Figure 16
demonstrate that stopes oriented perpendicularly to geological structures (or if no structures are found
nearby) result in significantly less overbreak than stopes with hangingwall surfaces oriented parallel to
the structure. Using the cumulative frequency plot (Figure 16), a 15% OB threshold indicates that only
55% of stopes with a HW or FW parallel to a structure do not exceed this threshold, compared to 75%
and 85% for stopes with no structures nearby or oriented perpendicularly to the structure. Similar
results are obtained using the cumulative frequency plot with a 0.5 m HW ELOS threshold.

Drill and Blast


The drill factor and the global drilling pattern parameters both control stope performance. As shown in
Figure 18, the global drilling pattern has a significant impact on the overbreak. The dice 5 drilling
pattern clearly shows lower ELOS for all percentile values. This is confirmed by the fact that higher drill
factors (stope tonnage per drilled metres) have a negative impact on stope ELOS (Figure 17). Stopes
drilled using the dice 5 method have the same burden as other stopes, but an extra hole is added, thus
increasing the drill metres (lowering of the drill factor). On the other hand, the fanned stopes have less
drill metres (greater drill factors). In summary, both the global drilling pattern and drill factor lead to
the same conclusion.

Figure 15. Scatter plot of overbreak against hangingwall average dip with superimposed
bar charts indicating relative frequency of stopes exceeding acceptable threshold.

13
Figure 16. Box-and-whisker diagrams of hangingwall overbreak (top left) and hangingwall
ELOS (top right) according to structure orientation. Cumulative frequencies of OB and
ELOS for each possible structure orientation (bottom).

Temporal
The temporal property influencing stope performance response is stand-up time. The scatter plot alone
does not truly allow a clear trend to be observed (Figure 19). However, superimposing a bar chart (10-
day bins) clearly demonstrates the effect of stand-up time on overbreak. Stopes lasting less than 10 days
are likely to succeed (ELOS < 0.5 m) 72% of the time, compared to 36% and 17% for stand-up times of
30 days and 60 days, respectively.

Figure 17. Scatter plot of overbreak against drill factor with superimposed bar charts
indicating relative frequency of stopes exceeding acceptable threshold.

14
Figure 18. Box-and-whisker diagrams (above) and cumulative frequency plots (below) of
hangingwall overbreak and ELOS relative to the global drilling pattern.

Figure 19. Scatter plots of overbreak against stand-up time and bar charts indicating
relative frequency of stopes exceeding the acceptable threshold.

DISCUSSION

The controlling parameters on stope performance identified in the previous section are the spatial
localization of the stope (mining horizon, mining lenses, central location within the orebody), the
mining method (longitudinal, intersection, or transverse), the orientation of the major structures relative
to stope orientation, the global stope drilling pattern, and stand-up time.

These current site-specific analyses help identify root causes of stope overbreak. The analyses evidently

15
do not take parameter interactions into account (e.g. multiple regression), although these relatively
straightforward plots can help confirm or disprove relations and influences. Such analyses are universal
and practical, as they can be applied elsewhere, and most importantly, can be explained to planners,
engineers, supervisors, managers, etc. Although not all parameters can be controlled during the stope
design phase, they do provide meaningful insights into possible cost-effective improvements in the
engineering design.

CONCLUSION

A database of 105 stopes mined at the Goldcorp Eleonore mine was created. The database consists of
more than 20 parameters defining spatial, mining, geometrical, geological, drill-and-blast, and temporal
stope properties. Univariate and bivariate statistical investigations were conducted using the database
to better understand the root causes of stope performance. Equivalent linear overbreak/slough (ELOS)
and overbreak percentage do not always quantify stope performance in a similar fashion. The
controlling parameters on stope performance are the spatial localization of the stope (mining horizon,
mining lenses, central location within the orebody), the mining method (longitudinal, intersection, or
transverse), the orientation of the major structures relative to stope orientation, the global stope drilling
pattern, and the stand-up time.

ACKNOWLEDGEMENTS

The authors thank Goldcorp for permission to publish this paper. The financial support of the Fonds de
recherche du Québec (Nature et Technologie) is greatly appreciated. The assistance by personnel of
Goldcorp Eleonore mine is also acknowledged.

REFERENCES

Bewick, R. and Kaiser, P. (2009). Numerical assessment of factor B in Mathews’ method for open stope
design. Proceedings of the 3rd CANUS Rock Mechanics Symposium, Toronto, May 2009. Diederichs,
M. and Grasselli, G. (eds). Paper 3996.
Cepuritis, P.M. and Villaescusa, E. (2006). Back analysis techniques for assessing open stope
performance. Proceedings of the Australian Mining Technology Conference, Hunter Valley, NSW.
Hood, M., Lever, P., and Knights, P. (eds). Australasian Institute of Mining and Metallurgy,
Carlton, Victoria.
Clark, L.M. (1998). Minimizing dilution in open stope mining with a focus on stope design and narrow
vein mining. MSc thesis, University of British Columbia, Vancouver.
Deswik Software Solutions. (2016). Deswik.CAD v. 2016.1.
Diederichs, M. and Kaiser, P. (1999). Tensile strength and abutment relaxation as failure control
mechanisms in underground excavations. International Journal of Rock Mechanics and Mining
Sciences, 36 (1), 69-96.
Golder Associates. (2009). Geomechanical design study for the underground mine - Eleonore project.
Internal Report, Montreal.
Hughes, R. )2011). Factors influencing overbreak in narrow vein longitudinal retreat mining. MSc thesis,
McGill University, Montreal.
Mathews, K., Hoek, E., Wyllie, D., and Stewart, S. (1980). Prediction of stable excavations for mining at
depth below 1000 meters in hard rock. CANMET report. DSS serial no. OSQ80-0081. Department of
Energy, Mines and Resources, Ottawa.
Miller, F., Potvin, Y., and Jacob, D. (1992). Laser measurement of open stope dilution. CIM Bulletin, 85,
96-102.
Potvin, Y. (1988). Empirical open stope design in Canada. PhD thesis, University of British Columbia,
Vancouver.

16
Potvin, Y., Grant, D., Mungur, G., Wesseloo, J., and Kim, Y. (2016). Pratical stope reconciliation in large-
scale operations. Part 2, Olympic Dam, South Australia. Proceedings of the Seventh International
Conference and Exhibition On Mass Mining, Sydney. Australasian Institute of Mining and
Metallurgy: Melbourne. pp. 501–510.
Suorineni, F. (2010). The stability graph after three decades in use : Experiences and the way foward.
International Journal of Mining, Reclamation and Environment, 24, 307-339.
Suorineni, F., Tannant, D., Kaiser, P., and Dusseault, M. (2001). Incorporation of a fault factor into the
stability graph method: Kidd mine case studies. Mineral Resources Engineering, 10 (3), 3-37.
Vallejos, J., Delonca, A., Fuenzalida, J., and Burgos, L. (2016). Statistical analysis of the stability number
adjustment factors. International Journal of Rock Mechanics and Mining Sciences, 87 (1), 104-112.
Wang, J. (2004). Influence of stress, undercutting blasting and time on open stope stability and dilution.
PhD thesis, University of Saskatchewan, Saskatoon.

17

View publication stats

You might also like