Professional Documents
Culture Documents
Block Cave Production Scheduling Using PCBC
Block Cave Production Scheduling Using PCBC
Preprint 10-097
using or planning to use CMS. CMS aims to generate a draw order for LSQ
each draw point every day or shift. It uses the recent historical (actual) The LSQ tool is intended for operating mines. Once a mine has
tonnages to adjust and manage the draw and provides the supporting been in operation for a few years, it will likely have a draw point
database, reporting and user interface to facilitate this process. At De sampling program. The draw point assay values can be stored and
Beers Finsch mine, CMS has been closely integrated with the Sandvik sorted per draw point and then composited into 10m or 15m intervals
Automine system. to provide some averaging of the highly variable assays.
Figure 9 shows how CMS fits in between the historical tons mined Subject to a variety of constraints, a least squares trend line is put
and the requirement to adjust the plan of the next few months (using through the composites and then this can be extrapolated for a short
PCBC schedules) in a process called Catch-up to fit in with the long distance up the draw column into what is essentially the un-mined part
term plan. (Diering, 2004) of the column (Figure ).
This becomes particularly useful when the draw point assays
suggest that the draw point should remain open (usually after 100%
draw) when the slice file values suggest that the draw point should be
closed. For draw points where the sample trend differs from the slice
file, then the slice file values are replaced with the sampled values for
selected draw points. This is somewhat similar to the open pit practice
of taking blast-hole samples to improve the local grade of a bench
about to be blasted and mined.
Figure 12 shows a single draw column with sampled values at
various heights (HOD) above the draw point. The graph shows these
together with the trend line and some extrapolated points. Maximum
and minimum grade values are set so that steep up or downward
trends do not generate unrealistic grade values.
Figure 9. How CMS fits in between historical tons and future plans.
Each draw point can be categorized in a variety of ways (Figure
10), including over-draw, under-draw, normal, draw-bell development,
wet muck (which is a safely concern) or as requiring special treatment.
The tonnage for each category is set accordingly.
The residual slice file model can be used to re-estimate a block specified which allows this material to be mixed with material higher up
model which can be used either in a multi-lift mining situation or each draw column.
as part of a new block model for a super-pit which some mines
are considering. Table 1. Material mixing options in PCBC.
Method Ease of use Linear? Comment
No mixing base case (In-
No mixing Easy Yes
situ)
Pre-Vertical PCBC Default (includes
Easy Yes
mixing pre-erosion)
Laubscher
Easy Yes Uses Laubscher tables
mixing
Older method, includes
Sequential mixing Harder No
toppling
Most flexible option
Template Mixing Harder No
available
Not yet generally
REBOP interface Harder No
available
Toppling
Rilling
Draw
points
Figure 17. Sandbox example with toppling, rilling and vertical mixing.
1.00
Cu%_M3P3
0.80
Figure 18. REBOP results displayed in PCBC (Markers left and cones
to right). 0.60 Cu% Tm4
0.40
As a separate, but related project, Gemcom worked with Rio Tinto 0.20
to calibrate PCBC and REBOP against one another. Figure 19 shows
0.00
the geometry of the calibration problem. 50 fictitious draw points were
Jan-08
Jan-09
Jan-10
Jan-11
Jan-12
Jan-13
Jan-14
Jul-08
Jul-09
Jul-10
Jul-11
Jul-12
Jul-13
Jul-14
The original version of PCBC used what we term Laubscher during a production schedule. Figure 26 shows a plot of HOD for the
mixing (Laubscher, 1994). This was replaced by pre-vertical and same mining step.
sequential mixing options in 1994 in PCBC. However, there are still
projects (or people) who like to be able to compare back against the
Laubscher mixing. So it was re-introduced into PCBC in 2008. It is
also useful for comparison against Footprint Finder results which use
the same mixing.
Figure 21 shows an example comparing dilution entry from a
single draw column. PCBC pre-vertical mixing has more of an S-curve
dilution entry where Laubscher mixing has a straight line dilution entry.
In addition to the static displays available, a more dynamic display Figure 27 shows the proximity of the block cave draw columns to
of selected information for individual draw points by right click or the large open pit.
mouse movement over draw points can be very useful (Figure 24).
Figure 28 gives an idea of the variability of the orebody edges and
Other options are size based plots (Figure 13), 3D draw columns also alludes to the difficulties in sequencing and scheduling such a
(Figure 14) and plotting of draw points in appropriate shapes. large orebody (grid size above is 200m!).
Figure 27. Grasberg block cave in close proximity to the large open
pit
Figure 24. CMS control panel with right click and display
information.
Figure 26. Height of draw profile at Freeport DOZ mine from Playback
tool.
Salvador
The Salvador mine in Chile has used PCBC both for the detailed
scheduling of individual mining panels (Figure 30) as well as for Figure 29. Northparkes E26 mining.
8 Copyright 2010 by SME
SME Annual Meeting
Feb. 28-Mar. 03, 2010, Phoenix, AZ
Figure 33. Plan of Andina third panel showing grizzly and LHD
sectors and existing development.
Palabora
The Rio Tinto Palabora mine in South Africa started block cave
production in 2000. (Moss, Russell, & Jones, 2004)
As the scheduler can work with historical tonnages as easily as
forward looking tonnages, the playback tool can thus also be used for
historical analysis or reconciliation purposes.
Palabora went through a period during which it was difficult to
achieve good draw control. However, more recently, the draw control
has been much improved. (Pretorius & Ngidi, 2008) This is shown
clearly in Figure 34.
Figure 34. Monthly tonnage display (poor draw control (top) and good
draw control (bottom).
Figure 35, also from Palabora, shows a plot of seismic events for
one month together with draw points and the cave Height of Draw
profile (which is different from cave back profile).
Ridgeway
The Ridgeway Deeps Mine of Newcrest used PCBC with the
Template Mixing option to study rilling and how this impacted the
mineable reserve and overall production schedule (Burgio & Diering,
Figure 32. View of three panels (lifts) at Andina mine. 2008) (Figure 36)
9 Copyright 2010 by SME
SME Annual Meeting
Feb. 28-Mar. 03, 2010, Phoenix, AZ
Figure 44 the observed rock types vs time. A detailed study of the graphs show how mixing parameters were adjusted to improve both
differences between the two can be very informative and lead to ways the model results and the confidence in other forecast results.
to improve the model which may not be apparent from the grade model
/ assays. The geological modeling thus provides another dimension
into the calibration process.
Palabora
Work has been done at Palabora to improve the short term grade
estimates using the LSQ tool described above. Figure 45 shows the
improvement in the short term comparing the PCBC LSQ forecast vs
Samples. Figure 45. Measured vs PCBC standard and LSQ adjusted grades.
Figure 46 shows a similar set of graphs, but comparing hang up Example of block model adjustment
frequency. In this case, there was no initial model forecast for hang- In this example (Figure 49, Figure 35), various attempts were
ups, but based on the LSQ approach, a reasonable forecast for short made to calibrate the PCBC results with the observed mill feed grades.
term hang up frequency was achieved. However, the PCBC grades were too high irrespective of the mixing
parameters used. This is an example in which the underlying block
Salvador model is at fault. Re-estimation of the block model with different
Various calibration runs were done at Salvador mine. Two interpolation parameters has largely resolved this discrepancy.
examples are shown in Figure 47 and Figure 48. A detailed
explanation of the curves is beyond the scope of this paper, but the
Figure 48. Various PCBC runs vs assayed results for ICE sector,
Salvador mine for 17 months.
not always the material mixing which required the most adjustment.
Careful attention is also required in areas of past mining, or for open pit
failure material or even to the geological block model itself.
The calibration examples also clearly show the benefits to be
gained from doing a calibration exercise using both grades and rock
types. In each case, a clearer understanding caving mechanisms is
gained from the work. This type of calibration also strongly justifies the
effort of taking draw point samples for grade and rock types.
The development process for PCBC has been significantly
enhanced by collaborative projects with key clients and this assistance
is gratefully acknowledged.
ACKNOWLEDGEMENTS
The authors would like to thank the following mining companies
for permissions to publish information and figures pertaining to their
projects in this paper: Freeport-McMoRan Copper & Gold Inc., Rio
Tinto, Newcrest Mining Limited, De Beers Consolidated Mines
Finsch mine, Codelco Divisin Salvador, Codelco Divisin Andina and
Palabora Mining Company.
The authors also gratefully acknowledge assistance with the
development of the software from Freeport-McMoRan Copper & Gold
Inc., PT Freeport Indonesia, Rio Tinto, De Beers Consolidated Mines
Finsch mine, Codelco Divisin Andina and Palabora Mining Company.
REFERENCES
1. Brannon, C., Casten, T., & Johnson, M. (2004). Design of the
Grasberg block cave mine. MassMin, (pp. 623 - 628). Santiago.
2. Burgio, N., & Diering, T. (2008). Simulating irregular cave
propagation using PCBC. MassMin, (pp. 1033 - 1042). Lulea.
3. Diering, T. (2004). Combining long term scheduling and daily
draw control for block cave mines. MassMin, (pp. 486 - 490).
Santiago.
4. Diering, T. (2000). PC-BC: A block cave design and draw control
system. MassMin, (pp. 469-484). Brisbane.
5. Diering, T. (2007). Template Mixing: A Depletion Engine for Block
Cave Scheduling. APCOM, (pp. 313 - 320). Santiago.
6. Laubscher, D. (1994). Cave Mining: State of the Art. SAIMM ,
October, 279 - 293.
7. Moss, A., Russell, F., & Jones, C. (2004). Caving and
Fragmentation at Palabora: Prediction to Production. MassMin,
(pp. 585 - 590). Santiago.
8. Pretorius, D., & Ngidi, S. (2008). Cave management ensuring
optimal life of mine at Palabora. MassMin, (pp. 63 - 72). Lulea.
9. Richter, O., & Diering, T. (2004). Production Scheduling at Finsch
Diamond Mine. MassMin, (pp. 453 - 458). Santiago.
10. Ross, I. (2008). Northparkes E26 Lift 2 block cave A case study.
MassMin, (pp. 25 - 34). Lulea.
11. T. Casten, L. R. (2008). P.T. Freeport Indonesia's Deep Ore Zone
mine - expanding to 80,000 tonnes per day. MassMin. Lulea.
12. Villa, D., Prasetyo, R., & Diering, T. (2008). Calibration of mixing
model to predict grade at Freeports DOZ Mine. Massmin, (pp.
1053 - 1062). Lulea.