USOO8082167B2

(12) United States Patent (10) Patent N0.: US 8,082,167 B2

Menabde (45) Date of Patent: Dec. 20, 2011

(54) PROGRAM
METHOD, APPARATUS
FOR SCHEDULING
AND COMPUTER
THE 7,853,439
, , B2 *i 12/2010 gVinrller
Menabde
royan gteta.
etal.1
al.................
.. .. 703/2
EXTRACTION OF A RESOURCE AND FOR 7,925,474 B2 * 4/2011 Froyland et al. 703/1
DETERMINING THE NET PRESENT VALUE 7,957,941 B2 * 6/2011 Froyland et a1. 703/1
OF AN EXTRACTION SCHEDULE 2002/0120480 A1 * 8/2002 Kroeger ........... .. 705/7
2006/0190219 A1* 8/2006 Froyland et al. ................ .. 703/1

(75) Inventor: Merab Menabde, Cheltenham (AU) (Continued)

(73) Assignee: BHP Billiton Innovation Pty Ltd. (AU) FOREIGN PATENT DOCUMENTS
RU 2004803 C 12/1993
( * ) Notice: Subject to any disclaimer, the term of this (Continued)
patent is extended or adjusted under 35
11-3-0 154(b) by 1328 days- OTHER PUBLICATIONS
(21) Appl_ No; 11/610,617 Caccetta L, Hill SP. An application ofbranch and cut to open pit mine
scheduling. Journal of Global Optimization 2003; 27:349-65.*
(22) Filed: Dec. 14, 2006 (Continued)
(65) Prior Pumica?on Data Primary Examiner * Johnna Loftis
US 2007/0185750 A1 Aug. 9, 2007 Assistant Examiner * Brett Feeney
_ _ _ _ _ (74) Attorney, Agent, or Firm * Brinks Hofer Gilson &
(30) Foreign Application Priority Data Lione; 6 peter Nichols
Jun. 21, 2004 (AU) .............................. .. 2004903360 (57) ABSTRACT
Mar. 8, 2005 (AU) .............................. .. 2005901100 , . . . . .
The scheduling of extraction of a resource Within a mine is
(51) Int_ CL disclosed Which comprises obtaining drill hole grade data
G06F 9/44 (2006.01)
relating to the resource
'
and creatin' g a P lurality of different
'
(52) us. Cl. .................................... .. 705/7.11; 705/7.12 block models, Wlth each block In each model haYmg a
(58) Field of Classi?cation Search None resource grade and each of the block models honoring the
See a lication ?le for com lete' drill hole grade data. An extraction schedule is determined
pp p ry' based on a cut-off grade policy to increase expected net
56 R f Ct d present value of the mine. The cut-off grade policy can be a
( ) e erences l 6 variable cut-off grade and the scheduling using the variable
Us PATENT DOCUMENTS cut-off grade is converted from a non-linear representation to
. a linear representation and a mixed integer programming
2 i """ formulation is used for simultaneous optimization of the
535243077 A * 6/1996 Faaland et 31, ,,,, n 705/8 extraction schedule and the cut-off grade.
6,122,621 A * 9/2000 Shimada ............ .. . 705/7.22
6,128,540 A * 10/2000 Van Der Vegt et al. ....... .. 700/36 69 Claims, 6 Drawing Sheets

BLOCK Schedule with

to maximise
\ NP V

301
 US 8,082,167 B2
Page 2

U.S. PATENT DOCUMENTS Ramazan S and Dimitrakopoulos R. (2004). Recent applications of

2006/0265342 A1* 11/2006 Froyland et al. ............ .. 705/500
operations research in open pit mining. Transactions of the Society
2008/0046272 A1* 2/2008 Menabde et al. .. 705/1 for Mining, Metallurgy and Exploration 2004; 316:73-8.*
2009/0306942 A1* 12/2009 Froyland et al. ................ .. 703/1 Ramazan S and Dimitrakopoulos R. (2003). Production Scheduling
Optimisation in a Nickel Laterite Deposit: MIP and LP Applications
FOREIGN PATENT DOCUMENTS and Infeasibility in the Presence of Orebody Variability. Mining
RU 2204719 C 5/2003 Planning and Equipment Selection. April, pp. 1-5.*
W0 WO 2004/033854 A 4/2004 Tvedt J (2001). Ownership Structure and Optimal Field Develop
ment. DnB Markets and Norwegian School of Economics and Busi
OTHER PUBLICATIONS ness Administration. 1-14.*
Hochbaum et al, Performance Analysis and Best Implementations Blending and Schedule Optimization with MinMAX Scheduler.
MineMAX MineMaX Ltd. (2004). Slides 1-36.*
of Old and New Algorithms for the Open-Pit Mining Problem, MineMAX Scheduler screen shots web.archive.org (Jun. 4, 2004).
Department of Industrial Engineering and Operations Research, Uni PP 1-8~*
versity of California, Berkley, Jan. 16, 1998.* Whittle Strategic Mine Planning. Brochure Apr. 3, 2004*
Godoy and Dimitrakopoulos (2004). Managing Risk and Waste Min MineMAX Scheduler: Mine Scheduling Optimization (2000). White
ing in Long-Term Production Scheduling of Open-Pit Mines. Society Paper. MineMaX. pp. 1-13.*
for Mining, Mettalurgy, and Exploration. 316: 43-50.* Bezdek, J .C. et al., Convergence Theory for Fuzzy c-Means:
Gershon ME (1987). Heuristic Approaches for Mine Planning and Counterexamples and Repairs, IEEE Transactions on Systems, Man
Production Scheduling. International Journal of Mining and Geo and Cybernetics, vol. SMC-17, No. 5, pp. 873-877, Sep./Oct. 1987.
logical Engineering. 5: 1-13.* Brealey, SC. et al., Opencast Mining, The Mining Engineer, I.M
Gershon ME ( 1983). Optimal mine production scheduling evaluation E.(UK), No. 99, pp. 147-163, Dec. 1968.
of large scale mathematical programming approaches. International Caccetta, L. et al., An Application of Branch and Cut to Open Pit
Journal of Mining Engineering. 1: 315-329.* Mine Scheduling, http://rutcor.rutgers.edu/~do99/ENSHill.doc,
Hack DR (2002). Comparison of Commercial Mine Planning Pack pp. 1-24, 1999.
ages Available to the Aggregate Industry. GeoNumerics, Inc. pp. International Search Report for corresponding PCT Patent Applica
1-9.* tion U.S. Appl. No. PCT/AU2005/000761 dated Jul. 8, 2005.
Hochbaum DS and ChenSource A (2000). Performance Analysis and Kaas, L.M. et al., Computer Techniques for Production Scheduling
Best Implementations of Old and New Algorithms for the Open-Pit and Blending of Taconite, Proceedings of the 26h Annual Mining
Mining Problem. Operations Research. 48 (6): 894-914.* Symposium, University ofMinnesota, pp. 135-144, Jan. 1965.
ILOG CPLEX 9.0 (2003). ILOG. pp. 1-564.* Lerchs, H. et al. Optimum Design of Open-Pit Mines, Transac
Ramazan S and Dimitrakopoulos R. (2004). Stochastic Optimisation tions, C.I.M, vol. LXVII, pp. 17-24, 1965.
of Long-Term Production Scheduling for Open Pit Mines With a New Pana, MT, The Simulation Approach to Open-Pit Design, Trans
Integer Programming Formulation, Orebody Modelling and Strate actions ofthe Short Course and Symposium on Computers and Com
gic Mine Planning Spectrum Series 14: 359-365. *Available on puterApplications in Mining and Exploration, University ofArizona,
Google Scholar. Do not have permissions to attach document.* pp. 1-23, Mar. 1965.
Kanevski M, Parkin R, Poznukhov A, Timonin V, Maignan M, Underwood, R. et al, A Mathematical Programming Viewpoint for
Yatsalo B and Canu S (2004). Environmental Data Mining and Solving the Ultimate Pit Problem, EJOR, pp. 1-18, 1998.
Modelling Based on Machine Learning Algorithms and Wharton, C. et al., The Effect of Minimum Mining Width on NPV,
Geostatistics. Environmental Sciences and Arti?cial Intelligence. 19 Proceedings of the 1997 Whittle Conference, Optimizing with
(9): 845-855.* Whittle, Perth, WA, pp. 173-178, Apr. 1997.
Knowles TW (1999). Optimization Models for Mine Planning. Com
puters & Industrial Engineering 37: 469-472.* * cited by examiner
US. Patent Dec. 20, 2011 Sheet 1 0f6 US 8,082,167 B2

10
11 11 1212 11 I/ 11 11

FIGURE 1

120
grade 12b

; ; cos
I |
I |
|
l| ll g
: : i A
i I : l 120
E E l 1
O I 2 3 time

FIGURE 2
303
302 /
BLOCK Schedule with
I I vurable COG
DATA MODEL Aggreqmion to Imuximise
) NPV
301 FIGURE 3
US. Patent Dec. 20, 2011 Sheet 2 0f6 US 8,082,167 B2

total 17
tonnes in J
block with -
grade greater
than 9 15
COG

9 grade of block

FIGURE 4

f 5030
Schedule with ?xed
-'-' COG to maximise
NPV
MULTIPLE
DATA > BLOCK -- Aggregation
MODEL (
Schedule with
) 502 __> variable COG
501 / gianqximise
503: L J
T
FIGURE 5 503
US. Patent Dec. 20, 2011 Sheet 3 0f 6 US 8,082,167 B2

panel
order
--P Average of 10
optimise using valuations of
10 Conditional ___.__ grades of all
Simulations , . the Conditional
10 Sif?UthlO?S I, Simulations
cosh) '
NPV $788M

FIGURE 6A

, _ Calculate Average Ptlmi$e Sing

10 Condltlonal hm, fades ____-|- only average
Simulations E_upe) grade

panel order

l coca)
Average of 10 Value of the
valuations of Overqqe grade
the Conditional
Slrnulations
(E"'dt'li'e)
m e

NPV $739M NPV $671M

FIGURE 68
US. Patent Dec. 20, 2011 Sheet 4 0f6 US 8,082,167 B2

35,000,000

30. 000.000 -

25.000.000 '

Rodc?nea) 20,000,000 "

15.000.000

10,000.000

5.000.000 -*

0%

FIGURE 7

8600.000.
3500.000000
00,000,0ch
$300,000,000
; $200,000,000
900.000.0004
30
$100,000.000
~$200,000,000'
0300,000,000
-$400.000.000
Periods

FIGURE 8
US. Patent Dec. 20, 2011 Sheet 5 0f6 US 8,082,167 B2

35,000,000

30,000,000 __ _.

25.000, 000 -

Rock?na) GCruntdoef?)
0 t0
20.000. 000

FIGURE 9

$600,000.00
$500,000,000J
$400,000,0001
$300,000,000
% 8200. 000. 000
3100.000000

$1 00,000, 000 -

$200,000,000

-$300,000.000'
4400000000

FIGURE 10
US. Patent Dec. 20, 2011 Sheet 6 0f6 US 8,082,167 B2

30.000000 1

30.000.0004 -'+1-

25,000,000- P10 5?
85 20,000,000* L 0.9 1;5
$1 a:
15.ooo,ooo-
m
ms 83
10,000,000 ~ 07

5,000,000- P05

01 ~05

FIGURE 11

$500,000,000
$500,000,000}
$400,000,000
$300,000,000
; $200,000,000
$100.000.ooo~
$0
4100.000000
-- $200.00CWQO

-Q300.000.000'
-$+00.000.00G
Periods

FIGURE 12
 US 8,082,167 B2
1 2
METHOD, APPARATUS AND COMPUTER whether a particular block is sent for processing to extract the
PROGRAM FOR SCHEDULING THE resource, sent to waste, or stockpiled for later processing.
EXTRACTION OF A RESOURCE AND FOR Because the drillholes are generally drilled some distance
DETERMINING THE NET PRESENT VALUE apart, the drillhole data is usually sparse and therefore this
OF AN EXTRACTION SCHEDULE introduces inherent errors in the deterministic block model.
To some extent, this can be overcome by providing more data
This application claims priority to PCT application PCT/ by drilling more holes. However, as is explained above, the
AU2005/000761 published in English on Dec. 29, 2005 as drilling of the drillholes is expensive and therefore, this is not
WO 2005/ 124104 and to Australian application nos. desirable.
2004903360 ?led Jun. 21, 2004, and 2005901100 ?led Mar. Thus, traditionally open pit mine planning is based on the
8, 2005, the entire contents of each are incorporated herein by block model which is built up using some kind of interpola
reference. tion technique such as the Kriging procedure so that a single
model is produced. This single model is assumed to be a fair
FIELD OF THE INVENTION representation of reality and is used for mine design and
optimisation. The design process consists of three main steps:
The present invention relates to a method and apparatus for (a) ?nding the block extraction sequence which produces the
scheduling the extraction of a resource and for determining best net present value whilst satisfying geotechnical slope
the net present value of an extraction schedule. Typically the constraints;
resource is an ore body which is to be mined in an open cut (b) designing the practically mineable mine phases (so-called
mine. 20 push backs) which are roughly based on the optimal block
sequence; and
BACKGROUND OF THE INVENTION (c) optimising the mining schedule and cut-off grades.
The cut-off grade (COG) is de?ned as the threshold such
Typically, the mining of a resource may take place over a that the blocks with a grade above it are sent to the processing
period of 15-30 years before the mine is exhausted to a posi 25 plant and with a grade below it are treated as waste. It can be
tion where further mining is not economic. constant for the whole life of mine, or can be variable, i.e.
The process of mine development and design and long dependent on the period of extraction.
term scheduling is based on spatial interpretation of drillhole In practice, the open mine is divided into a number of the
data. Thus, drillholes are drilled over the region to be mined mining phases, which are mined bench by bench, each bench
and data relating to the grade of the resource is obtained. The 30 being represented by a horizontal layer of blocks within the
drillholes are generally drilled a reasonable distance apart given mining phase and having the same elevation. A bench
because this procedure is relatively expensive. A block model within a mining phase is sometimes referred to as a panel
of the region to be mined is created and typically, the block (one or more layers of blocks). The mining phases can be
model may contain between 50,000 to 1,000,000 blocks mined one by one from top to bottom. However, this kind of
which are to be scheduled for mining over the period of 15-30 35 schedule is usually sub-optimal. Mining several phases
years. A block is that material enclosed by a rectangular prism simultaneously and applying a variable cut-off grade can
in the ground and may contain air to a volumetric percentage produce much better results. There are several commercially
strictly less than 100%. A block model is a collection of available packages which claim to optimise the schedule and
non-intersecting blocks that are usually, but not necessarily, cut-off grade using a single block model representation of the
spatially connected and which contain no less than all mate 40 resource. However, it is dif?cult to estimate their effective
rial considered to have economic value in a mining enterprise. ness as the upper theoretical limit on the net present value
The objective of the scheduling procedure is to ?nd the block remains unknown.
extraction sequence which produces the maximum possible The standard optimisation technique widely used in many
net present value (NPV) and obeys a number of constraints. industrial applications is the linear and integer programming
The constraints include: 45 (e.g. Padberg, 2003). However, in order for this program to
(a) geotechnical slope constraints which are modelled by a set operate satisfactorily, the problem to be solved needs to be
of precedence rules constraining the order of extraction of formulated as a linear one.
individual blocks;
(b) mining constraints, i.e. total maximum amount of rock SUMMARY OF THE INVENTION
which can be mined in one time period (usually 1 year); 50
(c) processing constraints, i.e. maximum amount or ore which A ?rst aspect of the invention is concerned with scheduling
can be processed through a given processing plant in one the extraction of a resource to improve the net present value.
time period; The invention may therefore be said to reside in a method
(d) and the market constraints, i.e. the maximum amount of of scheduling the extraction of a resource within a medium
metal, which can be sold in one time period; 55 comprising the steps of:
(e) Any other constraints salient to the practical mining opera obtaining drillhole grade data relating to the resource con
tion including but not restricted to maximum limits in tained within the medium;
sinking-rate and available ore. creating a plurality of different block models, each formed of
A schedule is a period of extraction for each block and a a plurality of blocks, each block in each model having a
destination for each block (waste, stockpile or process plant. 60 resource grade, each of the block models honouring the
The ore body model which is built up from the drillhole drillhole grade data; and
data is a deterministic model created by spatial interpretation scheduling extraction of blocks based on a cut-off grade
of the data using some kind of so-called Kriging procedure. policy to increase expected net present value having regard
This enables each of the blocks in the model to be assigned a to all of the block models.
resource grade (i.e. the amount of the resource present in the 65 Thus, by using a plurality of different block models, a
block). The resource grade information is then used to deter much larger number of possible representations of the pres
mine the scheduling of the mining operation, and also ence of the resource in the medium are obtained. Each of
 US 8,082,167 B2
3 4
those representations is equally likely to occur because each In the preferred embodiment, the step of scheduling using
honours the drillhole grade data which have been measured. a variable cut-off grade is converted from a non-linear repre
Thus, by using all of those possibilities, a more accurate sentation to a linear representation, and a mixed integer pro
representation of the actual resource distribution within the gramming formulation is used for simultaneous optimisation
medium can be obtained, and therefore a better indication of of the extraction schedule and the cut-off grade.
whether a region of the medium represented by a block should Preferably the scheduling is determined by
be forwarded to processing. For example, if the cut-off grade
of the resource in a particular period is 0.6%, the method is
1 N T (1)
able to see that a particular block may in some realities be
below this cutoff grade (in which case it is sent to waste), and
Maxzmzse
Mt MO 1:1
in some realities is above this cutoff grade (in which case it is
processed). By using all of the models, such as by averaging
different block valuations (processed or wasted) from differ subject to the following constraints:
ent realities we more accurately model the process in reality.
This is because at the time of extraction extra blast-hole data
are available to enable determination of a more accurate l N P c (2)
quanti?cation of the block grade and hence the selectivity foij, 5 R5), for all I
n:l [:1 jil
option encoded in the use of multiple conditional simulations
can be gainfully exploited in calculating both the schedule l N P c (3)
and undiscounted block valuation and hence ultimately in 20 ngt-j, 5 Q), for all I
obtaining a more accurate determination of the net present n:l [:1 j:l

value of each block. Thus, a schedule is obtained which gives yiykl s y;,, for all i and I (4)
an increased and more accurate net present value, compared
to existing techniques, of a mine containing the resource and r G (5)
the medium. 25 ZZXU-T s y;,, for all i
In the preferred embodiment of the invention, the creation 1:1]:1

of the plurality of different block models is a conditional c T (6)

simulation technique which provides the generation of a plu y;, 5 222cm, for all i, I and k c Si
rality of equally probable block model realisations, all of j:l 1:1
which honour the data and ?rst and second order statistics of 30
the resource represented, respectively, by probability distri 0 (7)
Zaj, = 1, for all 1
bution function and variogram. j:l
In one embodiment of the invention, the number of blocks
in each of the block models is reduced by aggregating blocks xi], 5 6],, for all i, j, and I (8)
prior to the step of scheduling extraction. Instances of block 35
aggregations are assumed to have homogeneous properties
within each grade bin. wherein;
A collection of one or more blocks that are members of a
T is the number of scheduling periods;
particular aggregate where the grade attribute of each block N is the number of simulations;
P is the total number of panels;
falls within de?ned upper and lower limits is called a grade 40
G is the number of all possible cutoff grades;
bin. Grade bins are used to practically apply cut-off grades, R1. is the total rock in the panel i in simulations n.
viZ: all blocks in grade bins above the cut-off grade are sent to Qij is the total ore in the panel i, simulation n, when mined
process plant. with the COG j;
In one embodiment of the invention, the cut-off grade is a Vij is the value of the panel i, simulation n, when mined and
?xed cut-off grade. However, in the preferred embodiment of 45 processed with the COG j;
the invention, the cut-off grade is a variable cut-off grade. RtO is the maximum mining capacity in period t;
In the preferred embodiment of the invention, the plurality QtO is the maximum processing rate in period t;
of block models are used to provide an average of resource Sl- is the set of panels that must be removed before starting the
grade and block valuations and the average of the resource panel i;
grade and block valuations is used to produce the scheduled 50 d is the time discount factor;
extraction of blocks and the cut-off grade value to increase xijt is the fraction of the panel i is extracted with the COG j in
expected net present value. period t;
In one embodiment of the invention, the scheduled extrac yit is a binary variable equal to 1 if the extraction of the panel
tion of blocks may be optimised by considering the average i has started in periods 1 to t, and equal to 0 otherwise;
valuation produced from the conditional simulations. 55 61115 a binary variable controlling the selection of the COG
However, in the preferred embodiment of the invention, all applied in period t.
of the conditional simulations are optimized simultaneously The invention may also be said to reside in an apparatus of
to produce a single extraction schedule with a variable cut-off scheduling the extraction of a resource within a medium from
grade policy, and wherein all the conditional simulations are which drillhole grade data relating to the resource contained
valued using this schedule and the valuations are averaged to 60 within the medium has been obtained comprising:
produce the expected net present value, which is substantially a processor for creating a plurality of different block mod
better than the one obtained by optimiZing the average grade els, each formed of a plurality of blocks, each block in
block model. each model having a resource grade, each of the block
In the preferred embodiment of the invention, the extrac models honouring the drillhole grade data; and
tion schedule determines whether resource and medium cor 65 a processor for scheduling extraction of blocks based on a
responding to each block is sent for processing, sent to waste cut-off grade policy to increase expected net present
or stockpiled for later processing. value having regard to all of the block models.
 US 8,082,167 B2
5
In the preferred embodiment of the invention, the processor -continued
is for creating conditional simulations which provides the 6 (7)
generation of a plurality of equally probable block model 251-, = 1, for all 1
j:l
realisations, all of which honour the data and ?rst and second
order statistics of the resource represented, respectively, by xij, s 6j,, for all i, j, and I (8)
probability distribution function and variogram.
In one embodiment of the invention, the processor is also
for reducing the number of blocks in each of the block models wherein;
by aggregating blocks prior to the step of scheduling extrac T is the number of scheduling periods;
tion. N is the number of simulations;
In one embodiment of the invention, the cut-off grade is a P is the total number of panels;
?xed cut-off grade. However, in the preferred embodiment of G is the number of all possible cutoff grades;
the invention, the cut-off grade is a variable cut-off grade. Ri is the total rock in the panel i in simulations n.
In the preferred embodiment of the invention, the processor Qij is the total ore in the panel i, simulation n, when mined
is for providing an average of resource grade and block valu with the COG j;
ations from the plurality of models, and the average of the Vij is the value of the panel i, simulation n, when mined and
resource grade and block valuations is used to produce the processed with the COG j;
scheduled extraction of blocks and the cut-off grade value to RtO is the maximum mining capacity in period t;
increase expected net present value. 20 QtO is the maximum processing rate in period t;
In one embodiment of the invention, the processor is for Sl- is the set of panels that must be removed before starting the
scheduling extraction of blocks optimised by considering the panel i;
average valuation produced from the conditional simulations. d is the time discount factor;
However, in the preferred embodiment of the invention, all xijt is the fraction of the panel i is extracted with the COG j in
of the conditional simulations are optimized simultaneously 25 period t;
to produce a single extraction schedule with a variable cut-off yit is a binary variable equal to 1 if the extraction of the panel
grade policy, and wherein all the conditional simulations are i has started in periods 1 to t, and equal to 0 otherwise;
valued using this schedule and the valuations are averaged to 61,15 a binary variable controlling the selection of the COG
produce the expected net present value, which is substantially applied in period t.
better than the one obtained by optimiZing the average grade 30 The invention may therefore be said to reside in a computer
block model. program for scheduling the extraction of a resource within a
In the preferred embodiment of the invention, the extrac medium comprising:
tion schedule determines whether resource and medium cor code for receiving drillhole grade data relating to the resource
responding to each block is sent for processing, sent to waste contained within the medium;
or stockpiled for later processing. 35 code for creating a plurality of different block models, each
In the preferred embodiment, the processor is for convert formed of a plurality of blocks, each block in each model
ing from a non-linear representation to a linear representation having a resource grade, each of the block models honour
when a variable cut-off grade is used, and a mixed integer ing the drillhole grade data; and
programming formulation is used for simultaneous optimisa code for scheduling extraction of blocks based on a cut-off
tion of the extraction schedule and the cut-off grade. 40 grade policy to increase expected net present value having
Preferably the scheduling is determined by the processor regard to all of the block models.
by In the preferred embodiment of the invention, the code for
creation of the plurality of different block models is code for
performing a conditional simulation technique which pro
(1) 45 vides the generation of a plurality of equally probable block
2
1:1 1
model realisations, all of which honour the data and ?rst and
second order statistics of the resource represented, respec
tively, by probability distribution function and variogram.
In one embodiment of the invention, the program further
subject to the following constraints:
50 comprises code for reducing the number of blocks in each of
the block models by aggregating blocks prior to the step of
1 N P c (2)
scheduling extraction.
fom 5 R5), for all I In one embodiment of the invention, the cut-off grade is a
":1 z:1 jil ?xed cut-off grade. However, in the preferred embodiment of
55 the invention, the cut-off grade is a variable cut-off grade.
1 N P c (3)
In the preferred embodiment of the invention, the code for
Hz 2 Z ij-xij, 5 Q9, for all 1 scheduling includes code to provide an average of resource
":1 z:1 F1
grade and block valuations and the average of the resource
y;,,21 s y;,, for all i and r (4) grade and block valuations is used to produce the scheduled
60 extraction of blocks and the cut-off grade value to increase
r G (5)
z ijr i y;,, for all i
expected net present value.
1:1]:1 In one embodiment of the invention, the code for schedul
ing further comprises code for optimiZing extraction by con
G T (6) sidering the average valuation produced from the conditional
yir 5 222cm, for all i, l and k c Si 65 simulations.
1:1 1:1
However, in the preferred embodiment of the invention, the
code for scheduling further comprises code forusing all of the
 US 8,082,167 B2
7 8
conditional simulations to optimize simultaneously to pro 61115 a binary variable controlling the selection of the COG
duce a single extraction schedule with a variable cut-off grade applied in period t.
policy, and wherein all the conditional simulations are valued The invention may therefore be said to reside in a method
using this schedule and the valuations are averaged to pro of mining a mine comprising a resource within a medium, the
duce the expected net present value, which is substantially 5 method comprising:
better than the one obtained by optimiZing the average grade removing medium and resource from the mine using an
block model. extraction schedule created by the following steps;
In the preferred embodiment of the invention, the extrac obtaining drillhole grade data relating to the resource con
tion schedule determines whether resource and medium cor
tained within the medium;
responding to each block is sent for processing, sent to waste 10
creating a plurality of different block models, each formed
or stockpiled for later processing.
of a plurality of blocks, each block in each model having
In the preferred embodiment, the code for scheduling using
a resource grade, each of the block models honouring the
a variable cut-off grade converts from a non-linear represen
tation to a linear representation, and further comprises a drillhole grade data; and
mixed integer programming formulation is used for simulta scheduling extraction of blocks based on a cut-off grade
neous optimisation of the extraction schedule and the cut-off policy to increase expected net present value having
grade. regard to all of the block models.
Preferably the code for scheduling is to In the preferred embodiment of the invention, the creation
of the plurality of different block models is a conditional
20 simulation technique which provides the generation of a plu
(1) rality of equally probable block model realisations, all of
which honour the data and ?rst and second order statistics of
1
the resource represented, respectively, by probability distri
bution function and variogram.
subject to the following constraints: In one embodiment of the invention, the number of blocks
in each of the block models is reduced by aggregating blocks
prior to the step of scheduling extraction. Instances of block
1 N P c (2) aggregations are assumed to have homogeneous properties
foij, 5 R9, for all I within each grade bin.
n:l [:1 jil 30
In one embodiment of the invention, the cut-off grade is a
1 N P c (3) ?xed cut-off grade. However, in the preferred embodiment of
Qflxijr 5 Q), for all I the invention, the cut-off grade is a variable cut-off grade.
n:l [:1 j:l
In the preferred embodiment of the invention, the plurality
yiykl s y;,, for all i and I (4) of block models are used to provide an average of resource
grade and block valuations and the average of the resource
r G (5)
grade and block valuations is used to produce the scheduled
z ijr i y, for all i
1:1]:1 extraction of blocks and the cut-off grade value to increase
expected net present value.
c T (6) In one embodiment of the invention, the scheduled extrac
40
y i: i 22 xk-T,
1 for all i, I and k c Si
tion of blocks may be optimised by considering the average
j:l 1:1
valuation produced from the conditional simulations.
G (7) However, in the preferred embodiment of the invention, all
251-, = 1, for all 1 of the conditional simulations are optimized simultaneously
j:l
to produce a single extraction schedule with a variable cut-off
xij, s 6],, for all i, j, and I (8) grade policy, and wherein all the conditional simulations are
valued using this schedule and the valuations are averaged to
produce the expected net present value, which is substantially
wherein; better than the one obtained by optimiZing the average grade
T is the number of scheduling periods; 50 block model.
N is the number of simulations; In the preferred embodiment of the invention, the extrac
P is the total number of panels; tion schedule determines whether resource and medium cor
G is the number of all possible cutoff grades; responding to each block is sent for processing, sent to waste
R1. is the total rock in the panel i in simulations n. or stockpiled for later processing.
Qij is the total ore in the panel i, simulation n, when mined 55
In the preferred embodiment, the step of scheduling using
with the COG j; a variable cut-off grade is converted from a non-linear repre
Vij is the value of the panel i, simulation n, when mined and sentation to a linear representation, and a mixed integer pro
processed with the COG j; gramming formulation is used for simultaneous optimisation
RtO is the maximum mining capacity in period t; of the extraction schedule and the cut-off grade.
QtO is the maximum processing rate in period t; 60 Preferably the scheduling is determined by
Sl- is the set of panels that must be removed before starting the
panel i;
d is the time discount factor; 1 N T (1)
xijt is the fraction of the panel i is extracted with the COG j in
period t; 65
Maxzmzse
Mt MO
J 1:1

yit is a binary variable equal to 1 if the extraction of the panel

i has started in periods 1 to t, and equal to 0 otherwise;
 US 8,082,167 B2
10
subject to the following constraints: In one embodiment of the invention, the cut-off grade is a
?xed cut-off grade. However, in the preferred embodiment of
the invention, the cut-off grade is a variable cut-off grade.
l N P c (2) In the preferred embodiment of the invention, the plurality
Rfxij, 5 R5), for all I
11:1 1:1 j:l
of block models are used to provide an average of resource
grade and block valuations and the average of the resource
l N P c (3) grade and block valuations is used to produce the scheduled
QZ-xt-j, 5 Q), for all I extraction of blocks and the cut-off grade value to increase
11:1 1:1 j:1
expected net present value.
yiykl s y;,, for all i and I (4) In one embodiment of the invention, the scheduled extrac
tion of blocks may be optimised by considering the average
1 G (5)
z xii-T s y;,, for all i
valuation produced from the conditional simulations.
1:1]:1 However, in the preferred embodiment of the invention, all
of the conditional simulations are optimized simultaneously
c T (6)
to produce a single extraction schedule with a variable cut-off
y;, g xkjr, for all i, I and k c Si
j:1 1:1 grade policy, and wherein all the conditional simulations are
valued using this schedule and the valuations are averaged to
0 (7) produce the expected net present value, which is substantially
z 51-, : 1, for all 1 better than the one obtained by optimiZing the average grade
j:1 20
block model.
xi], 5 6],, for all i, j, and I (8) In the preferred embodiment of the invention, the extrac
tion schedule determines whether resource and medium cor
responding to each block is sent for processing, sent to waste
wherein; 25 or stockpiled for later processing.
T is the number of scheduling periods; In the preferred embodiment, the step of scheduling using
N is the number of simulations; a variable cut-off grade is converted from a non-linear repre
P is the total number of panels;
sentation to a linear representation, and a mixed integer pro
G is the number of all possible cutoff grades;
R, is the total rock in the panel i in simulations n. gramming formulation is used for simultaneous optimisation
Qij is the total ore in the panel i, simulation 11, when mined 30 of the extraction schedule and the cut-off grade.
with the COG j; Preferably the scheduling is determined by
Vij is the value of the panel i, simulation 11, when mined and
rocessed with the COG j;
Rt is the maximum mining capacity in period t; N P c T (1)
QtO is the maximum processing rate in period t; 35 Maximise Vgxijrdt
11:1 1:1 j:l 1:1
S,- is the set of panels that must be removed before starting the
panel i;
d is the time discount factor; subject to the following constraints:
xijt is the fraction of the panel i is extracted with the COG j in
period t; 40
yit is a binary variable equal to 1 if the extraction of the panel l N P c (2)
i has started in periods 1 to t, and equal to 0 otherwise; Rfxij, 5 R9, for all I
6]., is a binary variable controlling the selection of the COG 11:1 1:1 j:l

applied in period t. l N P c (3)

The invention may therefore be said to reside in a method 45
QZ-xt-j, 5 Q), for all I
of determining the net present value of a mine comprising a 11:1 1:1 j:1
resource within a medium, the method comprising the steps
of: yLH s y;,, for all i and I (4)
obtaining drillhole grade data relating to the resource con 1 G (5)
tained within the medium; 50 22 xii-T s y;,, for all i
creating a plurality of different block models, each formed of 1:1]:1
a plurality of blocks, each block in each model having a
c T (6)
resource grade, each of the block models honouring the y;, g xkjr, for all i, I and k c Si
drillhole grade data; and j:1 1:1
scheduling extraction of blocks based on a cut-off grade 55
policy to provide the net present value having regard to all 0 (7)
of the block models. Z 51-, : 1, for all 1
j:1
In the preferred embodiment of the invention, the creation
of the plurality of different block models is a conditional xi], 5 6],, for all i, j, and I (8)
simulation technique which provides the generation of a plu 60
rality of equally probable block model realisations, all of
which honour the data and ?rst and second order statistics of wherein;
the resource represented, respectively, by probability distri T is the number of scheduling periods;
bution function and variogram. N is the number of simulations;
In one embodiment of the invention, the number of blocks 65 P is the total number of panels;
in each of the block models is reduced by aggregating blocks G is the number of all possible cutoff grades;
prior to the step of scheduling extraction. R, is the total rock in the panel i in simulations n.
 US 8,082,167 B2
11 12
Qij is the total ore in the panel i, simulation n, when mined However, in the preferred embodiment of the invention, all
with the COG j; of the conditional simulations are optimised to produce a
Vij is the value of the panel i, simulation n, when mined and plurality of extraction schedules, each having valuations and
processed with the COG j; the valuations are averaged to produce the net present value,
RtO is the maximum mining capacity in period t; and the extraction schedule comprises the extraction schedule
QtO is the maximum processing rate in period t; which is closest to that giving the net present value.
Sl- is the set of panels that must be removed before starting the In the preferred embodiment of the invention, the extrac
panel i; tion schedule determines whether resource and medium cor
d is the time discount factor; responding to each block is sent for processing, sent to waste
xijt is the fraction of the panel i is extracted with the COG j in or stockpiled for later processing.
period t; Preferably the scheduling is determined by
yit is a binary variable equal to 1 if the extraction of the panel
i has started in periods 1 to t, and equal to 0 otherwise;
6]., is a binary variable controlling the selection of the COG 1 N T (1)
applied in period t.
A further aspect of the invention concerns reducing the
Maximise
Mn: Mo 1:1

dif?culty of processing information in order to provide

extraction schedules and net present value estimations.
subject to the following constraints:
According to a further aspect, the invention may be said to 20
reside in a method for scheduling the extraction of a resource
contained within a medium, comprising the steps of: 1 N P c (2)
obtaining drillhole grade data relating to the resource con Rfxij, 5 R9, for all i
tained within the medium; ":1 [:1 jil
creating a block model with each block in the model having 25 l N P c
a resource grade; z z ngt-j, 5 Q), for all i (3)
scheduling extraction of blocks based on a variable cut-off Nn:l [:1 F1
grade value to increase expected net present value, the yiki 5 ya, for all i and 1 (4)
variable cut-off grade being represented by a non-linear r G

function; and 30 2 zxijr 5 yin for all i (5)

converting the non-linear representation to a linear repre 1:1]:1

sentation and using a mixed integer programming for c T

mulation for a simultaneous optimisation of an extrac y s 2 2%, for all i, I and k c Si

j:l 1:1 (6)
tion sequence provided by the scheduled extraction and
cut-off grade policy. 35
c

Thus, the method according to this aspect of the invention z 61': :1, for all 1 (7)
F1
simpli?es processing because of the linearisation of the prob
Xi]: 5 51-1, for all i, j, and 1 (8)
lem and also therefore enables a variable cut-off grade to be
used in the optimisation instead of a ?xed cut-off grade value.
In the preferred embodiment of the invention, the non 40
linear function is converted to a plurality of linear values by wherein;
producing a plurality of step values relating to the non-linear T is the number of scheduling periods;
function, and selecting one of the values to provide the cut-off N is the number of simulations;
grade decision by the mixed integer programming function. P is the total number of panels;
In one embodiment of the invention, the block model is a 45 G is the number of all possible cutoff grades;
deterministic model built by spatial interpretation of the data Ri is the total rock in the panel i in simulations n.
using a Kriging procedure. Thus, in this embodiment, only a Qij is the total ore in the panel i, simulation n, when mined
single model is provided. with the COG j;
However, in the preferred embodiment of the invention, the Vij is the value of the panel i, simulation n, when mined and
block model is a conditional simulation producing a plurality 50 processed with the COG j;
of different block models providing a number of equally RtO is the maximum mining capacity in period t;
probably realisations of block models honouring the drillhole QtO is the maximum processing rate in period t;
data and ?rst and second order statistics of the resource dis Sl- is the set of panels that must be removed before starting the
tributed in the medium represented, respectively, by probabil panel i;
ity distribution and variogram. 55
d is the time discount factor;
In one embodiment of the invention, the number of blocks
xijt is the fraction of the panel i is extracted with the COG j in
in each of the block models is reduced by aggregating blocks
prior to the step of scheduling extraction. period t;
In the preferred embodiment of the invention, the plurality yit is a binary variable equal to 1 if the extraction of the panel
of block models are used to provide an average of resource 60 i has started in periods 1 to t, and equal to 0 otherwise;
grade valuations and the average of the resource grade valu oij is a binary variable controlling the selection of the COG
ations is used to produce the scheduled extraction of blocks applied in period t.
and the cut-off grade value to increase expected net present According to a further aspect, the invention may be said to
value. reside in an apparatus for scheduling the extraction of a
In one embodiment of the invention, the scheduled extrac 65 resource contained within a medium, comprising:
tion of blocks may be optimised by considering the average a processor for receiving drillhole grade data relating to the
valuation produced from the conditional simulations. resource contained within the medium;
 US 8,082,167 B2
13 14
a processor for creating a block model with each block in the subject to the following constraints:
model having a resource grade;
a processor for scheduling extraction of blocks based on a
l N P c (2)
variable cut-off grade value to increase expected net Rfxij, 5 R9, for all I
present value, the variable cut-off grade being represented ":1 z:1 jil

by a non-linear function; and 1 N P c

z z ngm 5 Q), for all I (3)
a processor for converting the non-linear representation to a Nn:l z:1 F1
linear representation and using a mixed integer program yiki 5 y, for all i and z (4)
ming formulation for a simultaneous optimisation of an r G

extraction sequence provided by the scheduled extraction 2 zxijr 5 yin for all i (5)
and cut-off grade policy. 1:1]:1
c T
In the preferred embodiment of the invention, the processor y s 2 21%, for all i, 1 and k c Si
is for converting the non-linear function to a plurality of linear j:l 1:1 (6)
values by producing a plurality of step values relating to the c

non-linear function, and selecting one of the values to provide z 61': :1, for all 1 (7)
F1
the cut-off grade decision by the mixed integer programming
function. Xi]: 5 51-1, for all i, j, and z (8)
20
In one embodiment of the invention, the block model is a
deterministic model built by spatial interpretation of the data wherein;
using a Kriging procedure. Thus, in this embodiment, only a T is the number of scheduling periods;
single model is provided. N is the number of simulations;
25 P is the total number of panels;
However, in the preferred embodiment of the invention, the G is the number of all possible cutoff grades;
block model is a conditional simulation producing a plurality R1. is the total rock in the panel i in simulations n.
of different block models providing a number of equally Qij is the total ore in the panel i, simulation n, when mined
probably realisations of block models honouring the drillhole with the COG j;
data and ?rst and second order statistics of the resource dis 30
Vij is the value of the panel i, simulation n, when mined and
processed with the COG j;
tributed in the medium represented, respectively, by probabil R,0 is the maximum mining capacity in period t;
ity distribution and variogram. QtO is the maximum processing rate in period t;
In one embodiment of the invention, the number of blocks Sl- is the set of panels that must be removed before starting the
in each of the block models is reduced by aggregating blocks panel i;
35 d is the time discount factor;
prior to the step of scheduling extraction. xijt is the fraction of the panel i is extracted with the COG j in
In the preferred embodiment of the invention, the processor period t;
is for averaging of resource grade valuations from the block yit is a binary variable equal to 1 if the extraction of the panel
models and the average of the resource grade valuations is i has started in periods 1 to t, and equal to 0 otherwise;
used to produce the scheduled extraction of blocks and the
40 61,15 a binary variable controlling the selection of the COG
applied in period t.
cut-off grade value to increase expected net present value. According to a further aspect, the invention may be said to
In one embodiment of the invention, the scheduled extrac reside in a computer program for scheduling the extraction of
tion of blocks may be optimised by considering the average a resource contained within a medium, comprising the steps
valuation produced from the conditional simulations. 45 of:
code for receiving drillhole grade data relating to the resource
However, in the preferred embodiment of the invention, the contained within the medium;
processor is for optimiZing all of the conditional simulations code for creating a block model with each block in the model
to produce a plurality of extraction schedules, each having having a resource grade;
valuations and the valuations are averaged to produce the net 50 code for scheduling extraction of blocks based on a variable
present value, and the extraction schedule comprises the cut-off grade value to increase expected net present value,
extraction schedule which is closest to that giving the net the variable cut-off grade being represented by a non-linear
present value. function; and
code for converting the non-linear representation to a linear
In the preferred embodiment of the invention, the extrac 55 representation and using a mixed integer programming
tion schedule determines whether resource and medium cor formulation for a simultaneous optimisation of an extrac
responding to each block is sent for processing, sent to waste tion sequence provided by the scheduled extraction and
or stockpiled for later processing. cut-off grade policy.
Preferably the scheduling is determined by the processor In the preferred embodiment of the invention, the non
by the following: 60 linear function is converted to a plurality of linear values by
producing a plurality of step values relating to the non-linear
function, and selecting one of the values to provide the cut-off
(1) grade decision by the mixed integer programming function.
Maximise In one embodiment of the invention, the block model is a
65 deterministic model built by spatial interpretation of the data
using a Kriging procedure. Thus, in this embodiment, only a
single model is provided.
 US 8,082,167 B2
15 16
However, in the preferred embodiment of the invention, the Vij is the value of the panel i, simulation n, when mined and
block model is a conditional simulation producing a plurality processed with the COG j;
of different block models providing a number of equally RtO is the maximum mining capacity in period t;
probably realisations of block models honouring the drillhole QtO is the maximum processing rate in period t;
data and ?rst and second order statistics of the resource dis S1. is the set of panels that must be removed before starting the
tributed in the medium represented, respectively, by probabil panel i;
ity distribution and variogram. d is the time discount factor;
In one embodiment of the invention, the number of blocks xijt is the fraction of the panel i is extracted with the COG j in
in each of the block models is reduced by aggregating blocks period t;
prior to the step of scheduling extraction. yit is a binary variable equal to 1 if the extraction of the panel
In the preferred embodiment of the invention, the plurality i has started in periods 1 to t, and equal to 0 otherwise;
of block models are used to provide an average of resource 61115 a binary variable controlling the selection of the COG
grade valuations and the average of the resource grade valu applied in period t.
ations is used to produce the scheduled extraction of blocks
and the cut-off grade value to increase expected net present BRIEF DESCRIPTION OF THE DRAWINGS
value.
In one embodiment of the invention, the scheduled extrac Embodiments of the invention will be described, by way of
tion of blocks may be optimised by considering the average example, with reference to the accompanying drawings in
valuation produced from the conditional simulations. 20 which:
However, in the preferred embodiment of the invention, all FIG. 1 is a diagram showing an open cut mine in which a
of the conditional simulations are optimised to produce a resource is distributed in a medium (such as an ore body in
plurality of extraction schedules, each having valuations and earth and rock) in which a block model has been overlaid;
the valuations are averaged to produce the net present value, FIG. 2 is a graph showing COG against time;
and the extraction schedule comprises the extraction schedule 25 FIG. 3 is a block diagram according to one embodiment of
which is closest to that giving the net present value. the invention;
In the preferred embodiment of the invention, the extrac FIG. 4 is a graph showing the non-linear representation of
tion schedule determines whether resource and medium cor variable cut-off grade;
responding to each block is sent for processing, sent to waste FIG. 5 is a block diagram showing net present value esti
30 mations according to embodiments of the invention;
or stockpiled for later processing.
Preferably the scheduling is determined by FIGS. 6A and 6B are diagrams showing various valuations
of net present value;
FIG. 7 is a graph showing mining schedule optimised with
IN G (1) a marginal cut-off grade;
35
FIG. 8 is a graph showing net present value of the schedule
1
optimisation with the marginal cut-off grade;
FIG. 9 is a graph showing a mining schedule optimised
subject to the following constraints: with a mean grade model;
40 FIG. 10 is a graph showing net present value of the sched
uled optimisation with the mean grade model;
l N P c (2) FIG. 11 is a graph showing mining schedule optimised
Rfxj, 5 R5), for all I with a set of conditional simulations;
":1 [:1 jil
FIG. 12 is a graph showing net present value of the sched
1 N P 0
45 ule optimised with the set of conditional simulations.
?x 2 z QZ-XU, 5 Q), for all z (3)
":1 [:1 F1
DETAILED DESCRIPTION OF THE
yiki i y for all i and z (4)
EMBODIMENTS
r G

Ext-? _ y, for all i (5)

1:1]:1 50 With reference to FIG. 1 an open cut mine 10 is shown
c T which is comprised of a resource such as an ore body which
y s 2 Zxk? for all i, 1 and k c Si is distributed in a medium such as earth and rock.
j:l 1:1 (6) The open cut mine 10 is to be mined over a signi?cant
c period, such as 10-30 years, and the preferred embodiment of
251', =1 for all I (7) 55 the invention is concerned with optimising return from the
F1 mine by scheduling mining phases to maximum net present
X1]: 5 511 for all i, j, and z (8) value of the mine 10.
The mine 10 is initially surveyed by drilling a plurality of
drillholes represented by the lines 11 to provide data relating
wherein; 60 to the resource contained within the mine. The data comprises
T is the number of scheduling periods; the grade of the resource at particular locations of the drill
N is the number of simulations; holes 11. Typically the data will provide a value which rep
P is the total number of panels; resents the grade of the resource as a percentage such as 0.1%,
G is the number of all possible cutoff grades; 0.6%, 1%, etc.
Ri is the total rock in the panel i in simulations n. 65 A block model of the mine 10 is created which is formed
Qij is the total ore in the panel i, simulation n, when mined from a plurality of blocks 12. Typically, the block model may
with the COG j; contain up to half a million blocks.
 US 8,082,167 B2
17 18
The blocks 12 provide an ore body model of blocks built up The use of a variable cut-off grade introduces a signi?cant
by the spatial interpolation of the data obtained from the disadvantage in that optimisation of the schedule to obtain
drillholes 11. maximised net present value using variable cut-off grade in its
Typically, it is desirable to mine the most valuable parts of direct formulation leads to a non-linear problem. This is much
the mine as quickly or as early as possible. Also typically, the harder to solve than a linear problem or linear value which
most valuable parts of the mine are likely to be well below occurs if a ?xed variable cut-off grade is used in the formu
ground level, which means that a large amount of rock and lation. To overcome this disadvantage, the preferred embodi
earth needs to be removed before the most valuable deposits ment of the invention provides a linearisation of the problem,
are uncovered. The preferred embodiments of the invention making it possible to use a mixed integer programming for
therefore seek to optimise the mining schedule so that regions mulation for a simultaneous optimisation of the extraction
of the mine represented by various ones of the blocks are sequence and cut-off grade.
removed to enable the mo st valuable deposits to be uncovered FIG. 4 is a graph representing the non-linear cut-off grade
as quickly as possible. The blocks which are mined prior to 15 which is a plot of total tonnes in a block with a grade
those blocks need to be considered to determine whether they greater than grade plotted against grade. In order to linearise
are to be forwarded to waste, stockpiled for later processing, the problem, the function 15 is divided into a number of linear
or forwarded direct to processing. By interpolating the grade steps represented at 17, each providing a discrete grade value
of the ore in each of the blocks, a decision can be made at a such as G1, as shown in FIG. 4. The grade value G1 is selected
particular time as to whether it is desirable to forward the 20 using a mixed integer programming formulation for the
mined material for processing or whether that is uneconomi simultaneous optimisation of the extraction sequence and the
cal and the material should merely be sent to waste or stock cut-off grade.
piled for later processing. Thus, this enables a schedule of block removal to be deter
In order to determine whether a block is forwarded for mined using a variable cut-off grade which maximises the net
processing, stockpiled or sent to waste, a cut-off grade value 25 present value of the mine. Thus, return from the mining
is established. For example, the cut-off grade at a particular operation is optimised at any particular time.
time may be 0.6%. Thus, any blocks which have a grade FIG. 5 is a block diagram showing a still further embodi
above 0.6% are sent for processing. However, if the grade is ment of the invention, and the mo st preferred embodiment of
below 0.6%, a decision is made as to whether the block is the invention. In FIG. 5 the data previously described is used
30 to create multiple block models because the drillhole data is
simply sent to waste or whether it is stockpiled for later
typically too sparse to support a unique and deterministic
processing. FIG. 2 is a graph which illustrates cut-off grade
block model in accordance with the embodiment of FIG. 3.
with respect to time and shows three blocks which may need
Thus, the likelihood of the block model produced with refer
to be mined at particular times. For example, block 1211 which ence to FIG. 3 matching actual reality is uncertain and ques
is mined at time T0 may have a grade which does not justify 35
tionable. Thus, a better approach is to use multiple block
processing at time T0 but which will justify processing at a models which are created by conditional simulation tech
later time T1. Thus, the block 1211 can be stockpiled for niques. The conditional simulation techniques allow the gen
processing at the later time T1. Block 12b which is mined at eration of a number of equally probably block models, all of
T2 has a grade which justi?es processing at the time T2, and which honour the drillhole data obtained from the drillholes
40
therefore may be sent immediately to processing. Block 120 11 and ?rst and second order statistics of the ore body repre
has a grade which is below the cut-off grade and at no future sented, respectively, by probably distribution and variogram
time has a grade which equals the cut-off grade and therefore, (e.g. Isaaks and Srivastava, 1989).
is sent to waste. The preferred embodiment of the invention Once again, the number of blocks in each of the models 501
targets all phases of mine optimisation, including the net 45 may be reduced by aggregation 502 in the manner referred to
present value optimal block extraction sequence, pushback in the aforementioned International applications. The block
design, and simultaneous cut-off grade and mining schedule aggregation formed according to the algorithm set forth in the
optimisation. aforesaid International application, preserves the required
One embodiment of the invention is shown in FIG. 3. With slope constraints and is very ?exible allowing the user to fully
reference to this ?gure, the data acquired from the bore holes 50 control the size and shape of the aggregation. The optimisa
11 is used to create a deterministic block model 201 to thereby tion according to the embodiments of the invention can be
provide the blocks 12 having assigned grade values. The applied to any aggregation of blocks with a set of precedence
deterministic block model 201 is created by spatial interpo arcs, prescribing which blocks should be extracted before a
lation of drillhole data, preferably using a Kriging procedure. particular given block.
The number of blocks in the block model 301 may be At step 303, a schedule 503 of block removal is created
reduced by aggregation of the blocks at step 302. The aggre dependent upon a cut-off grade value to maximise the net
gation of the blocks is disclosed in our International Patent present value of the mine. The cut-off grade value which is
Application Nos. PCT/AU2003/001298 and PCT/AU2003/ used may be a ?xed cut-off grade value 503a shown in FIG. 5
001299. The contents of these two International applications 60 or, in the most preferred embodiment of the invention, a
are incorporated into this speci?cation by this reference. variable cut-off grade value 5031) which is the same as the
However the invention is applicable to any sort of aggre variable cut-off grade value described with reference to FIG.
gation that is spatially connected and respects geometrical 3.
block extraction precedence constraints. The simplest and most straight forward use of the multiple
At step 303, a block removal schedule is created based on 65 block models 501 is to estimate the variability in the projected
a variable cut-off grade to maximise net present value of the net present value associated with the ore body uncertainty by
mine 10. valuing the optimal schedule obtained from the Kriged model
 US 8,082,167 B2
19 20
through each of the conditionally simulated realisations. This 1. COG is less than all conditional simulation grades.
is shown in FIG. 6A in which, for example, ten conditional COG:0.03% Cu, equivalent to a processing cost of 0.60
simulations are generated and those ten conditional simula $/tonne.
tions are immediately averaged at step 601. 2. COG is greater than only one of the conditional simulation
FIGS. 3 and 5 schematically show an apparatus in the form 5
grades. COG:0.05% Cu, equivalent to a processing cost of
of a processor which may be a personal computer or the like 1.00 $/tonne.
for performing the method of the preferred embodiment.
3. COG is greater than two of the conditional simulation
FIGS. 6A and 6B are an actual example of valuations of a
grades. COG:0.08% Cu, equivalent to a processing cost of
known pit. 1.60 $/tonne.
10
FIG. 6A shows the preferred embodiment of the invention
and is a schedule determination which gives the best estimate 4. COG is greater than all conditional simulation grades.
of the net present value of a mine. In FIG. 6A it is noted that COG:0.12% Cu, equivalent to a processing cost of 2.40
the optimization uses all ten simulations of the conditional $/tonne.
simulation represented in FIG. 6A. The block valuation will be performed for each situation.

Conditional
Conditional Simulation
Block Conditional Simulation Simulation Value
COG Processing Value $ Average Etype Premium

Situation (% Cu) Cost ($) 1 2 3 Value (35) Value (35) (35)

1 0.03 6000 2000 6000 16000 8000 8000 0
2 0.05 10000 0 2000 12000 4667 4000 667
3 0.08 16000 0 0 6000 2000 0 2000
4 0.12 24000 0 0 0 0 0 0

FIG. 6B shows optimization using only an average grade This example demonstrates that whenever the conditional
rather than all ten simulations, and shows that an increased net 30 simulated grades for a block are found to be distributed on
present value is still obtained if the average of the ten valua both sides of the COG, the block valuation for the average of
tions of the conditional simulations are used later rather than the conditional simulations will be greater than the valuation
earlier in the process. of the average grade (Kriged) model.
In FIG. 6B the term E-type represents the average of the
conditional simulations, equivalent to the Kriged model. The following observations may be helpful:
Here, we explain the difference between the $671 M and In situation 2, the E-type valuation effectively attributes nega
$739M valuations that use identical extraction and cut-off tive value to the ?rst conditional simulation, whereas when
grade (COG) schedules. In following this explanation, it is the conditional simulations are valued separately this
important to realise that each conditional simulation is an instance is attributed exactly zero valueiits value as
equally likely representation of the true block grade. waste.
40
The extra value arises because the conditional simulations In situation 3, the E-type valuation ignores the value of one of
can be valued individually using the prescribed COG, then the conditional simulations.
their value averaged, as compared to the case with the aver The valuation using the conditional simulations is more
age E-type model where the averaging is done earlier, at the realistic because when a block is mined, blast hole assays can
grade level. A very simple single block model with no dis 45 be used to determine a close estimate of the actual block
counting will be used to illustrate this. grade. With this near-perfect information about block grade,
Single Block Model Example we can make an accurate assessment of whether to process or
The following data is used-the average value corresponds waste a block. This is equivalent to valuing each conditional
to the E-type block model. This is a single block model with simulation individually as if each were an actual reality.
only three conditional simulations. 50 Thus, by using the multiple simulations to produce mul
tiple valuations and then averaging the multiple valuations in
order to produce the net present value, a higher expected net
Block mass (tonnes) 10,000 present value is obtained than if the multiple simulations are
Number ofblocks 1 ?rst averaged and simply that average is then used to produce
Conditional simulations 3 55 a valuation.
Grade in each conditional simulation (% Cu) [0.04, 0.06, 0.11]
Average block grade (% Cu) [Etype value] 0.07 As previously mentioned, the preferred embodiment of the
Tonnes of Cu in each conditional simulation [4, 6, 11] invention uses mixed integer programming formulation to
Average block tonnes of Cu [Etype value] 7 include the multiple conditional simulations and variable cut
Revenue per tonne of Cu (35) 2,000 off grades. This approach allows one to estimate the gap
Block value as waste (35) (Zero mining cost!) 0
60 between the obtained solution and the upper theoretical limit.
To illustrate this, an example of how optimisation is generated
In valuing this block, we will use a marginal COG (Pro will be given by reference to a simple case in which one
cessing CostIRevenue), and look at four situations, repre rock-type contains one metal type, which can be processed
sented by four different processing costs. The marginal COG through one processing plant, generalisation to the case of
is calculated thus: (Block Tonnes)*COG*(Revenue/tonne): 65 multiple rock types, metals and processing streams is cum
(Block Tonnes)*(Processing Cost/tonne) COG:(Processing bersome but straight forward. For simplicity, we consider
Costitonne)/(Revenue/tonne) below only the case of a discrete set of cut-off grades that is
 US 8,082,167 B2
21 22
possible to generalise the results to the continuous cut-off sch and Grossmann, 1965) and the procedure similar to that
grade case. The following notations are used: used in Whittle Four-X software. The ultimate pit contains
T is the number of scheduling periods; 191 million tonnes of rock and 62912.7 million tonnes of ore
N is the number of simulations; (above the marginal COG:0.6%). The undiscounted value in
P is the total number of panels; the ultimate pit (if processed with the marginal COG) is $
G is the number of all possible cutoff grades; (1,316199) million. It was divided into 6 mining phases and
R, is the total rock in the panel i in simulations n. scheduled over 12 years. The mining rate was set to 30
Qij is the total ore in the panel i, simulation n, when mined MT/year and the processing rate to 5 MT/year. The initial
with the COG j; capital investment was assumed to be $300 million, and the
discount rate 10%. The base case optimisation was done
Vij is the value of the panel i, simulation n, when mined and
processed with the COG j; using the marginal COG and produced the discounted cash
?ow $(704131) million, and the NPV was $(404131) million.
RtO is the maximum mining capacity in period t; The mining schedule and the NPV are shown, respectively, in
QtO is the maximum processing rate in period t; FIGS. 7 and 8. The second optimisation was done using the
Sl- is the set of panels that must be removed before starting the variable COG, but was based on the mean grade block model,
panel i; i.e. it was similar to the one which can be generated by using
d is the time discount factor; one deterministic model. The schedule was then evaluated
xijt is the fraction of the panel i is extracted with the COG j in against all 10 realisations of ore body model and produced the
period t; NPV:$(485:40) million, an increase of 20% over the base
yit is a binary variable equal to 1 if the extraction of the panel 20 case. The results are shown in FIGS. 9 and 10. The third
i has started in periods 1 to t, and equal to 0 otherwise; optimisation was done using the algorithm described above,
6]., is a binary variable controlling the selection of the COG and produced the NPV:$(505:43) million, a further increase
applied in period t; of 4.1% over the case of mean grade based optimisation. The
The MIP formulation is: results are shown in FIGS. 11 and 12. The relative variability
25 of NPV in all cases was roughly the same, about 8%. Another
important result of the variable COG policy is that the pay
11v c (1) back period (de?ned here as the time when the cumulative
EX 1:1 1
NPV becomes equal to zero) is decreased from 5 to 3 years.
The increase of 4.1% in NPV may be not seen as a very
30 substantial, but it should be mentioned that the block model
subject to the following constraints: considered does not have a high variability. The relative vari
ance in the undiscounted value of the ultimate pit is only
7.6%. There are many deposits which can have variability of
l N P c (2) the order of 20-30%. For these kind of deposits the potential
EX 2 z Rfxij, 5 R9, for all I 35 improvement in the expected NPV may be substantially high.
":1 [:1 jil
Thus, according to the preferred embodiment of the inven
1 N P 0
tion, an extraction sequence and cut-off grade policy can be
EX 2 z QZ-xt-j, 5 Q), for all I (3) determined which, when evaluated through the whole set of
":1 [:1 j:l
conditionally simulated ore bodies, will produce the best
yiki 5 y for all i and I (4)
40 possible expected net present value. The degree of accuracy
r G
of this optimised schedule can be estimated precisely, in
2 21ng i y for all i (5) contrast to prior art techniques.
1:1]:1
c T
In the claims which follow and in the preceding description
y s 2 Zxk? for all i, 1 and k c Si of the invention, except where the context requires otherwise
j:l 1:1 (6) 45 due to express language or necessary implication, the word
c comprise, or variations such as comprises or compris
261': =1 for all I (7) ing, is used in an inclusive sense, ie. to specify the presence
j:l of the stated features but not to preclude the presence or
xv, s 6], for all i, j, and I (8) addition of further features in various embodiments of the
invention.
It is to be understood that the prior art publications referred
The objective function (1) represents the discounted cash to herein, do not constitute an admission that that the publi
?ow. Constraints (2) and (3) enforce the mining and process cation forms a part of the common general knowledge in the
ing limits on average. Constraints (4)-(6) enforce the panel art, in Australia or in any other country.
extraction precedence constraints, and constraints (7) and (8) 55 Since modi?cations within the spirit and scope of the
ensure that the same COG is applied to all panels extracted in invention may readily be effected by persons skilled within
any given time period. This MIP formulation is solved by the the art, it is to be understood that this invention is not limited
commercially available software package CPLEX version to the particular embodiment described by way of example
9.0, by lLOG Inc. hereinabove.
To test the algorithm we have chosen 10 conditional simu 60
lations of a block model containing one type of metal and The invention claimed is:
using one processing plant. Because of con?dentiality 1. A method of scheduling the extraction of a resource in a
requirements all the economic parameters were rescaled and mine medium comprising the steps of:
do not represent reality. However, all the relative character obtaining drillhole grade data relating to the resource con
istics which demonstrate the potential of the new method are 65 tained within the medium;
not affected by the resealing. The ultimate pit for the design creating a plurality of different block models of the mine
was chosen by using the Lersch-Grossmann algorithm (Ler with a processor, each formed of a plurality of blocks,
 US 8,082,167 B2
23 24
each block in each model having a resource grade, each 4. The method of claim 1 wherein the cut-off grade is a
of the block models honouring the drillhole grade data; variable cut-off grade.
and 5. The method of claim 1 wherein the plurality of block
generating a single schedule for extraction of blocks from models are used to provide an average of resource grade and
the mine with the processor based on a cut-off grade block valuations and the average of the resource grade and
policy to increase expected net present value having
block valuations is used to produce the scheduled extraction
regard to all of the plurality of different block models,
wherein the single extraction schedule is determined by of blocks and the cut-off grade value to increase expected net
the following: present value.
6. The method of claim 1 wherein the scheduled extraction
of blocks is optimised by considering the average valuation
(1) produced from the conditional simulations.
7. The method of claim 1 wherein all of the conditional
simulations are optimized simultaneously to produce the
single extraction schedule with a variable cut-off grade
policy, and wherein all the conditional simulations are valued
using the single extraction schedule and the valuations are
averaged to produce the expected net present value.
8. The method of claim 1 wherein the single extraction
1 N P G (2)
schedule determines whether resource and medium corre
Rfxj, 5 R5), for all I 20
Nn:l [:1 jil sponding to each block is sent for processing, sent to waste or
1 N P 0
stockpiled for later processing.
EX 2 z ijlxt-j, 5 Q), for all I (3) 9. The method of claim 1 wherein the step of scheduling
":1 [:1 j:l using a variable cut-off grade is converted from a non-linear
you i y, for all i and I (4) representation to a linear representation, and a mixed integer
25 programming formulation is used for simultaneous optimisa
r G

Ext-? i y;,, for all i (5) tion of the extraction schedule and the cut-off grade.
1:1]:1 1 0. An apparatus for scheduling the extraction of a resource
c T within a medium in a mine from which drillhole grade data
y s 2 Zxk?, for all i, 1 and k c s; relating to the resource contained within the medium has been
j:l 1:1 (6) 30
obtained comprising:
c a processor for creating a plurality of different block mod
26j1=1, for all I (7) els of the mine, each formed of a plurality of blocks, each
j:l
block in each model having a resource grade, each of the
xi], s 61,, for all i, j, and 1 (8) block models honouring the drillhole grade data; and
a processor for generating a single schedule for extraction
of blocks based on a cut-off grade policy to increase
wherein; expected net present value having regard to all of the
T is the number of scheduling periods; plurality of different block models, wherein the single
N is the number of simulations; extraction schedule is determined by the following:
P is the total number of panels; 40
G is the number of all possible cutoff grades;
R1. is the total rock in the panel i in simulations n, (1)
Qij is the total ore in the panel i, simulation n, when mined
with the COG j;
Maximise
iii Mo Me
Vij is the value of the panel i, simulation n, when mined and 45
processed with the COG j;
RtO is the maximum mining capacity in period t; subject to the following constraints:
QtO is the maximum processing rate in period t;
Sl- is the set of panels that must be removed before starting the
1 N P G (2)
panel i; 50
EX 2 z Rfxij, 5 R9, for all I
d is the time discount factor; ":1 [:1 jil
xijt is the fraction of the panel i is extracted with the COG j in 1 N P 0
period t; NZ 2 z ngm 5 Q), for all I (3)
yit is a binary variable equal to 1 if the extraction of the panel ":1 [:1 j:l

i has started in periods 1 to t, and equal to 0 otherwise; 55 yiJl 5 ya, for all i and I (4)
6]., is a binary variable controlling the selection of the COG r G
applied in period t. 21ng i y, for all i (5)
2. The method of claim 1 wherein the creation of the 1:1]:1
plurality of different block models is a conditional simulation c T

technique which provides the generation of a plurality of 60 ya i 2 21cm, for all i, I and k c S;
j:l 1:1 (6)
equally probable block model realisations, all of which
honour the data and ?rst and second order statistics of the c

resource represented, respectively, by probability distribution 2 511:1, for all I (7)

j:l
function and variogram.
3. The method of claim 1 wherein the number of blocks in 65 xi], s 61,, for all i, j, and 1 (8)
each of the block models is reduced by aggregating blocks
prior to the step of scheduling extraction.