LerchsGrossmann algorithm with variable slope angles
R. Khalokakaie, P. A. Dowd and R. J. Fowell
Synopsis
One of the most important design factors in open-pit
mining is determination of the optimal pit. Pits may be
redesigned many times during the life of a mine in
response to changes in design parameters as more
information is obtained and to changes in the values of
technical and economic parameters. Over the past 35
years the determination of optimum open-pits has been
one of the most active areas of operational research in
the mining industry and many algorithms have been
published. The most common optimizing criterion in
these algorithms is maximization of the overall profit
within the designed pit limits subject to mining (access)
constraints.
Almost all algorithms use a block model of the ore-
body, i.e. a three-dimensional array of identically sized
blocks that covers the entire orebody and sufficient
surrounding waste to allow access to the deepest ore
blocks. Of these, the LerchsGrossmann algorithm,
based on graph theory, is the only method that is guar-
anteed always to yield the true optimum pit. However,
the original algorithm assumes fixed slope angles that
are governed by the block dimensions. None of the sub-
sequent attempts to incorporate variable slope angles
provides an adequate solution in cases where there are
variable slopes controlled by complex structures and
geology.
A general method of incorporating variable slope
angles in the LerchsGrossman algorithm is presented.
It is assumed that the orebody and the surrounding
waste are divided into regions or domain sectors within
which the rock characteristics are the same and that
each region is specified by four principal slope angles
north, south, east and west face slope angles. Slope
angles can vary throughout the deposit to follow the
rock characteristics and are independent of the block
dimensions.
The size, location and final shape of an open-pit are impor-
tant in the planning of the location of waste dumps,
stockpiles, processing plant, access roads and other surface
facilities and for development of a production programme.
The pit design also defines minable reserves and the associ-
ated amount of waste to be removed during the life of the
operation. The pit design, which is a function of numerous
variables, may be re-evaluated many times during the life
of the mine as design, technical and economic parameters
change or more information is obtained during operation.
The use of computer methods is necessary to redesign the pit
as rapidly as possible and to implement complex algorithms
on large block models.
Manuscript received by the Institution of Mining and Metallurgy on
10 May, 2000. Paper published in Trans. Instn Min. Metall. (Sect. A:
Min. technol.), 109, MayAugust 2000. The Institution of Mining
and Metallurgy 2000.
With the advent and widespread use of computers a num-
ber of algorithms have been developed to determine optimum
open-pits. The main objective of these algorithms, almost all
of which are based on block models, is to find groups of
blocks that should be removed to yield the maximum overall
profit under specified economic conditions and technological
constraints. The most common methods are graph theory
LerchsGrossmann method),1 network or maximal flow tech-
niques,2,3 various versions of the floating or moving cone,4
the Korobov algorithm,5 the corrected form of the Korobov
algorithm,6 dynamic programming7,8 and parameterization
techniques.9,10 Of these, the algorithm developed by Lerchs
and Grossmann 1 based on graph theory is the only algorithm
that can be proved, rigorously, always to generate the true
optimum pit limit. The original algorithm was, however,
limited to only one slope angle defined by the block dimen-
sions and was incapable of taking into account variable slope
angles.
Many factors govern the size and shape of an open-pit.
The pit slope is one of the key factors that govern the amount
of waste to be removed to gain access to ore and it is not
restricted to a constant gradient. Small changes in slope angle
can change the amount of waste to be removed and have a
significant effect on the degree of selectivity in mining opera-
tions. It is often important to change slope angles either for
geotechnical reasons or to follow different structures and rock
types in the deposit, apart from the overriding need to keep
the total amount of waste as small as possible. Any truly opti-
mal pit design algorithm must, therefore, take into account
variable slope angles. Incorporation of variable slope con-
straints into the LerchsGrossmann algorithm makes it much
more flexible, practical and reliable.
Many attempts6,1113 have been made to overcome the
difficulties of incorporating variable slope angles within
the LerchsGrossmann algorithm, but none provides an
adequate solution for cases in which variable slopes are
controlled by complex structures and geology. Alford and
Whittle14 reported the incorporation of variable pit slopes
into the algorithm, but gave no details. Whittle also reported
a solution elsewhere,15 but again the details were not suffi-
cient to enable objective assessment. Lipkewich and
Borgman12 proposed a knights move pattern to approxi-
mate a conical expansion to the surface. Zhao and Kim 13
defined a method based on cone templates. Dowd and Onur6
used the idea of cone templates to derive a general technique
to deal with the problem, but the algorithm does not always
give the correct solution.
LerchsGrossmann algorithm
Despite the rigorously optimal nature of the Lerchs
Grossmann algorithm, it suffers from the disadvantages of
complexity of the method, long computing times and diffi-
culty in incorporating variable pit slopes. The method
converts the revenue block model of a deposit into a directed
graph, which is a simple diagram consisting of a set of nodes,
or vertices, and a set of connecting arcs (lines with direction)
used to indicate the relationship between the vertices. Each
A77

Fig. 1 Directed graph representing vertical section
block is represented by a vertex; each vertex is assigned a
mass equal to the net value of the corresponding block.
Vertices are connected by arcs in such way as to represent the
mining, or access, constraints. These arcs indicate which
blocks should be removed before a particular block can
be mined. Fig. 1 shows a directed graph for a simple, two-
dimensional example in which the pit slope angle is 45 and
the blocks are squares. In this example to mine block 10 it is
first necessary to remove blocks 2, 3 and 4.
In graph theory notation the vertices are denoted as xI and
the arc connecting vertices xi and xj is denoted (xi,xj), the
order defining the direction of the arc. If the set of all vertices
is denoted X and the set of all arcs is denoted A, a graph G =
(X, A) is defined as the set of all vertices X together with the
set of all arcs A. Vertex xj is said to be the successor of vertex
xi if there exists an arc with its initial extremity in xi and its
terminal extremity in xj. The set of all successors of a vertex xi
is denoted Gxi. For example, in Fig. 1 the set of all successors
of vertex number 18 is Gx18 = {x10, x11, x12, x2, x3, x4, x5, x6}.
A set of vertices constitutes a closure of the graph if the
successors of all vertices in the set also belong to the set, i.e. if
the set of blocks, represented by the vertices, satisfies access
constraints for all blocks in the set. Thus, the vertices 2, 3, 4
and 10 constitute a closure. A closure is defined as a subset of
vertices Y X such that if x Y, Gx Y. The value of a clo-
sure is the sum of the masses of the vertices within it. The
optimal open-pit is defined by the closure with the maximum
value. The algorithm thus involves finding the maximum
closure of the graph that represents the block model of the
orebody.
Mining and access constraints
For a deposit represented as a grade, or revenue, block model
pit slopes are specified in terms of blocks that must be
removed to provide access to each block within the block
model. In the LerchsGrossmann algorithm directed arcs
impose these restrictions. They indicate which blocks should
be removed before a particular block can be mined. Consider,
for example, Fig. 1, in which each block has three immediate
successors. The immediate successor blocks (vertices) of any
specified block (vertex) must be removed before that block
can be mined. The various procedures used to specify mining
and access constraints for block models can be classified into
(1) non-cone-based methods and (2) cone-based methods.
The first category, non-cone-based methods, involves the
use of a pattern or a set of blocks to define mining slopesa
1 : 5 block configuration, a 1 : 9 block configuration or a com-
bination of these, known as a 1 : 5 : 9 pattern.
A78
(a)
(b)
Fig. 2 Non-cone-based patterns: (a) 1 : 5, five overlying blocks
must be removed to mine one; (b) 1 : 9, nine overlying blocks must
be removed to mine one
In the original formulation of the LerchsGrossmann algo-
rithm the 1 : 5 block pattern is used to specify mining slopes.
In this pattern, to gain access to one block, five overlying
blocksone up and one over, as illustrated in Fig. 2(a)
must first be removed. This pattern requires the use of
five arcs pointing away from each vertex (block) to satisfy
the mining constraints. As indicated by Lipkewich and
Borgman,12 if this pattern is carried up over several levels, an
undesirable wall slope will be obtained. For example, in a
cubic block model the average slope angle would approxi-
mate 4555. The second configuration is a 1 : 9 block
pattern, in which nine overlying blocks must be removed to
mine one block (Fig. 2(b)). This approximation to slopes pro-
duces a cone with slopes ranging from 35 to 45 in a cubic
block model. A close approximation to a 45 slope in the
cubic block model is obtained by combining a 1 : 5 block pat-
tern for the first level above the base block with a 1 : 9 pattern
for the second level. The use of this 1 : 5 : 9 pattern in the
LerchsGrossmann algorithm is exemplified in a previously
published program.16
One of the main disadvantages of use of the first category is
the difficulty of establishing optimum pit outlines with vari-
able slope angles. The slope angles are assumed to be defined
by the dimensions of the blocks. For example, if a 1 : 5 : 9
pattern is used in the general rectangular revenue block
model of an orebody, with 10 m 10 m 5 m blocks slope
angles of 25 would be obtained. Thus, when this procedure
is used, different slope angles will require different sizes for
the blocks in the orebody block model, but these may not
correspond to the required bench heights. The grades of
blocks of different sizes estimated from a given configuration
of data would have different estimation errors and, thus, cre-
ate difficulties in assessment of the reliability and confidence
levels associated with the final pit values (the optimal pit is
commonly used to define minable reserves with stated levels
of confidence). In addition, different parts of the orebody
may require different slope angles. It is impossible in this
method to have different angles for different parts of the pit.
In cone-based methods a cone is used in a variety of ways
to define the mining slope.12,13 Dowd and Onur6 used the
idea of cone templates to derive a technique to establish the
optimum pit with variable slope angles. This method involves
construction of a cone, or extraction volume, from the block
on a given level to the surface by joining rings or envelopes of
blocks corresponding to the pit slope angles. If the mid-point
of any block lies within the extraction cone, it is assumed that
it must be removed before removal of the base block.
However, the algorithm that was developed does not give the
correct solution in all cases. In the work presented here the
method of Dowd and Onur6 has been modified to derive a
general technique for variable slope angles. This procedure is
incorporated in the LerchsGrossmann algorithm given later.
deposit: x dim, block dimension in the x direction (west
east); y dim, block dimension in the y direction (south
north); z dim, block dimension in the z direction (vertical);
num x, number of blocks in the x direction (westeast);
num y, number of blocks in the y direction (southnorth);
and num z, number of blocks in the z direction or number of
levels.
Variable slope angles
Pit slopes can be approximated by constructing a cone that
represents an extraction volume. This can be done by creat-
ing rings or envelopes from the mid-point of the base block
and extending them to the surface (Fig. 4) in such a way that
the side angles of the cone are equal to the four principal
slope angles.

LerchsGrossmann algorithm with variable

slope angles

To incorporate variable slope angles in the Lerchs

Grossmann algorithm it is assumed that the orebody and sur-
rounding waste have been divided into regions or domains on
the basis of the geotechnical information. It is further
assumed that within each region or domain the rock charac-
teristics are the same and can be characterized by a set of
slopes and that each region can be approximated by a poly-
gon. Depending on the number of regions, the problem is
treated in one of two ways: (1) variable slope angles in which
only one region or domain sector is specified to define the pit
slopes; and (2) multiple variable slope angles in which more
than one region or domain sector is specified to define the pit
slopes. For each region or domain sector pit slopes are
assumed to be defined by four principal slope angles in four
principal directions: a north face slope, east face slope, south
face slope and west face slope.
Fig. 4 Construction of cone from base block

Fig. 3 Block model of deposit and coordinate systems
Two types of coordinate system, as illustrated in Fig. 3, are
used. The first is the X,Y,Z Cartesian system in which the X-
axis runs westeast, the Y-axis runs southnorth and the
Z-axis is vertical. The origin of the system is located in the
southwest of the uppermost level, the shaded block shown in
Fig. 3. The second system is an i,j,k coordinate index system.
The i, j and k coordinates increase along the line of increasing
X, Y and Z coordinates, respectively. In addition, the follow-
ing parameters are used to define the block model for the
If the pit wall slopes in the four principal directions are not
the same, the upper area of the cone on each level (intersec-
tion of the cone with the level), will consist of four quadrants
of different ellipses. If the pit wall angles are the same, the
upper area of the cone will be a circle. Fig. 5 shows the
extraction cone and the blocks within it on the first level and
on the two cross-sections. On each level the values of the two
semi-major axes and two semi-minor axes depend on the four
principal slope angles and the vertical distance of the mid-
point of the base block from the overlying blocks. These
parameters can be found by use of trigonometric functions.
The number of blocks in the principal directions on any level
above the base block can be calculated by dividing these para-
meters by the corresponding block dimensions. Consider a
block Xi,j,k on level k: the parameters and the numbers of
blocks in the principal directions, as illustrated in Fig. 5, can
be calculated from the equations
dx1 =
(k - t ) z dim (1)
tan (west face angle)
dy1 =
(k - t ) z dim (2)
tan (south face angle)

A79
 where t is level above the base block and varies from 1 to k1;
m1 is number of blocks from the base block to the east; n1 is
number of blocks from the base block to the north; m2 is
number of blocks from the base block to the west; and n2 is
number of blocks from the base block to the south.
When the numbers of blocks within the upper area of the
cone on any levelsay, the tth level above the base block
are calculated in the four principal directions all the blocks
Xm,n,kt, where m = i m2, i + m1 and n = j n2, j + n1, must
be examined to determine whether they are within the extrac-
tion volume. This can be done by use of the ellipse equation:

a = xdim (im) (9)

b = ydim (jn) (10)

If m is equal to or greater than i and n is equal to or greater

than j
(a)
a2 b2 (11)
Value = +
( dx1) (dy1)
2 2

If m is equal to or greater than i and n is less than or equal to j

a2 b2 (12)
Value = +
(dx1) (dy2 )
2 2

If m is less than or equal to i and n is less than or equal to j

(b)
a2 b2 (13)
Value = +
(dx2 ) 2
(dy2 ) 2

If m is less than or equal to i and n is equal to or greater than j

a2 b2 (14)
Value = +
(dx2 )2 (dy1)2
where a and b are the horizontal distances from the mid-point
of the block under consideration to the base block measured
in the westeast and southnorth directions, respectively, as
illustrated in Fig. 6. If the value is less than or equal to 1, it
is assumed that the block is within the extraction cone and it
(c) must be removed before the base block. Otherwise it is
Fig. 5 Extraction cone of base blocks showing all blocks within assumed that the block is outside the extraction cone. Blocks
cone: (a) upper area of cone on first level; (b) northsouth section that lie within the extraction cone are submitted to the graph
AA; (c) eastwest section BB algorithm. The program was written in such a way that
extraction cones are established only for ore blocks. This pre-
vents unnecessary increases in computing time and prevents
dx2 =
(k - t ) z dim waste blocks from being considered several times.
tan (east face angle) (3) With this procedure pit slopes are no longer fixed and are
not limited to one-up and one-over patterns. They can vary in

dy2 =
(k - t ) z dim
tan (north face angle) (4)

dx1
m1 =
x dim (5)

dy1
n1 = (6)
y dim

dx2
m2 = (7)
x dim

dy2 (8)
n2 =
y dim Fig. 6 Value of parameters a and b

A80
the principal directions and are independent of block dimen- z dim
dx11 =
sions. Fig. 7 illustrates a directed graph that represents a
northsouth section in a cubic block model in which the east
[
tan west face angle of block (k , j ) ] (15)

face angle and the west face angle are assumed to be 60 and
45, respectively. In this graph vertices 4, 5, 6, 7, 14 and 15
z dim
are in the extraction cone of block 23. dx12 = (16)
[
tan east face angle of block (k , j ) ]

dx11 (17)
m11 =
x dim

dx12
m12 - (18)
x dim

Fig. 7 Directed graph representing northsouth cross-section in
cubic block model with east face angle of 60 and west face angle of
45
Multiple variable slope angles
In complex cases in which the pit slopes vary in different parts
of the orebody on account of slope stability requirements it is
necessary to divide the orebody into regions or domain sec-
tors within which the rock characteristics are the same and to
use different slope angles for each region. In these cases slope
angles are assigned to each block in the four principal direc-
tions within each region; this is discussed later.
In the case of multiple variable slopes an extraction volume
is constructed level by level by creating rings or envelopes
from the base block and extending them to the surface, with
regard for pit slopes that have already been assigned to
blocks. The extraction volume is constructed from the base
block to the next overlying block and is then constructed
from the point of intersection of the cone with this level to the
second level above the base block. This procedure is conti-
nued to the surface (Fig. 8).
where z dim and x dim are the block dimensions in the verti-
cal and horizontal directions, respectively. On the first level
above, the blocks Xk1,m where m = j m21, j + m11 are consid-
ered as part of the extraction cone.
There are two intersection points of the extraction cone
with the level above the base block (lines drawn from the
mid-point of the base block to the next overlying level). The
extraction cone is extended from these two points to the next
overlying block (second level above the base block) by using
the slope angles of the blocks that contain the points of inter-
section. To determine the slope of the block to be used (in
other words, to find the block in which the intersection lies)
the values of the parameters dx11 and dx21 are divided by the
block dimension and the result is rounded up. This means
that a value of 0.5 is added to the result of division and then
the integer part is taken, i.e.
dx11 (19)
ml11 = + 0 .5
x dim
dx12 (20)
ml21 = + 0 .5
x dim
The values of the parameters dx11 and dx21 and the number of
blocks in both directions m12 and m22 on the second level
above the base block are determined as

z dim
dx12 = dx11 +
[
tan west face angle of block (k - 1, j + ml11 ) ]
(21)
z dim
dx22 = dx12 +
[ ]
tan east face angle of block (k - 1, j - ml21 )

(22)
dx12 (23)
m12 =
x dim
Fig. 8 Extraction cone of block for two-dimensional example with
three different regions
dx22
m22 = (24)
Consider the construction of an extraction cone in the two- x dim
dimensional case (northsouth section) shown in Fig. 8. For
a block on level k (row) and column j lines are drawn from Again, blocks Xk2,m, where m = j m22, j + m12, are consid-
the mid-point of the base block to the left and right with ered as part of the extraction cone on the second level above
slopes equal to the east and west face angles of the base block, the base block. This procedure is continued to the surface.
respectively, and the lines are then extended to the level The procedure presented for multiple variable slopes in
above. The values of parameters dx11, dx21 and the number of two dimensions can be applied to the three-dimensional case.
blocks to the east, m11, and to the west, m21, on the first level As with the procedure used for variable slope angles, the pit
above the base block are determined by the equations shape is assumed to be defined by an irregular, elliptical out-
A81
line on each level. The outline on each level consists of four dx1t
quadrants of different ellipses defined by the pit slope angles m1t = (33)
x dim
in the four directions. The values of the two semi-major axes,
two semi-minor axes and the number of blocks in the prin-
cipal directions on any level above the base block should be dy1t
calculated in both sections in the same way as described n1t = (34)
previously for one section. When these parameters are deter- y dim
mined, again by use of the ellipse formula, any block whose
mid-point lies inside the ellipse is considered to be part of the dx2t (35)
m2t =
cone. The values of the four axes and the number of blocks in x dim
the four directions for the tth level above the base block
(block Xi,j,k) can be found from the equations dy2t
n2t = (36)
t -1 y dim
dx i
1
ml1t -1 = i =1 + 0.5 (25) When the numbers of blocks in the four principal direc-
x dim tions (m11, n11, m21 and n21) have been calculated the blocks are
examined according to the procedure described for variable
t -1
slope angles to determine whether they are within the extrac-

dy i
1
tion volume.

nl1t -1 = i =1 + 0 .5 (26) Assigning slope angles to blocks

y dim
If more than one region or domain sector is specified to
t -1 define mining slopes, it is necessary to assign slope angles to

dx i
2
each block. To assign slope angles to the blocks the first step
is to determine which blocks are inside the particular region.
ml2t -1 = i =1 + 0.5 (27) A block is deemed to be inside a region if its mid-point lies
x dim
within that region. Blocks deemed to be within a given region
have the slopes of that region assigned to them. Different
t -1 methods can be used to determine whether a point is inside,
dy i
2
outside or on the boundary of a polygon. The approach
adopted here is the angle sum method based on coding origi-
nl 2t -1 = i =1 + 0 .5 (28) nally written by Dowd.17 In this method lines are drawn from
y dim
the point in question to each of the vertices that define the
boundary of the polygon and the angles between each succes-
t -1 sive pair of lines are then summed. Angles measured in the
dx1t = dx
i =1
i
1
clockwise direction are positive and those in the anticlockwise
direction are negative. The point is inside the polygon if the
sum of the angles is 2p radians and outside if the sum is zero.
z dim
+ The basic requirement for this method is the signed angle
[
tan west face angle of block (i + ml1t -1 , j , k - t + 1) ] between pairs of lines from the point to successive pairs of
vertices that define the boundary of the polygon. These
(29) angles can be determined by use of either the dot product of
t -1
two vectors or the equation of a triangle. The signs of angles
dy1t = dy i
1
can also be determined from the cross-product of two
vectors:
i =1
z dim Triangle equation
+
[
tan south face angle of block (i , j + nl1t -1 , k - t + 1) ] a2 = b2 + c2 2bc cos q
(30)
Dot product of two vectors
t -1
dx2t = dx
i =1
i
2 v1.v2 = v1v2 cos q

z dim Cross-product of two vectors

+
[
tan east face angle of block (i - ml 2t -1 , j , k - t + 1) ] v1v2 = v1v2 sin q
(31)
A block whose mid-point lies within a slope region is
t -1

dy
assigned the slopes of that region. The method is imple-
dy2t = i
2 mented by first imposing a bounding box around the
i =1 regionthis is the smallest rectangle that contains the region
z dim or polygon. Then the mid-points of all blocks in the first level
+
[
tan north face angle of block (i , j - nl 2t -1 , k - t + 1) ] of the region that are inside the bounding box are examined
to see whether or not they are inside the region. If the mid-
(32) point of any block lies inside the region, slope angles are
A82
assigned to all blocks at this location from the minimum to method of estimation has been detailed elsewhere.19 The
the maximum depth of the region. deposit is divided into 15 m (eastwest) 10 m (northsouth)
5 m (vertical) blocks and the recoverable tonnage is based
Case study on a selective mining unit of 5 m (eastwest) 4 m
(northsouth) 5 m (vertical). The physical and economic
The method of variable slope angles has been incorporated parameters for this case are: specific gravity of ore and waste,
into the LerchsGrossmann algorithm and has been coded 2.71 t/m3; cost of mining of ore and wasteas given Table 1;
into an interactive Windows software package.18 Data from a processing cost, SEK 52/tonne of ore; price of gold,
real orebody were used to illustrate and test the application of SEK 90/g; and recovery, 91%.
the software in the determination of optimum open-pit limits.
The data are from the Bjrkdal low-grade gold deposit Table 1 Cost of mining of ore and waste
located approximately 35 km northwest of Skellefte in the
north of Sweden.19 Level, m Cost of mining, SEK/t
Gold mineralization in the Bjrkdal area occurs within From To Waste Ore
a network of steeply dipping quartz veins in the contact
between older granodiorite and limestone/acid volcanic rocks. 0 120 11.0 11.0
120 130 11.30 11.30
The gold is erratically distributed but is mainly concentrated
130 140 11.60 11.60
in and around high-grade quartz veins. It occurs as both fine
140 150 11.90 11.90
and coarse grains and is free-milling. 150 160 12.20 12.20
The block grade model of the deposit contains 101 82 160 170 12.50 12.50
36 blocks in the eastwest, northsouth and vertical direc- 170 180 12.80 12.80
tions, respectively. Each block is assigned the estimated 180 200 13.20 13.20
(kriged) recoverable tonnage of ore above a cutoff grade and
the estimated (kriged) average grade of this tonnage. The
The slope regions and associated principal slope angles are
shown in Fig. 9 and Table 2. The slopes used here are solely
for the sake of example and do not necessarily correspond to
actual slopes. The overall results of pit optimization are
shown in Table 3. Two cross-sections through the optimal pit
are shown in Fig. 10. The application of the software to the
case study has enabled a much more realistic pit design that is
able to accommodate real slope angles within a traditional
block model for a complex, low-grade gold orebody.

Table 2 Slope angles applied to example shown in Fig. 9

Region North face East face South face West face

1 30 40 42 38
2 41 37 50 46
3 35 35 35 35
Fig. 9 Deposit and surrounding waste subdivided into four geo-
4 39 39 46 46
technical regions

Fig. 10 Cross-sections at (a) easting 150 m and (b) northing 120 m

A83
 Table 3 Optimum pit

Level Number of blocks Tonnage, t Value, SEK 10000 Mean

no. Pit Ore Waste Ore Waste Ore Waste grade, g/t

1 0 0 0 0 0 0 0 0
2 7 0 7 0 14 227.5 0 15.7 0
3 20 0 20 1 261.2 39 388.8 0 44..2 0
4 47 6 41 8 321.0 87 206.5 28.2 72.1 1.534
5 89 19 70 18 943.7 161 948.8 71.4 133.7 1.516
6 98 24 74 23 070.9 176 114.1 120.4 149.2 1.642
7 116 23 93 26 635.0 209 135.0 172.8 180.7 1.787
8 115 22 93 30 349.5 203 388.0 223.7 183.4 1.831
9 132 31 101 38 982.5 229 307.5 313.7 208.1 1.890
10 188 70 118 88 491.2 293 618.8 986.6 243.1 2.241
11 304 109 195 149 660.5 468 219.5 1638.9 406.3 2.195
12 381 160 221 229 524.3 544 858.2 2661.4 462.7 2.258
13 652 186 466 257 278.2 1 067 911.8 2804.6 1009.1 2.179
14 833 192 641 274 901.7 1 418 170.9 2903.7 1398.5 2.131
15 912 199 713 294 251.1 1 559 389.0 3259.6 1543.8 2.193
16 884 213 671 290 347.6 1 506 382.4 2632.3 1443.5 1.996
17 899 183 716 230 409 9 1 596 807.6 2006.2 1523.7 1.956
18 853 181 672 216 524.7 1 517 197.8 2190.4 1428.1 2.140
19 793 181 612 224 820.8 1 386 951.8 2453.2 1295.3 2.227
20 730 182 548 225 121.9 1 258 603.1 2529.7 1171.2 2.257
21 734 212 522 270 892.7 1 220 962.2 3666.3 1133.8 2.536
22 723 228 495 286 762.7 1 182 734.8 3542.2 1045.6 2.405
23 663 209 454 285 203.3 1 062 344.2 3118.6 986.9 2.217
24 638 233 405 312 822.6 983 912.4 3298.2 862.0 2.173
25 650 281 369 389 399.2 931 725.8 4327.1 766.2 2.244
26 652 344 308 464 991.1 860 198.9 5262.1 619.0 2.268
27 625 324 301 441 324.8 828 987.8 5652.3 606.6 2.460
28 579 342 237 478 614.2 698 203.3 6984.8 456.9 2.667
29 479 296 183 415 863.8 557 703.7 5937.7 381.9 2.623
30 384 241 143 338 962.9 441 517.1 4642.5 290.7 2.554
31 300 223 77 295 231.1 314 518.9 3920.7 153.8 2.516
32 220 181 39 249 775.2 197 374.8 4036.9 66.2 2.856
33 150 137 13 195 474.3 109 400.7 3117.2 28.0 2.816
34 100 94 6 140 512.9 62 737.1 2686.1 12.1 3.192
35 57 56 1 83 122.6 32 729.9 1964.2 2.7 3.741
36 23 23 0 35 842.0 10 905.5 700.1 0 3.230
Total 15 030 5405 9625 7 313 691.1 23 234 784.1 89 853.7 20 324.9 2.380

Conclusion computers and operations research in the mineral industries (APCOM)

The LerchsGrossmann algorithm is well known for being
the only method that can be proved, rigorously, always to
yield the true optimum pit. However, when the algorithm was
first introduced it was based on a fixed slope angle governed
by the block dimensions. The methods presented here have
been incorporated into the algorithm to overcome this limi-
tation and to take account of variable slope angles. As
demonstrated by a case study, the algorithm is able to gener-
ate an optimal open-pit with variable slopes. The method can
be used for both cubic and rectangular block models. Slope
angles can vary in different parts of the orebody without
change to the block dimensions, which are independent of
the slope angles.
Methods for the determintion of varying slope angles for
incorporation into pit design algorithms are described in the
accompanying contribution.20
References
1. Lerchs H. and Grossmann I. F. Optimum design of open pit
mines. CIM Bull., 58, 1965, 4754.
2. Johnson T. B. and Barnes R. J. Application of the maximal flow
algorithm to ultimate pit design. In Levary R. R. ed. Engineering
design: better results through operations research methods (Amsterdam:
North Holland, 1988), 51831.
3. Yegulalp T. M. and Arias J. A. A fast algorithm to solve the ulti-
mate pit limit problem. In Proc. 23rd symposium on the application of
(Littleton, Colorado: AIME, 1992), 3917.
4. Lemieux M. Moving cone optimizing algorithm. In Weiss A. ed.
Computer methods for the 80s in the mineral industry (New York: AIME,
1979), 32945.
5. Korobov S. Method for determining optimal open pit limits
(Montreal: Ecole Polytechnique de lUniversit de Montral, 1974),
24 p. Technical report EP74-R-4
6. Dowd P. A. and Onur A. H. Open-pit optimizationpart 1:
optimal open-pit design. Trans. Instn Min. Metall. (Sect. A: Min.
industry), 102, 1993, A95104.
7. Wilke F. L. and Wright E. A. Determining the optimal ultimate
pit design for hard rock open pit mines using dynamic programming.
Erzmetall, 37, 1984, 13944.
8. Yamatomi J. et al. Selective extraction dynamic cone algorithm
for three-dimensional open pit designs. In Proc. 25th symposium on the
application of computers and operations research in the mineral industries
(APCOM) (Brisbane: Australasian Institute of Mining and
Metallurgy, 1995), 26774.
9. Matheron G. Paramtrage des contours optimaux
(Fontainebleau: Centre de Gostatistique et de Morphologie math-
matique, 1975), 54 p. Internal report N-403; Note gostatistique 128
10. Franois-Bongaron D. and Guibal D. Algorithms for parame-
terizing reserves under different geometrical constraints. In Proc. 17th
symposium on the application of computers and operations research in the
mineral industries (APCOM), (New York: AIME, 1982), 297309.
11. Chen T. 3D pit design with variable wall slope capabilities. In
Proc. 14th symposium on the application of computers and operations
research in the mineral industries (APCOM) (New York: AIME, 1976),
61525.
12. Lipkewich M. P. and Borgman L. Two- and three-dimensional
pit design optimization techniques. In Weiss A. ed. A decade of digital
computing in the mineral industry (New York: AIME, 1969), 50523.

A84
13. Zhao Y. and Kim Y. C. A new optimum pit limit design algo-
rithm. Reference 3, 42334.
14. Alford C. G. and Whittle J. Application of LerchsGrossmann
pit optimization to the design of open pit mines. In Large open pit
mining conference AusIMMIE Aust Newman Combined Group,
1986, 2017.
15. Whittle J. The facts and fallacies of open pit optimization (North
Balwyn, Victoria, Australia: Whittle Programming Pty., Ltd., 1989).
16. Dowd P. A. The optimal design of quarries. Mineral resource
evaluation II: methods and case histories. Spec. Publ. geol. Soc. Lond.,
no. 79, 1994, 14155.
17. Dowd P. A. Un ensemble de sous-programmes pour raliser le
krigeage dun bloc irrgulier (Montreal: Ecole Polytechnique de
lUniversit de Montral, 1973), 18 p. Rapport Technique EP-73-R
18. Khalokakaie R. Computer-aided optimal open pit design with
variable slope angles. Ph.D. thesis, University of Leeds, 1999.
19. Dowd P. A. Bjrkdal gold-mining project, northern Sweden.
Trans. Instn Min. Metall. (Sect. A: Min. industry), 104, 1995, A14963.
20. Khalokakaie R., Dowd P. A. and Fowell R. J. Incorporation of
slope design into optimal pit design algorithms. Trans. Instn. Min.
Metall. (Sect. A: Min. technol.), 109, 2000, A706.
Authors
R. Khalokakaie graduated from the University of Tehran with a
master of science degree in mining engineering. He completed a
Ph.D. in optimal open-pit design at the University of Leeds in 1999
and has recently taken up a post as lecturer in mining engineering at
the University of Shahroud, Iran.
P. A. Dowd Fellow is Professor of Mining Engineering and head of
the School of Process, Environmental and Materials Engineering at
the University of Leeds. He was President of the Institution of
Mining and Metallurgy for 199899.
Address: Department of Mining and Minerals Engineering,
University of Leeds, Leeds LS2 9JT, England.
R. J. Fowell Fellow was formerly a reader in the University of
Newcastle upon Tyne, where he gained his Ph.D., and is now
Reader in Mining Engineering at the University of Leeds.

CALL FOR PAPERS & EXPRESSION OF INTEREST

Third Cardiff Mineral Resource Evaluation Conference:

DEALING WITH THE NUGGET EFFECT IN PRACTICE

NUGGET 2001
34 May 2001, Cardiff, Wales
Following the successful Cardiff MRE21 meeting, an international two-day symposium on the theme Dealing with the
nugget effect in practice: NUGGET 2001 is being planned in association with the Institution of Mining and
Metallurgy. The nugget effect poses a major challenge to precious-metal mining companies during exploration, evalua-
tion and exploitation. This meeting will cover all aspects of dealing with the nugget effect, from deposit geology through
grade control. The first days programme will comprise presentation of invited and offered papers and the second will
consist of workshops of case studies etc. and a panel-led discussion session.

Submissions
Synopses of approximately 500 words are invited both as conference papers and for workshop sessions. Synopses may
refer to case histories, research results and/or experiences from actual projects. The deadline for receipt of synopses is
1 December, 2000.

Expression of interest
Anyone interested in attending the meeting should contact the convenor to be placed on the mailing list. Please provide
e-mail address if possible.

Sponsorship
The organizers are also seeking corporate sponsors for the meeting.

Enquiries
Dr. Simon Dominy, Department of Earth Sciences, Cardiff University, P.O. Box 914, Cardiff CF10 3YE, Wales, UK.
Tel: +44 (0)29 2087 4924; Fax: +44 (0)29 2087 4326; e-mail: dominys@cardiff.ac.uk

A85