A hybrid intelligent optimization method for multiple metal grades optimization

(Yu, Shiwei; Zhu, Kejun; He, Yong)

Sept2012;SOURCE: Neural Computing & Applications;Sep2012, Vol. 21 Issue 6, p1391)

Theory and method of genetic-neural optimizing cut-off grade and grade of crude
ore(Conference Paper)
(Yong He, Kejun Zhu, Si -wei Gao, Ting Liu, Yue Li)-----ELSEVIER
Expert Systems with Applications (Volume 36, Issue 4, May 2009, Pages 76177623)
Volume 36, Issue 4, May 2009, Pages 76177623/// Source: DBLP
School of Economics and Management, China University of Geosciences, Wuhan Hubei 430074, China
Expert Syst. Appl 01/2009; 36:7617-7623. DOI:10.1016/j.eswa.2008.09.018

Using a Combination of Genetic Algorithm and the Grid Search Method to Determine Optimum
Cutoff Grades of Multiple Metal Deposits
International Journal of Surface Mining, Reclamation and Environment
Volume 18, Issue 1, 2004
DOI:10.1076/ijsm.18.1.60.23543
M. Ataei & M. Osanloo
pages 60-78

The grade selection rule of the metal mines; an empirical study on copper mines
Takayoshi Shinkumaa, , Takashi Nishiyamab ---- ELSEVIER
Resources Policy Volume 26, Issue 1, March 2000, Pages 3138

Methods for Calculation of Optimal Cutoff Grades in Complex Ore Deposits
M. Ataei, M. Osanloo
Journal of Mining Science, September 2003, Volume 39, Issue 5, pp 499-507

THE CHOICE OF CUT-OFF GRADE IN MINING
Rd Cairns, T Shinkuma------ ELSEVIER
Resource Policy 29(2003), 75-81 / SCIENCE DIRECT

Cut-off Grade Economics - Introduction
Kelsey, R. D. ---SME, 16th Application of Computers and Operations Research in the Mineral Industry -
1979 ,PAGES: 7 /PUBLICATION DATE-1/1/79, ISBN: 0-89520-261-1

GENERAL BACKGROUND THEORY OF CUT-OFF GRADES, Taylor H.K.(1972)
Instituitions Of Mining And Mettallurgy Transactions,A160-179

CUT-OFF GRADE-SOME FURTHER REFLECTIONS
Taylor H.K.(1985)/ Instituitions Of Mining And Mettallurgy Transactions,A204-216

Determination of Optimum Cut-off Grade for Metalliferous Deposit using Mathematical Programming
Technique: Journal of Mines, Metals and Fuels, Vol.52,No.12, pp.405-409, Dec.2004.

Development of a Computer Model for Determination of Cut -off Grade for Metalliferous Deposits:
Journal of Mines, Metals and Fuels, Vol. 54, No. 6&7, pp. 147-152, June-July, 2006.


Feasibility Study of Strategic Mine Planning: A Review
on Underground Coal Mines Transition into Opencast
Coal Mines

Pritam Biswas
1



Abstract

This paper presents an overview of selecting optimum cut-off grade while doing mine
production planning, which is a very important subject of mine design process. One of the
most important issues in mine production planning is the cut-off grade which is simply a grade
used to distinguish between ore and waste. Waste materials may either be left in place or sent
to waste dump. Ore is sent to the mill for further processing. Lower cutoff grade causes higher
amounts of ore to be processed and subsequently lower amounts of waste materials to be
dumped resulted in fluctuations in the cash flow of a mining project. Dumping waste is
accompanied with the rehabilitation cost which will affect the overall cost of final production
and also the optimum cutoff grade. Rehabilitation cost is the cost per tone of rehabilitating
material of a particular type of rock after it has been dumped as waste. One of the most
popular algorithms for determination of the optimum cut-off grade is Lanes method. Lane
formulated the cut-off grade optimization, but he did not consider rehabilitation cost during
optimization process. This cost item should be evaluated first, and then considered during cut-
off grade optimization process. In this paper rehabilitation cost is inserted directly into cut-off
grade optimization process using Lanes theory. The cut-off grades obtained using suggested
method will be more realistic rather than ones by using the original form of the Lanes
formulations.

Keywords: Strategic, Optimisation, Opencast, Underground, Transition

Introduction

In India the demand of coal is increasing day by day but the number of opencast mines is being
the same and its getting reduced day by day and the future is going to be depend on the
underground coal mines and on the abandoned underground coal mines, but the phenomenal
growth and demand rate can be attained and full fill if and only if the coal production in the
country laying greater emphasis on the Opencast mining. Both from the considerations of
volume of production as well as cost of production, the Opencast Mining prove its supremacy.
So, for the same reason and continuity this paper exhibits a case study on transition of existing
underground mines into an opencast mine of BHP Billiton Mine in Australia and some
guidelines that to be follow.
Snowden completed a study of a large ore body with proposed opencast through an existing
underground operation. The study considered the best combined mine plan, include showing
which areas of the underground mine should be extracted from underground methods ahead
of opencast mining.
Then, a dynamic programming mathematical program is used to evaluate the optimum value
of a combined opencast and underground operation. The results generated in the case study
presented herein provide a clear focus and direction for the next level of detailed mine design
and planning.
For this study, earlier investigations had established that a change in the opencast pushback
sequence to accommodate a conventional underground mining option resulted in a significant
loss of overall project value. Therefore, each block mined from underground must carry an
opportunity cost corresponding to the value lost from the opencast.

Factors To Be Consider:

Grade Profile

The mining of an opencast through the high grade with an underground mine requires there to
be sufficient grade material to be available after underground mining in order to make
opencast mining viable.

Underground Mining Method

Mining an opencast through an underground mine requires that the voids be left stable after
underground mining. Caving methods are unlikely to allow for future opencast mining. Often,
the underground mining voids will need to be backfilled, increasing the initial underground
mining cost.

Timing

Given that the high grade would have been extracted by the opencast at some stage,
underground mining can be considered a double up of mining costs. Although this is true,
the benefit of underground mining is its potential to access higher grades much earlier than
the opencast. In this case, the discounted revenue benefits of early, high grade will partially
compensate for the additional costs. The earlier the high grade is mined, the more attractive
underground mining becomes.

Capital Constraints

Underground mining may not provide a better economic outcome overall than opencast
mining; especially as underground production rates are often significantly lower than opencast
production rates (for the same deposit). However, a lower production, high grade
underground mine may be employed as a starter case to reduce the capital funding required.
Cash flows generated from the underground mine may be used to fund an opencast
expansion. Without this staged approach the deposit may not be developed at all.

Methodology

The optimisation process used for this process is iterative. At this stage, there is no single tool
can be simultaneously optimise an opencast and underground mine plan. For this paper it is
considered that the overall value of the project is equal to the value of open cast as a stand-
alone consideration, plus the value of underground plan, minus the lost through depleting
underground resources from the opencast plan.
In general a three step process is followed (B. Roberts et al., 2013):
i. Opencast optimisation, wherein an optimal opencast only sequence and schedule is
determined.
ii. Underground with pit optimisation: use the schedule to determine the discounted
value of the each block being mined from the potential underground mining value from
the potential underground mining value and the underground is optimised on the net
objective.
iii. Combined opencast and underground optimisation: the opencast and underground
sequences are integrated into an overall project schedule to determine a combined
actual value.
iv. 1 no restrictions
1 all restricted
3 one restricted
3 two restricted

Opencast Optimisation

The standalone opencast is optimised using a combination of Blastor pit optimisation tool and
the COMET cut off grade and schedule optimiser (King, 2001; Wooller, 2007). Blastor (Stone
et al., 2007) originated as a tool for completing pit optimisation of multiple complex iron ore
deposits. The tool has been successfully applied to other types of projects, such as base metal
deposits. Blastor uses a mixed integer programming formulation (MIP) to produce an
optimised pushback sequence. These aggregations are too small to form practical pushbacks,
so another algorithm is applied to generate pushback designs (termed Shells) of a more
suitable size. The key strength of this tool is its ability to optimise the under blending
constraints and to incorporate discounting. For this project, the Blastor output (in the form of
an optimum mining sequence) is exported to a software tool called COMET for optimisation of
the mining schedule as clearly shown in the Fig. 1. (B. Roberts et al., 2013)






Fig. 1 Blasor Shells (plan view) forming optimum opencast sequence
Underground with pit optimisation

Block Ranking

When optimising a schedule for a single mining operation the value ranking of a block for
mining is determined by variables such as the net value per tonne, net mill return or net
smelter return. In this project, however, the ranking variable must include the opportunity
cost represented by opencast mining. If the discounted value of a block is greater when mined
by opencast then it should not be extracted from underground. Therefore, in order to rank
blocks correctly for an optimised underground-with-pit strategy, a measure of incremental
value (IV) was developed. When considering a single block, the discounted value that the
block adds to the overall operation, its IV, will be equal to the discounted value when mined
by the underground mine, minus the discounted value of the block if it were to be mined by
the opencast (B. Roberts et al.,2013 ),see equation


(1)

The methodology for calculating OPDV and UGDV is shown in Equations 2 and 3, below. All
price and cost parameters applied are assumed to be constant over the life of the mine (B.
Roberts et al., 2013).


(2)

(3)

The variables used in Equation 1 are defined below (B. Roberts et al., 2013):


Opencast mining cost is not included in the opencast discounted value. This is because the
opencast mining blocks have already been committed for extraction. The decision is whether
the block is worth more to mine and process from underground or from Opencast. An
unknown in Equation 2 is the timing of underground mining, which is required in order to
determine its UGDV and hence its IV. Obtaining this data would require the underground to
already have been scheduled, creating a circular problem. In order to circumvent this circular
problem, an undiscounted underground value, UGV0, is used to calculate the value measure,
IV0 (i.e. the incremental value assuming underground extraction in year 0). In the presence
of constant parameters, this measure represents the maximum possible UGDV and IV for each
block. Any block with a positive IV0 value is a potential candidate for extraction by stoping.
However, having a positive IV0 does not guarantee that a block will add value if taken by the
underground operation. The later the block is taken in the underground mine schedule, the
lower the UGDV will become due to the effect of discounting.
There will come a point in time (before a block is taken by the opencast) when the OPDV of a
block exceeds its UGDV, at which time the IV of that block will become
negative. The use of IV0 as a ranking measure allowed all potentially value-adding
underground mining blocks to be identified (B. Roberts et al., 2013).
The higher the IV0, the more likely the block will add value to the operation being extracted
from underground. The differences between OPDV (top), IV0 (middle) and UGV0 (bottom) for
a section of the ore body are shown in the Fig. 2. The opencast is revealed to have more
potential ore than the underground operation due to a lower cut-off grade. However, much of
this material has very low discounted value due to the long mine life. The high value blocks are
concentrated in the upper levels of the south end of the ore body. The UGV0 and IV0 of blocks
in the north end of the ore body are very similar due to the low OPDV in this area. The key
difference between UGV0 and IV0 is in the south end of the mine where the pit commences.
Hence, the IV0 parameter biases towards high grade blocks mined late in the opencast mine
life coming from the picture.



Fig. 2 Representation of resource (looking west) for a) OPDV>0, b) UGV0>0 and c) IV0>0, (B.
Roberts et al., 2013)

Stope optimisation

The process of creating mining outlines from the IV0 resource model is conducted using
snowdens stopesizor software (Myers et al., 2007). Stopesizor modifies a geological block
model to identify the optimum mining outline for a range of cut-off values (usually grade-
based). This stope resource generation is achieved by constructing selective mining blocks
(SMBs), where the SMB represents a minimum practical mining geometry. Each SMB
comprises a contiguous group of resource blocks that honour minimum dimension, bearing
and dip constraints for each axis. Stopesizor has the flexibility to optimise mining outlines for a
variety of
Underground mining methods by allowing the user to specify minimum practical mining
shapes and orientations. Mining capacity constraints and the effect of discounting mean that
not all of this material can be mined from the underground mine in an optimum combined
underground and opencast operation. However, the analysis does indicate that there is
potential to increase the overall value of the operation by mining a significant amount of
South Area ore from underground (B. Roberts et al., 2013).

Production activities

The basic underground production unit is the ventilation district, for which production and
development constraints are well understood. This project included the design of conceptual
development layouts for the South Area and the creation of suitable new ventilation districts
for an underground-with-pit scenario. In order to reduce the optimisation problem to a
manageable size for efficient computation, stope blocks are accumulated into a series of
groups or bins. It is assumed that within each group, the contained blocks are to be depleted
at the same rate. In order to produce the best possible approximation of a block by block
optimisation, each group contains blocks with similar properties. The blocks are grouped
according to the ventilation district, IV0 outline, and the year in which the blocks were planned
to be extracted by the opencast. The tonnage and grade for each group is determined by
summing the tonnages and weight-averaging the grades of the contained resource blocks.
Each ventilation district is a distinct geometric zone; hence, grouping blocks by this property
means that they may be able to be practically extracted at the same rate, due to the location
of development and mining equipment (B. Roberts et al., 2013).
The stope optimisation process generates a series of nested mining outlines at IV0 cut-offs
between $0/t and $100/t intervals. The number of tonnes added to the outline at each
interval was substantial. As such, it is assumed that material within each of the intervals can be
grouped together. Material contained within the highest IV0 interval and so on. Geotechnical
sequencing constraints are not considered in this high level study, but will need to be
incorporated in subsequent studies in order to verify their impact on the optimal solution. In
addition to being grouped by ventilation district, blocks are also grouped by their planned
opencast extraction year. The extraction year is the governing factor which is determines at
what time the IV value of a block becomes negative. This grouping allows those blocks that
had a negative IV due to the year of underground extraction to be omitted from the optimal
outline (B. Roberts et al., 2013).

Development Activities

In order to optimise the schedule, it is necessary to provide development parameters for each
ventilation district. Because the development designs cannot be finalised until after the
analysis had generated a final stoping inventory, it was decided to use conceptual designs to
generate indicative parameters in the form of (B. Roberts et al., 2013).

Y = A + (B + C)* X (4)

Where Y is total development metres, A is fixed capital development metres (development
required to access the highest IV0 cut-off stope in a ventilation district), B is variable capital
development (metres per tonne of stope ore), C is variable operating development (metres
per tonne of stope ore), and X is stoping ore tonnes. The fixed capital development is further
subdivided into two categories: common capital development and access capital development.
Common capital development is development required by more than one ventilation district.
Access capital development is development required to access the highest IV0-cut-off stope in
a ventilation district. Framing the development required for each ventilation district to be
estimated and adjusted during the schedule optimisation process. Common capital
development is identified and appropriate dependencies are imposed to account for the
shared nature of this development (B. Roberts et al., 2013).

Underground Optimisation Technique

Schedule optimisation is conducted using Snowdens Evaluator software package. Evaluator is
a high level strategic scheduler based on MIP formulation solved by a CPLEX engine (ILOG
CPLEX, 2007). The scheduler optimises the timing and cut-off grades for both stoping and
development activities. MIP involves determining the optimal value of a series of decision
variables which reflect the mathematical model generated. In this optimisation, there was a
decision variable for each activity in each period (except where explicitly stated otherwise by
the user).The output for each decision variable reflects the portion of each activity that is
completed in a period. After optimising the schedule for incremental value using Snowdens
Evaluator software, the depleted opencast and underground schedules were combined. This
was bought forward metal (and hence revenue) for the planned operation (B. Roberts et al.,
2013).

Combined open pit and underground optimisation

Objective of combined schedule optimisation

The work described above is aimed at optimising an underground operation around the fixed
open pit schedule. A further modelling stage using COMET was required to (B. Roberts et al.,
2013):

i. Determine the relative value of alternative underground opencast combinations. For
example, depletion of ore from the pit schedule may significantly reduced value.
ii. Allow the COMET optimiser to choose whether to fill pit feed shortfalls from the
opencast stockpile or from the underground mine.
iii. Indicate optimum life for the underground operation
iv. Determine the processing destinations for underground ore.


COMET modelling

A comet model is constructed to model the current proposed ore processing configuration,
stockpiles, operating costs, capital costs and operational constraints. The opencast mine is
represented as a set of pushback block models and each underground case as a dummy
opencast pushback. Blocks mined in each underground opt on are excluded for the resource
available to the opencast. For the final optimisation run, COMET is set up so that if all
underground material available in a particular period has to be mined, COMET can choose
whether to process the ore in that period or stockpile it for later processing. This approach
ensures that the underground schedules were feasible. In this stage of this project, COMET is
used to determine the optimum combined opencast and underground schedule from the twin
inputs of optimum underground schedule and optimum pushback sequence. COMET used a
more conventional block value ranking variable as the problem of completion between
underground and opencast has now been resolved; the variable chosen is net return per mill
hour.

Final economic results

COMET generated the mining and processing scheduled that yielded the maximum discounted
value (NPV) under the given constraints. The best value is yielded by case UGI in Fig. 3 shows
how this additional value is achieved.


Fig. 3 Total copper production schedule, (B. Roberts et al., 2013)
The combined discounted value of this was 9% higher than the previous base case, which
ceased underground production much earlier. Interestingly, no resources that were contained
within the first third of the opencast life were selected for mining by the underground. So
clearly, discounting differential plays a large part in the economic decision and the opencast
must have a long life in order to justify such a strategy. In another case analysed by Snowden,
which had a 10 year opencast life, there were no blocks that were preferred for underground
mining (Snowden, 2013).

CONCLUSION

In tough times, it is worth exploring all options to reduce unit costs and minimise risk.
Considering a conversion of mining methods to lower scale underground mining may provide
one such avenue.

1. Increases in NPV are achieved by mining high grade underground ore earlier in the
project and by making potential shortfalls in an opencast ore production (B. Roberts et
al., 2013).
2. The underground operation should employ a declining cut-off strategy in order to
maximise value (B. Roberts et al., 2013).
3. NO stoping should take place within the 30 year pit limits.
4. The underground optimiser gives planners a guideline for the production rate in each
mine area in any given period and what the target average grade should be (B. Roberts
et al., 2013).
5. Value could be improved by repeating the process with knowledge of how the quality of
material produced by the underground operation should be scheduled in order to better
complement the opencast schedule.
6. Analysis of the results indicated a suitable time for closing the underground mine in
favour of an opencast only operation.

Acknowledgements

I would like to thank ISM librarian for granting permission to access the books and journals
available in the library for successfully peruse my thesis paper.

References


ILOG CPLEX, 2007. v.10, ILOG Inc, http://www.ilog.com/products/cplex/,accessed on
6.12.2013.

Myers, P., Standing, C., Collier, P. and Noppe, M. 2007. Assessing underground mining
potential at Ernest Henry Mine using conditional simulation and stope optimisation, Orebody
modelling and strategic modelling, AusIMM Spectrum series v. 14, 2
nd
edn, 191-200.

Roberts B., Elkington T., Olden van K. And Maulen M., 2013, optimising combined open pit and
underground strategic plan, Mining Technology v.122, 2
nd
edn, 94-100.

http://www.snowdengroup.com/news/newsletters/september-2013/open-pit-underground-
or-both-part-2, accessed on 5.12.13

King, B. 2001. Optimal mine scheduling, in Mineral resources and ore reserve estimation The
AusIMM guide to good practices, (ed. A. Edwards), 451-458, Melbourne, The Australasian
Institute of Mining and Metallurgy.

Stone, P., Frroyland, G., Menabde, M., Law, B., Pasyar, R. And Monkhouse, P. 2007. Blasor
blended iron ore mine planning optimisation at Yandi, Orebody modelling and strategic mine
planning, AusIMM Spectrum series v. 14, 2
nd
edn, 133-136.