Caving 2018 – Y Potvin and J Jakubec (eds)

© 2018 Australian Centre for Geomechanics, Perth, ISBN 978-0-9924810-9-4

https://papers.acg.uwa.edu.au/p/1815_31_Arancibia/

Developing projects using CaveLogicTM

M Arancibia Maptek, Chile

F Soto Maptek, Chile

Abstract
This paper presents a new long-term planning software tool to develop projects exploited by mass caving
methods. There are no easy to use applications currently available on the market. The software packages
available commonly act as black boxes, are complex to use, and generally require highly specialised
knowledge and training. The fundamental idea behind CaveLogicTM was to develop a friendly and intuitive
tool, leaving aside the black boxes, to allow the user to concentrate on cave mine planning and not the
software learning process. The application provides tools for the entire cycle from the block model deposit
to the generation of long-term production plans.
CaveLogicTM integrates the possibility of generating footprints through a genetic algorithm, incorporating
restrictions and caving parameters, as well as evaluating multiple scenarios such as multiple undercut level
planning. CaveLogicTM enables the simulation of previously caved levels, allowing the planner to
interactively modify the optimal solutions presented by the software. This paper involves a case study for a
mining project planned for a fictitious mine transitioning from open pit to mining by caving, enhancing the
management of large analysed nodes and block volumes.
CaveLogicTM uses an algorithm based on image processing and geomorphology applications, applying
genetic algorithm guidelines. The algorithm accomplishes the commissioned task successfully and has a
processing time per envelope in the order of seconds. The software includes tools to simulate previously
caved sectors, and enables the mine engineer to interactively modify the solutions considered optimal by
the software.
Keywords: caving, mine planning, software, genetic algorithm

1 Introduction
CaveLogicTM is a holistic software tool from MaptekTM designed to support the strategic planning of mining
projects operated under caving methods. It was designed to be an easy to use model, with a quick learning
curve, user customisation and enhanced objective traceability so the mine planning engineer can use their
time and skills to find the best possible project, rather than setting up a model. This tool integrates the
possibility of generating footprints through the genetic algorithm, incorporating the constraints and
parameters of this type of mining and assessing multiple scenarios and plans with multiple caving levels.
The planning process is explained in different stages, opening the possibility of managing multiple
scenarios. Figure 1 shows a typical process for a cave mining project, where a block model and topography
are required as an entry point.

Figure 1 CaveLogic workflow

Caving 2018, Vancouver, Canada 407

Developing projects using CaveLogicTM M Arancibia and F Soto

Each component represents independent units, thus the composition of a group generates a specific
project, enabling one to repeat some components in a scenario and create new ones as desired.
Within the planning process of a cave mining project, the value is directly associated with the definition of
the economic envelope. The envelope for caving methods is complex to define as the model must
incorporate the mixing process of the material with a high degree of uncertainty (Laubscher 2003),
including rock mechanics constraints such as connection height, critical hydraulic radius, and maximum
draw height governing this type of exploitation (Brown 2003), thus becoming a problem difficult to resolve
through deterministic algorithms. There are several approximation algorithms leading to an acceptable
solution through an efficient exploration of the solution space. The genetic algorithm is a method for
solving optimisation problems based on the process of natural selection of biological evolution, exploring
the solution space efficiently (Talbi 2009). The economic envelope problem boils down to, with precise
block models, the creation of a series of algorithms that define blocks which maximise the return on the
project and that, in turn, meet the series of constraints mentioned (Arcos Troncoso 2012).

2 Theory remarks
This section presents some theoretical principles for developing a caving project in CaveLogicTM, considering
the basics for caving projects and the explanation of the envelope definition and scheduling algorithm.

2.1 Economic and dilution model

The economic model follows the guidelines for any kind of exploitation, but also includes the cost of
development and repair of the infrastructure.
The process of gravitational flow is the dynamics of a set of particles, which are subjected to the interaction
between gravity and forces arising from movement of these particles through the drawpoint. In caving
methods, it is critical to know the flow of the particles forming the material being removed. This is because
the determination of the exact moment in which the particle is in the extraction point helps in assessing the
exact composition of material being extracted at a given period of time (Pierce 2010). This cannot be
estimated through deterministic methods given the number of variables interfering with the gravitational
flow, as well as the geological uncertainty of each of these variables (Vargas Vierling 2014). Therefore,
different models were developed to approximate the trajectory of the particles.
The volumetric model is an empirical model designed by Heslop and Laubscher, which through a single
parameter known as the entry point of dilution (PDE), tries to predict the composition of the mixed
material for the whole life of the mine (Laubscher 1994).

2.2 Envelope defined by genetic algorithm

CaveLogicTM uses genetic algorithm guidelines as a framework for envelope definition, which is a method
for solving optimisation problems based on the process of natural selection of biological evolution,
exploring the solution space efficiently. The algorithm repeatedly modifies a population of individuals,
where each represents a solution to the problem. At each iteration, the algorithm randomly selects
individuals from the current population to produce children of the next generation. After several
generations, the population evolves towards a solution considered optimal.
The objective of this type of algorithm is to be flexible, allowing the integration of restrictions such as the
degree of caveability of the rock mass, column height minimum and maximum, and interactivity between
the columns. The workflow chart of the process can be seen in Figure 2.

408 Caving 2018, Vancouver, Canada

 Mine planning

Figure 2 Workflow used by the footprint definition algorithm used by CaveLogic TM

The optimal economical column height is calculated automatically. However, in order to keep a reasonable
operational gradient between columns, it is necessary to smooth them. The smoothing process is done
completely with the genetic algorithm assisted by geomorphology principles (Pike et al. 2008) and image
processing (Solem 2012), while the contouring of the envelope is made by marching squares, which is an
algorithm that generates contours for a two-dimensional scalar field. An envelope before and after being
subjected to the process of definition is shown in Figure 3. Processing time in defining an envelope is in the
order of seconds.

Figure 3 Schematics for (a) The footprint with the optimum draw height without considering any
constraints; and, (b) An envelope based on the first footprint after the process of definition
considering constraints

2.3 Schedule algorithm

The software uses the CaveLogic SchedulerTM for scheduling the global extraction of the ore deposit. The
algorithm uses the following information as input parameters:
 Behaviour of each column in terms of maximum extraction rate and checked-out percentage.
 Time in which a column will be put into production.
 Maximum gradient that can be formed between the neighbouring columns.
 Desired global extraction rate and duration of each period.

Caving 2018, Vancouver, Canada 409

Developing projects using CaveLogicTM M Arancibia and F Soto

The algorithm takes as its main function the maximum tonnage that can be removed in a specific period in
a column considering its current state.
𝑡𝑒
𝑇𝑚𝑎𝑥 = ∫𝑡𝑐 𝑓𝑑𝑟𝑎𝑤 (𝑡)𝑑𝑡 (1)
where:
𝑇𝑚𝑎𝑥 = maximum tonnage possible to extract in the current period.
𝑡𝑒 = extraction time plus current time in maximum extraction of the column.
𝑡𝑐 = current maximum extraction time of the column.
𝑓𝑑𝑟𝑎𝑤 = specific function of tonnage of a column in a period of time t.
The scheduling engine completes the following operations to develop a production plan, which satisfies the
parameters and constraints in a specific time period:
 Calculation of maximum possible extraction of active columns without considering draw control.
 Identification of columns meeting the extraction criteria, to be extracted at the fastest possible rate.
 The remaining columns will be subjected to analysis of total tonnes extracted to achieve draw
control constraints.
The algorithm operates under an environment that allows a mass balance. This means, for instance, if a
goal of production with only the columns that are in over-extraction is met, the engine will not extract from
the other columns. The convergence time of the proposed algorithm is in the order of seconds.

3 Case study
The case study considered the development of a mining project transitioning from open pit to caving in
order to mine remaining underground reserves, all based on fictitious data. The following sections show the
methodology, characteristics of the fictitious deposit, details of the economic and dilution models plus the
rock mechanics constraints, and finally the results of the developed project.

3.1 Methodology
The methodology for this case study involved a three-stage process, as follows:
Preparing the model that represents the mining project in question.
Defining the economic envelopes or footprints.
Establishing the sequencing and development of the production plan.
The workflow is shown in Figure 4.

(a) (b) (c)

Figure 4 Diagram showing (a) The necessary information for input; (b) The typical planning process for a
long-term caving project; and, (c) The useful information as the output of the process

410 Caving 2018, Vancouver, Canada

 Mine planning

The case study starts with the incorporation of the topography data and a block model which represents
the ore deposit. It is possible to handle several topographies like different open pit phases in order to
decide which phase will be the last ‘best one’ before starting a caving operation, which is a process to
evaluate the transitional open pit to caving phase. The study continues with the creation of an economic
model and a dilution model. The software also allows one to create multiple models of each class.
The dilution model considers a reference topography, the PDE, and the grades of the material located
above the topography. The template of the diluted model is shown in Figure 5. The economic model is
compounded by the base valorisation of any block model and the addition of the cost of developing a
square metre of footprint. The collection of models is shown in Figure 6, where it is possible to add or
update as the project advances.

Figure 5 Schematics for (a) Dilution template in the software; and, (b) Visualisation of grade shells, the
topography selected for the dilution model

Figure 6 Collection of (a) Topographies; (b) Block models; (c) Economic models; and, (d) Dilution models

After defining the base data needed, the construction of the economic envelope or footprint uses the
CaveLogic Automatic EnvelopeTM. This module allows one to determine a series of envelopes at different
levels based on a dilution model and an economic model using the genetic algorithm as the framework to
accomplish this task. The constraints for the module include the critical hydraulic radius, minimum value of
a column, maximum and minimum draw height, and the maximum gradient that can exist between
neighbour columns. The main genetic parameters include the number of generations, which means the
number of stages that will cross the algorithm before converging to a solution considered optimum, and
the size of the population, which is the number of possible solutions being evaluated in each generation.
The greater these numbers are, the larger the universe of possibilities the algorithm will explore, but at a
slower rate. The inputs and the value of the envelopes generated at different levels is shown in Figure 7.

Caving 2018, Vancouver, Canada 411

Developing projects using CaveLogicTM M Arancibia and F Soto

Figure 7 (a) Template for caving parameters; (b) Template for genetic parameters; and, (c) Chart of
economic value of determined envelope and undercut level

A footprint is selected at a desired level, based on the economic value or other criteria, unique for the
project in question. An additional two or more lifts could be evaluated to make the process generated an
excavated topography for considering the volume caved by the upper lifts. After that, the CaveLogic
Automatic EnvelopeTM process is repeated. Figure 8 shows some output of this process.

(a) (b)
Figure 8 Schematics for (a) A range of fictitious topographies generated to simulate previously
excavated levels; and, (b) Different envelopes generated by the defining algorithm

Once the footprints that will be used in the project are identified, they are sequenced using the sequencing
module and placed in subareas as desired. Figure 9 shows a template for computing the sequence. Each
subarea requires a starting point which determines the beginning of the exploitation process and the
direction of its progress.

412 Caving 2018, Vancouver, Canada

 Mine planning

Figure 9 Template for making the sequence in a specific envelope. The colour map goes from yellow
(first columns initiated) to blue (last columns initiated)

Finally, after sequencing the footprints, CaveLogic SchedulerTM is used to determine the production plan
based on the areas of incorporation per year in each footprint, the mill capacity, and draw boundaries.
Some output of the scheduler is shown in Figure 10.

(a) (b)
Figure 10 (a) Image of a production plan using the scheduling algorithm; and, (b) Representation of a
specific period of the scheduling process and the subsidence generated

3.2 Ore deposit description

In order to demonstrate the application of CaveLogicTM on a project transitioning from open pit to caving, a
case study of a porphyry copper orebody was simulated with fictitious data. Using the Lerchs–Grossman
algorithm (Lerchs & Grossman 1965), a final pit shell was created based on simple cost–geometrical data
defined as follows:
 Open pit mining cost: 2.3 USD/t.
 Flotation processing cost: 8.1 USD/t.
 Recovery: 85%.
 Price: 6,612 USD/t of copper.
 Selling cost: 440 USD/t of copper.
 Slope angle: 45°.

Caving 2018, Vancouver, Canada 413

Developing projects using CaveLogicTM M Arancibia and F Soto

The block model has approximately six million blocks measuring 20 × 20 × 15 m. The open pit reserves and
available remaining caving resources are shown in Tables 1 and 2, and Figure 11.

Table 1 Open pit mined reserves

Copper cutoff (%) Copper (%) Tonnes

0.30 0.49 861,417,114
0.40 0.57 537,521,239
0.50 0.67 305,921,595

Table 2 Remaining reserves available for caving

Copper cutoff (%) Copper (%) Tonnes

0.30 0.47 4,313,752,239
0.40 0.58 2,361,023,459
0.50 0.69 1,275,385,615

The resultant open pit surface with the geological unit is shown in Figure 11.

Figure 11 Schematic of the fictitious copper deposit showing final pit shell and remaining ore reserves

3.3 Data
For the case study, the pre-mixing volumetric model from Laubscher and Heslop was used to simulate the
dilution. This uses a unique parameter called point dilution entry (PDE) (Laubscher 1994). The PDE is 60%, and
the grade of the broken material was assumed to be 0% – a conservative approach. The economic parameters
and rock mechanics constraints are specified in Tables 3 and 4. Figure 12 is a schematic of the constraints.

Table 3 Economic parameters for the case study

Parameter Value
Surface development cost 2,500 USD/m2
Treatment cost 8.1 USD/t
Mining cost 8 USD/t
Investment cost 2 USD/t
Refining/selling cost 661 USD/t
Copper price 6,504 USD/t
Days per period 360

414 Caving 2018, Vancouver, Canada

 Mine planning

Table 4 Rock mechanics constraints for the study case

Parameter Value
Maximum connection height 400 m
Maximum draw height 250 m
Minimum draw height 100 m
Mining rock mass rating (MRMR) 48
Maximum draw columns gradient 60°
Critical hydraulic radius 25 m
Extraction angle 35°
Subsidence angle 65°

Figure 12 Schematics for envelope’s constraints showing maximum height connection, MRMR, undercut
level, subsidence angle, and maximum and minimum draw height

4 Results
Using the conditions and constraints of the case study, and applying the algorithm for automatic
determination of an economic envelope, two footprints were identified located at distinct levels – levels
1,590 and 1,205 m. The statistics for each level’s footprint are shown in Table 5.

Table 5 Statistics of the defined envelopes for the project

Parameter Footprint 1,590 Footprint 1,205

Tonnage 311.458 Mt 452.7 Mt
2
Area 0.518 km 0.716 km2
Mean height 231.080 m 243.043 m
Mean cut 0.622% 0.534%

Caving 2018, Vancouver, Canada 415

Developing projects using CaveLogicTM M Arancibia and F Soto

4.1 Sequence results

The selected envelopes were sequenced in such a way that the caving areas are in the high grade sector.
This economic criterion was taken under the assumption that no constraint was exceeded; for instance, a
major failure in the rock mass. Figure 13 shows the envelope schematic and the sequencing direction.

Figure 13 Image of the two project footprints (1,590 and 1,205) showing (a) The grade of copper, where
hot colours represent higher grades; and, (b) The sequence of the envelope, where colder
colours show the sector of initiation of the caving

4.2 Scheduling results

The exploitation schedule was determined after several iterations where the ideal extraction rate is
25.92 Mt/year and the development of the footprint area is 43,200 m2/year. The final schedule and the
areas involved in the project are shown in Figures 14 and 15. The quick convergence of the scheduling
algorithm allowed several possible configurations to be evaluated, providing a robust result.

Schedule
30,000,000.00 0.80
25,000,000.00 0.70
0.60
20,000,000.00
Grade [%]
Tonnage

0.50
15,000,000.00 0.40
10,000,000.00 0.30
0.20
5,000,000.00 0.10
0.00 0.00
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33
Period

Tonnage 1205 Tonnage 1590 Cut Cut 1205 Cut 1590

Figure 14 Case study schedule. The second lift begins at year 13, and the operations end at year 33

416 Caving 2018, Vancouver, Canada

 Mine planning

The schedule shows a harmonic extraction, with a decreasing grade as a result of the convenient sequence
selected and increasing dilution introduced with the caving zone, with five years of ramp-up and a mean
grade of copper over the years equivalent to 0.57%.

Mine Area
500000 0.4

Velocity [t/day/m²]
400000 0.3
Area [m²]

300000 0.2
200000 0.1
100000 0
0 -0.1
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
Period

Active area Incorporated area Depleted area Velocity

Figure 15 Results of the project activity areas in terms of incorporated, depleted and active areas

In the area report (Figure 15), instability in the second lift in year 18 can be seen. The active area began to
increase in size over the years.

5 Conclusion
CaveLogicTM is a comprehensive, user-friendly software package aimed at assisting mine planning engineers
in assessing cave mining projects for strategic analysis. The system has a fast learning curve and no
requirement for mine engineers to build complex software programs or scripts to run black boxes. The
planning process is compiled in different stages, opening the possibility of managing multiple scenarios.
Defining an envelope is difficult to achieve by deterministic algorithms due to the complex constraints
involved with cave mining. Nevertheless, the approximation methods allow solving the defined problem.
The speed of completion and convergence of a mine plan schedule is in the order of seconds. Hence,
multiple scenarios are possible to provide a view of the best options.
The case study presented the development of a caving project using fictitious data. The methodology
considers preparing the working environment, defining the economic envelope and realising the
scheduling. The economic envelope analysis resulted in two footprints at the 1,590 and 1,205 m levels with
a tonnage of 764.16 Mt, 1.23 km2 of footprint, an average height of 238.17 m, and a mean copper grade of
0.57%, all developed considering the parameters and constraints of the project. The sequence of these
envelopes was initiated in the higher grade sector due to there being no constraints. The scheduling
assumed an extraction rate of 25.92 Mt/year and the development of a footprint area of 43,200 m2/year.
Using CaveLogicTM for strategic mine planning of caving projects allows one to increase the productivity of
mine planners based on the fast performance and easy to use interface.

References
Arcos Troncoso, CA 2012, Metodología para la Selección del Piso de Hundimiento en Panel Caving, bachelor’s thesis, Universidad de
Santiago, Santiago.
Brown, ET 2003, Block Caving Geomechanics, Julius Kruttschnitt Mineral Research Centre, Indooroopilly, and The University of
Queensland, Brisbane.

Caving 2018, Vancouver, Canada 417

Developing projects using CaveLogicTM M Arancibia and F Soto

Laubscher, D 1994, ‘Cave mining – the state of the art’, Journal of the South African Institute of Mining and Metallurgy, vol. 94,
no. 10, pp. 279–293.
Laubscher, D 2003, Cave Mining Handbook, De Beers, Johannesburg.
Lerchs, H & Grossman, I 1965, ‘Optimum design of open pit mines’, CIM Bulletin, vol. 58, no. 633, pp. 47–54.
Pierce, ME 2010, A Model for Gravity Flow of Fragmented Rock in Block Caving Mines, PhD thesis, The University of Queensland,
Brisbane.
Pike, RJ, Evans, IS & Hengl, T 2008, ‘Geomorphometry: a brief guide’, in T Hengl & H Reuter (eds), Geomorphometry: Concepts,
Software, Applications, Elsevier, Amsterdam.
Solem, JE 2012, Programming Computer Vision with Python, O'Reilly Media, Sebastopol.
Talbi, EG 2009, Metaheuristics: From Design to Implementation, John Wiley & Sons, Hoboken.
Vargas Vierling, EA 2014, Cálculo Envolvente Económica para Minas de Caving Bajo Incertidumbre Geológica, MSc thesis,
Universidad de Chile, Santiago.

418 Caving 2018, Vancouver, Canada