School of Mining Engineering 

 

OPTIMISING THE USE OF THREE-DIMENSIONAL DATA TO 

LOWER GOLD GRADE DILUTION BY CONTROLLING 

STOPE WIDTH IN THE MINING OF ULTRA-DEEP COMPLEX 

ORE BODIES. 

Anne-Marie Olivier 

A research dissertation submitted to the Faculty of Engineering and the Built 

Environment, University of the Witwatersrand, Johannesburg, in partial 

fulfilment of the requirements for the degree of Master of Science in 

Engineering. 

 

 

Johannesburg, 2023 



i 

DECLARATION 

I declare that this report is my own, unaided work. I have read the University 

Policy on Plagiarism and hereby confirm that no plagiarism exists in this 

report. I also confirm that there is no copying, nor is there any copyright 

infringement. I willingly submit to any investigation in this regard by the 

School of Mining Engineering and I undertake to abide by the decision of 

any such investigation. 

. 

__________________    09 January 2023 

Signature of Candidate    Date 

  



ii 

ABSTRACT 

Excessive hanging wall or footwall dilution in narrow, tabular gold mines 

affects the payability of the mine. Three-dimensional (3D) scanning of stope 

faces is an option for quickly creating face maps to incorporate into a mine’s 

digital map. As this would make data capturing less time-consuming, the 

geology team now has more time to guide production regarding stopping 

width specifications and possibly reduce associated dilution. Unfortunately, 

Mponeng Mine’s current 3D scanner is bulky, expensive and unavailable for 

direct use by the geology team. 

Mponeng Mine investigated more convenient alternatives for 3D scanning 

of stope faces. Since March 2020, certain Apple products have incorporated 

a LiDAR scanner. The iPad Pro scanner, with a scanning range, of five 

metres, was tested in the mine mock-up at the University of the 

Witwatersrand. A 3D-printed iPad holder was created, holding a compact 

yet powerful LED light. DigiMine tunnels and stopes were scanned using 

this hardware and a variety of software. The 3D Scanning App supplied with 

the iPad Pro was found adequate for the identified purposes. Underground 

testing was then carried out in the ultra-deep stopes of the Mponeng Gold 

Mine. These scans can be geo-referenced into the mine's existing digital 

models. In addition, they can capture the contacts of the ore body. However, 

the iPad Pro is not waterproof, and the large screen makes it susceptible to 

damage. So, the mine is now testing LiDAR-equipped iPhone 13 Pro as a 

more robust, compact option.  



iii 

ACKNOWLEDGMENTS 

The contribution of Senior Lecturer Dr C. Birch is appreciated. Thank you to 

Mponeng Mine’s team from Harmony Gold for assisting and supporting the 

testing of this technology and for making time and technology available to 

conduct the test work. 

R.E. Olivier for editing the document. 

I am grateful for all the support on my journey. 

  



iv 

TABLE OF CONTENTS 

DECLARATION ........................................................................................... i 

ABSTRACT ................................................................................................. ii 

ACKNOWLEDGMENTS ............................................................................ iii 

LIST OF FIGURES ...................................................................................... i 

LIST OF TABLES ...................................................................................... vii 

LIST OF SYMBOLS ................................................................................... ix 

LIST OF ABBREVIATIONS ........................................................................ x 

INTRODUCTION ....................................................................................... 1 

1.1 Problem Statement ................................................................... 4 

1.2 Justification for Research .......................................................... 5 

1.3 Previous Work ........................................................................... 6 

1.4 Research Methods .................................................................... 6 

1.5 Sources of Data ........................................................................ 8 

1.6 Structure of the Research Report ............................................. 8 

GEOLOGICAL THREE-DIMENSIONAL MODELLING ............................ 11 

1.7 Section Overview .................................................................... 11 

1.8 Manual Visualisation ............................................................... 16 

1.9 Computer 3D Modelling .......................................................... 18 

1.10 Section Summary .................................................................... 31 

THREE-DIMENSIONAL SCANNING ....................................................... 32 



v 

1.11 Section Overview .................................................................... 32 

1.12 Scanning methods .................................................................. 42 

1.13 Case Studies ........................................................................... 61 

1.14 Section Summary .................................................................... 72 

RESEARCH METHODOLOGY ................................................................ 73 

1.15 Section Overview .................................................................... 73 

1.16 Deliberation of Devices and Methods ...................................... 75 

1.17 Data Collection and Analysis ................................................ 116 

1.18 Section Summary .................................................................. 121 

RESULTS AND OBSERVATIONS ......................................................... 123 

1.19 Section Overview .................................................................. 123 

1.20 Apple iPad Pro 11-inch in the DigiMine at Wits ..................... 123 

1.21 Apple iPhone Pro Max in On-Surface Conditions ................. 137 

1.22 Underground Stope Scanning ............................................... 146 

1.23 Section Summary .................................................................. 168 

MINE THREE-DIMENSIONAL SCANNING WORKFLOWS .................. 169 

1.24 Section Overview .................................................................. 169 

1.25 Geological Scanning Workflow ............................................. 169 

1.26 Production Scanning Workflow ............................................. 171 

1.27 Section Summary .................................................................. 172 

CONCLUSIONS AND RECOMMENDATIONS ...................................... 174 



vi 

1.28 Introduction ........................................................................... 175 

1.29 Summary of Chapters ........................................................... 176 

1.30 Research Observations ......................................................... 177 

1.31 Research Contributions ......................................................... 179 

1.32 Recommendations for Future Research Work ...................... 180 

1.33 Conclusions .......................................................................... 181 

REFERENCE LIST ................................................................................ 183 

 

  



i 

LIST OF FIGURES 

Figure 1-1: Location of Mponeng Mine, South Africa ................................. 2 

Figure 2-1: Filled block model framework ................................................ 12 

Figure 2-2: Block model blocks rotated to Z-axis (Modified) .................... 13 

Figure 2-3: Unrotated blocks (Modified) ................................................... 14 

Figure 2-4: 3D geological model created with Leapfrog modelling software

 ................................................................................................................. 20 

Figure 2-5: Geology modelled in 2D using drawings and tables of raw data

 ................................................................................................................. 22 

Figure 2-6: A Geologist busy mapping a panel using the traditional method.

 ................................................................................................................. 23 

Figure 2-7: Mapping sheet that informs the production team ................... 24 

Figure 3-1: Manually operated arm and strip 3D scanner ........................ 33 

Figure 3-2: Typical laser triangulation system comprises a laser source . 35 

Figure 3-3: 3D laser scanning of land, buildings and objects ................... 36 

Figure 3-4: Structured optical light scanners ............................................ 37 

Figure 3-5: Cross-sectional view of a 12K SLM modulated light scanner 38 

Figure 3-6: Modulated light scanner - 1x12288 Linear Array Liquid Crystal 

on Silicon scanner .................................................................................... 39 

Figure 3-7: Computerized tomography (CT) scan .................................... 40 

Figure 3-8: SIGNA 7.0T, a magnetic resonance imaging (MRI) scanner . 41 



ii 

Figure 3-9: Summary of the different scanning methods ......................... 42 

Figure 3-10: Concept of image sectioning ............................................... 43 

Figure 3-11: Camera offset geometry ...................................................... 44 

Figure 3-12: Line JK is normal to AB ....................................................... 45 

Figure 3-13: Illustration of a very dense point cloud with errors and resulting 

mesh ........................................................................................................ 46 

Figure 3-14: Illustration of less dense point cloud with errors and resulting 

mesh ........................................................................................................ 46 

Figure 3-15: Leica BLK2GO Imaging laser scanner ................................. 47 

Figure 3-16: ZEB Revo RT - Real-Time handheld scanner ..................... 48 

Figure 3-17: Marker on a scanned image. ............................................... 49 

Figure 3-18: The principle of tacheometric Laser Scanning ..................... 50 

Figure 3-19: The different types of tacheometric scanners ...................... 51 

Figure 3-20: Leica RTC360 Laser Scanner ............................................. 52 

Figure 3-21: RIEGL VZ-6000 laser scanner ............................................. 53 

Figure 3-22: Camera field of view relationship ......................................... 54 

Figure 3-23: F-number to focus depth relationship .................................. 55 

Figure 3-25: Point cloud of a scanned building ........................................ 57 

Figure 3-26: Picked points are shown as A0, A1 and A2 in 3D and in a table 

CloudCompare software .......................................................................... 58 

Figure 3-27: Reference point coordinates ................................................ 59 



iii 

Figure 3-28: End scan at the scan start point .......................................... 60 

Figure 3-29: East Kalimantan Province .................................................... 61 

Figure 3-30: Compared point cloud difference between DTMs ................ 63 

Figure 3-31: Measurement of discontinuity spacing ................................. 66 

Figure 3-32: Delineation of significant discontinuity planes for export and 

capture in the Acquire geotechnical database ......................................... 67 

Figure 3-33: Excel face mapping report sheet from Acquire geotechnical 

database .................................................................................................. 68 

Figure 3-34: Schematic data flow for geotechnical face mapping data at 

Sishen Mine ............................................................................................. 69 

Figure 3-35: Terrestrial laser scanner using a vehicle-mounted set-up (left) 

and a high-resolution point cloud for use in face mapping (right) ............. 70 

Figure 4-1: The Occipital Structure Sensor (Mark II)) .............................. 77 

Figure 4-2: Apple iPad Pro 11-inch .......................................................... 78 

Figure 4-3: iPhone 12 Pro ........................................................................ 79 

Figure 4-4: Total Station N7 ..................................................................... 82 

Figure 4-5: TS07 3” R500 Total Station ................................................... 83 

Figure 4-6: Face section from Deswik Mapping application ..................... 85 

Figure 4-7: Brass spad with numbered copper disc ................................. 86 

Figure 4-8: Occipital Structure Sensor (Mark II) ....................................... 88 

Figure 4-9: OBJ file open in Microsoft 3D Builder .................................... 91 



iv 

Figure 4-10: File structure of a PLY file format......................................... 93 

Figure 4-11: An object in a PLY file format with colour attributes assigned to 

each vertex format ................................................................................... 94 

Figure 4-12: Scanned object and room by the ‘Scanner’ application ....... 95 

Figure 4-13: Scanned room by ‘Room capture’ application ..................... 98 

Figure 4-14: The iPhone accelerometer axes ........................................ 101 

Figure 4-15: Scanned object by the Scaniverse application .................. 105 

Figure 4-16: Polycam 3D scanned model with measurements .............. 108 

Figure 4-17: Focus3D X 330 laser scanner ........................................... 110 

Figure 4-18: Stone pillar that was created with LiDAR Mode and Photo 

Model ..................................................................................................... 111 

Figure 4-19: Selection of the most suitable scanners and applications for 

the research. .......................................................................................... 117 

Figure 5-1: Leapfrog Geo Model of the Wits Mine Tunnel showing UG1 and 

UG2 stratigraphy as well as the fault and dyke relationship ................... 124 

Figure 5-2: The Wits DigiMine control room and life-size mining stope . 124 

Figure 5-3: Initial scan of stope panel at DigiMine before light-fitting 

modification; iPad using SiteScape to export Polygon Model File at 

University of the Witwatersrand in May 2021. ........................................ 126 

Figure 5-4: Image of ‘Mobile LED Video Light with Smartphone Bracket'

 ............................................................................................................... 127 



v 

Figure 5-5: Ultimaker S5 printer ............................................................. 128 

Figure 5-6: The back view of the iPad Pro 11-inch holder was designed 

using Shapr3D software. ........................................................................ 129 

Figure 5-7: Front view photograph of iPad Pro 11-inch holder during printing.

 ............................................................................................................... 130 

Figure 5-8: Cross-section through the stope panel of DigiMine scanned after 

light-source insufficiency was mitigated; with iPad at the University of the 

Witwatersrand in May 2021.................................................................... 131 

Figure 5-9: ‘Polygon Model File’ of stope area in the DigiMine exported to 

CloudCompare to assess iPad scanning using the SiteScape application 

during May 2021. ................................................................................... 132 

Figure 5-10: ‘Polygon Model File’ format of a portion of crosscut at the 

DigiMine to assess iPad scanning using the SiteScape application during 

May 2021, illustrating geological features. ............................................. 133 

Figure 5-11: ‘Object File’ format of the crosscut at the DigiMine exported to 

Deswik CAD to assess iPad scanning using the ‘3D Scanner’ application 

during May 2021, illustrating geological features. .................................. 135 

Figure 5-12: ‘Object File’ format processed in Deswik CAD showing 

Multislice lines of the crosscut at the DigiMine (as black lines) in Deswik 

CAD to assess iPad Pro scans using the ‘3D Scanner’ application in May 

2021. ...................................................................................................... 136 

Figure 5-13: HILTI PD-I distometer ........................................................ 138 



vi 

Figure 5-14: Rooms scanned with the 3D Scanner App on iPhone Pro. 141 

Figure 5-15: Enlargement of 1m mark measured from the face of the 

DigiMine tunnel. ..................................................................................... 142 

Figure 5-16: Bottom view of the ‘3D Scanner App’ scan captured in the 

DigiMine tunnel. ..................................................................................... 143 

Figure 5-17: Visible ground formation in the 113-68-East 5 stope panel.

 ............................................................................................................... 147 

Figure 5-18: Sampling report for 113-68-East 5. .................................... 150 

Figure 5-19: 20cm measured CW of the VCR in the 113-68-East 5 stope 

panel. ..................................................................................................... 151 

Figure 5-20: 113-68 East 2 panel of the VCR. ....................................... 153 

Figure 5-21: Scan accurate to within 20 cm. .......................................... 154 

Figure 5-22: Survey point (target) that was placed below the survey peg.

 ............................................................................................................... 155 

Figure 5-23: Mapping report of the 113-68 East 2 panel of the VCR. .... 157 

Figure 5-24: The 120-80 E1A panel of the CLR. .................................... 158 

Figure 5-25: Tributary area requirements for rockfall conditions ............ 159 

Figure 5-26: Stope plan view indicating minimum distance behind the face

 ............................................................................................................... 160 

Figure 5-27: Mapping report of the CLR’s 120-80 E1A panel from 15 

September 2021. ................................................................................... 162 



vii 

Figure 5-28: Mapping report of the CLRs from 7 October 2021. ............ 164 

Figure 5-29: Sampling report of the panel of the CLRs 120-80 E1A from 8 

November 2021. .................................................................................... 166 

Figure 5-30: Mapping report of the CLR’s 120-80 E1A panel from 12 

November 2021. .................................................................................... 167 

Figure 6-1: Geological Scanning Workflow. ........................................... 170 

Figure 6-2: Production Scanning Workflow. ........................................... 172 

 

LIST OF TABLES 

Table 3-1: Volume calculation using Global Mapper and Surfer .............. 64 

Table 4-1: Object File Format .................................................................. 90 

Table 4-2: 3D software that supports PLY file format ............................... 92 

Table 4-3: Scanned models exported to 3D file formats. ....................... 104 

Table 4-4: Supporting file formats - Deswik CAD for viewing scans ...... 114 

Table 4-5: CloudCompare supported file formats .................................. 115 

Table 5-1: Distometer measurements .................................................... 139 

Table 5-2: Distometer versus scan - dimension comparison and variance in 

rooms measured. ................................................................................... 140 

Table 5-3: Accuracy test results from the 3D Scanner application (.OBJ). 

from DigiMine. ........................................................................................ 144 

Table 5-4: Measurements from 113-68-East 5 stope panel of the VCR. 149 



viii 

Table 5-5: Volume calculation of 113-68-East 5 stope panel of the VCR.

 ............................................................................................................... 152 

 

  



ix 

LIST OF SYMBOLS 

Centimetre cm 

Cubic metres (Volume) m3 

Degrees ° 

Dollar $ 

Gold Au 

Grams per tonne g/t 

Inches " 

Kilogram per cubic metre kg/m³ 

Megapascal  MPa 

Metres m 

Millimetres mm 

Parts per million ppm 

Percentage % 

Rand R 

Three-dimensional 3D 

  



x 

LIST OF ABBREVIATIONS 

Allocated Stoping Width ASW 

Carbon Leader Reef CLR 

Channel Width CW 

Computer-Aided Design CAD 

Digital Terrain Model DTM 

Explosion protected Ex 

Global Navigation Satellite System GNSS 

Harvest with Advanced Regeneration HARP 

Identification Application ia 

Inertial Measurement Unit IMU 

Infrared IR 

Light Detection and Ranging  LiDAR  

Magnetic Resonance Imaging MRI 

Mineral Resource Estimation MRE 

Mineral Resource Management MRM 

Red Green Blue RGB 



xi 

Simultaneous localisation and mapping SLAM 

Stoping Width SW 

Structured Query Language SQL 

Subscriber Identity Module SIM 

Tool Command Language TCL 

University of the Witwatersrand Wits 

Ventersdorp Contact Reef VCR 

  



1 

INTRODUCTION 

This research project was initiated to identify orebody dilution in an ultra-

deep South African gold mine. Mponeng Mine is southwest of 

Johannesburg (shown in Figure 0-1) and mines the Ventersdorp Contact 

Reef (VCR) and Carbon Leader Reef (CLR). The narrow, tabular stopes are 

currently over 3,500 metres (m) below the surface. The VCR is highly 

variable as it forms an unconformity with the underlying sediments of the 

Witwatersrand Supergroup. The orebody is typically 120 cm thick and dips 

approximately 18 degrees towards the south. 



2 

 

Figure 0-1: Location of Mponeng Mine, South Africa (Harmony Gold 

Mining Company Limited., 2020). 

Predicted virgin stresses, at these depths, are between 95 MPa and 135 

MPa with a rock density of 2700 kg/m³ (Malan & Basson, 1998). Predicted 

virgin stresses create challenges concerning the practical support of 

workplaces. The rolling of the reef occurs on the strike and dip of the ore 



3 

body (Roberts & Schweitzer, 1999). For this reason, mining heights need 

constant adjustments to avoid grade dilution. Other mining operations that 

mine in the same region and under similar circumstances, for example, 

Kloof Gold Mine, may benefit from the research results (Manzi, et al., 2014). 

Mine planning requires three-dimensional (3D) spatial measurements of the 

orebody. Currently, on Mponeng Mine, the mine survey, geology and 

valuation departments gather this information on a regular grid. However, 

this information was unavailable in a format easily incorporated into the 

mines’ 3D model. The research aims to investigate how to best supply 

dynamic 3D information to the underground mine planning department and 

highlight any excessive dilution. Light detection and ranging (LiDAR) has 

proved to be a technology that will assist with gathering and disseminating 

3D information concerning the orebody. 

Mineral resource management (MRM) is the information hub for extracting 

minerals in underground mining (Blaauw & Trevarthen, 1987). Limited 

information on the ore body may influence mining decisions. Therefore, 

exploring more innovative mining techniques could increase precision and 

decrease the lag in relaying information about dilution is essential. One way 

to do this is by utilising and optimising dynamic 3D details. 

  



4 

1.1 Problem Statement 

Orebody/reef dilution affecting the grade destroys the economic value of 

mining operations. In addition, unwanted waste mining causes dilution, 

affecting the ore grade and increasing the mining and processing costs 

(Rogers & Kanchibotla, 2013). For this reason, Mponeng Mine strives to 

reduce dilution by controlling the stoping width (SW). As a result, Mponeng 

Mine’s SW was, on average, 12% greater than the allocated stoping width 

(ASW) for the past ten years (AngloGold Ashanti, 2020). 

With the information available, it is possible to make informed decisions. 

There may be room for the MRM department at Mponeng Mine to optimise 

the 3D data they supply to the production team. Currently, geologists and 

samplers capture information at the stope face. Geologists and samplers 

then provide the production team with a report showing the results in the 

form of a plan and a section view of only one panel on a sheet of A4 paper. 

This report may sufficiently guide the production team in the short term. 

Unfortunately, this information is unavailable for monthly planning in a 

holistic 3D view. The MRM department cannot verify this non-integrated 

information internally. As found in other Harmony Gold Mines, the geological 

sections are not copied and included in the monthly planning sessions. As 

the VCR ore body's complexity poses challenges in effectively controlling 

SW, Mponeng Mine faces challenges in reducing associated grade dilution. 

Grade dilution is the result of mining unwanted waste material. This 

translates to added waste tonnes needing transportation to the plant for 



5 

processing. Additional waste tonnes result in increased production and 

processing costs, with a concomitant reduction in profitability. 

The research hypotheses show that integrated 3D data from samplers, 

geologists and mine surveyors can help the production team better manage 

SW control to reduce grade dilution. 

1.2 Justification for Research 

“Some gold loss and dilution will always occur during blasting and 

transporting broken ore” (Xingwana, 2016, p. 149). Gold loss and ore 

dilution are evident in sampling results from Mponeng Mine. However, 

sampling results show that it is possible to increase the average gold grade 

by 31% if the recommended SW is honoured (AngloGold Ashanti, 2020). 

Improving ore grades has added benefits such as reducing support costs. 

Currently, the mine has a LiDAR scanner available from the mine survey 

department on Mponeng Mine. However, this instrument is bulky and 

unsuitable for use in stopes, prevalent on the mine. This research report 

explored simple, low-cost solutions that will allow for 3D scanning by 

geologists, valuation officers and production personnel. The aim is to 

provide the underground mine planning department with more frequent 3D 

dynamic information. 



6 

1.3 Previous Work 

The MSc candidate had the privilege of taking part as a co-author in a 

publication that inspired, developed alongside, and received a contribution 

from work done in this MSc research project. The leading author, C. Birch, 

presented findings from the research paper “Narrow, Tabular Stope 3D 

Scanning in Deep-Level Gold Mines Using An iPad Pro LiDAR” at the 2022 

International Multidisciplinary Conference of Engineering Technology 

(IMCET) in Turkey. The work complements this research project's purpose, 

specifically to explore the potential of implementing 3D scanning to map 

faces and develop a workflow that could improve associated reconciliation. 

Correspondingly, the work was supported by Wits and on-site investigations 

performed at Mponeng Mine. 

1.4 Research Methods 

Research design: The research methodology aimed to investigate using a 

cost-effective 3D LiDAR scanner for underground use. The research design 

comprises both a quantitative and a qualitative method. The quantitative 

portion short-lists scanning technology. The qualitative portion focused on 

the quality of the scans and the value that the 3D scanner offers. This 

involved the following steps: 

• Stock-taking exercise to short-list suitable hardware and software; 

• The literature survey process supports the evaluation of the short-

listed hardware and software. 



7 

The following criteria were applied in short-listing and then identifying the 

choice of scanner for the experimental underground conditions of narrow 

tabular mining: 

• Ease of use; 

• Accuracy and quality of scanners; 

• Cost of 3D scanners and software. The compatibility of software 

considers the currently used software of Mponeng mine. Price 

played an essential role since the project was in the test phase; and 

• Test short-list of scanners and software. 

 

The aim was to test what influence light may have on the quality of scans. 

Hence, the intention was to try these scanners at DigiMine at the University 

of the Witwatersrand. The DigiMine at the University of Witwatersrand is a 

simulated environment of an underground gold mine where scanners were 

utilised for the following: 

• Comparing and analysing the quality of the scans captured; 

• The best quality scanner and the sensor were to be tested 

underground at Mponeng Mine; 

• Through practical observation of geological practices, original data 

was collected by 3D LiDAR scans of stope panels, using built-in 

features of the iPad Pro for comparisons; 

• Comparing and analysing the quality of the scans captured at 

Mponeng Mine; and 



8 

• Decide what scanning applications would be best suited to the 

research study in the future. 

1.5 Sources of Data 

The 3D LiDAR scanning process used the following data sources: 

• 3D scans of stope panels; 

• Mponeng Mine’s survey peg database; 

• Data/information used from Mponeng Mine; 

• Mine survey actuals measured; 

• Sampling results; and 

• Geological mapping. 

 

Data capturing applied the following procedures: 

• Capture time, date, and workplace names for each scan; and 

• Create relevant workflows for both geological scanning and 

production scanning. 

1.6 Structure of the Research Report 

Chapter 1 is the introduction with the research background and the problem 

statement. Chapters 2 and 3 are the literature review, focusing on 



9 

background knowledge and geological modelling and 3D scanning case 

studies, respectively. Chapter 4, the research methodology, consists of the 

following points: 

• Further literature survey content about current mine survey 

equipment on Mponeng Mine; 

• The appraisal of potential scanners; and 

• Justifies considerations for subsequent experimental procedures 

(thereby lending itself to initiating the qualitative portion of the 

research method). 

 

Chapter 5 is the quantitative short listing of scanning technology and the 

reporting of preliminary testing in the DigiMine on the West Campus of the 

University of the Witwatersrand. DigiMine is a mock mine with a control 

room. The simulated mine has a life-size tunnel that imitates geological 

features on the sides of the tunnel. This mock mine also has a life-size 

narrow reef stope and lamp room. (University of the Witwatersrand, 2020). 

Chapter 5 compares underground observations of the following on-site 

manual measurements: 

• iPad Pro scans with measurements processed in Deswik CAD (data 

obtained on the same date in the presence of the MSc candidate); 

and 

• Mponeng Mine’s sampling team performed mining survey 

measurements. 



10 

Chapter 6 presents the new projections of a Geological and a Production 

Scanning Workflow, incorporating the chosen scanning technology and 

allowing a baseline suggestion for future optimisation. Chapter 7 

summarises the research conclusions and recommendations for furthering 

this research of reducing dilution, which is currently associated with the 

inherent limitations in informational flow between geological and production 

departments on the Mponeng Mine.  



11 

GEOLOGICAL THREE-DIMENSIONAL MODELLING 

1.7 Section Overview 

Geologists create a geological model from the core extracted from 

geological boreholes. Geological boreholes reveal, among other things, the 

ore body's physical characteristics in terms of size, shape, geological 

structures and geo-mechanical properties of ore and waste rock (Pandey, 

2022). The ore body features determine which mining method can be 

applied to mine safely and economically. 

Geological mapping is a manual process that traces contacts between 

different rock formations, often on other planes. This manual process 

enables the geologist to project and interpret geological features. In turn, 

this led to a better understanding of the ore body. 

3D modelling computer systems, for example, Datamine, Leapfrog, MineRp 

and Deswik, were introduced in the mining sector to visualise and 

understand the complexity of ore bodies. With Datamine software 

packages, namely Strat3D and Fusion, it is possible to create 3D block 

models. South African mines, and Mponeng Mine, in particular, have 

adopted computerised 3D modelling. 

Geological 3D block models are rectangular 3D X Y Z grid systems. Figure 

0-1 illustrates a filled block model framework where “N” is the number of 

blocks in the framework. A block origin defines the cell positions within the 

block model framework (Poniewierski, 2019). 



12 

 

Figure 0-1: Filled block model framework (Poniewierski, 2019). 

Grade dilution can start with the 3D block model creation process. 3D block 

models are created as rotated or unrotated models. A rotated model rotates 

cells according to the coordinate system (Poniewierski, 2019). For example, 

Figure 0-2 illustrates a stratified orebody dipping or plunging. Waste from 

the host rock does not dilute rotated block model cells (Poniewierski, 2019). 



13 

 

Figure 0-2: Block model blocks rotated to Z-axis (Modified) (Poniewierski, 

2019). 

Figure 0-3 shows a standard orthogonal unrotated block model applied to a 

stratified ore body that is dipping or plunging. With this unrotated block 

model, grade dilution occurs as the grade value of the block model cells is 

diluted by the waste from the surrounding host rock (Poniewierski, 2019). 

  



14 

 

Figure 0-3: Unrotated blocks (Modified) (Poniewierski, 2019). 

Each block in the 3D model represents a particular value of width, grade, 

tonnes and geological entity of the ore body. An ore body model must satisfy 

the following four conditions: 

• The parameters of the chosen model should allow estimation; 

• The model must be able to provide an answer to a relevant 

question; 

• The model must be compatible with data; and 

• An experienced geologist should very or check the model’s 

predictions (Pandey, 2022). 

 

The 3D geological model guides the mineral resource estimation (MRE) and 

classification process. Mine planners also use it to plan excavations at an 

accurate mining height for minimal grade dilution. The following geological 

aspects are used to generate the 3D geological model: 



15 

• Exploration drilling; 

• Sampling; 

• Geological mapping; 

• Interpretation of geological features; and 

• Ore grade (Chanderman, et al., 2017). 

Gold deposit in Mali 

“The workflow adopted in this study is based on a gold deposit containing 

an oxide zone (oxides) and a deeper sulphide zone (sulphides) comprised 

of unweathered (fresh rock) material, located in southwestern Mali. 

Exploring the potential for additional oxide resources is thus a natural step 

to increase the life of the mine” (Chanderman, et al., 2017, p. 189). The gold 

deposit is located on the West African craton in the Malian portion of a 

Paleoproterozoic inlier known as the Kedougou-Kenieba window 

(Chanderman, et al., 2017). 

In Mali, using 3D geological modelling and geostatistical evaluation 

techniques enabled the resource estimation of a gold deposit to increase by 

7200 ounces. Stochastic approaches to ore body modelling and estimation 

should be considered to fully characterise the geological uncertainty 

(Chanderman, et al., 2017). This information is based on 3D geological 

modelling and geostatistical evaluation techniques as informed by newly 

drilled advanced grade-control holes. 



16 

Gemfields Resources PLC in Zambia 

Accurate ore body modelling depends on accurate drilling information. If drill 

hole practices do not support the uniqueness of the ore body, precise 

estimation of tonnes and grade will not be possible. Research indicates that 

it is challenging to sample economic concentration zones. Examples of such 

challenges can be seen with the Kagem emerald deposit of Gemfields 

Resources PLC in Zambia. The MRE process depends on available results 

of bulk samples and historical production statistics for reporting on the 

volume of economic talc-chlorite-tremolite magnetite (SRK Consulting (UK) 

Limited, 2012). This paper stresses the importance of accurate ore body 

modelling to estimate tonnes and grades accurately. Dynamic 3D LiDAR 

scanning may positively contribute to this cause. 

1.8 Manual Visualisation 

Over the years, geologists have used different methods to represent an ore 

body visually. Geologists must communicate geological information 

effectively between departments (Donnelly, 2008). 

Traditional geological mapping is a manual process that traces contacts 

between different rock formations, groups, types and planes. The geologist 

would take measurements with a tape and clinometer along a single centre 

line that runs across the width of a stope panel or against the sidewall of a 

development excavation. A clinometer measures the dip and strike of the 



17 

reef and fault plane (Lisle, et al., 2011). This information is then drawn and 

projected on a plan and section view. 

A representation of the ore body was often constructed from layering 

transparent Perspex sheets. Geological sections were drawn onto individual 

Perspex sheets that were joined together to create a physical model of the 

ore body. Over the years, as 3D laser cutting technology developed, it 

became possible to create plastic models from digital data (Lisle, et al., 

2011). 

As technology developed, geologists used cameras to photograph stope 

faces while mapping. These photographs were a reference in writing 

geological reports (Lisle, et al., 2011). 

Notably, time at a particular face is always limited and would seldomly 

exceed 30 minutes. Therefore, the geologist must capture as much 

information about the stope as possible with maximum efficiency and 

accuracy. In complex ore bodies, the grade may fluctuate over short 

distances and for this reason, it is also vital to map geology regularly. The 

mining process can be more efficient and cost-effective when a well-

designed grade control programme is applied. Developing a value-adding 

grade control programme may be well received by stakeholders (Dominy & 

Platten, 2012). 

  



18 

1.9 Computer 3D Modelling 

Data given directly or indirectly to a specific location is known as “spatial 

data”. Geospatial data is often about a geographical area (Information 

Commissioner's Office, 2015). Geometric data is global data mapped on a 

two-dimensional flat surface (Zola & Fontecchio, 2021). Geographic data is 

information mapped around a sphere. Most often, the globe is planet Earth. 

Geographic data highlights the latitudinal and longitudinal relationships 

relative to a specific object or location. A global navigation satellite system 

(GNSS) is a familiar example of where geographic data is fully implemented 

(Zola & Fontecchio, 2021). Geological subsurface models are based on rock 

units' spatial distribution and deformation. The geologist maps information 

about rock types, contacts and depositional or magmatic flow features. In 

summary, the geologist maps the following structures: 

• Folding - dip, strike, deformation, the orientation of grains; 

• Joints – attitude, size, open or closed; and 

• Faults - look for slickensides, fault gouge, breccia and visible 

displacements (Balasubramanian, 2017). 

 

The uncaptured geological information about a stope face is accepted as 

lost after the stope face has advanced. Regular mapping of stope faces 

provides a complete picture of the ore body's behaviour. The greater the 

distance between mapping data, the greater the interpretation required of 

the geologist. The data needed to estimate local grade, and undertake 



19 

optimal SW, is provided by the mapping and sampling of these exposures 

(Dominy & Platten, 2012). Geological mapping information assists the 

geologist when they recommend the maximum SW of a stope panel. 

Leonida (2016) reported from an interview with the chairman of the mining 

software firm Maptek that ‘spatial modelling’ is a concept that has been 

already accepted and incorporated since the late 1970s. In the 1990s, 

however, mining became revolutionised by what is now called ‘3D 

modelling’. Computing capabilities and enhanced software produced user 

interfaces that present the data and allow user interaction within the 3D 

space. (Leonida, 2016). Maptek now uses Silicon Graphics technology 

initially used for movies and entertainment in the mid-1990s (Leonida, 

2016). 

Christchurch, New Zealand-based ARANZ was founded in the early 2000s. 

ARANZ started its operations with 3D medical imaging technology that 

supported accurate information for medical diagnostics. In 2004, geological 

science was the first to apply this technology. The software enabled faster, 

dynamic processing of data. As a result, the software became better known 

as Leapfrog Geo and has impacted mineral resources positively by enabling 

better decision-making. Seequent acquired ARANZ Geo in 2018 (Seequent 

Limited, n.d.). Figure 0-4 is an example of a 3D model of an orebody created 

with Leapfrog Geo software (Orefind, 2017). 



20 

According to borehole data, Figure 0-4 depicts the different attributes 

associated with the 3D model. These attributes may include the following: 

• Variables; 

• Mineral grades; 

• Contaminant concentrations; 

• Geomechanically properties; 

• Characteristics; 

• Lithology; 

• Mineralogy; and 

• Coordinates (University of the Witwatersrand, n.d., p. 14). 

 

 

Figure 0-4: 3D geological model created with Leapfrog modelling software 

(Orefind, 2017). 

Datamine software has provided time-saving automation and ease-of-use 

digital mining technology to the mining industry since 1981 (Datamine, 



21 

2022b). MineMapper 3D is part of Datamine's Fusion Suite, which enables 

the geologist to capture mapping digitally while at the workplace. This 

mapping then informs mine planning and modelling systems of the 

characteristics of the orebody and ore grade (Datamine, 2022a). 

Surpac was initially developed in Perth, Western Australia, by its founder 

Bebb in 1978 (Mining Patch Associates, 2017). By the late 1990s, Surpac 

was used primarily for non-seam or layered mining projects. In the mid-

1990s, Surpac introduced the tool command language (TCL), which 

contains macros that allow users to automate repetitive activities. In later 

years, Surpac joined forces with Minex Company and Gemcom Software. 

Since 2012, these groups have been part of Dassault Systems (Mining 

Patch Associates, 2017). 

The 1960s marked the beginning of 3D models. At that stage, 3D modelling 

was reserved only for computer engineering and automation professionals. 

Instead, they worked with mathematical models and data analysis 

(Architectural CGI, 2016). One of the pioneers of 3D graphics is Sutherland, 

the creator of Sketchpad. This revolutionary program helped create the first 

3D objects and significantly made 3D what it is today. Along with his 

colleague, Evans, Sutherland opened the first department of computer 

technologies at the University of Utah (Architectural CGI, 2016). 

Before this computerised technology became available, geologists were 

entirely dependent on mapping and modelling in 2D, using drawings and 



22 

tables of raw data. Figure 0-5 illustrates Geology modelled in 2D using 

drawings and tables of raw data (Leonida, 2016). 

 

Figure 0-5: Geology modelled in 2D using drawings and tables of raw data 

(Leonida, 2016). 

Due to its variability, the currently available 3D modelling technology 

enables geologists to understand an ore body’s behaviour and the entire 

mine’s performance, due to its variability (Leonida, 2016). 



23 

Currently, the geologists working on Mponeng Mine do manual stope 

mapping using tapes and a clinorule. Figure 0-6 shows a geologist mapping 

a stope panel at Mponeng Mine. 

 

Figure 0-6: A Geologist busy mapping a panel using the traditional 

method. 

Figure 0-7 is an example of a typical plan and section view of a mapping 

sheet that informs the production team of the following: 

• SW measured; 

• CW measured; 

• If there is reef in the hanging wall or reef in the footwall; and 

• The ASW. 



24 

 

Figure 0-7: Mapping sheet that informs the production team (Mponeng 

Mine, 2021). 

The mapping is plotted on elevation in Deswik CAD software. Interpretations 

enable a 3D geology view of lines and polygons with very little data. 



25 

Geologists at Mponeng Mine assign the following attributes to lines and 

polygons: 

• Reef type; 

• Mining level; 

• Raise line number; 

• The grain size of rock formation; 

• Colour; 

• Rock type; 

• Stratigraphic position; and 

• Stratigraphic colour. 

 

Geologists at Mponeng Mine use Fusion software from Datamine to create 

3D geological models from the above-mentioned geological mapping, 

sampling and drill hole data. 

The geology department is the custodian of the 3D geological and grade 

models for a mine. 3D modelling is crucial in evaluating an ore deposit to 

guide production in the mine planning process. 3D modelling software has 

been used successfully to model thin tabular reef deposits. Although 

negative perceptions about the accuracy of computer-generated 3D 

geological models exist, these perceptions have been critiqued as baseless 

and said to generally stem from a misunderstanding of software applications 

(Johnstone, 2003). Multi-dimensional modelling software should 



26 

complement traditional mapping techniques, thereby improving the 

professional output of geologists in the mining industry (Johnstone, 2003). 

A study at Namibia's Navachab Gold Mine showed that 3D implicit modelling 

reduces user-based modelling bias by generating open or closed surfaces. 

It uses geochemical, lithological, and structural data without requiring 

manual digitisation or linking. Instead, mathematical interpolation is used to 

visualize patterns and trends in large drillhole datasets. This research also 

proposed that examining existing drillhole datasets using 3D implicit 

modelling is a powerful tool for spatial analysis of mineralisation patterns. 

Furthermore, when used in conjunction with fieldwork, this approach can 

potentially improve the structural understanding of various ore deposits 

(Vollgger, et al., 2015). 

Mponeng Mine, previously owned by AngloGold Ashanti, was acquired by 

Harmony in October 2020 (Harmony, 2022). After the holding period of three 

months elapsed, Harmony implemented new systems and software at 

Mponeng Mine. Harmony's software was upgraded across various mines to 

meet the requirements of the different software companies. Consequently, 

block model calculations were processed much faster. In addition, 3D block 

modelling enables the combination of separate data sets that can be 

analysed and updated. As a result, the block models are more realistic and 

can be updated based on information from ongoing drilling (Fallara, et al., 

2006). 



27 

Traditional wireframing requires a good understanding of the ore body and 

demands more significant manual labour and interpretation of the ore body. 

On the other hand, implicit geological modelling involves a single 

mathematical function capable of determining the structural composition of 

the ore body. Therefore, the mathematical function decreases the time 

needed for creating 3D block models (Birch, 2014). 

Geological modelling software used in South African mines 

The most common block models in the mining industry are Datamine, 

Vulcan, Surpac, Micromine and MineSight (Poniewierski, 2019). The British 

Geological Survey developed the G-EXEC relational database 

management system in the 1970s. “The G-EXEC system consisted of an 

integrated collection of applications built around a relational database 

engine. The first was operational in 1973 and served the needs of its 

geological users for over ten years” (Resources Computing International 

Ltd, n.d.). Datamine has used the G-EXEC system since its establishment 

in 1981 (Poniewierski, 2019). Datamine software is used for accurate 

resource modelling and reporting for large and small deposits on all 

commodities (Datamine, 2021c). 

Surpac clusters identical blocks until further grouping is impossible. The 

effect is smaller model sizes. Surpac uses an exact method for sub-

blocking, resulting in parent blocks in fractions of a half, quarter, eighth and 

so forth. It is also necessary to specify the sub-blocking size during block 



28 

model creation. The actual division of parent blocks only occurs when more 

detailed data is required. The result is that a Surpac block model always 

uses the minimum blocks possible. 

For this reason, a Surpac block model can be much smaller than a Datamine 

block model. Conversion between Surpac and Datamine block models is 

possible when the data is available in comma-separated values (CSV) 

format. A CSV “file is a text file with a specific format which allows data to 

be saved in a table structured format” (Google, 2022). It is possible to 

convert a Surpac block model into Datamine format using Deswik software 

(Poniewierski, 2019). 

MineSight block models use a whole block modelling system, meaning no 

sub-celling is present. This method allows extensive mines to be modelled 

within computing memory and storage limitations at the time. In addition, 

this method identifies the percentages of the block within geological domain 

contacts. Sub-blocking/sub-celling have been available in MineSight since 

2013, which generates an additional file associated with the 3D block model 

for sub-blocked items and areas (Poniewierski, 2019). 

Here are the differences between the different types of block models: 

• How sample grades are interpolated/ extrapolated into a block used 

to populate the blocks within a block model; 



29 

• It may be possible to estimate within a block; for example, it is 

possible to estimate the magnitude of a sample by weighing four 

sampled points placed around it; and 

• The physical constructed blocks represent the size and rotation of 

blocks (Poniewierski, 2019). 

 

The following are different types of geological block models: 

• Inverse distance models; 

• Ordinary Kriged models; 

• Linear versus non-linear methods; 

• Multiple Indicator Kriged models; 

• Localised indicator Kriging / Uniform conditioning; 

• Conditional Simulation models; 

• Gridded seam models; and 

• Harvest with Advanced Regeneration (HARP) models 

(Poniewierski, 2019). 

Geological modelling software used at Mponeng Mine 

The Mponeng Mine facilitates communication between the production team 

and the MRM department. It is then the daily responsibility of the mine 

overseer to communicate guidance from the geology department back to 

the miners. Geological mapping reports are available to the production team 



30 

on a central system, Syncromine. Geological instructions are uploaded on 

Syncromine, and mining teams must acknowledge when they have read the 

instructions. This way, the geologist can keep track of the information flow 

between the Geology and Mining departments. 

Mponeng Mine uses two software packages from Datamine to create two-

dimensional block models, namely Strat3D and Fusion. The Strat3D 

software automatically builds a structural model from drillhole and seam 

description data (Datamine, 2019). 

A geostatistical method interpolates data points taken at different locations. 

Sample points are measurements from correlating attributes, for example, 

reef type, elevation and rock type. The main interpolation techniques are 

deterministic and geostatistical. “Deterministic techniques use 

mathematical functions for interpolation. On the other hand, geostatistics 

relies on both statistical and mathematical methods, which can be used to 

create surfaces and assess the uncertainty of the predictions” (Esri, 2021). 

The Fusion software stores all the geological information that informs 

Strat3D, including the following geological aspects: “completely 

configurable templates that capture all geological, geotechnical, 

geophysical, geochemical, downhole survey, mapping, quality assurance 

and quality control (QA/QC) and sample data” (Datamine, 2021b). 



31 

1.10 Section Summary 

From the literature review above, it is clear that accurate 3D modelling plays 

a critical role in determining resource estimations and mining costs. The 

current geological mapping approach (tapes and clinometers) makes it 

difficult and time-consuming to put geological mapping into a format suitable 

for the 3D geological model or even to display on the digital plans used 

during the planning sessions. Accurate 3D geological models increase the 

precision of mining and, therefore, higher gold grade if mining takes place 

according to plan. The accuracy and prediction of mining are further 

improved when these models are provided more frequently, specifically 

before blasting commences. 

  



32 

THREE-DIMENSIONAL SCANNING 

1.11 Section Overview 

The first 3D scanners, developed during the 1960s, used lights, cameras 

and projectors to scan objects for research and design. The first 

computerised model was created using a contact probe in the 1980s. A 

probe is a 3D contact scanner at the end of an articulated mechanical arm. 

“The arm may be robotically or manually manipulated over the part's 

surface. As the probe contacts the object's surface, the scanner records the 

X, Y, Z position of the probe by taking positional measurements of the 

armature” (Ebrahim, 2011, p. 9). As scanners evolved after 1985, they 

became capable of capturing surfaces using white light, lasers, and 

shadows. The development of optical technology using light began in the 

mid-1990s. Compared to prior scanning technologies, this technology was 

much faster. In 1994, REPLICA, which uses the laser stripe scanning 

method, was launched. See Figure 0-1. 

  



33 

 

Figure 0-1: Manually operated arm and strip 3D scanner (Ebrahim, 2011). 

Introduced in 1996, ModelMaker was able to create 3D models by 

combining a manual arm with a 3D stripe scanner. ModelMaker was a fast, 

flexible system that could produce complex models with colour and texture 

(Ebrahim, 2011). 

It is possible to classify 3D non-contact active laser scanning into three main 

categories. These categories are the time of flight, triangulation and 

structured light (Ebrahim, 2011). Time-of-flight 3D laser scanning uses laser 

light to detect an object with laser rangefinder technology and measures 

accurately within millimetres. A laser emits a light pulse, and the rangefinder 

determines the distance to a surface by the amount of time taken (‘round-

trip time’) before the detector receives the reflected light (Ebrahim, 2011). 

The total round-trip distance, d, from the scanner to an object is given by 

Equation 1. 



34 

� = (� × �) ÷ 2 Equation 1 

Where c is the speed of light and t is the time of flight 

(Moberg, 2017). 

 

The Leica Geosystems time-of-flight 3D laser scanner, ScanStation 2, can 

scan at a maximum instantaneous scan speed of 50,000 points/second 

(Diversified Communications, 2022). 

3D laser scanners that use the triangulation method use laser light to probe 

the environment. 3D laser scanner technology shines a laser on an object 

that uses a camera to look for the location of the laser dot. The laser dot, 

camera and emitter form a triangle, hence the term ‘triangulation’ (Ebrahim, 

2011). The diagram in Figure 0-2 illustrates the concept of triangulation. 

  



35 

 

Figure 0-2: Typical laser triangulation system comprises a laser source 

(Academic library, 2022). 

High-resolution scans take more points per second in comparison to low-

resolution scans. A disadvantage of this is some motion distortion. However, 

some laser scanners have level compensators built to counteract scanner 

movement during scanning (Ebrahim, 2011). The following explains 

standard active non-contact 3D Structured-light scanners. This includes 

laser scanners, structured optical light scanners, modulated light scanners, 

computer tomography (CT) scanners and magnetic resonance imaging 

(MRI) scanners (Ebrahim, 2011). 



36 

1.11.1 Laser scanners 

3D laser scanning produces detailed and accurate measurements for 2D 

drawings and 3D models of land, buildings, and objects (E-Architect, 2020). 

 

Figure 0-3: 3D laser scanning of land, buildings and objects (E-Architect, 

2020). 

1.11.2 Structured optical light scanners 

Structured-light 3D scanners project a pattern of light onto a subject. A liquid 

crystal displays (LCD) projector or a multi-laser projects the line onto the 

subject. “An LCD projector is a type of projector based on liquid crystal 

displays which can display images, data or video” (Techopedia Inc., 2022). 

A camera placed slightly away from the pattern projector observes the 

shape of the line, using the triangulation technique to calculate the distance 

from each point on the line. Structured-light 3D scanners scan multiple 

points simultaneously, thus reducing distortions caused by motion. Figure 

0-4 is an example of a structured optical light scanner. 



37 

 

Figure 0-4: Structured optical light scanners (Taubin, et al., 2014). 

  



38 

1.11.3 Modulated light scanners 

Modulated light 3D scanners convert digitised data into visual information 

by shining a continually changing light on a subject in a sine wave pattern. 

“A camera detects the reflected light, and the amount the pattern is shifted 

by determines the distance the light travelled. Modulated light also allows 

the scanner to ignore light from sources other than a laser, so there is no 

interference” (Pethe, 2008). Figure 0-5 shows a cross-sectional view of a 

spatial light modulator (SLM) scanner. 

 

Figure 0-5: Cross-sectional view of a 12K SLM modulated light scanner 

(Pethe, 2008). 

Figure 0-6 shows a Silicon scanner's 1x12288 Linear Array Liquid Crystal. 

It is possible to electronically control the speed at which the light of the SLM 

moves to operate at very high speeds, on the order of a few hundred frames 

per second (Pethe, 2008). 



39 

 

Figure 0-6: Modulated light scanner - 1x12288 Linear Array Liquid Crystal 

on Silicon scanner (Pethe, 2008). 

1.11.4 Computer tomography scanners 

“A computerised tomography (CT) scan combines a series of X-ray images 

taken from different angles around your body and uses computer 

processing to create cross-sectional images (slices) of the bones, blood 

vessels and soft tissues inside your body” (Mayo Foundation for Medical 

Education and Research (MFMER), 2022). X-rays are high-energy 

electromagnetic radiation that can pass through any object, including the 

body, to produce images of the shadows of objects (National Institutes of 

Health, n.d.). CT scans provide doctors with more detailed and accurate 

images than X-rays, allowing them to diagnose diseases and injuries and 



40 

plan medical, surgical, and radiation treatments. Figure 0-7 shows a CT 

scanner. 

 

Figure 0-7: Computerized tomography (CT) scan (Canon medical systems 

USA, inc., n.d.). 

1.11.5 Magnetic resonance imaging scanners 

Magnetic resonance imaging (MRI) scanners produce 3D detailed 

anatomical images. Living tissues contain water that contains protons 

whose rotational axis constantly changes. Sophisticated technology 

measures this change (Medical Device Network, 2022). 

Powerful magnets produce a strong magnetic field that forces protons in the 

body to align with the magnetic field. A radiofrequency current stimulates 



41 

protons, causing them to spin out of equilibrium, straining against the 

magnetic field. The MRI sensors detect the energy released as the protons 

realign with the magnetic field once the radiofrequency field is turned off. 

“The time it takes for the protons to realign with the magnetic field, as well 

as the amount of energy released, changes depending on the environment 

and the chemical nature of the molecules” (Medical Device Network, 2022). 

This allows physicians to detect and diagnose illness. Figure 0-8 is an MRI 

scanner developed for neurological and musculoskeletal imaging functions 

(Medical Device Network, 2022). 

 

Figure 0-8: SIGNA 7.0T, a magnetic resonance imaging (MRI) scanner 

(Medical Device Network, 2022). 

  



42 

1.12 Scanning methods 

Figure 0-9 summarises the different scanning methods. 

 

Figure 0-9: Summary of the different scanning methods (Moberg, 2017). 

During the process of photometric scanning, a single hand-held camera can 

capture several overlapping images of an object to analyse the shadows 

and the “surface normal” determined. With surfaces normal, computer 

software can recreate objects in three dimensions and with surface textures 

of high quality (Moberg, 2017). Photometric scanning requires four different 

software steps: 

• Micro-controller code written in C for 3d scanning; 

• Raspberry Pi code is written in Python or 3d scanning; 

• Raspberry Pi code is written in Python for mapping image pixels to 

sample points; and 

• Merging and meshing clouds in open-source software (Moberg, 

2017, p. 22). For example, CloudCompare computer software can 

recreate the object in three dimensions and with surface textures of 

high quality (Moberg, 2017). 



43 

Images are distributed according to the camera's angle of view to 

synchronise LiDAR samples with photographed pixels. Figure 0-10 shows 

images captured with some overlap due to the camera's viewing angles, 

62.2° horizontal and 48.8° vertical. 

 

Figure 0-10: Concept of image sectioning (Moberg, 2017). 

Raspberry Pi calculates pixels from the images depending on angles phi (φ) 

(φ) and theta (θ). When the correct image is found, the pixels are organised 

in columns and rows. The correct pixel is mapped according to the offset 

between the LiDAR and the camera (Moberg, 2017). Figure 0-11 illustrates 

the relationship between the camera, LiDAR and the image. 



44 

 

Figure 0-11: Camera offset geometry (Moberg, 2017). 

The system’s range is limited by how much diffused light is reflected and 

absorbed by the sensor (Moberg, 2017). Therefore, fewer samples result in 

a less detailed point cloud but a correct mesh. “While a dense point cloud 

captures the object shape very well, the inaccuracy of the LiDAR sensor 

reduces the final mesh quality” (Moberg, 2017, p. 35). Mathematically 

speaking, the word 'normal' refers to being at right angles or meeting at 90 

degrees (Math Open Reference, 2011). Figure 0-12 shows line JK as 

perpendicular to AB. 



45 

 

Figure 0-12: Line JK is normal to AB (Math Open Reference, 2011). 

The LiDAR sensor calculates normals between a set of points resulting in 

small planes between points (Moberg, 2017). The calculated normals 

determine the normal to a point on a surface to estimate the normal of a 

plane tangent to the surface resulting in a least squares plane fitting 

estimation problem (Sphinx, n.d.). Least Squares Fitting is “a mathematical 

procedure for finding the best-fitting curve to a given set of points by 

minimising the sum of the squares of the offsets ("the residuals") of the 

points from the curve” (Wolfram Research, Inc., 2022). 

The LiDAR sensor captured points on a wall with a minimal gap. With a 

smaller spacing between points, the tiniest opening in the wall causes a 

jagged shape instead of a flat wall; see Figure 0-13. 



46 

 

Figure 0-13: Illustration of a very dense point cloud with errors and 

resulting mesh (Moberg, 2017). 

Figure 0-14 shows the result of a scan with fewer sample points resulting in 

a less detailed point cloud but a correct mesh. The algorithm (Poisson 

surface reconstruction) averages points to create the most suitable plane 

(Moberg, 2017). 

 

Figure 0-14: Illustration of less dense point cloud with errors and resulting 

mesh (Moberg, 2017). 

“The ‘Poisson surface reconstruction algorithm’ uses point cloud normals to 

build a 3D surface. The algorithm considers all the points at once and is 

therefore highly resilient to the noise” (IGI Global, 2022). It is crucial to scan 



47 

objects from a straight linear perspective (Moberg, 2017). Moberg (2017) 

found that a faster and less detailed scan produces the best result. 

Hand-held scanners use triangulation range finders which typically range 

between 25-30 m. Examples of hand-held scanners that use triangulation 

range finders of are the Leica BLK2GO and the ZEB REVO RT. Leica 

BLK2GO handheld scanner with a range of 25 m at an accuracy of 6-15 

mm, designed for indoor and outdoor use (Leica Geosystems, 2019). Figure 

0-15 shows an example of the Leica BLK2GO Imaging laser scanner. 

 

Figure 0-15: Leica BLK2GO Imaging laser scanner (iF Design, 2022). 

The ZEB REVO RT with a scanning range of 30m and an accuracy of 1-

3cm for indoor and outdoor use (Optron group, 2018). See Figure 0-16 for 

an example of the ZEB Revo RT scanner. 



48 

 

Figure 0-16: ZEB Revo RT - Real-Time handheld scanner (Optron Group, 

2018). 

As a result of high-resolution scanning, it is possible that the beam will strike 

an edge of an object, causing data to appear noisy just behind the edge. 

Although scanners with a smaller beam width help to solve this problem, 

they are limited by range as the beam width will increase over distance 

(Trebuňa, et al., 2018). “Some laser scanners have level compensators built 

into them to counteract any movement of the scanner during the scan 

process” (Trebuňa, et al., 2018, p. 3). 

An Inertial Measurement Unit (IMU) “is an electronic device that measures 

and reports a body's specific force, angular rate, and sometimes the 



49 

magnetic field surrounding the body, using a combination of accelerometers 

and gyroscopes, sometimes also magnetometers” (IoT ONE, 2022). 

determines the position between the handheld scanner and the object being 

scanned. The IMU allows it to decide where data is collected with the GNSS 

(Trebuňa, et al., 2018). One should typically use computer software to 

correlate geo-referenced markers on the scanned image (Figure 0-17) to 

the pre-defined coordinates of the scanned markers (Trebuňa, et al., 2018). 

 

Figure 0-17: Marker on a scanned image. 

Tacheometric measurements combine measured distances and angles 

(Staiger, 2003). For example, tachymetric Laser Scanners measure one 

oblique distance (s’) to each point of an object, and two orthogonal angles 

(W1 and W2) see Figure 0-18. 



50 

 

Figure 0-18: The principle of tacheometric Laser Scanning (Staiger, 2003). 

Tacheometric Laser Scanners can be divided into three groups (Figure 

0-19): 

• Panorama-Scanner; 

• Hybrid Scanner; and 

• Camera Scanner (Staiger, 2003). 



51 

 

Figure 0-19: The different types of tacheometric scanners (Staiger, 2003). 

With Panorama-Scanners, the instrument's base limits the FOW, including 

its tripod (Staiger, 2003). In addition, panoramic images taken with a 3D 

scanner “contain depth information. That makes it possible to take 

measurements within the images, inspect coordinates and add markups, 

and even include 3D models as augmented reality objects within the 

imagery” (Arrival 3D, Inc., 2022). An example of a Panorama-Scanner is the 

Leica RTC360 Laser Scanner with a range between 0.5 - 130 m (SCCS 

Survey, 2022). 



52 

 

Figure 0-20: Leica RTC360 Laser Scanner (SCCS Survey, 2022). 

With a Hybrid Scanner, one rotation axis is without restrictions (often the 

horizontal movement); the second rotation axis is limited to, for example, 

60° due to the use of mirrors (Staiger, 2003). An example of this scanner 

includes the RIEGL VZ-6000 with a scanning range of 6km that can operate 

in poor visibility caused by dust, haze, rain and snow, see Error! Reference 

source not found. (RIEGL Laser measurement systems, 2020). 



53 

 

Figure 0-21: RIEGL VZ-6000 laser scanner (RIEGL Laser measurement 

systems, 2020). 

Camera Scanner works on the photogrammetry principle, which “is a 3-

dimensional coordinate measuring technique that uses photographs as the 

fundamental medium for metrology (or measurement)” (Geodetic Systems, 

Inc., 2020). The Camera Scanner has a limited Field of View (FOW); see 

Figure 0-22. In addition, this type of scanner captures a view from outside 

of the object(s) (Staiger, 2003). 



54 

 

Figure 0-22: Camera field of view relationship (Geodetic Systems, Inc., 

2020). 

The range of acceptable sharpness determines the optimal distance that a 

camera should be used from an object refers to the depth of focus (Geodetic 

Systems, Inc., 2020). Figure 0-23 demonstrates the focus depth 

relationship. 



55 

 

Figure 0-23: F-number to focus depth relationship (Geodetic Systems, 

Inc., 2020). 

With a Hybrid Scanner, one rotation axis is without restrictions (often the 

horizontal movement); the second rotation axis is limited to, for example, 

60° due to the use of mirrors (Staiger, 2003). An example of this scanner 

includes the RIEGL VZ-6000 with a scanning range of 6km that can operate 



56 

in poor visibility caused by dust, haze, rain and snow, see Error! Reference 

source not found. (RIEGL Laser measurement systems, 2020). 

1.12.1 Point clouds 

Point clouds create 3D surfaces from scanned objects. “Point clouds are 

often converted to polygon mesh or triangle mesh models or CAD models 

through a process commonly referred to as surface reconstruction” 

(Trebuňa, et al., 2018, p. 5). Figure 0-24 is an example of point clouds of a 

scanned building.   



57 

 

Figure 0-24: Point cloud of a scanned building (Trebuňa, et al., 2018). 

The point set registration process allows the alignment between point 

clouds and 3D models (Trebuňa, et al., 2018). “CloudCompare is a free & 

open-source software package that allows the 3D registration of point 

clouds based on (at least) 3 known points” (Rock Mapper Helpdesk, 2022). 

Align point cloud based on known points using the [FaceID]_pointcloud CSV 

file from a 3D scan. “CloudCompare automatically detects the correct 

columns for X/Y/Z and RGB (colour)” (Rock Mapper Helpdesk, 2022). Once 

the [FaceID]_pointcloud CSV file is opened in CloudCompare, a window 

appears that allows picking a minimum of three known points for 



58 

georeferencing on the point cloud; this is referred to as “to align entities”. 

The picked points will be shown as A0, A1 and A2 in 3D and in a table; see 

Figure 0-25. 

 

Figure 0-25: Picked points are shown as A0, A1 and A2 in 3D and in a 

table CloudCompare software (Rock Mapper Helpdesk, 2022). 

The next step is to enter surveyed reference coordinates of points A0, A1 

and A2 in the table with the heading “show reference entities” reference 

point coordinates are numbered R0, R1 and R2, see Figure 0-26 (Rock 

Mapper Helpdesk, 2022). 



59 

 

Figure 0-26: Reference point coordinates (Rock Mapper Helpdesk, 2022). 

Once the align button is pressed, the point cloud will be aligned to match 

the reference point coordinates (Rock Mapper Helpdesk, 2022). Both 

publications by Morris and Langari (2012) and Yashchuk (2009) note that 

all the calibrations and specifications of an instrument are only valid under 

specific, controlled environmental conditions – dependent on variables such 

as temperature, pressure and humidity. Environmental changes affect 

instruments: zero drift and sensitivity drift (Morris & Langari, 2012). Zero drift 

can be corrected by recalibration of the scanning instrument. Manufacturers 

usually calibrate laser scanners and recommend that users re-calibrate 

them regularly (Mijic, 2015). Sensitivity drift measures the error per each 

environmental constraint to which the instrument characteristics are 

sensitive (Morris & Langari, 2012). It is possible to reduce drift error by 

ensuring that scanning occurs within the range of the scanning instrument 

(Wang, et al., 2021). 



60 

Simultaneous localisation and mapping (SLAM) are an algorithm that uses 

data from the mapping system’s onboard sensors – 3D LiDAR, RGB 

camera, and IMU to determine the trajectory as the scanner moves through 

an area (Higgins, 2020). The most popular process for correcting errors is 

finishing back where one started (Higgins, 2020); see Figure 0-27. 

 

Figure 0-27: End scan at the scan start point (Higgins, 2020). 

Coordinated control points can be used when the scanned environment 

does not allow for a closed loop. For example, when the scan is geo-

referenced, the control points are used to snap the scanned image to 

corresponding coordinate points. This method should be used where 

survey-grade accuracy is of the utmost importance (Higgins, 2020). 

  



61 

1.13 Case Studies 

1.13.1 3D Scanning in an open pit coal mine in Indonesia 

3D LiDAR technology was used in a study in the East Kalimantan Province 

of Indonesia for more effective mine mapping as an alternative to traditional 

mine mapping methods. Mining in the East Kalimantan Province consists of 

open pit coal mining (Septarini, et al., 2013). 

 

Figure 0-28: East Kalimantan Province (Septarini, et al., 2013). 

3D LiDAR data was obtained in point clouds, which were then filtered and 

grouped according to the needs of getting a digital terrain model (DTM). A 

DTM is a 3D “digital file of a detailed representation of the topographical 

variations in the surface of the Earth. A DTM does provide a 3D image of 

the land surface” (Septarini, et al., 2013, p. 4). The DTMs of 2010 and 2012 

were used to calculate the volume difference between the two DTMs 

(Septarini, et al., 2013). The DTMs were stacked so that the volume 

difference could be calculated. The DTMs were obtained by generating the 



62 

ground points (after the classification process) using Global Mapper 

software (Septarini, et al., 2013). The “Unsigned Subtraction” method 

(positive numbers) (O’Reilly Media, Inc., 2022), was used to compare the 

volume difference between the DTMs between 2010 and 2012. Global 

Mapper and Surfer software were used to compare the result from the 

volume of the mine excavation in 2010 and 2012. Point cloud cleaning was 

done to eliminate spikes in the DTM (Septarini, et al., 2013) to ensure an 

accurate volume calculation. The point data was classified into two main 

groups: ground and non-ground. “The orange points mean ground points, 

and the white points mean non-ground points” (Septarini, et al., 2013, p. 5). 

In this instance, non-ground classified points may represent vegetation and 

the black points unclassified data in the data set (Prerna & Singh, 2015). 

Figure 0-29 display data from 2010 and 2012. 

  



63 

 

 

Figure 0-29: Compared point cloud difference between DTMs (Septarini, 

et al., 2013). 

Table 0-1 shows the difference in volume by using two different software 

programs, Global Mapper v14 and Surfer v10, with an applied cut factor and 

fill factor. “The cut factor is used to adjust the volume of excavated material 

to account for the expected amount of the material to swell when it is 

excavated. The fill factor is used to account for the additional volume of 

material that would be required due to the compaction of the soil when it is 

placed” (Septarini, et al., 2013, p. 6). 



64 

Table 0-1: Volume calculation using Global Mapper and Surfer (Septarini, 

et al., 2013). 

 Software  

 Global Mapper v14 
(m³) 

Surfer v10 
(m³) 

Volume Difference 
(m³) 

Cut 
Volume 

101,931 101,938 6.3 

Fill 
Volume 

0.2 0 0.2 

 

The result of the scan is primarily used to report the volume of blasted rock. 

Scans are captured before and after blasts to calculate the volume of 

blasted rock, and therefore, the advance of excavations can be determined 

once the blasted material has been removed (Ahamad & Ojha, 2015). In 

addition to the advantage of calculating volume from excavations, 

geological and geotechnical mapping are also available from the scans. 

Because scans do not take long to capture in the workplace, this has an 

added safety aspect as mine personnel can spend less time on the face and 

thus reduce the probability of accidents and injuries. This is possible as 

technology has enabled remote scanning of workplaces (Ahamad & Ojha, 

2015).  



65 

1.13.2 3D Scanning in an underground room-and-pillar limestone 

mine in South Korea 

Daesung MDI Donghae Limestone Mine is in Donghae-si, Gangwon-do, 

South Korea. Daesung MDI Donghae Limestone Mine is an underground 

operation that uses a hybrid room-and-pillar mining method. The case study 

aims to determine if it is possible to maximise ore recovery while ensuring 

the stability of stopes by using horizontal and vertical safety pillars. This 

limestone mine used 3D LiDAR scanning technology to analyse the 

orientation of a joint in a vertical safety pillar (Lee & Choi, 2019). 

The 3D point cloud data obtained from the 3D LiDAR scan made it possible 

to measure the dip directions of joint sets. The study collected data on 79 

joints with distinctive orientations from the entire stope area under 

investigation (Lee & Choi, 2019). 

1.13.3 Using laser scanner face mapping to improve geotechnical 

data confidence at Sishen Mine, an iron ore open-pit mining 

operation 

Sishen Mine is a large open-pit mining operation in South Africa that 

requires reliable geotechnical data to design and evaluate pit wall stability. 

The primary data sources to achieve this are “geotechnical borehole data, 

face mapping data, geotechnical laboratory testing data and implicit 

structural models” (Russell & Stacey, 2019, p. 11). Face mapping has 

traditionally been done via direct contact with the face through a technique 



66 

referred to as line mapping (Russell & Stacey, 2019). Line mapping 

measures the spacing between structural discontinuities, see Figure 0-30. 

 

Figure 0-30: Measurement of discontinuity spacing (Russell & Stacey, 

2019). 

Digital mapping of faces to capture geological discontinuity and structural 

orientation has become more prevalent in recent years. In 2013, Sishen 

Mine acquired a terrestrial laser scanner with the resolution, photographic 

capabilities and software capable of carrying out geotechnical face mapping 

(Russell & Stacey, 2019). During the study by Russell and Stacey, a 

mapping procedure was set up based on accepted face mapping methods, 

the data requirements of the mine and the capabilities of the laser scanner 

system. Mapping data from the scans was integrated into the Acquire 

geological data management system. Acquire is a purpose-designed 

structured query language (SQL) database system that stores the mine’s 



67 

geotechnical data (Russell & Stacey, 2019). This method integrates scanner 

software, Microsoft Excel, and the Acquire geological database system. The 

scanner software makes information available in text or CSV format. Sishen 

uses an Excel template to analyse and delineate dominant discontinuities 

and rock mass grading before entering data into the mine's Acquire 

geotechnical database. Figure 0-31 shows the delineation of significant 

discontinuity planes ready for export and capture in the Acquire database. 

 

Figure 0-31: Delineation of significant discontinuity planes for export and 

capture in the Acquire geotechnical database (Russell & Stacey, 

2019). 

Acquire, a geotechnical database, allows for data to be queried based on 

various parameters and for statistical analysis to be carried out. “Rock mass 

parameters from either face mapping or borehole logging datasets can be 

queried with the required lithological and spatial constraints applied. This 

feeds into the geotechnical design process by providing up-to-date rock 

mass parameters” (Russell & Stacey, 2019, p. 18). Likewise, the mapping 



68 

report sheet feeds into the geotechnical design process by providing up-to-

date rock mass parameters (Russell & Stacey, 2019). Figure 0-32 is an 

example of the output Excel face mapping report. 

 

Figure 0-32: Excel face mapping report sheet from Acquire geotechnical 

database (Russell & Stacey, 2019). 

Figure 0-33 is a schematic data flow for geotechnical face mapping data at 

Sishen Mine. 



69 

 

Figure 0-33: Schematic data flow for geotechnical face mapping data at 

Sishen Mine (Russell & Stacey, 2019). 

The system integrates the scanner software, Microsoft Excel, and the 

Acquire geological database system. The scanner software allows for 

exporting the various individual parameters in text or CSV format. For data 

management at Sishen, an Excel template was set up to carry out kinematic 

analysis, delineation of predominant discontinuities and rock mass rating 

before importing the relevant data into the mine’s Acquire geotechnical 

database. 

A terrestrial laser scanner can create a high-resolution point cloud covering 

several hundred square metres in a couple of minutes. These scanned point 

clouds produce fast and efficient digital face mapping (Russell & Stacey, 

2019). Figure 0-34 shows a high-resolution point cloud for face mapping 

(Russell & Stacey, 2019). 



70 

 

Figure 0-34: Terrestrial laser scanner using a vehicle-mounted set-up (left) 

and a high-resolution point cloud for use in face mapping (right) 

(Russell & Stacey, 2019). 

A surveyed reference point needs to be positioned on the face and two 

camera tripod positions need to be accurately surveyed (Russell & Stacey, 

2019). This is important if the scanned image needs to be geo-referenced 

to feed into Sishen Mine’s geological model. Sishen Mine realised the 

following advantages of laser scanning: 

• The laser scanner provided faster and more accessible data 

collection; 

• The laser scanner provided faster data processing and was less 

demanding on software systems; 

• More accurate discontinuity orientation measurements were 

obtained using the laser scanner (up to 15° difference in dip 

measurements between the two techniques was observed); and 

• Planes oblique to the exposed face were more readily observable 

with the laser scanner (Russell & Stacey, 2019). 



71 

(Russell & Stacey) determined that the laser scanner would give reliable 

roughness measurements for discontinuity traces greater than 2 m in length 

for typical scan ranges between 50 m and 200 m. The research showed that 

scanner inaccuracy would prevent the reliable measurement of discontinuity 

roughness for shorter traces in instances where the height of the surface 

irregularities is less pronounced (Russell & Stacey, 2019). At Sishen, the 

Micromine 3D modelling software package has been adopted to analyse 

designs and incorporate spatial geotechnical data into the design and 

analysis process (Russell & Stacey, 2019). 

The literature from Sishen Iron Ore Mine showed that the terrestrial laser 

scanner provides a practical, faster and safer way to collect face mapping 

data. The results obtained from the terrestrial laser scanner were compared 

to manual techniques and a stereo photosystem such as Sirovision (Russell 

& Stacey, 2019). The system is safer and allows inaccessible rock faces to 

be mapped because no contact with the face is needed. The photographic 

overlay of the 3D mapping face was an accurate means of interpreting 

structural and rock mass features on the underlying scan surface (Russell 

& Stacey, 2019). 

The terrestrial laser scanner system at Sishen Mine has proved invaluable 

for geotechnical data capture, geotechnical hazard assessment and 

structural mapping. The laser scanner also makes the rapid collection of 

data possible when surface support such as wire mesh or shotcrete is to be 

applied to the rock face soon after excavation. Russell and Stacey were 



72 

confident that the additional data the laser scanning made available would 

enhance the quality of designs and reduce risk accordingly. They, therefore, 

regard the scanner as “a tool of strategic importance” (Russell & Stacey, 

2019, p. 20). 

1.14 Section Summary 

3D laser scanning offers more advantages to the mining industry, which, as 

discussed, will improve the overall accuracy and efficiency of mining. This 

is primarily due to the reduced time taken to capture large surface areas 

and the geological mapping process not being dependent on traditional, less 

accurate mapping methods. Geospatial data are also stored on a digital 

system whereby the data is always readily available to all production team 

members. Monitoring the flow of blasted ore becomes a traceable process 

with higher accuracy, increasing ore grade estimations and affecting profit 

calculations. The frequency of scanning and surface mapping is increased 

when utilising 3D scanners. This means that the entire pre-blasting planning 

and preparation process could be accelerated. As per the case studies, 3D 

scanning systems commonly used on open pit mines are suitable for the 

narrow, tabular stopes found on Mponeng. 

  



73 

RESEARCH METHODOLOGY 

1.15 Section Overview  

The quantitative and qualitative research methodology aimed to investigate 

using a cost-effective 3D LiDAR scanner for underground use. The primary 

purpose of this scanner will be to capture underground geological mapping 

information without interrupting production. Furthermore, the data captured 

underground should be distributable among relevant stakeholders; the data 

is processed as soon as an employee returns to the surface. Therefore, 

finding a fast and easy mechanism to share the scanned underground 

geological mapping information with the production team was essential. The 

reason for this is to guide the production team on SW control. 

The priorities used to guide the weighting between device suitability and 

appraisal of performance in the testing phase are introduced in this chapter. 

The initial approach was developed during the research to short-list eligible 

hardware and software. A literature survey process supports evaluating the 

short-listed laser scanner hardware and software. 

The research describes applied data collection methodologies and then 

discusses how the collected information is analysed. Finally, the results 

illustrate obstacles encountered during the research. 



74 

1.15.1 Priorities of the Research Criteria 

The following criteria were applied in the short-listing and the identifying of 

the choice of a laser scanner for the practical underground conditions of 

narrow tabular mining: 

• Ease of use. The scanner must be used daily to acquire and share 

information productively. The buy-in and willingness of underground 

personnel to use the device optimally was an additional and 

important consideration; 

• Accuracy and quality of laser scanners. It is vital to build the 3D 

block model from accurate information. Therefore, the 

manufacturer's accuracy specifications had to be considered; and 

• Cost of 3D scanners and software. The evaluation included the 

financial outlay of purchasing and maintaining the laser scanning 

system, which would consist of software and the cost of software 

upgrades. 

Cost played an important factor since the project was in the test phase. It 

was essential to keep all costs as low as possible to ensure that 

underground use of the proposed 3D LiDAR scanner was accepted upon 

presentation of the research results. Therefore, the availability of tools and 

devices dictated research feasibility, the research intended to find 

technology that could be advantageous to both the production and geology 

disciplines. 



75 

The production department may benefit from the use of non-georeferenced 

scans. Non-geo-referenced scans and the reef widths can be compared to 

each other to identify grade dilution. The scanning will enable the production 

team to scan the workplace during the underground shift, plot daily 

advances, plan equipment and complete legal documents such as safety 

reports while underground. This may help reduce extra surface time, which 

is additionally desirable as computers are limited for the use of shift boss. 

The scans may also indicate substandard cleaning of stope panels and 

include a record of roof support for safety standard reporting. 

However, the primary goal is for the geology department to use geo-

referenced scans to digitise the underground stope mapping directly onto 

Deswik CAD. This mapping will then update the 3D geological block model. 

This will ease the workflow and help proactive action towards better 

controlling the SW and grade dilution. 

1.16 Deliberation of Devices and Methods 

Before conducting a literature review to compile a short list of 3D scanner 

technologies, the current devices used for 3D mapping underground on the 

Mponeng Mine were reviewed. The following total stations and electronic 

distance measurement (EDM) capabilities are the primary items of 

equipment investigated in this project: 

• South Total Station - N7, a product from V.I. Instruments; and 



76 

• TS07 3 R500 Total Station, produced by Leica Geosystems 

Additionally, information on traditional measuring (using tapes), map 

processing and software requirements (including file formats) currently used 

for integration with the 3D block model were compiled. Finally, it was 

necessary to determine if the proposed research methods and shortlist of 

alternative scanners would benefit Mponeng Mine. It was also essential to 

ensure that shortlisted alternative scanners would yield a smooth integration 

of the workflows discussed in Chapter 6. 

Three devices were short-listed according to the prerequisite criteria for the 

practical use of a 3D scanner underground. The prerequisite criteria 

consider range, accuracy, weight, ease of use, safety, durability and cost-

effectiveness. The devices are listed as follows: 

• Structure Sensor Mark II, built by Occipital, a spatial computing 

company, see Figure 0-1; 

• The non-cellular Apple iPad Pro is 11-inch with Wi-Fi and 256 GB 

storage capability, see Figure 0-2; and 

• Apple iPhone 13 Pro Max with Wi-Fi 128 GB storage capability; see 

Figure 0-3. 



77 

 

Figure 0-1: The Occipital Structure Sensor (Mark II)) (3D Printing Systems 

South Africa, 2021). 

  



78 

 

Figure 0-2: Apple iPad Pro 11-inch (iStore, 2021). 

  



79 

 

Figure 0-3: iPhone 12 Pro (Apple Inc., 2022). 

All these devices have hardware specifications and compatible software to 

which the data is downloaded and presented with limitations. These 

limitations are discussed in more detail in the research. The shortlist of 

devices additionally has system requirements and particular applications 

(apps) considered in selecting the most optimal choice(s). Discussion in this 

section will highlight all information necessary for comparison of the relevant 

3D scanning technologies and to justify the decisions on those submitted to 

the testing phase of the study. 



80 

Photogrammetry 3D scanning technology is the process whereby photos of 

a 3D object are stitched together with software to produce a 3D model that 

represents the colour and texture of the scanned object (Aniwaa Pte. Ltd., 

2021). Another significant influence in the decision-making process was the 

total cost of the applications. Therefore, they were purchasing these 

applications. The costs that apply when exporting 3D scans, as well as the 

cost of additional software requirements (that is, other apps necessary for 

data manipulation, display and conversion. Furthermore, the file formats in 

which exportable data becomes available must be compatible with 

Mponeng’s established software. Finally, data storage and scan-capturing 

mechanisms were noted, as some applications may require an internet 

connection. 

1.16.1 Current technologies used at Mponeng Mine 

Total Stations 

Mponeng Mine currently uses two types of Total Station. They are the South 

Total Station - N7 from V.I. Instruments and the Leica TS07 3” R500 Total 

Station from Leica Geosystems to survey underground workings for 

depletion calculations of the reserve block model. Mine surveyors go 

underground daily to survey development ends that have been blasted. 

Mine surveyors also survey stope panels during the month, and a final 

measurement is reported for payment purposes. Both the development and 

stope advances are plotted on Syncromine software. The mine surveyors’ 

scans are exported in the form of points. Once the mine surveyor has 



81 

downloaded the scanned points, they connect them manually to determine 

the outlines of excavations. 

The South Total Station - N7 device measures an ultra-fast speed of under 

three seconds in fine mode over 600 metres. It works with a rechargeable 

lithium battery that lasts six hours in temperatures ranging from -20° Celsius 

to 50° Celsius. The dimension of the Total Station is 196 mm x 192 mm x 

360 mm, and it weighs 6.2 kg. The accuracy of scans is ± (3 + 2 ppm x 

Distance) mm (South VI Instruments, n.d.). Figure 0-4 is an image of the 

South Total Station - N7. The cost of a new Total Station is currently R 145k 

(V. I. Instruments, 2022). 



82 

 

Figure 0-4: Total Station N7 (South VI Instruments, n.d.). 

The Leica TS07 3” R500 Total Station typically measures at the speed of 

2,4 seconds in precise mode over a range of 3,5km. It works with an 

exchangeable lithium battery. The GEB361 model battery for the TS07 has 

an operating time of up to 30 hours. The GEB331 model battery has a 

working time of up to 15 hours. It can withstand temperatures between -20° 

Celsius and 50° Celsius. The Total Station weighs between 4,3kg and 

4,5kg. The accuracy of scans is 1,5 ppm (SCCS Survey, 2015). Figure 0-5 

is an image of the TS07 3” R500 Total Station. The cost of a new Total 

Station is currently R 200k (Leica-geosystems, 2022). 



83 

 

Figure 0-5: TS07 3” R500 Total Station (SCCS Survey, 2015). 

As a scanner alone weighs 4,5 – 6,2 Kg (SCCS Survey, 2015), these 

instruments quickly become an arduous and heavy load to carry. At 

Mponeng Mine, underground transportation by rail-bound man-carriage is 



84 

not always available. As a result, survey equipment must be carried to and 

from worksite locations. Travelling distances are often more than 4 km one-

way, and accessing workplaces usually involves climbing through steep 

areas or through confined spaces. Large, heavy equipment can therefore 

be cumbersome to carry into workplaces. Choosing the most suitable 

instrument also requires consideration of the vertical angle measurement 

capability of the instrument, which makes observations over certain 

angles/degrees impossible and complicates checking backsight pegs. This 

poses a problem in confined excavations, for example, box holes/ore 

passes, steep raises and stope panels. 

Deswik CAD software used at Mponeng Mine 

Mponeng Mine uses Deswik CAD to capture and export geological mapping 

information. Deswik CAD supports a wide variety of file formats. The file 

formats of interest to the research were OBJ, DXF and DWG. 

“Deswik Mapping is a Microsoft Windows tablet application that allows 

geologists to create geological maps and photography in their underground 

workplace” (Anderson & Gall, 2020, p. 4). Figure 0-6 shows a face section 

scanned and mapped on a tablet using the Deswik Mapping application. 



85 

 

Figure 0-6: Face section from Deswik Mapping application (Anderson & 

Gall, 2020). 

Although Mponeng Mine uses Deswik software, Mponeng Mine did not 

purchase the Deswik Mapping application, which potentially could 

streamline the mapping process. Therefore, Mponeng Mine is unable to use 

the Deswik Mapping application. Deswik Mapping uses survey pegs to align 

scanned images with corresponding survey pegs when georeferencing 

scanned images. For example, Figure 0-7 shows a mine surveyor 

measuring from a roof peg (spad with numbered copper disc). 



86 

 

Figure 0-7: Brass spad with numbered copper disc(Nagel, 2022). 

Mponeng Mine does not use Deswik's mapping tools. Instead, the geologist 

uses a clinometer, tape, and photographs to capture underground 



87 

measurements in notebooks. Once the geologist is on the surface, they plot 

the measurements in Deswik CAD on a flat surface at the correct elevation 

(Z). This information is fed into the 3D geological model. The geologist's 

photographs of the stope face underground are used in the mapping report 

to explain and show the position of the CW with the hanging wall and 

footwall of the stope panel; the norm is to use two to four photographs. 

Unfortunately, photographs are not an accurate representation of the 

complete stope panel. In addition, the next blast may result in the loss of 

valuable geological data of the stope panel in 3D space. 

1.16.2 Technologies investigated in the research 

Occipital Structure Sensor (Mark II) 

The Occipital Structure Sensor (Mark II) is an upgrade of the original 

Structure Sensor (TC Join TechCrunch, 2021). The Occipital Structure 

Sensor (Mark II) scanner in the medical field captures 3D models of any 

body part for custom-fit prosthetics. (Occipital, Inc., 2021f)The Canvas 

application, which uses Apple LiDAR devices, can capture small and 

medium-sized residential rooms but not extensive industrial facilities. 

(Occipital, Inc., 2020c). Therefore, Occipital Inc. recommends using 3D 

LiDAR-based scanners when scanning extensive industrial facilities is 

required (Occipital, Inc., 2020c). 

In Figure 0-8, this scanner is a hardware fitting for an iPad and is powered 

by the fitted mobile device's iPhone Operating System (iOS). 



88 

 

Figure 0-8: Occipital Structure Sensor (Mark II) (Birch, 2021). 

The scanner sensor projects an infrared (IR) speckle pattern to the scanned 

object(s). Once it reaches dark and reflective objects, the pattern distorts, 

this technology application is called structured light (Occipital, Inc., 2020c). 

Black objects absorb the IR laser emissions and as a result, lasers emitted 

by this device reflect randomly, causing objects or features to disappear. 

Some users have solved this problem by applying paint or powder to the 

scanned objects to help the sensor capt