Towards Understanding Breakage and Flow in Sublevel Caving (SLC)

DOC TOR A L T H E S I S
Department of Civil, Environmental and Natural Resources Engineering

Division of Mining and Geotechnical Engineering
Lule University of Technology 2012
Matthias Wimmer
ISSN: 1402-1544 ISBN 978-91-7439-509-9
Towards Understanding Breakage and

Flow in Sublevel Caving (SLC)
Towards Understanding Breakage and Flow in Sublevel Caving (SLC)
Development of new measurement techniques

and results from full-scale tests
Matthias Wimmer
DOCTORAL THESIS
Towards Understanding Breakage and Flow in
Sublevel Caving (SLC)
Development of new measurement techniques and results from full-scale tests
Matthias Wimmer
Division of Mining and Geotechnical Engineering

Department of Civil, Environmental and Natural Resources Engineering
Lule University of Technology
SE-971 87 LULE
Sweden
Printed by Universitetstryckeriet, Lule 2012

ISSN: 1402-1544
ISBN 978-91-7439-509-9
Lule 2012
www.ltu.se
DOKTORSAVHANDLING
kad Frstelse av Snderbrytning och
Malmflde i Skivrasbrytning
Utveckling av nya mtmetoder och resultat frn fullskalefrsk
Matthias Wimmer
Avdelningen fr Geoteknologi
Institutionen fr Samhllsbyggnad och Naturresurser
Lule Tekniska Universitet
SE-971 87 LULE
Sweden
when you can measure what you are speaking about,

and express it in numbers, you know something about it
Lord Kelvin, 1883
ACKNOWLEDGEMENT
This thesis would not have been possible without financial support, supervision and encouragement
from various individuals and organizations throughout the years.
I was able to carry out full-scale research and to advance knowledge at the front line at a worldrenowned underground mine. Considerable time and effort was involved. The Hjalmar Lundbohm
Research Centre (HLRC), a research foundation at LTU (Lule University of Technology) endowed
by LKAB (Luossavaara-Kiirunavaara Aktiebolog), is thanked for being the major financial sponsor.
The LKAB mining company is thanked for all their support and time in the completion of the project.
I am pleased to thank those who contributed in the reference group meetings and promoted the project.
Foremost, Professor Finn Ouchterlony is greatly acknowledged for providing supervision, twentyfour-seven support, encouragement and immense knowledge during this research work. Thank you!
I would like to express my gratitude to my honored teacher and co-supervisor, Professor Peter Moser
for initiating the contacts with LKAB and Swebrec (Swedish Blasting Research Centre) and for his
interest in following up this research.
I am greatly indebted to my colleague, Anders Nordqvist, who helped in both practical and technical
matters. In particular, his tireless input in writing and improving the software RingCave was
invaluable.
Furthermore, I would like to thank my former colleagues, Ulf Nyberg and Mats Olsson at Swebrec,
who helped in the field of measurement engineering and with whom I had countless stimulating
discussions.
I thank those who were of great help and support in practical matters. In particular, I would like to
thank the engineers Joel Kangas and Ola Sllstrm, and the technicians Leif Keskitalo, Jonny
Olofsson and Pertti Pannula. I also express my gratitude to the LKAB team of surveyors represented
by Anders Berg for their assistance.
LTU`s central workshop and colleagues at Complab are thanked for numerous and valuable
discussions in connection with the development of different measurement equipment. The personnel at
the company 3G Software & Measurement in Graz, Austria is thanked for their innovative support.
I would also like to thank all my former colleagues from the Division of Mining and Geotechnical
Engineering at LTU for supporting me in all kinds of ways during these years.
During the project many international contacts could be established and they provided valuable input
to various aspects of this work. I have encountered a very open mind which is much appreciated.
I thank Peter Travis for proof-reading and commenting on the manuscript of all reports and the thesis.
Finally, Id like to thank my family. My parents encouraged my interest in mining at an early age. My
partner, Michaela, has been backing me all the way and showed endless patience throughout the years:
being part of my life, Ive the strength and courage to turn a new leaf and stride forward! More
recently, little Arvid loves to keep me on my toes with his inquisitive mind.
Matthias Wimmer
Kiruna, October 2012
ii
PREFACE
The interest and growing use of cave mining techniques have mainly been initiated by the depletion of
ore bodies suitable for open pit operations. Cave mines with production rates in excess of 100.000
tonnes/day are being planned and constructed. Existing underground mines have been continuously
adapted to larger scale operation in the past. Sublevel caving (SLC) geometries have e.g. been
increased tremendously. Improved techniques, for instance in drilling and blasting, have allowed
increased mining productivity and reduced development and mining costs. Production has advanced to
greater depths at which the mining method is untested. In-situ rock stresses and confining stresses on
the caving debris increase which ultimately affect blast performance. Based on that, it could be
expected that flow behaviour has changed during the years. Planning of caving on this scale and in this
time frame is difficult as there are not many active caving mines and those existing mainly operate on
the basis of empirical data. Under these circumstances it is critical to understand the impact of
breakage on material flow if further improvements in SLC performance are to be obtained.
Measurements are therefore indispensible, but difficult to perform.
R&D work related to cave mining needs to be considered in a long perspective: it started several
decades ago and will continue to be of importance. Attempts to measure in this environment are often
hindered by the physical scale and the time frame. Blasting takes place rapidly while the draw of
material may last several days for an individual SLC ring. Incorporating flow from levels above
extends the process to months or even years. In addition, often several parameters are changed
simultaneously, which renders reliable analysis difficult. This thesis contributes with new
measurement techniques and, based upon the controlled full-scale tests, with an increased
understanding of breakage and flow in SLC. The thesis is presented in two parts as:
I. An extended summary which covers (i) introduction, (ii) literature review, (iii) review on drill
and blast practice at the Kiruna mine; followed by chapters relating to system development and
results from full-scale tests in the field of (iv) blast function, (v) fragmentation and (vi) gravity
flow, and concluded by a chapter (vii) concluding remarks with discussion and conclusions and
proposed future work and research.
II. An appendix which covers both papers and reports.
APA referencing style (American Psychological Association, 2009) is used for references. A number
of keywords, acronyms and symbols used throughout the thesis are explained on pages xxiii xxvi.
iii
iv
ABSTRACT
Blast function, fragmentation and gravity flow are core elements for sublevel caving (SLC). For the
ore recovery to be as high and the waste rock dilution as low as possible all three elements have to
work as planned. In the past some boundary conditions were changed: The scale and layout of SLC
changed tremendously and production advanced to greater depths. As a consequence, blast
performance and material flow are believed to have changed. The difficulties in simulating SLC relate
to the physical scale as well as the broad range of time-scales involved, which results in a large
number of unknowns and uncertainties. The present thesis contributes with the development of new
measuring techniques, full-scale tests and analysis of results in all the above three core elements.
Based upon that a better understanding of breakage and flow is obtained and this supports future
process improvements.
A remotely operated 3D photogrammetry system was designed and used to study the blast results from
SLC rings in opening and hang-up situations. Various blasting situations were observed and this
allowed a deeper study of the i) geometry of blasted cavity, ii) over- and underbreak and its effect on
subsequently blasted rings, iii) interaction effects between adjacent holes/rings, iv) mapping of
geological structures and its influence on the blast result and v) drilling accuracy based on identified
boreholes. An attempt to identify malfunctions and categorize breakage problems was made. Some of
the identified problems for initiation were unexpected and require a closer examination to increase
blasting efficiency and to mitigate nitrate leakages. Potential improvements for blasting initial
openings and scaling-up of rings are presented.
Confined drift wall blasting tests in which the blasted burden was instrumented with various sensors to
study movement and breakage were carried out. The main focus was the development of measuring
equipment that could be scaled-up to full-scale SLC blasting and installed behind the rings. Results
from an unconfined and confined blast situation are presented. A new measurement system to measure
burden movement was developed, which combined the initial movement recorded by an accelerometer
designed to minimize zero-shift with the later slower movement recorded by a photoelectric sensor
(fibre-optic zebra gauge). In addition, time domain reflectometry (TDR) proved to be a reliable
method for the verification of burden breakage (over- and underbreak).
Fragmentation characteristics for blasted ore loaded from a specific ring were studied by sieving. For
comparison lab-scale data from crushing, grinding and blasting, and also historical full-scale data from
the mine, were analyzed. This data confirms that the material follows the NBC criteria (NBC =
v
natural breakage characteristics) in the fines region (0 10 mm). However, a relative flattening of
the sieving curves due to selective breakage in the mid-range (25 - 75 mm) and increased amount of
fines was observed.
The main conclusion is that magnetite behaves like normal hard rock from a fragmentation point of
view despite this flattening deviation from Swebrec distribution behaviour as this is likely to be caused
by the internal flow mechanisms in the SLC process.
Gravity flow and the effects from confined blasting in hang-up situations were studied with the 3D
imaging system and allowed the extension of conceptual models for breakage and flow. Evidence is
given that gravity flow was disturbed and occurred in shallow, crescent-shaped flow zones. The front
part of the burden was fractured, but immobilized. For the subsequent blast this meant that swell and
compaction was limited to that zone with only marginal dilation and disaggregation of the burden at
the sides. The operating figures for the blasted rings with observed flow zones indicated a disturbancefree extraction. Inflow of caving debris might occur either from above or in front when the loader
penetrates into the flow zone of the previous ring. As a result of flow disruption in this narrow channel
a temporary hang-up might evolve and the flow scheme be altered so that material is then loaded from
previous blasts. With continued extraction, the cavity enlarges and finally becomes unstable, collapses
and both previously frozen ore and waste rock enters the draw point.
At the end a pivotal question remains: Are the recent observations of disturbed flow an inherent part of
blasting SLC rings? This raises also the follow up question: Can we expect an undisturbed gravity
flow if all borehole charges and rings detonate and break as planned? With continued investigations an
answer seems to be obtainable in the near future. It is also recommended that the methods developed
in this thesis be used in instrumented confined blasting tests in full-scale rings and be combined with
gravity flow measurements.
Keywords: sublevel caving, blast function, fragmentation, gravity flow, full-scale tests,
photogrammetry, photoelectric sensor, time-domain reflectometry, burden movement, burden
breakage, fragmentation, Swebrec function, secondary fragmentation, disturbed gravity flow.
vi
SAMMANFATTNING
Sprngsalvors funktion, fragmentering och rasflde r viktiga delprocesser i skivrasbrytning. Fr att

malmutbytet skall bli s stort som mjligt och grbergsinblandningen s lg som mjligt mste alla tre
delprocesser fungera enligt plan. Frutsttningarna har ndrats med tiden. Brytningsskalan har kat
betydligt under rens lopp, utformningen av raskransarna har ndrats och brytningen sker p allt strre
djup. Svrigheten att modellera skivrasbytningen beror bde p den fysiska skalan och p de olika
tidsskalor som ingr i frloppet, vilket medfr stora oskerheter och ven oknda storheter. Denna
avhandling bidrar med utveckling av nya mtmetoder, fullskalefrsk och analys av resultat frn alla
tre delprocesserna. Detta leder till en bttre frstelse av snderbrytningen och rasfldet vilket
mjliggr en framtida frbttring av hela processen.
Ett fjrrstyrt system fr 3D-fotogrammetri har konstruerats, byggts och anvnts fr att studera
kaviteter vid svl ppningar som hng (stopp i rasfldet) i skivrasbrytningen. En kvantifiering och
kartlggning har gjorts av i) den utsprngda kavitetens geometri, ii) baktbrytning och ofullstndig
lossbrytning, iii) hur brytningen samverkar mellan nrliggande sprnghl och skivraskransar, iv) hur
strukturerna pverkar sprngningsresultatet och 5) felborrningen baserad p identifierade kvarstende
borrhl. En ansats att identifiera felfunktioner i lossbrytningen och kategorisera problem har gjorts.
Problemen med initiering av laddningarna var ovntade och de behver underskas nrmare fr att
ka salvornas effektivitet och att minska kvvefrluster. Frslag ges till frbttringar av sprngning av
de frsta hlen och efterfljande kransar i en skivrasppning dr kransarnas hjd kar snabbt.
Sprngning av ortvggar mot mothll frn inlastade rasmassor har utfrts. I och bakom frsttningen
har olika instrument installerats fr att mta rrelser och snderbrytning. En viktig utgngspunkt har
varit att instrumenten skall kunna anvndas i fullstora skivraskransar dr de installeras frn baksidan.
Resultat frn mttligt och stort mothll presenteras. Fr att kunna mta frsttningens rrelse under
fltfrhllanden har ett nytt mtsystem utvecklats. Det kombinerar en accelerometer med minimal
nollpunkts-frskjutning fr registrering av det transienta initialfrloppet med en fotoelektrisk givare
(s.k. fiberoptisk zebragivare) som mter det lngsammare efterfljande frloppet. Dessutom anvnds
s.k. TDR-teknik, dvs. ingjutna koaxialkablar genom vilka pulser skickas och dr reflektioner frn olika
strningar lngs kabeln registreras. Denna teknik ger en god uppfattning om var grnsen fr
frsttningens lossbrytning gr.
Styckefallet hos sprngd magnetitmalm frn en utlastad skivraskrans har studerats genom siktning.
Som jmfrelse har data frn malm som krossats och malts i labbskala samt historiska data frn
vii
raskransar analyserats. Data styrker att malmen i finomrdet 0 - 10 mm har s.k. NBC-egenskaper
(NBC = natural breakage characteristics). I mellanomrdet 25 - 75 mm uppstr en relativ
avflackning i siktkurvan jmfrt med en s.k Swebrec-funktion, frmodligen pga selektiv neddelning
under rasfldet, och en strre andel finmaterial bildas. Trots avvikelsen frn Swebrec-beteendet hos
siktkurvan r nd slutsatsen att magnetiten upptrder som grberget ur fragmenteringssynpunkt.
Rasfldet och effekterna av sprngning mot mothll (rasmassor) har studerats med 3D-fotogrammetri
under hngsituationer. Detta har gjort det mjligt att frbttra befintliga snderbrytnings- och
fldesmodeller. Resultaten visade att fldet varit strt (lokaliserat) och skett inom smala
mnskreformade kanaler. Frsttningens frmre del hade spruckit snder men inte brutits upp, dvs
kunde inte brja rasa. Fr efterfljande raskransar blir svllningen av frsttningen och tfljande
kompaktering av mothllsmaterialet i rasmassorna begrnsad med mycket liten dilatation och
uppluckring av sidodelarna av frsttningen. Produktionsuppfljningen (lastade bergmngd mm) fr
kransarna med strt flde visade inte ngot onormalt innan hngen upptrde. Inflde av grberg frn
rasmassorna sker normalt uppifrn eller framifrn om lastmaskinen kan penetrera rasmassorna
tillrckligt lngt in. I en hngsituation har fldet uppifrn genom kanalen upphrt. Hnget kan vara
tillflligt och nr lastaren brjar lasta i rasmassorna frn fregende krans kan det kollapsa och
tidigare frusen malm och grberg rasa ner i lastorten.
En viktig frga kvarstr: r det strda fldet som observerats en integrerad del av rasfldet efter
sprngning av skivraskransar? Och fljdfrgan: Kan vi frvnta oss ett ostrt rasflde om alla
sprnghlen i en skivraskrans detonerar och bryter upp frsttningen som planerat? Svar p dessa
frgor kan fs inom en nra framtid om underskningar fortstter. Det rekommenderas ocks att
mtmetoderna som utvecklats i denna avhandling anvnds i instrumenterade sprngfrsk i storskaliga
skivraskransar (under mothll) och kombineras med mtningar av gravitationsflde.
Nyckelord: Skivrasbrytning, sprngsalvors funktion, rasflde, fullskalefrsk, fotogrammetri,
fotoelektrisk givare, pulsreflektionsmtningar (TDR), frsttning, frsttningens rrelse och
snderbrytning, styckefall, Swebrec-funktion, sekundr snderbrytning, strt rasflde.
viii
APPENDICES
Several papers and reports were written during the course of the project and are appended. The reports
document the experimental work in detail. The layouts of measurement systems and experiments
should be reproducible.
The papers are appended in printed form and the reports are contained on a CD attached to this thesis.
For the papers and reports with several authors, the order in which the names appear reflect the
contributions made by each author. The contributions from co-authors are specified as follows.
Furtney, J.: Numerical modeling of drift wall blasting tests (Blo-Up software in HSBM, Hyrbid
Stress Blast Model)
Kangas, J.: Support in field tests (fragmentation measurements)
Lenz, G.: Development of specific components within the SMX3D (ShapeMetriX3D) software
Moser, P.: Discussions related to the fragmentation measurements (NBC, natural breakage
characteristics)
Nordqvist, A.: Development of the RingCave software, discussions and support in various field tests
Nyberg, U.: Support in various laboratory and field tests
Ouchterlony, F.: Discussions, supply of specific figures and analysis and proofreading of reports
Selldn, H.: Discussions related to gravity flow of broken rock
ix
APPENDED PAPERS
Paper A:
Wimmer, M., Ouchterlony, F. & Moser, P. (2008). The fragment size distribution of
Kiruna magnetite, from model-scale to run of the mine. In H. Schunnesson & E.
Nordlund (Eds.), 5th International Conference and Exhibition on Mass Mining (pp.
691-703). Lule, Sweden: University of Technology.
Paper B:
Wimmer, M., Ouchterlony, F., Moser, P., Nordqvist, A. & Lenz, G. (2009).
Referenced 3D images from inside cavities and behind rings in sublevel caving. In
J.A. Sanchidrin (Ed.), 9th International Symposium on Rock Fragmentation by
Blasting (pp. 91-100). London, England: CRC Press.
Paper C:
Wimmer, M., Nordqvist, A., Ouchterlony, F., Selldn, H. & Lenz, G. (2012). 3D
mapping of sublevel caving (SLC) blast rings and ore flow disturbances in the LKAB
Kiruna mine. In G. Baiden & Y. Bissiri (Eds.), 6th International Conference and
Exhibition on Mass Mining (on CD). Sudbury, Canada: Laurentian University.
Paper D:
Wimmer, M., Nordqvist, A., Ouchterlony, F., Nyberg, U. & Furtney, J. (2012).
Burden movement in confined drift wall blasting tests studies at the LKAB Kiruna SLC
mine. Accepted for presentation at 10th International Symposium on Rock
Fragmentation by Blasting.
APPENDED REPORTS
Report 1:
Wimmer, M., Kangas, J. & Ouchterlony, F. (2008). The fragment size distribution of
Kiruna magnetite loaded from a draw point Evaluation and analysis of a full-scale
test (Swebrec Report 2008:P2). Lule, Sweden: Lule University of Technology. Basis
for paper A.
Report 2:
Wimmer, M. & Ouchterlony, F. (2008). Application of time domain reflectometry

(TDR) for block- and sublevel caving mines State-of-the-art and preliminary
laboratory shear tests (Swebrec Report 2008:P3). Lule, Sweden: Lule University of
Technology.
Report 3:
Wimmer, M. & Ouchterlony, F. (2008). Application of borehole geophysics to identify

variations in sublevel caving debris Field tests in drifts filled with ore and waste
(Swebrec Report 2008:P4). Lule, Sweden: Lule University of Technology.
Report 4:
Wimmer, M. & Ouchterlony, F. (2009): 2D image analysis using WipFrag software

compared with actual sieving data of Kiruna magnetite loaded from a draw point
Report 5:
Wimmer, M. (2010): Gravity flow of broken rock in sublevel caving State-of-the-art

Report 6:
Wimmer, M. & Ouchterlony, F. (2011). Study of burden movement in confined drift

wall blasting tests in block 12, 691 m level, Kiruna mine (Swebrec Report 2011:P1).
Lule, Sweden: Lule University of Technology. Basis for paper D.
Report 7:
Wimmer, M. & Nordqvist, A. & Ouchterlony, F. (2011). Burden movement while

blasting under constraints Tests with new gauges in block 9, 741 m level, Kiruna
mine (Swebrec Report 2011:P2). Lule, Sweden: Lule University of Technology.
Basis for paper D.
Report 8:
Wimmer, M., Nordqvist, A., Ouchterlony, F. & Selldn, H. (2012). 3D mapping of

sublevel caving (SLC) rings and flow disturbances in the LKAB Kiruna mine (Swebrec
Report 2012:P1). Lule, Sweden: Lule University of Technology. Basis for paper C.
xi
xii
CONTENTS
PART 1 EXTENDED SUMMARY ..................................................................................................................... 1

1
INTRODUCTION ........................................................................................................................................ 3
1.1
PROBLEM STATEMENT ............................................................................................................................ 5
1.2
OBJECTIVES ............................................................................................................................................ 6
1.3
SCIENTIFIC CONTRIBUTIONS ................................................................................................................... 7
BREAKAGE AND GRAVITY FLOW IN SLC ........................................................................................ 9

2.1
SUBLEVEL CAVING (SLC) .................................................................................................................... 11
2.2
BREAKAGE IN SLC ............................................................................................................................... 12
2.2.1
2.2.1.1
Small- and full-scale experiments .......................................................................................................... 13
2.2.1.2
Modeling approaches.............................................................................................................................. 18
2.2.2
2.3
Blasting into compressible materials .............................................................................................. 13
Ring blasting effects ........................................................................................................................ 19

GRAVITY FLOW IN SLC ........................................................................................................................ 19
2.3.1
Controllable factors and their effects on gravity flow ..................................................................... 21
2.3.2
Conceptual flow models .................................................................................................................. 24
SLC DRILL AND BLAST PRACTICE (KIRUNA MINE).................................................................... 27

3.1
SLC LAYOUT AND RING DESIGN ........................................................................................................... 29
3.2
CHARGING AND BLASTING ................................................................................................................... 32
BLAST FUNCTION .................................................................................................................................. 35

4.1
3D MAPPING OF SLC BLAST RINGS ....................................................................................................... 37
4.1.1
Measurement procedure on-site ...................................................................................................... 37
4.1.1.1
Photogrammetry system ......................................................................................................................... 37
4.1.1.2
Evaluation process .................................................................................................................................. 39
4.1.2
Blast results ..................................................................................................................................... 41
4.1.2.1
Initial opening blasts............................................................................................................................... 41
4.1.2.2
Scale-up of SLC rings ............................................................................................................................ 43
4.1.2.3
Categorization and discussion of breakage problems ............................................................................. 46
4.1.3
Overbreak........................................................................................................................................ 48
4.1.4
Drilling accuracy ............................................................................................................................ 52
4.2
BURDEN INSTRUMENTATION OF SLC BLAST RINGS .............................................................................. 54
4.2.1
Measurement problem ..................................................................................................................... 54
4.2.2
Dynamics of burden movement ....................................................................................................... 55
4.2.2.1
Confined drift wall blasting tests ............................................................................................................ 55
xiii
4.2.2.1.1
Instrumentation .................................................................................................................................. 57
4.2.2.1.2
Unconfined situation (test 2).............................................................................................................. 60
4.2.2.1.3
Confined situation (test 4).................................................................................................................. 61
4.2.2.1.4
Numerical modeling .......................................................................................................................... 64
4.2.2.2
4.2.3
4.2.2.2.1
Accelerometers .................................................................................................................................. 66
4.2.2.2.2
Fibre-optic zebra gauge ..................................................................................................................... 67
4.2.2.2.3
Laboratory tests ................................................................................................................................. 67
4.2.2.2.4
Field tests ........................................................................................................................................... 69
Breakage of burden ......................................................................................................................... 72
4.2.3.1
Pretests............................................................................................................................................... 72
4.2.3.1.2
Field tests ........................................................................................................................................... 73
4.2.3.1.3
Development of measuring method ................................................................................................... 73
4.2.3.2.1
Laboratory tests ................................................................................................................................. 73
4.2.3.2.2
Field test ............................................................................................................................................ 76
5.1
NBC APPROACH AND SMALL-SCALE BLASTS ........................................................................................ 79
5.2
PREVIOUS FULL-SCALE FRAGMENTATION TESTS................................................................................... 81
5.3
RECENT FULL-SCALE FRAGMENTATION TESTS ...................................................................................... 83
5.3.1
Test layout and working procedure ................................................................................................. 83
5.3.2
Sieving results ................................................................................................................................. 85
5.3.2.1
Sieving curves for 0 - 63 mm fraction samples ...................................................................................... 85
5.3.2.2
Evaluation of complete sieving curves ................................................................................................... 87
2D image analysis (WipFrag software) .......................................................................................... 89
GRAVITY FLOW ...................................................................................................................................... 93

6.1
SLC FLOW DISTURBANCES ................................................................................................................... 95
6.1.1
Occurrence of hang-up situations ................................................................................................... 95
6.1.2
Observations from 3D mapping ...................................................................................................... 96
6.2
Low-frequency (low-f) TDR measurements ........................................................................................... 73
FRAGMENTATION.................................................................................................................................. 77
5.3.3
6
High-frequency (high-f) TDR measurements ......................................................................................... 72
4.2.3.1.1
4.2.3.2
Measurement system for burden movement ........................................................................................... 65
CONCEPTUAL MODELS........................................................................................................................ 100
6.2.1
Blast dynamics .............................................................................................................................. 100
6.2.2
Disturbed gravity flow ................................................................................................................... 102
6.2.3
Similarities with, and differences, from other models ................................................................... 103
CONCLUDING REMARKS ................................................................................................................... 105

7.1
DISCUSSION AND CONCLUSIONS ......................................................................................................... 107
7.1.1
Development of new measurement techniques .............................................................................. 107
7.1.1.1
3D mapping of SLC blast rings ............................................................................................................ 107
7.1.1.2
Burden instrumentation of SLC blast rings .......................................................................................... 107
7.1.2
Understanding breakage and flow ................................................................................................ 109
7.1.2.1
Initial opening blasts............................................................................................................................. 109
xiv
7.2
7.1.2.2
Unconfined ring blasts (scale-up of SLC rings) ................................................................................... 109
7.1.2.3
Confined ring blasts (hang-ups)............................................................................................................ 111
7.1.2.4
Fragmentation measurements ............................................................................................................... 112
PROPOSED FUTURE WORK AND RESEARCH.......................................................................................... 113
PART 2 APPENDIX ......................................................................................................................................... 129
xv
xvi
FIGURES
Figure 1. Terminology and coordinate systems for an SLC blast ring, schematic horizontal xy-section
(1), vertical zx-section (2) and zy-section (3). ............................................................... xxvi
Figure 2. System description of the SLC extraction process................................................................... 5
Figure 3. SLC exemplified by the LKAB iron ore mines in Kiruna and Malmberget (by courtesy of
LKAB). ............................................................................................................................. 11
Figure 4. Impact of breakage on gravity flow and SLC performance. .................................................. 12
Figure 5. High-speed photography of burden movement (Rustan, 1970). ............................................ 13
Figure 6. Apparatus to measure movement inside compressible material versus time (Volchenko,
1977). ................................................................................................................................ 13
Figure 7. Crosscuts at different extraction states with opening method A/B, block 9, 792 m level. .... 29
Figure 8. N-S section showing change from flat (level 849 m) to silo-shaped ring designs (907 m
level). ................................................................................................................................ 31
Figure 9. Standard charge column. ........................................................................................................ 32
Figure 10. Blast design, standard SLC ring. .......................................................................................... 32
Figure 11. Survey at a draw point in SLC crosscut, hang-up situation. ................................................ 37
Figure 12. Measurement procedure on-site, opening situation (not to scale)........................................ 37
Figure 13. Measuring equipment mounted at a remote head. ............................................................... 38
Figure 14. Geometric arrangement of the laser modules. ..................................................................... 38
Figure 15. 3D image mapping at the roof inside an opening, KI-16-964-150-10. ................................ 40
Figure 16. Overview, RingCave software. ............................................................................................ 40
Figure 17. Geo-referenced, merged 3D model (wireframe view) with preceding ring outlines, ring no.
10, drift 150, 964 m level. ................................................................................................. 41
Figure 18. 3D model (rendered view) with different sections, view direction west, ring no. 10, drift
446, 935 m level................................................................................................................ 41
Figure 19. Cross-sections showing centre ring no. 5 within initial opening blasts, drift 443, 935 m
level................................................................................................................................... 42
Figure 20. Initiation of opening blast, section at 925 m, ring 1 8, drift 443, 935 m level. ................. 42
Figure 21. Burden (m) vs. nominal time delay (ms) for holes in centre rings of initial opening blasts. 42
Figure 22. Ring outlines with measured max. breakages (red cubes), ring nos. 1-23, drift 150, 964 m
level................................................................................................................................... 43
Figure 23. Planned increase in area (%) and height (m) for successive SLC rings versus broken area
(%). ................................................................................................................................... 44
xvii
Figure 24. Effective burden, ring no. 13, hole no. 5, drift 150, 964 m level. ........................................ 46
Figure 25. Categorization of breakage problems, 244 analyzed boreholes. .......................................... 47
Figure 26. Overbreak (red) and underbreak (blue), ring no. 12, drift 150, 964 m level. ....................... 48
Figure 27. Charge and initiation pattern with specific charge (kg/m3), ring no. 12, drift 150, 964 m
level................................................................................................................................... 48
Figure 28. Cumulative distribution of overbreak (OB) and affected relative area, 36 rings (openings &
hang-ups). ......................................................................................................................... 50
Figure 29. Fraction (%) of half casts in charged/uncharged part versus hole type (), 30 rings
(openings). ........................................................................................................................ 52
Figure 30. Borehole deviations and definitions of error components (not to scale).............................. 52
Figure 31. Future burden instrumentation of SLC blast rings (not to scale). ........................................ 54
Figure 32. Test site 4, 3D model (wireframe view) with blastholes (red), measurement holes (green)
and measurement points (blue cubes) within burden, (1) top-view and (2) isometric view.
.......................................................................................................................................... 56
Figure 33. Anchor, single components and force-fit mounting with pulling rod. Test 6, hole M1. ...... 57
Figure 34. Incremental displacement sensor: Coaxial cable spear, back side of the protection casing.
Test 4, hole M16. .............................................................................................................. 58
Figure 35. Continuous displacement sensors: Draw-wire and magnetostrictive sensor mounted at the
borehole collar. Test 2, hole M5. ...................................................................................... 58
Figure 36. Test site 2 with height isolines for filling. ........................................................................... 60
Figure 37. Blast result with excessive material movement, test site 2. ................................................. 60
Figure 38. Horizontal velocity and displacement, test 2. ...................................................................... 60
Figure 39. Test site 4 with height isolines for filling. ........................................................................... 62
Figure 40. Blast result with void between the blasted burden and pillar, test 4. ................................... 62
Figure 41. Horizontal velocity and displacement, test 4. ...................................................................... 63
Figure 42. Cross-sectional view of model predictions for burden swell. .............................................. 64
Figure 43. Comparison of burden velocity measurements and Blo-Up model predictions................... 65
Figure 44. Measurement system installed behind the burden (not to scale).......................................... 65
Figure 45. Main components of the measurement system. Test 6, hole M1. ........................................ 66
Figure 46. Velocity and displacement: PE and PR accelerometer compared with fibre-optic data;
abrupt stop at 150 mm. ..................................................................................................... 68
Figure 47. Velocity (v) and displacement (d): PE accelerometer compared with fibre-optic data,
suspended stop at 400 mm. ............................................................................................... 68
Figure 48. Test 5, velocity (v) and displacement (d): coaxial cable spear and fibre-optic data. ........... 69
Figure 49. Test 6, velocity (v) and displacement (d): laser distance sensor (Sick OD2-250W150) and
fibre-optic data. ................................................................................................................. 69
Figure 50. Test 5, velocity (v) and displacement (d): PE accelerometer and fibre-optic data. ............. 70
xviii
Figure 51. Test 6, velocity (v) and displacement (d): PE accelerometer and fibre-optic data. ............. 71
Figure 52. Testing machine with clamped sample, Complab, LTU. ..................................................... 74
Figure 53. Different shear failure modes............................................................................................... 74
Figure 54. Reflection coefficient versus shear displacement. ............................................................... 75
Figure 55. Fragment size distribution for NBC tests and model-scale blasts, M3 ore type. ................. 80
Figure 56. Fragment size distribution for NBC tests and model-scale blasts, M4 ore type. ................. 80
Figure 57. Energy register functions for both ore types M3 and M4. ................................................... 81
Figure 58. Average fragment size distribution for caving debris, Maripuu`s data (1968). ................... 83
Figure 59. Working procedure for full-scale sieving test. ..................................................................... 84
Figure 60. Sieving curves for 0 - 63 mm fraction samples, log-log scales. .......................................... 85
Figure 61. Average fragment size distribution for 0 - 63 mm fraction samples with Swebrec function
fit. ...................................................................................................................................... 86
Figure 62. Continuous GGS exponent curves for actual full-scale sieving and Maripuu`s data........... 86
Figure 63. Constructed, complete sieving curves for all samples. ........................................................ 87
Figure 64. Influence of image resolution. Data count (24/12/18/15/12/9). ........................................... 90
Figure 65. Segregation effects. Data count (16/7/7).............................................................................. 91
Figure 66. Hang-up statistics, year 2002 - 2011. ................................................................................... 95
Figure 67. Hang-up observation (type B) with waste lenses, drift 135, 792 m level. ........................... 97
Figure 68. Section at 776.8 m and projection of waste lens from ring 7, drift 135, 792 m level. ......... 97
Figure 69. Plausible sequence of events in the blasting of a SLC ring (section perpendicular to ring
plane at 20 m height, not to scale). ................................................................................. 101
Figure 70. Conceptual model, shallow draw (vertical section). .......................................................... 102
Figure 71. Conceptual model, temporary hang-up (vertical section). ................................................. 102
xix
xx
TABLES
Table 1. Experimental work: Blasting into compressible materials, chronology (1/3). ........................ 15
Table 4. Modeling approaches, models A - H: Blasting into compressible materials, chronology. ..... 18
Table 5. Full-scale observations of operational changes and their effect on material flow. ................. 23
Table 6. Conceptual models of gravity flow mechanisms in SLC. ....................................................... 25
Table 7. Overview of different opening methods. ................................................................................. 30
Table 8. Correlation matrix, blast result (overall analysis), 30 rings (openings). ................................. 45
Table 9. Relative broken ring area (overall analysis), statistics for F-test, 30 rings (openings). .......... 45
Table 10. Overbreak (ring face/brow area), statistics for F-test, 36 rings (openings & hang-ups). ...... 49
Table 11. Correlation matrix, overbreak (spatial analysis), 8 rings (openings) with 3606/995 data
points................................................................................................................................. 49
Table 12. Overview of drift wall blasting experiments. ........................................................................ 55
Table 13. Blast design, test 2 and 4 with blastholes B2 B5................................................................ 56
Table 14. Overview on different tested sensors. ................................................................................... 59
Table 15. Previous fragmentation measurements for blasted rock in Kiruna........................................ 82
Table 16. Surveyed hang-ups classified according to type (vertical section)........................................ 96
Table 17. Data for surveyed hang-ups. .................................................................................................. 98
Table 18. Data for mapped flow lenses of previously extracted rings. ................................................. 98
xxi
xxii
KEYWORDS, ACRONYMS AND SYMBOLS
Keywords and definitions

Brow, edge of the roof in a production drift, defined as 4 m wide arc above the roof
Compaction, (%), decrease in volume for the caving debris
Flow zone, volume inside which material moved during extraction
Hang-up, temporary stop in gravity flow caused by boulders or material arching in/above the draw point
Opening blast, initiates the mining of a new draw point closest to the hanging wall, consists of several rings fired
in one blast.
Opening method, describes the procedure to open a new draw point, i.e. amount and placement of opening
blasts.
Overbreak, (m), the outfall of rock behind the planned contour. Specifically, for SLC, the overbreak to the
subsequent blast ring may also be termed backbreak. Overbreak values are defined as negative in tables/figures.
Scale-up of SLC rings, rings are sequentially scaled up in height and width between an initial opening blast and
another opening blast in the same drift connecting to the caving debris above
Swell, (%), increase in the volume of burden rock due to fragmentation and compaction of caving debris relative
to the planned burden
Underbreak, (m), the rock which remains unbroken inside the planned contour. Underbreak values are defined as
positive in tables/figures.
xxiii
Acronyms (used in the extended summary)

ANFO
ammonium nitrate and fuel oil
Blo-Up
software component within HSBM
CAD
computer-aided design
DRAW
draw-wire sensor, see Table 14
DTH
down-the-hole hammer
FIBRE
fibre-optic zebra gauge, see Table 14
GGS
Gates-Gaudin-Schuhmann function
Giron
LKAB`s production database
HSBM
Hybrid Stress Blast Model
ICS
International Caving Study project
KR0500
Kimulux Repumpable 0500, bulk explosive product from Kimit AB
LASER
laser distance sensor or light barrier, see Table 14
LHD
load-haul-dump unit
MAGNET
magnetostrictive sensor, Table 14
MMT
Mass Mining Technology project
NBC
natural breakage characteristics
OCS
optimum comminution sequence
PE
piezoelectric accelerometer, Table 14
PETN
pentaerythritol tetranitrate
PR
piezoresistive accelerometer, Table 14
REBOP
software for rapid simulation of the flow of fragmented rock in cave mines
RFID
radio frequency identification
RingCave
software to visualize and evaluate blasting results from SLC rings
RMS
root mean square
SLC
sublevel caving
SLP-50
Sliding Primer-50, packaged explosive product from Kimit AB
SMX3D
software for 3D acquisition (photogrammetry) and assessment of rock surfaces
SPEAR
coaxial cable spear, see Table 14
TDR
time domain reflectometry
VLF
very low frequency
xxiv
List of symbols (used in the extended summary)

Ab, (m)
broken area of an SLC ring
Ai, (m)
area increase from one SLC ring to the next
As, (cm2/g)
specific outer surface
B, (m)
burden, i.e. shortest planned distance between centre line of hole and free surface
Beff, (m)
burden, effective i.e. shortest distance between centre line of hole and free surface
Bmax, (m)
burden, maximum i.e. largest distance between centre line of hole and free surface
d, (mm)
displacement
Es, (J/g)
net energy consumption
Hi, (m)
height increase from one SLC ring to the next
n, (-)
uniformity index in the Kuz-Ram model (xc = x50/(ln2)1/n)
OB, (m)
overbreak
PPV, (vmax), (mm/s)
peak particle velocity
q, (kg/m3) or (kg/t)
specific charge, consumption (planned or actual) of explosive per cubic metre or metric tonne of
rock
R, (cm2/J)
Rittinger coefficient
, (-)
reflection coefficient, i.e. amplitude ratio of reflected wave/incident wave
S, (m)
spacing, i.e. distance between adjacent holes in a row, and in the case of SLC a ring
UB, (m)
underbreak
v, (m/s)
burden velocity
VOD, (vd), (m/s)
velocity of detonation
xc, (mm)
characteristic size, size where 63.2 % of the particles pass
xmax, (mm)
maximum size
x10, x25, , (mm)
percentile sizes. x10 is the size of square sieve opening through which 10 % by weight of the
sample would pass.
x50, (mm)
median fragmentation size, size where 50 % of the particles pass
xxv
Conventions
The terminology and coordinate systems related to an SLC ring are explained in Figure 1.
Figure 1. Terminology and coordinate systems for an SLC blast ring, schematic horizontal xy-section (1),
vertical zx-section (2) and zy-section (3).
The mine coordinate system is shown in black. The East direction of the geodetic system (green) is rotated 10.7
relative to the x axis and towards the y coordinate of the mine coordinate system. In addition, a local coordinate
system (H/V) perpendicular to the planned or fitted ring plane is introduced (blue).
Stereographic projections are based upon the mine coordinate system and thus reflect the relative orientation of
geological structures to the ring faces.
xxvi
PART 1
EXTENDED SUMMARY
INTRODUCTION
The framework of the thesis is presented. The focus of the thesis is based upon the problem statement,
the main objectives, and a specification of scientific contributions.
Introduction
1.1
Problem statement
Sublevel caving (SLC) is a highly productive mining method with the major disadvantages of
irrepressible ore loss and dilution (Bull & Page, 2000). The SLC extraction process may be simplified
as in Figure 2. At the heart of the sublevel caving (SLC) process lie the three core elements: blast
function, fragmentation, and gravity flow.
Figure 2. System description of the SLC extraction process.
For the ore recovery to be as high as possible and for the waste rock dilution to be as low as possible
all three elements have to work as planned. These elements are affected by a number of given factors
and controllable factors. The effect of rock mass characteristics and a semi-confined blasting situation
are largely unknown. Rock mass characteristics are mainly the intact properties and rock mass
discontinuities of blasted ore and host rock, and in-situ stresses. The controllable factors are related to
mine planning (SLC layout and ring design), charge and initiation pattern, performance of drill and
blast work, and mucking (draw control).
The initial conditions for the ore flow after blasting are unknown, as both the blasted geometry and the
fragmentation itself are hidden. For example, without special tools it is not possible to say:
x How large the burden and spacings for the blastholes are,
x if all blastholes have effected complete breakage, or caused overbreak,
x if the detonators and charge columns in the next charged ring are undamaged,
5
Introduction
x what ore fragmentation and mobilization look like, and,

x if the ore being loaded is that which was blasted in the last ring.
All three elements, i.e. blast function, fragmentation, and gravity flow have been investigated in R&D
projects for a long period of time. Interest is still great as ore recovery, dilution, and flow disturbance
are direct consequences of flow behaviour. The use of best available technology (BAT), for instance in
drill and blast work, has resulted in increased mining productivity and has decreased development and
mining costs. The scale has increased tremendously, at the LKAB Kiruna mine, for example from a 12
m sublevel height in 1983 (Hustrulid, 2000) to 28.5 m in 2012. The SLC layout and ring design was
altered in 2006, see section 3.1. At the same time, mining is carried out at greater depths at which the
mining method has not previously been tested. In-situ and confining stresses increase at these depths
and this affects the blast performance. For these reasons it would be expected that flow behaviour has
changed during the years. The difficulties in simulating SLC relate to the physical scale as well as to
the time-scales involved: this results in a large number of unknowns and uncertainties. Measurements
are therefore indispensible, but difficult to obtain.
1.2
Objectives
The work presented in this thesis started with a project (Improved breakage and flow in SLC)
financed by the Hjalmar Lundbohm Research Centre (HLRC). It is part of the focus area Mining
methods with the following long-term objective:
To achieve a disturbance-free fragmentation and flow in sublevel caving, which leads to an increase
in ore recovery and a decrease in waste rock loading and nitrates leakage. Such an achievement would
rely on having a reliable breakage and flow model for the SLC, at least a reliable conceptual one, as
the existing conceptual ones have a limited experimental basis.
However, before a better conceptual model for SLC could be constructed, a better, experimentally
supported understanding of breakage and flow is required. The latter is the main objective of the work
presented in this thesis. Tests with newly developed and old, proven measuring techniques are made to
support the conceptual models that are presented.
Introduction
1.3
Scientific contributions
Field measurements in full-scale conditions are few world-wide and the hard evidence behind the
existing conceptual breakage and flow models often insufficient. To improve this basis for the
understanding of the SLC extraction process all three core elements have been investigated, see Figure
2.
The present project contributed with the following development of new measuring techniques and
better understanding of breakage and flow in SLC:
1. Development of new measuring techniques for use under SLC conditions
x
A camera crane for the acquisition of scaled and referenced 3D images of cavities, see
section 4.1.
A rugged embedded displacement gauge (fibre-optic zebra gauge) for dynamic

measurements inside the burden, see section 4.2.2.2.
The sensitivity of TDR cables under shear and tensile loading, see section 4.2.3.2.
2. Understanding breakage and flow in SLC

x
Quantification of breakage and flow conditions inside SLC cavities in opening and hangup conditions, see section 4.1 and section 6.1.
Quantification of the whole burden movement with its acceleration and later retardation
when blasting against the confinement of filled and reinforced drift walls, see section 4.2.
Demonstration that the fragmentation of magnetite ore from the Kiruna mine has NBC
(natural breakage characteristics) properties and has the same general blasting
characteristics as normal hard rock, see section 5.3.
The verification of the existence of the shallow draw phenomenon, see section 6.1.
A more detailed conceptual model for SLC breakage and flow, which has a better
experimental basis, see section 6.2.
BREAKAGE AND GRAVITY FLOW IN SLC
Research on breakage and flow in SLC caving mines has proceeded over a long period of time. In
recent years, in particular, international research cooperation (e.g. ICS, International Caving Study and
MMT, Mass Mining Technology projects; Chitombo, 2010) contributed to increase knowledge in the
field of mass mining technology. Some fundamental questions have remained throughout the years,
however, and there have been additional technical and economical challenges. In the next decade the
production of many open pits needs to be replaced by large-scale, low-grade cave mines and existent
operations may face new problems as they are mined at greater depths.
A separate report (Report 5) concerns the general state of knowledge on gravity flow and the present
chapter reviews the effects of confined blasting on breakage and gravity flow. The measurement
methods developed and used are specified.
10
Breakage and gravity flow in SLC
2.1
Sublevel caving (SLC)
Sublevel caving is a mass mining method based upon the utilization of gravity flow of blasted ore and
caved waste rock (Kvapil, 1998). It relies on the principle that ore is fragmented by blasting while the
overlying host rock fractures and caves under the action of mine induced stresses and gravity. Thereby
the caved waste originating from the overlying rock mass fills the temporary void created by ore
extraction. The method itself was initially applied in the early 1900s to extract soft iron ores found in
Minnesota and Michigan (Cokayne, 1982). At that time heavily timbered drift support was
sequentially removed at the end of a drift initiating caving of the ore and then the ore to be slushed
out. When dilution became excessive the next set of timbers was removed, and so on. Today many
uncertainties about fragmentation and ore cavability are eliminated since the ore is drilled and blasted
from drifts on successive sublevels. Breaking the ore by blasting removes the dependency on existing
fractures and joints as the mechanism for ore breakage, which dependency is shared by most other
methods of caving. For this reason SLC is strictly speaking not considered a caving method any longer
as far as the ore is concerned, but SLC does rely on the walls caving and thus the name is retained. As
practiced today the method should probably be given another name, such as sublevel retreat stoping,
continuous underhand sublevel stoping or something similar that better reflects the process (Hustrulid,
2000). SLC is nowadays usually applied in hard, strong ore in which the hanging wall progressively
caves, keeping pace with the retreating rings.
Key layout and design considerations are the achievement of high recovery with an acceptable amount
of dilution. Current SLC geometries, Figure 3, consist of a series of sublevels created at intervals of
between 20 and 30 m beginning at the top and working down.
Figure 3. SLC exemplified by the LKAB iron ore mines in Kiruna and Malmberget (by courtesy of LKAB).
11
A number of parallel drifts are excavated on each sublevel with drifts being offset sideways between
the individual sublevels. From each drift vertical or near vertical rings of boreholes are drilled upward
to the overlying sublevels. A new draw point is initiated closest to the hanging wall by an opening
blast (see section 3.1). Once contact with the level above is established, the rings are then blasted one
by one against the material lying in front consisting of ore and caved waste. The burden between the
rings is about 2 3 m. The extraction of ore from the blasted ring continues until dilution or some
other determining measure reaches a prescribed level. Thereupon the next ring is blasted and the
retreating process continues. Depending on the thickness of the orebody the technique may be applied
using transversal or longitudinal retreat.
2.2
Breakage in SLC
Blasting in sublevel caving (SLC) has been identified throughout the literature as having a significant
impact upon material flow characteristics and therefore on the overall performance of the SLC
method, see Figure 4. Brunton (2009) gives a comprehensive literature review.
Figure 4. Impact of breakage on gravity flow and SLC performance.
Blasting in SLC takes place in a situation with varying confinement. Blasted material swells while the
caved material compacts, and also, to a lesser extent, fills parts of the void of the production drift.
For SLC, blasting occurs under varying confined conditions. It may be described more precisely as
blasting into compressible materials being in-between the extremes blasting into a void volume
and blasting into non-compressible materials. The impact of void volume on blast performance was
investigated by a number of authors (Fedorenko & Kovtun, 1977; Jarlenfors & Holmberg, 1980;
Gorham-Bergeron et al., 1987; Olsson, 1987). The effect of blasting into non-compressible material
resembles a de-stressing or preconditioning of the blasted rock masses. For SLC, all phases between
the extremes are conceivable, but under regular production conditions a blasting situation into
compressible materials, i.e. caved rock, is most likely.
Blast function and fragmentation are dependent upon several parameters (see Figure 2). Given factors
are on the one hand properties of both the blasted and compressible material and confinement.
12
Controllable factors on the other hand include the SLC layout/ring design, explosives/initiation and
drill and blast work. A limited number of physical experiments and models were deployed in the past
to understand the blast function and to quantify fragmentation.
2.2.1 Blasting into compressible materials
2.2.1.1
Small- and full-scale experiments
An overview on both small- and full-scale experiments with respect to setup and results is shown in
Table 1 - Table 3.
Small-scale experiments primarily investigated the swell/compaction and fragmentation when blasting
into compressible materials (Belen`kii et al., 1969; Rustan, 1970 & 2012; Cullum, 1974; Volchenko,
1977; Kirpichenko, 1982; Zhang, 2004; Johansson, 2011; Petropoulos, 2011). While most of the
experiments considered only the final results, some also contained the study of burden movement. As
measurement methods, high-speed photography was applied as one side of the confined blasting
model was locked by plexiglass (see Figure 5). The interface between sample and compressible
material could be well identified and the swell of blasted material be mapped. Volchenko (1977)
determined the compaction zone with metal plates inserted in the compressible material (see Figure 6).
Telescopic tubes reached into different zones of the compressible material and transferred its
compaction to the outside of the blasting box. Time-distance data was received from the detection of
cable breaks.
Figure 5. High-speed photography of burden

1
movement (Rustan, 1970).
Figure 6. Apparatus to measure movement inside

compressible material versus time (Volchenko,
2
1977).
Swell of 50 % at time = 1.46 ms. Sample in scale 1:50, ring front angle 85, 4 holes with B = 24 mm, S/B = 1.
Dark rock pieces were inserted in the caving debris (limestone) to identify any movements therein.
2
Notation in figure: 1: blasting box, 2: sample, 3: first row of holes, 4: measurement arrangement, 5: reference
tags, 6: wire gauges, 7: strikers, 8: telescopic tubes. (Volchenko`s terminology).

13
Measurements of burden movement by Petropoulos (2011) on a small-scale were made on the basis of
the methods developed, and used in full-scale tests before (see section 4.2.2.1.1).
Blasting models incorporating the study of gravity flow were made by Rustan (1970 & 2012) and
Zhang (2004). The initial conditions are much more realistic than if gravity flow is studied in
simplistic bin models with homogenous material parameters. Many of the phenomena observed in fullscale studies such as overbreak and variations in waste rock content during the extraction flow could
also be recognized in these models. However, there are still discrepancies relating to particle size,
relative sizes between containers and particles, and boundary conditions such as high overburden
pressures in SLC.
Full-scale measurements are considered important in understanding blasting mechanisms in confined
situations and its resultant material flow behaviour, but few have been made. An experiment was
mentioned by Kirpichenko (1982), but no details of the measurement method were given. Newman et
al. (2008) measured rock mass damage and the final swell when blasting an ore slice against caved
material in a drift at Kiruna mine. Rubber mats were installed at the side wall of the drift prior to
production and prior to the experiment. With the help of diamond drill holes cored through the pillar
from the adjacent drift prior/after the experiment both damage and the pillar/material interface could
be monitored.
In summary, small- and full-scale experiments indicate that swell and fragmentation is negatively
affected when blasting under confinement: swell becomes limited and fragmentation gets coarser. The
dynamic behaviour relates to a number of parameters including SLC layout and ring design,
explosives and initiation, and the material properties of the blasted and compressible material.
14
Test
Small-scale
Meas.
Authors
Small-scale
Small-scale
Final result
Dynamic + final result
Final result
Belenk`ii et al. (1969)
Rustan (1970),
Rustan (1990), Rustan (2012)
Cullum (1974)
Impact of blast layout on fragmentation, swell and gravity

flow
SLC geometry, 1:75, LKAB blast ring
Magnetite mortar
Limestone, scaled fragmentation
SLC ring design, 12 holes, burden 24 mm
PETN in powder form
In order to achieve scaled fragmentation artificial weakness

planes (microscope glass plates) were inserted in the sample.
Markers were installed between blast rings.
Burden dynamics were studied by high-speed photography.
Impact of increased confinement on fragmentation and

swell
Description
Sample
Material
Buffer
Design
Explosives
Remarks
Description
Remarks
15
PETN, 0.91 kg/m3

Blast rows were retreated from a starting slot. As each slot
was fired, confinement for the next ring increased.
Remarks
Explosives
Detonating cord (DSH-A)
Explosives
Rows with 3-4 holes, S/B = 1.1
Parallel and ring shaped hole arrangements
Design
Buffer
Design
Crushed martite ore, defined fragmentation
Buffer
Cuboids
Concrete
Fragmentation coarsened with increased confinement level. No trend was observed

for the bulk density of the blasted material with increased confinement. It was
concluded that the bulk density of the blasted material is dependent on the method of
initiation, model size and rigidity.
It was observed that additional blasting slots with void ratio of 27 % and 13.5 %,
located each six rings back of the starting slot, could improve fragmentation. The
parameter x50 became gradually coarser moving away from the starting slot, became
significantly finer after the blasting slot and then gradually became coarser again. The
void ratio of the blasting slot had no impact on the fragmentation.
Martite ore
Material
Sample
Material
The finest fragmentation (x80) was achieved for inter-hole delays of 0.1-1.0 ms.
Maximum swell (40 %) was achieved for a delay of 0.1 ms. An inverted parabolic
relation existed between pressure exerted by the compressible material and blast
fragmentation x80, with the coarsest x80 obtained for a pressure of 800 kg/m2. An
increase in the pressure also resulted in less horizontal swell of the blasted material.
Burden dynamics were studied for confined bench model blasts. Max. velocity was
about ~24 m/s (after 1 ms) with final halt after 2.25 ms. Max. acceleration was ~40
m/s2 (after 0.75 ms) and retardation ~24 m/s2 (after 2 ms).
Several markers were found on the muck pile directly after blasting. Some waste rock
fragments were found on the muck pile directly after blasting. Extracted markers
primarily derived from the mid-part and closest to the blasted ring plane.
Cuboids (400400500 mm)
Sample
The level of compaction at any given distance from the blast increased, as the
compressible material size fraction became smaller. It was argued that this is due to
an increase in the specific volume of cavities as the compressible material became
finer. The depth of compaction of the compressible material occurred to a greater
depth for the ring hole geometry as compared to the parallel arrangement of holes.
Data from small-scale tests correlated well with full-scale blasts at mines (GigantGlubokaya).
Impact of blast design on compaction of compressible

materials
Results
Description
Setup
Table 1. Experimental work: Blasting into compressible materials, chronology (1/3).
Test
Small-scale
Small- & full-scale
Full-scale
Meas.
Authors
Final result
Volchenko (1977)
Kirpichenko (1982)
Newman (1997),
Newman et al. (2008)
Cuboids (120100200 mm and 300275245 mm)
Iron ore and concrete
A variety of compressible materials
Three parallel holes per row, S = 5.5 8.0 cm, B = 4.5 6.5
cm
1.4 g PETN / hole
Inter-hole and inter-row initiation timing varied. The zone of

compaction was determined by metal plates fixed in the
compressible material. An apparatus measured the burden
velocity of a telescopic tube as detection of cable breakages.
Measurement of movement of the blasted material
n.a., scale 1:100
Concrete
Crushed concrete
Row with parallel blast holes and single-hole shots
PETN
No details of the full-scale test are given.

High-speed photography of burden movement.
Measurement of burden swell and damage

Drift wall blasted against caved rock masses
Magnetite
Caved rock masses filling a drift
3 holes ( 115 mm), drilled within a vertical plane, fanning

about a horizontal plane, blasted at a burden of 2.5 3.5 m
KR0500
12 monitoring holes were diamond drilled from the adjacent

drift to document damages prior and after blasting.
Sample
Material
Buffer
Design
Explosives
Remarks
Description
Sample
Material
Buffer
Design
Explosives
Remarks
Description
Sample
Material
Buffer
Design
Explosives
Remarks
16
Impact of compressible material bulk density, blast hole

spacing/burden and initiation on swell and fragmentation
Description
Setup
Swell was measured to be 2 17 %.

Induced cracking was noted up to 1.7 m behind the burden.
Major damage, classified as an area of crushed core, was verified at a distance of
0.5 m behind the blast hole plane near the centre of the explosive charge.
Secondary damage was identified as a 20 - 50 cm wide zone, located between the
major damage plane and the pillar wall.
Full-scale experiments at the Tashtagol mine indicated a movement of the blasted

material of 1.6 2.3 m. It was at a maximum at the centre and at a minimum at
contacts with unbroken rock.
Small-scale experiments showed that with increased specific energy, an increased
swell of the burden occurred. Similarly the velocity, which was measured by highspeed photography, increased.
The displacement from a row with parallel blast holes was smaller than if a single
hole with equivalent diameter was blasted. It was argued that the energy would be
utilized more effectively when a single hole is used
Fragmentation varied dependent on bulk density of the compressible material, S/B

ratio and initiation timing. The finest fragmentation (x50) and lowest percentage of
oversize (> 40 mm) was achieved for an S/B ratio of 1.4 and a specific initiation
scheme (wave-2). Unfortunately, no details of the initiation scheme are given.
Swell of blasted material decreased and depth of compaction increased with
increasing bulk density of the compressible material. Greatest swell and compaction
was achieved for row by row initiation and a S/B ratio of 1.78.
Data from small-scale tests correlated well with full-scale blasts at mines
(Kuzmetkombinat group).
Results
Test
Small-scale
Meas.
Authors
Small- -scale
Small-scale
Final result
Zhang (2004)
Johansson (2011)
Petropoulos (2011)
Fragmentation became coarser for confined blasts. The properties of the debris had
a strong influence on the fragmentation and swell of the blasted
material/compaction of the compressible material.
A freezing-slicing method allowed to measure radial expansion at different heights
of the cylindrical sample.
The compressible material was described by means of acoustic impedance. This
value in combination with specific charge showed a great influence on
fragmentation and compaction. For compaction, the porosity of the compressible
material was found to be another influencing factor.
Prediction equations, both for fragmentation and compaction were presented.
The study of timing effects showed no distinct differences or improvements on
fragmentation when comparing delays in the time range of shock wave interaction
to no shock wave interactions. The parameter x50 decreased around 20 % at a delay
time ~ 1.1 ms/m burden compared with longer delays (2 ms/m). The minimum at
short delays was not found to be statistically significant.
Fragmentation was coarser under confined situations. It becomes finer for longer
delays (> 4.1 ms/m of burden) than for very short delay times. The fragmentation of
compressible material was altered as blast parameters (specific charge, delay time,
firing pattern) were changed.
Different methods were tested to measure burden dynamics.
Maximum burden velocity was around 29 m/s. A rebound effect of a couple of
millimeters was observed before its final halt.
The V-shaped firing pattern gives higher compaction by 15 % due to differential
movement of the two parts of the burden.
SLC geometry, 1:50, SLC interval 10 m, drift spacing 10 m

Concrete
Crushed rock
SLC ring design, two rows with each 7 holes, 60 mm burden
PETN detonating cord

Coloured model to represent different heights of the ring.
Effects of confinement and initiation delay on fragmentation

and compaction
Cylinders, 140x280 mm
Cuboids 650/660205300 mm (timing effects set-up)
Magnetite and magnetic mortar
4 different granular materials (crushed granite, non-magnetic

mortar) with defined properties and fragmentation
Cylinders with centered hole ( 5 - 12 mm)

Cuboids with 2 rows and each 5 holes ( 10 mm); S/B = 1.6
PETN cords of different strengths
Confinement was made by a steel cylinder, inner diameter 309

mm. To create a surrounding pressure from the material to the
sample a slotted steel cylinder was used.
For the cuboids a U-shape yoke made from high-strength
concrete was used to minimize the reflection of waves.
Accelerometers were installed at the burden-buffer interface.
Influence of different blast parameters (specific charge,

delays, firing pattern and confinement) on fragmentation
Cuboids, 660270210 mm
Magnetic mortar
Crushed non-magnetic mortar, defined size distribution

2 rows, 5 holes (S/B = 1.6) & 3 rows, 7 holes (S/B = 1.4)
PETN detonating cords, 2.6 kg/m3 (5 holes) or 4.1 kg/m3

Set-up was similar to cuboids blasted by Johansson (2011).
Sample
Material
Buffer
Design
Explosives
Remarks
Description
Sample
Material
Buffer
Design
Explosives
Remarks
Description
Sample
Material
Buffer
Design
Explosives
17
Fragmentation was finest at the bottom and became progressively coarser towards
the top of the ring. The blasted material swelled both vertically and horizontally.
Inter-mixing of blasted and compressible material was found to occur at the
interface between the two materials.
The draw body was of complex shape with a volume much smaller than the actual
blasted volume. It was pointed out that in order to solve the ore dilution problem
associated with the extraction process, the blast and ore extraction processes need
to be treated together.
Effects from confined blasting and gravity flow
Description
Remarks
Results
Setup
2.2.1.2
Modeling approaches
Table 4 gives an overview of analytical methods developed to study blast dynamics (model A - C) and
fragmentation (model D - G) when blasting into compressible materials. The empirical data on which
the models are based upon is very limited, and does not take into account a wide-range in blast designs
and different rock mass properties.
Recent developments in a 3D numerical model (Furtney et al., 2009) which also incorporates the link
between blast dynamics and fragmentation (model H) are promising. The model was validated with
the help of a series of unconfined test blasts (Onederra et al., 2009). Brunton & Chitombo (2009)
applied it to model the impact of SLC blast design and performance on material recovery. Its built-in
flex boundary logic, representing the caved material as non-linear hysteretic spring, was recently
validated against empirical data from confined drift wall blasting tests (see section 4.2.2.1.4).
Table 4. Modeling approaches, models A - H: Blasting into compressible materials, chronology.
Authors
Model type
Description
Markenzon
(1967)
Semiempirical
A three-stage model was proposed, predicting the swelling of the blasted ring,
compaction of the compressible material and creation of a temporary void behind
the blast (before relaxation of the blasted material and caving debris filling the void
again). The swell factor was assumed to be a constant (1.15). The compaction was
based upon a relationship proposed by Imenitov & Pustovalov (1964, cited by
Brunton, 2009). The temporary void was simply calculated by subtracting the swell
of the blasted material from the compaction.
Alford
(1978)
Semiempirical
Material expansion factors were assigned for various regions in an SLC ring.
Compaction of the compressible material was assumed to be linearly related to the
dimensions of the blasted zone.
Abramov &
Gorbunov
(1978)
Energy based
The movement of blasted material is predicted based upon input parameters

considered important for confined blasting, such as blast geometry, explosive
properties and material characteristics. The model is difficult to use due to the large
number of input parameters required which are difficult to assess.
Preston
(1995)
Stochastic
Explosive and rock mass properties were used to derive a stress attenuation curve.
Based upon this curve a maximum and minimum breakage region was calculated.
The fragmentation distribution was calculated using a modified Kuz-Ram approach.
Lith et al.
(2004)
Modified KuzRam
Fragmentation data from the LKAB SLC operation were used to develop this
model. The calculations of rock factor and uniformity index were modified to
consider factors unique to SLC blasting.
Onederra
(2004)
Two-comp.
mechanistic/
empirical
fragmentation
Scott (2004,
cited by
Brunton,
2009)
Furtney et
al. (2009)
A mechanistic approach was used to predict the fines, while a peak particle velocity
(PPV) criterion was adopted to predict the coarse end of the fragmentation
distribution. Empirical data from the Ridgeway SLC operation were used.
Energy based
A modified form of the crush zone fragmentation model (Kanchibotla et al., 1999)
was used to predict the fragmentation distribution for various parts of the SLC ring.
The fragmentation distribution of the entire ring was calculated using a weighed
average of each fragmentation region.
Continuous/dis
continuous
numerical
The key processes occurring in non-ideal detonation, rock fracturing and muck pile
formation are represented by a combination of continuous and discontinuous
numerical methods. A flex-boundary condition was developed to model the
behaviour at the boundary between intact burden and caved material in an SLC ring.
The compressible material is expected to show compaction behaviour in which the
modulus increases with displacement and unloading occurs along a stiffer modulus.
18
2.2.2 Ring blasting effects

Nitrate leakage measurements and function control measurements that are made at regular intervals in
the LKAB Kiruna mine indicate that 10 - 15 % of the holes within a single blasted ring do not detonate
as planned (Hedstrm, 2000; Fjellborg, 2002; Zhang, 2005).
If the holes then fail to break out, the breakage front becomes uneven. This provides an irregular
burden for the subsequent ring blast. Uneven breakage then probably causes an uneven fragmentation,
and, in extreme cases, fractured, but immobilized pillar remnants that obstruct a regular caving flow.
Verifying the actual reasons for poor breakage is important, but normally access to study effects from
ring blasts is not possible. For this reason, measurements are sparse and difficult to perform.
A dedicated test area in the LKAB Malmberget mine with blasts into an open stope allowed taking
photographs in steps from the bottom to the top of 3 ring faces (Rustan et al., 1988; Rustan, 1993).
The projected burden was 2.5 m and the ring front angle 80. A blast ring consisted of twelve
blastholes ( 104 mm) which were 14 19 m long and charged with ANFO (ammonium nitrate and
fuel oil). Within this test campaign only one or two holes were initiated in a ring and its damage to
adjacent holes studied. The interpretation of 2D images stitched together led to revised design rules.
To obtain undisturbed detonations the minimum distance between charged holes was required to be
larger than 2.5 m (1.5 m for side holes). To avoid unbroken parts the uncharged length should be less
than 1.7 times the burden. The number of boreholes could thereby be reduced from 12 to 8.
Recently the study of SLC ring blasting effects in a more controllable environment was tried in the
abandoned open pit at Svappavaara (Olsson et al., 2009). For two tests, parts of an upside down
standard production ring at Kiruna mine (see section 3.1) were drilled behind a bench face. In another
two tests, burden and spacing were varied to identify maximum breakage conditions. The bench was
~100 m long with heights of 10 20 m. About 40 holes were drilled ( 115 mm) and charged with
emulsion (KR0500). The geological conditions made a meaningful interpretation of the blast results
difficult. Maximum overbreak occurred for the ring blasts. The burden could be increased to 3.5 m,
but fragmentation was coarser. The maximum breakable spacing was found to be 7 m for the 3.0 m
burden. High-speed photography of the blasts indicated free-face velocities to be in the range 13 21
m/s.
2.3
Gravity flow in SLC
Material flow behaviour is complex in nature, being controlled by the interaction of a wide range of
factors (Rustan, 2000). Fundamental mechanisms and factors influencing flow behaviour of broken
rock were investigated in small-scale storage bin models throughout the years. Mine-based
experimental models were set up to model specific situations (Sjstrand, 1957; Koppanyi, 1960;
Finkel & Skalare, 1963; Airey, 1965; Janelid & Kavpil, 1966; Free, 1970; Janelid, 1972; Tessem &
19
Wennberg, 1981). For example, the impact of a specific mine layout and its orebody on material flow
behavior and subsequent ore recovery/dilution was quantified. Numerous limitations associated with
small-scale experiments related to issues of similitude (Free, 1970; Sandstrm, 1972; Alford, 1978;
Gustafsson, 1998; Power, 2004a), model design, and properties of ore and waste (Sandstrm, 1972;
Cullum, 1974; Panczakiewicz, 1977; Alford, 1978; Yenge, 1981; Stazhevskii, 1996; Hustrulid, 2000)
were discussed in the literature. Despite these limitations this work has provided quantitative and
qualitative results that have been applied in the design and operation of SLC mines over the years.
Further, due to the importance of material properties on flow behaviour the future direction of SLC
small-scale flow modeling needs to incorporate blasting (Rustan, 2000).
Full-scale monitoring of SLC material flow is normally done with the help of markers, i.e. objects
with unique identification numbers that are placed inside holes drilled in the burden. The markers
might be visually recovered at the draw point, magnetically separated during the material handling
process or automatically detected by RFID (radio frequency identification) technique (Brunton, 2009).
Details about the development and final shape of the extraction zone are obtainable based upon the
recovered markers. To follow up the flow movement in real time, synthetic instrumented rock pieces
were developed to sense rock flow and transmit their actual positions via VLF (very low frequency)
communication methods in real time (Baiden et al., 2008). There are, however, no actual test results
from a mine. There was also an earlier development of instrumented boulders (Hanisch et al., 2003) to
study the dynamics of debris flow on mountain slopes. The authors claim that the configuration and
the redundant layout of the built-in 3D accelerometers and the differentiation and combination of
translational and rotational movements could be computed with reasonably stable long-time behavior.
There are no field data to support these claims though.
Visual observations, which allow a sporadic insight into the caving flow (Selldn & Pierce, 2004;
Power, 2004a-b) are very important. The 3D imaging system pursues this approach, see section 6.1.2.
Production figures (extraction, recovery, fragmentation) are an indication of the status of gravity flow.
Although, the internal processes remain unknown, production figures could be suited to analysis of the
impact of operational changes, see section 2.3.1.
Based upon systematic research and sporadic observations a number of different conceptual models
were developed in the past and are presented in the following, see section 2.3.2.
A series of mathematical models evolved during the years addressing phenomena observed in smalland full-scale gravity flow research. An overview is provided in Report 5. The key motivation for
development of these models was to simulate the effects of different geometries and draw control on
the economic performance of SLC operations.
20
2.3.1 Controllable factors and their effects on gravity flow

Apart from the given factors, rock mass characteristics, and the semi-confined blasting situation, the
controllable factors (Figure 2) related to design issues and influencing breakage and flow in SLC are
discussed in the following. Layout criteria concerning overall SLC geometry, ring design, explosive
properties and timing are commonly based on site experience. A number of authors have outlined
guidelines for the design of SLC blast rings (Bull & Page, 2000; Hustrulid, 2000).
Table 5 gives an overview of full-scale observations of operational changes and their effect on
material flow. Due to site specific characteristics generalized conclusions are often difficult to make.
Foremost, the ring design (Table 5) is assumed to have a major impact on material flow. A design with
shorter and flatter side holes proved over the years to be impractical. Marker trials revealed that
interaction between adjacent extraction zones did not exist for the conventional fan-shaped ring design
(Larsson, 1998; Hollins & Tucker, 2004; Power, 2004a-b). To meet this concern, the layout at Kiruna
mine was changed to a silo-shaped layout, see Figure 8.
The number of blastholes, i.e. specific charge, is another important influencing factor. Brunton (2009)
found that with removal of one blasthole (8 reduced to 7) the extraction zone recoveries decreased. In
particular, the primary recovery from the marker ring plane closest to the ring plane was affected.
Documented full-scale tests on the effect of ring front angle have not been reported. However, a
number of model-scale tests were performed. Janelid & Kvapil (1966) found that the ring front angle
has the effect of reducing the intermixing of waste, or conversely, the intermixing of ore into waste. A
relationship between the ring front angle and the ratio of particle size of the ore and waste was
suggested. By contrast, a number of others suggest that the ring front angle should always be less than
or equal to 90 (Wregg, 1966; Trumbull, 1967; Free, 1970; McMurray, 1982). Altering the ring front
angle by only a few degrees from its optimum value could have the effect of greatly decreasing the
extraction efficiency (Free, 1970). In addition, ground control such as overbreak need to be considered
(McMurray, 1982).
Burden and spacing (Table 5) have a significant impact on material flow. Full-scale tests (Marklund,
1976; Kosowan, 1999; Quinteiro et al., 2001) showed that an increase in burden and spacing may
cause coarser fragmentation, a decrease in recovery and an increase in dilution. Small-scale tests (see
section 2.2.1.1) indicated a similar trend with respect to coarser fragmentation. However, they do not
take into account several geometric parameters related to SLC blast design, such as variable hole
spacing and the front angle of the blast ring.
Void (Table 5) for SLC ring blasting is provided by the compaction of caving debris and the
movement of blasted rock into the production drifts. The former may be termed dynamic void ratio
and defined as the percentage ratio of the void volume (existing void and dynamically created void) to
21
the pre-blast volume of the rock to be broken (Brunton, 2009). For the dynamic void ratio to be
utilized to improve material flow, an understanding of its interaction with factors such as blast
geometry, location of existing voids and initiation is required (Brunton, 2009). A minor effect of
increased recovery and reduced dilution was found if the drift width was increased. It was interpreted
to be caused by a widening of the extraction width and increasing of the swell of volume. The
efficiency of additional void volumes between blast rings was proven in small-scale tests (Cullum,
1974, see Table 1). A full-scale trial is not reported though.
The selection of appropriate explosive is part of the optimization process. The impact of explosive
properties on fragmentation and heave is well investigated for surface mines, but largely unexplored
for underground mines. Complicating factors for SLC operations are further decreases in explosive
performance due to blasthole damage, and desensitization (Fjellborg, 2002; Zhang, 2005). Boreholes
that consequently do not detonated (based on vibration monitoring records) may have a detrimental
impact on material recovery (Brunton, 2009). Timing of initiation (inter-row/hole, Table 5) has an
impact on fragmentation, swell and compaction (see section 2.2.1.1) and consequently material flow
characteristics (Brunton, 2009). However, its mechanisms are not clearly understood. Limited data
indicates that the effect of timing is related to the creation of additional swell of the blasted material
(Quinteiro et al., 2001). According to experience at the LKAB mines in particular, for blasting
openings or multiple ring blasting, correct timing is of utmost importance (Nordqvist, 2012).
Once the ring was fired, material flow could be influenced only with diligent draw control. For
instance this concerns extraction strategies at the draw point and also for a production area and bucket
penetration into blasted material. However, there are no comprehensive full-scale studies to date.
Andersson (2004) analyzed data from a remote controlled LHD (load-haul-dump) machine during
bucket filling and found among other observations that
x The penetration depth of the bucket could be up to 3.6 m.
x The longitudinal penetration was on average 0.8 m towards the hanging wall and the
transverse movement was 0.3 m towards the end of the mucking.
x After on average filling four buckets material the materials flow out > 0.5 m in the drift.
Towards the end of mucking the inflow occurs after fewer buckets.
The relevance of these results for gravity flow and correlation with, for example, waste rock inflow
was, however, not further investigated.
22
Factor
Ring design
Burden (B) and spacing (S)
Void
Timing
Ridgeway
Kiruna, Malmberget
Stobie
Kiruna
Kiruna
Kiruna
Kiruna
Brunton (2009)
Marklund (1976)
Kosowan (1999)
Quinteiro et al. (2001)
Hustrulid (2000)
Kiruna
Kiruna, Perseverance,
Ridgeway
Larsson (1998), Hollins &

Tucker (2004) and Power
(2004a-b)
Quinteiro (2004)
Observations
Author
23
Precise electronic initiation (i-konTM system,

Orica)
Fragmentation analysis (image analysis) did not show substantial

differences when compared with standard layout or top initiation layout.
Similar results were obtained regarding ore recovery and waste dilution.
Measurements indicated that fragmentation with a lower value of xc and
higher value of n gave better ore recovery and lower dilution.
A significant reduction in dilution was observed for a delay of 300 ms

between the inner and outer holes of the blast ring.
10 hole rings; four centre holes fired on a short

delay and after a delay of between 100 - 1000
ms, blasting the other holes with a delay of 50
ms between.
Ore recovery decreased by 32 % and dilution increased by 10 %.
Burden increased from 3.0 to 3.5 m. Specific

charge was not adjusted.
The majority of the ring will swell for the 1983 geometry, while for the
2000 geometry less than half the ring will swell into the drift. A detrimental
effect upon material flow is expected.
For larger B/S ratios fragmentation (photographic analysis) passing 250 mm

decreased from 65 to 32 %. Ore recovery decreased.
B/S = 1; 2.44, 2.74 and 3.05 m tested. Specific

charge ranged from 0.98 kg/t (2.44 m) to 0.42
kg/t (3.05 m).
Comparative study of the SLC blast geometry

from year 1983 and 2000.
Recoveries decreased and rate of blast freezing increased. In addition, due to

a higher specific charge, overbreak was increased. It was argued that double
ring blasts (2 1.2 1.5 m) could improve ore recovery by allowing a
reduction in burden for each ring.
Burden increased from 1.2 to 1.5 and 1.7 m.

Specific charge was adjusted.
Ore recovery increased by 7 % and dilution decreased by 7 %.
The primary recovery from the marker ring plane closest to the blast ring
plane was decreased.
Reduction of number of holes per ring from 8

to 7
Drift width increased from 7 to 11 m.
Interaction between adjacent extraction zones did not exist. A very small
number of markers originated from the sides whereas a large number came
from the central part of the ring and close to the ring plane.
Effects
Marker trials for conventional, fan-shaped

SLC ring designs
Changes
Table 5. Full-scale observations of operational changes and their effect on material flow.
2.3.2 Conceptual flow models

As SLC evolved during the years, it became evident that the uniform material flow behaviour assumed
by earlier authors was not an adequate description of reality anymore (Stazhevskii, 1996; Gustafsson,
1998; Kosowan, 1999; Rustan, 2000; Clout, 2004; Power, 2004a-b; Brunton, 2009) Changes to the
mining method concerned the scale (Hustrulid, 2000) and often also the ring design (Bull & Page,
2000). In addition, the stress state for the current SLC operations has changed.
Results from full-scale trials in modern SLC operations indicated an irregular and asymmetrical
shaped extraction zone (Gustafsson, 1998; Hollins & Tucker, 2004; Power, 2004a-b; Brunton, 2009).
As a direct consequence waste rock inflow measured at the draw point typically exhibits an episodic
character during extraction. Pulsation effects in dilution entry curves may be modeled but it was
shown that quite different approaches could yield the same response. Thus response curves by
themselves are not uniquely related to the conditions imposed and hence are insufficient to validate
these conditions. Reduction of such ambiguities can only be made by in-situ observations which aim
to understand the relevant flow mechanisms.
To summarize, the information acquired in the research of gravity flow in SLC is still so limited that a
generally accepted conceptual model cannot yet be constructed. Table 6 surveys the existing relevant
conceptual models and shows that characteristics could differ greatly. Conceptual models that consider
issues of disturbed flow are highlighted in the table.
It was found that several modes of flow as described by different models could occur in different rings
(Power, 2005). Also a combination of elements from individual models is possible (Brunton, 2009).
For instance, results from the most detailed marker trials to date at Ridgeway mine showed that
primary recovery consisted on an area of continuous shallow draw and fingers of recovery further
away from the ring plane. Extraction from an overbreak zone was found to be relatively common too.
Secondary, tertiary and quaternary recoveries occurred as relatively small discrete zones within the
blasted material.
Interestingly, Kvapil (2008) has also adopted the theory of draw bodies which says that ore fragments
would penetrate into the coarser caved rock due to blasting and form a compacted interface. This
would then allow for a preferential shallow, vertical and upwards orientated flow.
Concluding, material flow behaviour is complex due to the interaction of a wide range of parameters
associated with SLC layout and ring design, material properties, drill and blast factors and draw
control. It is therefore difficult to draw any general conclusions based on the observations from
different mines. In addition, the data set of available observations is limited to a few mines only.
24
Boulders are assumed to be the cause of waste rock

inflow. Ore above the boulder is blocked from
flowing and waste from previously blasted rings in
front is assumed to flow in below the boulder.
A slot between the blasted rock and the pillar opens

during blasting and waste rock from above and
become trapped.
Draw bodies diverge to the location of overbreak.

Markers are retrieved from unblasted rings.
Draw bodies consist of a lower compact part and

several long structures above. The complex shape of
draw body is thought to be caused by spatial
variations in the mobility of rock before mucking.
Waste rock inflow reflects the relative mobility of

the ore in the upper ring parts and the caved rock.
Cavities are formed in the draw body and waste rock
from the previous rings fill the void.
Continuous flow occurs near the ring plane,

progresses upwards and dilution entries from above.
Boulder
blockage
Blast heave
explanation
Overbreak
Palm- and
finger draw
body
Cavity
formation
Shallow
draw
phenomenon
Characteristics
25
theory
in-situ
in-situ
variations of in-situ ore grades
small-scale, Kiruna
lateral
above
size, location of ore boulders
theory
above
above
waste rock trapped within blasted material
Grngesberg, Kiruna, Perseverance, Ridgeway
lateral
above
overbreak at a ring face
Kiruna, Ridgeway
above/lateral
immobilized material within the round
"flow channels" from a main draw body
Kiruna
lateral
above
compacted ore in the upper zones
Kiruna, Ridgeway
above
frozen ore bands and flows lenses with waste
banding effect; possible gap between the blasted and
compressed material
Flow fields are time independent and cyclic

variations in waste inflow are caused by variations in
the in-situ ore grade.
Observations
Dilution
Ore losses
Characteristics
Observations
Dilution
Ore losses
Characteristics
Observations
Dilution
Ore losses
Characteristics
Observations
Dilution
Ore losses
Characteristics
Observations
Dilution
Ore losses
Characteristics
Observations
Dilution
Ore losses
Characteristics
Observations
Dilution
Ore losses
Geological
variations
shape of draw body
Characteristics
Full-scale observations by Selldn & Pierce (2004), Power (2004ab) and Brunton (2009). Mentioned also by Kvapil (2008), but no
details are given if this comprehension is based upon observations
or a theoretical consideration.
Full-scale observations by Gustafsson (1998) and Hustrulid (2000).
Full-scale observations by Gustafsson (1998). Brunton (2009)

observed "fingers" or recovery further from the ring plane.
Full-scale observations by Janelid (1972), Gustafsson (1998),

Hollins & Tucker (2004) and Brunton (2009).
Suggested first by Markenzon (1967) and modeled by Stazhevskii

(1996). Gustafsson (1998) argues that this explanation is physically
incorrect.
Observations from small-scale experiments (Stazhevskii, 1996).

Flow hindrances at the brow are described by Gustafsson (1998).
Suggested by Gustafsson (1998).
Further development of the elliposid theory (Frstrm, 1970).

Mathematical description as Bergmark-Roos equation (Bergmark,
1975; Hedn, 1976). Full-scale observations (Janelid, 1972; Chen
& Boshkow, 1981). Revised for a non-zero opening width (Kuchta,
2002).
Drop
hypothesis
Longtan
ideally none
zones beyond drop draw body
The extracted draw body has a near elliptical form

but with a maximum width above the upper half. By
use of the Bergmark-Roos equation the shape of the
extraction body can be constructed.
shape and eccentricity of both ellipsoids
Observations
Dilution
Ore losses
Studies of free extraction of granulated material through an outlet at

the bottom of a hopper and interpretation of full-scale data (Janelid
& Kvapil, 1966; Janelid, 1972).
Extraction& loosening
ellipsoid
Characteristics
References
small-scale, Grngesberg
ideally none
zones outside extraction ellipsoid
The ellipsoid of extraction defines the original

location of extracted material. The ellipsoid of
loosening is the boundary between stationary
material and material that moved from its original
location any time material is extracted.
Observations
Dilution
Ore losses
Description
Model
Table 6. Conceptual models of gravity flow mechanisms in SLC.
26
SLC DRILL AND BLAST PRACTICE (KIRUNA MINE)
All results presented in the thesis are in one way or another related to applied research in the SLC
mine in Kiruna. It appears therefore appropriate to briefly cover current drill and blast practice in SLC
production, which includes opening methods, standard ring design and charging and blasting.
27
28
SLC drill and blast practice (Kiruna mine)
3.1
SLC layout and ring design
The average thickness of the ore deposit at the LKAB Kiruna mine is between 80 - 100 m with a dip
of 60 - 70 towards east which makes a SLC layout with crosscuts suitable in most cases, see Figure 7.
Rock mechanics as well as quality control considerations have given rise to a mining sequence with a
diagonal stope face. The elongated ore deposit with a strike length of around 4 km is subdivided into
individual production blocks of about 400 m and sublevel intervals of 28.5 m. Staggered crosscut
(production drift) placing with a spacing of 25 m and a Bezier-shaped cross-section of 7.0 5.2 m is
applied. Caving in each production drift is started by blasting openings that connect the blast level to
the drifts and caving masses located in the level above. In general, two different opening methods are
applied nowadays, A and B, see Table 7.
Figure 7. Crosscuts at different extraction states with opening method A/B, block 9, 792 m level.
The opening blast looks alike for both methods A and B including variety B1. It is designed around an
open ( 700 mm) boxhole with eight surrounding cut holes arranged in a square pattern (2 2 m) plus
rings or parts of rings that break to the sides. It is fired in one blast, the subsequent rings one or two at
a time.
In the standard method (A), a single opening blast is located in a drift along the hanging wall
connecting two adjacent crosscuts.
In the other method (B), 1 - 2 opening blasts are placed in a single crosscut. This method needs to be
applied in situations of a flatly dipping orebody, e.g. if it has become progressively thicker with depth,
3
The access drift to this production block is at Y ~ 1000 m.

29
or if sudden changes in the strike direction occur. Also for rock mechanics reasons this method of
opening becomes more important (Malmgren, 2012). The SLC rings behind the opening blasts are
sequentially scaled-up both in height and width. The number of boreholes varies (6 9) and the side
angles for the rings are usually flatter (60 65) than for the standard SLC ring design.
Variety B1 is used only where the orebody becomes much thicker with depth and a series of similar
ring blasts are made without a substantial increase in height.
Table 7. Overview of different opening methods.
Method A
Method B
Variety B1
For the opening method B including B1, the ring blasts up to the second opening blast mostly take
place under the waste rock of hanging wall and the second opening blast is designed to create a slot
that connects to the caving masses above. A flattening of the ring front angle from 80 to 70 between
4
Isometric view and horizontal section. The opening blasts are marked in green, all other rings in grey.
30
the opening slots reduces the high toe confinement to some extent. A complete breakage is unexpected
for these boreholes though and a loosening might be the actual effect. Even if ore losses might
sometimes be large for this opening method, the second opening blast is critical for the success of the
SLC mine (Nordqvist, 2012). The second opening is intended to connect to the caved masses above.
The opening is thereby drilled either towards a drift floor or adjusted to the outline of a ring located at
the level above. Once the second opening has been blasted, the standard layout for the SLC rings is
applied (see section 3.1).
The SLC layout has in recent years undergone some basic changes, see Figure 2. Marker trials
(Larsson, 1998) have shown that the ring design with shorter and flatter side holes to initiate
interactive draw zones does not work well, but that flow occurs within a relatively slim vertical zone.
Therefore the design has, with a transition level in between been altered to a silo-shaped design.
Figure 8. N-S section showing change from flat (level 849 m) to silo-shaped ring designs (907 m level).
The present standard ring design involves drilling 8 holes with a diameter of 115 mm and a projected
burden along the drift of 3 m. The individual rings have a front angle of 80 and side angle of,
typically, 73. The specific drilling amounts to about 0.03 m/tonne for a full-sized ring, yielding a
tonnage of 10000 tonnes of ore. All drilling is performed with fully automated drill rigs with a
Wassara W100 DTH, down-the-hole hammer powered by a high-pressure water pump.
The 878 m level was a transition level between the two different ring designs.
31
3.2
Charging and blasting
Charging of the holes is done with the emulsion explosive KR0500 (Kimulux Repumpable 0500),
which is sensitized by glass micro-balloons and contains aluminum. Charging starts at the bottom of
the hole and the hose is retracted automatically, yielding a specific charge concentration of 11 - 12
kg/m. The operation is carried out via electrical and remote-operated hydraulic systems. Sometimes
packaged emulsion (Kimulux 82) needs to be used in the case of wet boreholes.
Initiation (Figure 9) consists of a system of two to three primers SLP-50 (Sliding Primer-50)
composed of cap sensitive emulsion packed in a hollow plastic casing.
Figure 9. Standard charge column.
Figure 10. Blast design, standard SLC ring.
A 25 g primer made of PETN is placed in the central hole of the SLP-50 and initiated by a non-electric
detonator (Austin Detonator) with a short lead (0.6 m) of shock tube. The SLP-50 casing also has a
guide notch for a 3.6 g/m detonating cord, which ensures a safe lateral initiation of the shock tubing.
Before charging a support claw is fixed at the end of the detonating cord, which is pushed to the
bottom of the hole. At designated lengths, the two SLP-50 primers are fixed to the cord by cable
straps. The charging hose is thin enough to pass the SLP casings.
32
Today, the following general drilling and charging restrictions apply (see Figure 10; cf. Rustan et al.
1988, data in section 2.2.2 for previous blast pattern):
x Minimum distance between charged parts of adjacent holes = 1.4 m to avoid negative
interaction effects,
x Maximum distance between adjacent holes = 4 m to ensure proper breakage, and
x Uncharged collar length = 2 m or more, exactly 2 m for holes 1 and 8 (the side holes).
The firing pattern is designed to open the round symmetrically from the two centre holes of the ring
with delay intervals of 25 ms. The holes are top-initiated with redundant lower primers set at + 25 ms.
Safety issues related to brow damage, ease of access in terms of the existing muckpile and nonclogged boreholes as well as efficiency are the reasons that typically 2 - 3 rings are charged at the
same time. Therefore, sleep times are typically one month, but extreme periods up to one year might
occur and crystallization within the emulsion droplets may occur. To some extent this is counteracted
by the multiple primer configuration, which with its additional PETN primer serves as a safety
precaution. When a ring is prepared for blasting the hidden end of the detonating cords are pulled
down from the borehole collar, bundled by a 5 g/m firing-cord, which is finally electrically initiated.
The assembly described makes individual placement of the multiple primer system possible according
to charging plans. Usually rings are individually blasted, yielding a specific charge of about 0.3 kg/ton
(1.35 kg/m3) for the standard 8-holes layout.
33
34
BLAST FUNCTION
Research in the field of blast function had two directions.

In the first one, a remotely operated 3D photogrammetry system was designed and used to map blast
results of SCL rings in both opening and hang-up situations (Papers B & C, Report 8). The blasting
condition of the SLC rings in the opening situation is still unconfined. By contrast, for hang-up
situations blasting was carried out in confinement, which gives also direct access to study gravity flow
mechanisms (see section 6).
In particular, drifts for which the SLC rings are scaled up between two opening blasts (method B and
variety B1, see section 3.1) were suitable for following up the blasting results. This made it possible to
study, i) broken geometry and height, ii) over- and underbreak and their effects on subsequently
blasted rings, iii) interaction effects between adjacent holes/rings, iv) mapping of geological structures
and their influence on the blast result, and v) drilling accuracy based on identified boreholes.
In the second one, confined drift wall blasting tests were carried out for which the blasted burden was
instrumented with various sensors to study movement and breakage (Paper D, Report 6 & 7). Main
focus was the development of measuring equipment that could be scaled-up to full-scale SLC blasting
and installed behind the rings. Results from an unconfined and confined blast situation are presented.
In addition, Blo-Up, a component of the Hybrid Stress Blast Model (HSBM; Furtney et al., 2009),
was used to model these blasting results (Paper D). The objective was to show that reasonable
predictions of fragmentation and burden movement under confined conditions can be made.
In the course of laboratory and full-scale experiments a new measurement system was developed
(Paper D, Reports 2, 6 & 7), which combines the initial movement recorded by an accelerometer
designed to minimize zero-shift with the later slower movement displacement recorded by fibre
photoelectric sensors (fibre-optic zebra gauge).
35
36
Blast function
4.1
3D mapping of SLC blast rings
4.1.1 Measurement procedure on-site

A photogrammetry system (see section 4.1.1.1) mounted on a rotating camera head with the option to
geo-reference the 3D images was designed and constructed together with the central workshop of
Lule University of Technology (LTU). The system was mounted at the top of a 12 m long lightweight crane. For the survey of openings the mobile crane was mounted on a car trailer. For the survey
of hang-up situations the camera crane was mounted with an adapter to the hammer of a scaling
machine. To enhance safety against a debris surge an additional pile of material was loaded in-front of
the scaling machine.
Figure 11. Survey at a draw point in SLC

crosscut, hang-up situation.
Figure 12. Measurement procedure on-site, opening

situation (not to scale).
An application of the system for the survey of a hang-up is shown in Figure 11 and the measurement
procedure in Figure 12. A single image pair yields a partial 3D model of the breakage face. A
complete 3D image of the entire breakage front requires merging of several overlapping models.
Therefore, image pairs are gathered in vertical columns starting from the bottom. A set of mini prisms
arranged around the digital camera allow its positioning in the mines coordinate system from total
station measurements made when acquiring the first images (yellow in Figure 12). Point lasers
mounted in parallel provide light spots on the cavity surface giving the length scale in each image pair.
4.1.1.1
Photogrammetry system
Digital photogrammetry can be regarded as a mature technology yielding reliable and reproducible
results with fast acquisition time and, compared with laser scanner technology low hardware costs.
The technique is increasingly used to survey and map rock faces (Tonon & Kottenstette, 2006). The
ShapeMetriX3D (SMX3D) software (version 3.0.2; Gaich et al., 2009) was used as a starting point.
The present system comprised a rotating head with additional equipment (see Figure 13). A customcalibrated digital camera with wide-angle conversion lens moveable along a 2.2 m long beam is used
to take the stereographic image pair. A laser distance measurement device measures the distance to the
37
Blast function
object which determines the camera spacing for the two images. The positioning of the camera
carriage is controlled by a surveillance camera and another camera with a fisheye lens gives an
overview of the cavity under investigation. Two halogen lamps provide sufficient illumination. Four
laser modules and 360 mini prisms are mounted on the crane head, which are both used for the
photogrammetry system described below. The need to remotely control all components of the
equipment from a safe distance (up to 100 m) required special solutions in terms of signal converting.
Figure 13. Measuring equipment mounted at a

remote head.
Figure 14. Geometric arrangement of the laser

modules.
3D images are generated by means of classical photogrammetry and extensions from computer vision.
A pair of images of the object of interest is taken from slightly different positions. A dense set of
corresponding image points is derived automatically from these images utilizing techniques from
multidimensional statistics. From this set the relative orientation of the 2 camera positions is
computed. Intersecting the rays of corresponding image points from these positions yields a 3D point
cloud. Finally these points are triangulated to form a surface mesh, which is directly linked to the
original image data. The imaging system is calibrated to take the distortion of the lenses and alignment
of the optical axis into account. The resulting 3D image represents the true shape of the surface, but
lacks the correct scale, position and orientation in 3D space.
The use of photogrammetry for the present task required the development of some customized
software extensions to the existing SMX3D software.
To achieve proper scaling a projection unit made up of 4 laser pointers was mounted on the camera
rig, see Figure 14. The laser modules are installed on a 50 50 cm frame and the unit emits parallel
beams normal to the imaging plane of the camera positions. The scaling factor (s) for the 3D image
can then be derived by the ratio of the known distance between the beams and the distance between
the normal projection onto the y'z'-plane of 2 observed laser points, e.g. s = (P4-P5) / [nproject(P11P12)]. To support geo-referencing of the 3D image, reflecting prisms were arranged in an equilateral
38
Blast function
triangle of 700 mm side length around the imaging system. By surveying the locations P1 to P3 of
these prisms in the object coordinate system (x,y,z) and their a priori known positions in the imaging
coordinate system (x',y',z') a transformation matrix M can be derived. M contains the translation and
rotation between these 2 systems, so all points from the correctly scaled 3D image can be transferred
to the global object coordinate system. Both the scaling and geo-referencing are implemented in the
software module LKAB Referencer. Surveying in a hang-up situation should be short for safety
reasons, so it was decided that geo-referencing of the 3D image was usually not needed. With another
custom extension of the SMX3D software, called the Interactive Referencer, a 3D image can be
roughly aligned to a given coordinate system by interactively rotating the 3D image, e.g. based upon
an identified half cast at the brow area and its planned coordinates. In this way, the 3D images could
be brought into the mine coordinate system and the hang-up relative to adjacent SLC rings visualized.
Due to the limited field of view of the imaging system and the arrangement of the cameras it is not
possible to create a full spherical 3D image from only one pair of images. The full image requires a
merging technique. This procedure takes structural information as well as topographic information
into account and does not need any reference points. Thereby corresponding points are determined
either interactively or by an automatic search algorithm in the Model Merger component of the
SMX3D software. To facilitate this, its capability had to be extended to support a larger number of
models, i.e. the merging of up to 15 instead of the default 6 individual 3D models.
Tests were made to quantify the resolution of detail and the geometric accuracy of the measurement
system (Report 8). The average 3D point spacing was found to be 2.8 1.8 cm. However, depending
on the actual size of surveyed opening, front angle of the ring, relative position of the rig and hence
viewing angle of the rig glancing intersections are sometimes problematic. An additional measure
considering the area and minimum angles within the triangulated surface reveals such problem zones,
see section 4.1.1.2. The absolute error for geo-referencing and scaling with the method described was
as small as 11 5 mm when compared with surveyed reference points. The relative error between a
large merged 3D image (740 m2, 12 individual 3D models) and another geo-referenced 3D model was
almost insignificant. At most it was in the range of 1 - 2 dm and can be related to the merging process.
4.1.1.2
Evaluation process
The 3D images document the local geology and allowed a structural mapping of the rock faces to be
made with the component JMXAnalyst within SMX3D. The mapped structures are presented in
stereographic projection plots, summarized for a given drift in combination with the orientation of the
respective ring faces, see Figure 15.
The blasting results of the surveyed SLC rings were visualized and evaluated with the RingCave
software (version 2012-05-20; Nordqvist, 2012), see Figure 16. It was written to handle the wealth of
new information obtained. Data is stored in a database (MS Access).
39
Blast function
Figure 15. 3D image mapping at the roof

inside an opening, KI-16-964-150-10.
Figure 16. Overview, RingCave software.
The following objects may be identified in the 3D image, separately imported into RingCave and
linked to the respective 3D CAD (computer-aided design) model:
x Point and line objects, model objects, i.e. borehole sockets, half casts etc. or
x Polylines, 3D polygons, i.e. waste lenses, geological discontinuities, etc.
The available drill, charge and blast data (rings) are then imported from LKAB`s production
database Giron for the surveyed SLC ring and its adjacent rings, see Figure 17. RingCave calculates
various sections through a 3D model relative to a defined plane or borehole, see Figure 18. This
enables a detailed analysis of the blasting results.
Information derived from 3D mapping, i.e. model objects, is used to further compute:
x Distances to a plane (over- and underbreak) relative to a best-fit plane
x Static / dynamic explosive distribution (Kleine et al., 1993)
x Breakage depths and spacing to adjacent holes
x Minimum distances (effective burden, Beff) to the 3D model
x Drilling deviations
6
A measured borehole (M-hole) is constructed from the orientation information of either a half cast or the best-
fit plane to all borehole objects (sockets, half casts) identified in the 3D image.
40
Blast function
Figure 17. Geo-referenced, merged 3D model

(wireframe view) with preceding ring outlines, ring
no. 10, drift 150, 964 m level.
Figure 18. 3D model (rendered view) with different

sections, view direction west, ring no. 10, drift 446,
7
935 m level.
Furthermore, quality measures, like the ratio area/minimum angle (m2/), can be calculated for all
triangles and hence areas of lower accuracy in a 3D image may be located.
All computation results including relevant attributes can be visualized both in 2D and 3D. The
capabilities of the RingCave software were illustrated based upon an example for which a series of
blast were analyzed (Paper C).
With 30 surveyed SLC blast rings inside openings and 6 inside hang-ups during the period 01/2008
04/2011 a reasonably large data set exists. A correlation matrix and the Anova F-test were used to
analyze parameters that may influence the results. The correlation matrix also revealed internal
correlations between input parameters, some of which are unexpected.
4.1.2 Blast results
4.1.2.1
Initial opening blasts
It was observed that a boxhole with eight surrounding parallel cut holes (see section 3.1), usually
broke to its full height. Even if earlier vibration records (Nordqvist, 2012) typically showed on
average missing detonations of 2 holes for the initial eight cut holes this seems to have no direct
influence on the height/geometry that was reached. There is a problem, though, when widening to the
sides. This problem is shown by the following example, see Figure 19 and Figure 20. The holes
intended to widen the initial cut, but which are longer than the boxhole (holes P1 P3 in Figure 19)
give rise to problems. Because of the small breakage angle at the toe there is realistically no chance for
the burden to break.
Section perpendicular to fitted ring plane at z = 918.5 m (A), section along borehole M5 (B), section
perpendicular to borehole M5 at z = 923.3 m (C) and vertical section in the middle of the ring (D).
41
Blast function
Figure 19. Cross-sections showing centre ring no. 5

within initial opening blasts, drift 443, 935 m level.
Figure 20. Initiation of opening blast, section at

925 m, ring 1 8, drift 443, 935 m level.
3.5
3.0
burden [m]
2.5
2.0
1.5
1.0
all data
0.5
height of boxhole not exceeded & breakage not completed
0.0
0
500
1000
1500
2000
2500
3000
3500
4000
nominal time delay [ms]
Figure 21. Burden (m) vs. nominal time delay (ms) for holes in centre rings of initial opening blasts.
However, for equally long box- and side-holes a problem related to spacing and initiation seems to
exist as well. The maximum broken burden versus nominal time delay (Figure 21) showed some
variation, but seems to exhibit a general trend. The data for which the height of boxhole was not
exceeded and the breakage not completed (filled diamonds) are examined more closely. The holes
directly adjacent to the boxhole/cut holes with a nominal time delay of 500 ms were able to break on
average 2.3 0.3 m. Successively detonating holes could break a much larger burden. With a delay of
around 2500 ms a burden of 3.0 0.3 m was broken. The maximum broken burden again decreases to
2.4 0.1 m when it comes to an even larger delay of 3500 ms.
42
Blast function
Concluding, it can be said that a short nominal time delay causes a smaller breakable burden and that
an increased burden requires a longer delay. For the planning of the adjacent holes to the boxhole/cut
holes a burden of 2.0 - 2.5 m should therefore not be exceeded. Subsequent holes further widening the
initial space could be with a burden of up to 3 m, but should also have a longer time delay.
4.1.2.2
Scale-up of SLC rings
Scale-up of SLC rings (see section 3.1) during adaption to the ore contact closest to the hanging wall
constitutes a difficult blasting situation, see Figure 22.
Figure 22. Ring outlines with measured max. breakages (red cubes), ring nos. 1-23, drift 150, 964 m level.
Results from the correlation analysis/Anova F-test are summarized in Table 8 and Table 9. Parameters
(P1 - P10) related to the ring design, structural geology, blast data and confinement possibly influence
the blast result. Some parameters (P7 - P10) were of binary character (either/or) and analyzed with the
F-test, which can analyze grouped data. As result parameters the broken area (R1) and extraction rates
(R2 - R3) were considered.
The large confinement, i.e. the large increase in area Hi (P4) or rather height Ai (P6) from one ring to
the next significantly reduced the broken area Ab, see Figure 23.
Isometric view (left) and side view (left). The opening blasts (initial/second) are marked in light green, the
surveyed rings in dark green colour and numbered.

43
Blast function
height increase Hi [m]

-2
-1
140
area increase
height increase
broken area Ab [%]
120
100
80
60
40
Ab = -1.28Ai + 89.06
R2 = 0.42
20
Ab = -15.07Hi + 91.55
2
R = 0.55
0
-20
-10
10
20
30
40
50
60
area increase Ai [%]

Figure 23. Planned increase in area (%) and height (m) for successive SLC rings versus broken area (%).
The opening method B (excluding variety B1), which implies the scaling-up of individual rings, broke
on average only 56 21 % of the planned ring area. The variety B1 improved breakage significantly.
After a successful initial opening blast, the following rings broke out the planned area, and, with the
addition of structural overbreak, sometimes more (104 18 %).
Despite large variations in the blast input data (P1 - P3) no explanatory effect for the investigated
result parameters was found. The orientation of rock mass structures (P8 - P10) relative to the ring
face did not have a significant effect on the broken ring area either. However, a pronounced effect of
structures on the blast result was observed in some cases.
The correlation analysis has shown that the extraction rates (R2 - R3) were a poor indicator of a
successful blasted ring in terms of broken area. The reason for this might be that the stop extraction
criteria are different than in standard SLC production. Possible criteria are stability considerations and
safety for the charging personnel which lead to the material of openings not being entirely loaded.
44
Blast function
Table 8. Correlation matrix, blast result (overall analysis), 30 rings (openings).
Correlation matrix
P1
P1
P2
P3
P4
specific charge, kg/m3

planned charge, %
sleep, days
area increase, %
1.00 P2
0.32 0.32
-0.07 -0.07
0.22 0.22
P5
P6
R1
R2
R3
width increase, %
height increase, %
broken area, %
extraction rate, %
accum. extract. rate, %
0.39 0.39 0.14 0.68 1.00 P6

0.04 0.04 0.26 0.83 0.39 1.00
-0.29 -0.29 -0.26 -0.59 -0.24 -0.56
0.30 0.30 -0.09 -0.02 0.10 -0.08
-0.20 -0.20 -0.05 0.13 0.20 0.13
P3
1.00
0.25
P4
1.00
P5
R1
1.00
-0.05
0.14
R2
1.00
0.29
R3
1.00
Classification: dark green |-0.8| rxy 1.0; light green |-0.6| rxy < |-0.8|; |-0.4| rxy < |-0.6|
10
Table 9. Relative broken ring area (overall analysis), statistics for F-test, 30 rings (openings).
quantities
mean, m
P7: opening method

method A
31.9
method B
55.8
variety B1
103.7
stdev, m
meas.
0.0
20.6
17.6
quantities
1
19
10
73.4
73.1
67.7
38.6
34.6
24.8
stdev, m
meas.
P9: dip of structures,

0-30
89.8
30-60
64.7
60-75
65.5
25.5
20.2
33.4
6
7
12
39.9
75-90
P8: strike of structures relative free face,

parallel
angled
perpendicular
mean, m
4
14
12
70.4
P10: blast with/against dip of structures

with
68.5
28.9
18
against
74.8
33.8
12
Classification: Groups which are significantly different are marked in light green; Light yellow marks a near
significant effect and with more measurements the differences would probably be significant.
1
A reliable conclusion could not be made, as the single observation (KI-12-792-127-9) is based on a remnant
pillar situation where the initial opening blast failed to connect to the cave above.
2
Parallel (0 x < 15), angled (15 x < 75) and perpendicular to the mining front (75 x 90).
With a series of blasts surveyed the initial conditions for the blasting of an SLC ring could be studied
hole-by-hole. With the post-blast information, the variations in effective burden (Beff) and respective
direction vector towards the 3D model for a specific borehole are calculated, see Figure 24.
The analysis showed that the effective burden (Beff), as calculated from measured holes (M-holes, see
Figure 16), could vary considerably along the broken ring face. It was on average 2.4 0.4 m, but
could be as large as 3.6 m (Bmax). An effective burden smaller than the nominal burden (3 m mostly)
should either depend on overbreak, for the ring faces on average 0.5 0.2 m (see section 4.1.3), and/or
on drilling accuracy (see section 4.1.4). The analysis also showed that a smaller effective burden at the
9
The complete table is found in Report 8.
10

45
Blast function
ring face is correlated with a somewhat larger breakage depth along the holes. However, the burden at
the maximum breakage depth does not explain why the breakage stops at this point. The cause is
rather an incomplete breakage of the preceding ring and the inability of increasing the ring breakage
heights any further. The toe breakage angle simply becomes too acute (< 70) when the hole passes
the previous maximum breakage depth and this results in an intact burden.
Figure 24. Effective burden, ring no. 13, hole no. 5, drift 150, 964 m level.
It was observed that side holes had difficulties in breaking straight forward. Instead, they broke out at
an angle relative to the centre. In addition, they tended to have a more acute breakage angle. The side
angle only marginally had an effect on the relative broken hole length, which implies that a somewhat
shorter broken length could be expected for side holes.
4.1.2.3
Categorization and discussion of breakage problems
Figure 25 is a list of the malfunctions for SLC ring blasts in openings and a categorization of possible
causes for unbroken boreholes. Problems observed in the 3D images are marked with a tick.
Furthermore, experience was used to identify other possible causes of the same symptom. These are
marked with a dot. Also the interaction of several factors is possible, but difficult to assess.
About 60 % of all boreholes were incompletely broken. The majority of the breakage problems are
probably related to the specific blasting situation in openings but a considerable proportion (10 % of
the incompletely broken holes) was caused by undetonated charges. Interaction effects between
boreholes of the same ring might be an important problem. The identified problems with initiation (i.e.
the existence of the detonating cord in the 3D image) are unexpected and their actual causes would
require a closer examination. Frequent problems with either intact or partly unbroken side holes might
46
Blast function
be related to an increased confinement for this type of hole. Excessive drilling deviations for side
holes might be another factor, but this was impossible to assess for unbroken holes.
Figure 25. Categorization of breakage problems, 244 analyzed boreholes.
Good drill and blast planning is important for an acceptable opening blast. With the special opening
method (B and variety B1) in mind, also a successful scaling-up of ring heights in-between the two
openings is a determining factor. Not only could production be increased, but also an improved blast
result for the second opening, which should establish a contact with the caving masses, could be
expected. At the moment, drill and blast plans are made independently of each other by personnel that
change frequently. This might be also be the reason that drill and blast layouts both for openings and
standard SLC rings vary largely and are rather more dependent on the human factor than on an
adaption to a specific situation. Drilling and charging restrictions (e.g. minimum distances between
charged parts, see section 3.2) are not always followed. A poor example of planning is for instance
also the placement of the upper primer at the bottom of the hole (top), where there may be no
explosives, instead of being set lower and embedded in the emulsion. However, this is normally
corrected by the charging personnel on-site. In addition, charging plans might also be changed on-site
and follow-up of the results is hence difficult.
A potential for the improvement of both the initial opening blast, and for the scaling-up of the rings
exist. Remedies and suggestions based upon the findings are discussed in Report 8.
47
Blast function
4.1.3 Overbreak
Overbreak, i.e. outfall of rock behind the blasted ring plane may cause specific problems, e.g. brow
damage, disturbed gravity flow and damage to pre-charged SLC rings.
Over- and underbreak was calculated for a 0.5 x 0.5 m grid as orthogonal distances from the best-fit
plane (see section 4.1.1.2, Figure 16) to the 3D model. An intensity graph is used to plot the 3D data
on a 2D area using color to represent the third dimension where each data value is rendered as a block
of color on the graph, see Figure 26.
Figure 26. Overbreak (red) and underbreak (blue),

ring no. 12, drift 150, 964 m level.
Figure 27. Charge and initiation pattern with specific

3
11
charge (kg/m ), ring no. 12, drift 150, 964 m level.
Overbreak was analyzed for all 36 surveys on a ring-by-ring basis with the parameters related to SLC
ring design (P1 - P7) and structural geology (P8 - P10), see Table 10. The result parameters are the
mean overbreak for the ring face12 (R1) and brow area (R2), which is defined as a 4 m wide arc above
the roof.
It was analyzed in more detail for 8 SLC rings for which complete data of all measured holes (Mholes, see Figure 16) were accessible, see Table 11. It could then be based upon the calculation grid
(0.5 0.5 m) and related to parameters like minimum distance to the charge (P11) and brow (P12) as
well as static and dynamic specific charge (static/dynamic q, P13 - P14), see Figure 27. The result
parameter is overbreak (R3).
11
Dynamic explosive distribution, calculation plane at a burden of 0.5 m. Details and input parameters are found
in Report 8.
12
Below called only overbreak.

48
Blast function
13
Table 10. Overbreak (ring face/brow area), statistics for F-test, 36 rings (openings & hang-ups).
category
quantities
mean, m
stdev, m
meas.
P1: block
stdev, m
meas.
north
-0.47/-0.56
0.21/0.38
22
opening blast
-0.78/-0.99
0.13/0.29
mid-part
-0.42/-0.45
0.14/0.22
unconfined ring
-0.45/-0.53
0.21/0.37
26
south
-0.64/-0.77
0.30/0.44
hang-up
-0.45/-0.47
0.21/0.22
P6: blast type
716-792 m
-0.47/-0.56
0.21/0.38
22
single ring
-0.45/-0.52
0.21/0.36
29
907-964 m
-0.52/-0.59
0.24/0.36
14
double ring
-0.45/-0.49
0.11/0.16
opening blast
-0.78/-0.99
0.13/0.29
P3: SLC ring dip,

70 -75
-0.47/-0.55
0.21/0.37
27
80-84
-0.54/-0.65
0.26/0.38
P7: initiation type

collar/bottom
-0.52/-0.64
0.49/0.43
22
bottom
-0.45/-0.47
0.17/0.21
14
P4: SLC ring dip direction,

270
-0.50-/0.58
0.22/0.37
35
others
-0.23/-0.25
0/0
P8 : strike of structures relative free face,

structural geology
mean, m
P5: blast situation
P2: level
SLC ring
quantities
P10: dip of structures,
parallel
-0.67/-0.89
0.13/0.21
0-30
-0.45/-0.56
0.30/0.56
angled
-0.48/-0.50
0.24/0.39
18
30-60
-0.43/-0.49
0.15/0.25
perpendicular
-0.44/-0.55
0.20/0.33
13
60-75
-0.54/-0.62
0.23/0.35
14
75-90
-0.50/-0.59
0.10/0.22
P9: blast with/against dip of structures

with
-0.46/-0.54
0.22/0.37
23
against
-0.53/-0.63
0.23/0.36
13
Classification: Groups which are significantly different are marked in light green; Light yellow marks a near
significant effect and with more measurements the differences would probably be significant.
1
Parallel (0 x < 15), angled (15 x < 75) and perpendicular to the mining front (75 x 90).
Table 11. Correlation matrix, overbreak (spatial analysis), 8 rings (openings) with 3606/995 data points.
Correlation matrix: Ring face (3606 data points)

minimum distance to charge, m
P11
minimum distance to brow, m
P12
static q, kg/m3
P13
P14
R3
dynamic q, kg/m3
overbreak, m
Correlation matrix: Brow area (995 data points)

minimum distance to charge, m
P11.1
minimum distance to brow, m
P12.1
static q, kg/m3
P13.1
3
P11
1.00
-0.42
-0.73
P12
1.00
0.48
P13
1.00
-0.71
-0.40
0.46
0.39
0.99
0.32
P11.1
1.00
-0.73
-0.65
P12.1
1.00
0.61
P13.1
1.00
P14
1.00
0.32
P14.1
1.00
R3
1.00
-0.63
0.56
0.99
dynamic q, kg/m
P14.1
R3.1
-0.43
0.33
0.27
0.28
1.00
overbreak, m
R3.1
Classification: dark green |-0.8| rxy 1.0; light green |-0.6| rxy < |-0.8|; |-0.4| rxy < |-0.6|
13

49
Blast function
Figure 28 shows a summary of measured overbreak for all 36 surveys (30 opening and 6 hang-up
surveys) in terms of overbreak (m) and affected relative area of the ring faces (%).
-2.5
100
OB, ring face
OB, brow area
OB > |-0.5| m
OB > |-1.5| m
90
-2.0
overbreak (OB) [m]
70
-1.5
60
50
-1.0
40
30
-0.5
affected relative area [%]
80
20
10
0
0.0
0
10
20
30
40
50
60
surveys [%]
70
80
90
100
Figure 28. Cumulative distribution of overbreak (OB) and affected relative area, 36 rings (openings &
14
hang-ups).
The mean overbreak calculated for the entire area of the blast rings (36 surveys) was -0.49 0.22 m.
The mean overbreak occurring at the brow only, was 16 % larger (-0.57 0.37 m). The increased
standard deviation for the mean overbreak at the brow damage indicates that the actual amount could
vary greatly in particular cases. The largest overbreak was -1.5 0.7 m for the brow specific area, and
could in some cases even reach the subsequent ring 3 m away. The mid part of the ring is most often
affected.
An overbreak behind -0.5 m occurred for 29 21 % of the broken area. Excessive overbreak higher up
on the ring face was seldom observed, and, if there was any, it was structurally limited to small areas
(< 10 m2 with a maximum less than 2/3 of the burden to the next ring). By contrast, it was found that
overbreak is often constant over large portions of the face (excluding brow damage). Most likely, large
radial blast cracks are the predominant cause of overbreak besides the effects of natural jointing.
The overall analysis showed that large damage identified at the brow area implied a general overbreak
problem at the entire ring face, rxy = 0.87 (see Table 13 in Report 8).
14
Based on sub-division of ring area into a 0.5 0.5 grid with one value each. Reference for the affected relative
area is the total broken area, which has an underbreak 0.5 m.

50
Blast function
It was found that opening blasts were in particular prone to cause overbreak in the next rings to be
blasted. Confinement and the blasting of several rings with reduced burden appear to be two important
reasons for this observation. No difference between the blast situation (P5) with sub-division in
unconfined (subsequent rings after the initial opening blast) and the semi-confined (hang-up)
situations was found. No difference between the blast types (P6), either single or double ring blasts,
was found either. With an increasing number of observations, the observed insensitivity to the
distinctions opening/hang-up and single/double ring blast might change though. The position of
initiation had a near significant influence as brow damage increased from |-0.47| 0.21 m to |-0.64|
0.43 m if initiation took place close to the collar.
The alignment of the ring faces, dip (P3) and direction (P4) did not have an explanatory effect on
overbreak. The grouping into different mining levels (P2) did not have any effect on overbreak either.
By contrast, a tendency for increased overbreak in the southern part of the mine (P1) was identified.
Some more measurements are required though for this to be a certainty.
Effects of unfavorable orientation of structures and regionally defined fallouts were observed from the
individual analyzes of the blasting results. To quantify this effect on the overall overbreak, predictors
related to the rock mass structures were tested. It was found that the strike direction of structures (P8)
had a near significant effect, as parallel orientated structures tend to give a larger overbreak. The
relative dip direction (P9) and dip angle (P10) did not have any explanatory effect though.
For the spatial analysis of overbreak (see Table 11) no significant correlations between overbreak and
minimum distance to charge or between overbreak and specific charge q (static/dynamic) were found.
By contrast, the effect of the brow acting as a free face came though clearly and it was observed that
up to a height of about 12 m an increased overbreak (> |-0.5| m) could be expected.
The damage of the ring face might also be described in terms of observed half casts, Figure 29. Half
casts will only appear in the transition zones between under- and overbreak. The actual charge and
initiation pattern (e.g. uncharged parts) will also influence their occurrence.
The half cast factor amounted on average to 6 % , i.e. 7 m of the 120 m drilled in an average SLC
opening ring. About 65 % of all half casts found were in the side holes and of these almost 80 % came
from charged parts. The much higher amount of half casts found for the charged parts of the side holes
might be due to the side holes being the last holes in the ring to detonate; no neighboring holes would
tear away existing half casts. The holes in the mid part contributed approx. 25 % of all half casts.
Fewer half casts than those were observed for the holes in-between (about 10 %).
51
Blast function
70
half casts in uncharged part
half casts in charged part
60
fraction [%]
50
40
50.6
30
20
12.4
10
14.3
2.6
7.9
12.3
0
side holes
in-between
mid holes
hole type [-]

15
Figure 29. Fraction (%) of half casts in charged/uncharged part versus hole type (), 30 rings (openings).
4.1.4 Drilling accuracy

Borehole deviations are dependent upon errors related to the set-out, collaring, alignment and in-hole
trajectory deviations, see Figure 30.
Figure 30. Borehole deviations and definitions of error components (not to scale).
15
Mid holes are defined with respect to the side angle, i.e. 80 side angle () 110, see Figure 1. Side holes
are the holes furthermost to the sides with a side angle of 60 - 65 and 115 - 120, see Figure 1.
52
Blast function
In-hole trajectory deviations for the Wassara DTH drilling system are normally within 1.0 - 1.5 % of
its length for 54 m long boreholes (Quinteiro & Fjellborg, 2008). As the lengths of holes were short
for the surveyed blast rings, i.e. typically between 10 - 20 m, trajectory deviations are therefore
assumed to be much smaller.
Borehole sockets and half casts can be identified in the 3D models. Their position and orientation
allow a comparison with the drill plans and hence of the drilling accuracy. A tentative analysis of
drilling deviations was made in Report 8. The main conclusions are as follows.
1. A systematic borehole deviation towards positive x (0.7 0.8 m), i.e. forward in the
longitudinal direction of the drift existed. A bias towards negative y (-0.1 0.8 m), i.e. north
in transverse direction of the drift, seemed to exist as well (Figure 50 in Report 8).
2. From a blasting point of view, the relative accuracy in between the SLC rings is a decisive
factor for completed breakage. Deviations within the drilled ring plane could be judged based
on the calculated RMS (root mean square) value of the best-fit plane to borehole objects (see
Figure 16 in section 4.1.1.2). A mean RMS of 0.10 0.05 m (40 measurements) illustrates
that they are rather insignificant. The systematic character of deviations between rings implies
therefore a constant offset with an essentially unchanged burden.
3. The borehole deviations in the x and y directions were mainly influenced by the collar
deviation. Errors in set-out and in collaring are both possible. However, the observed
systematic character of x, i.e. the constant offset between subsequent rings along the axis of
the drift, rather suggests a set-out problem. Further causes for the observed collar deviation
were structures whose dip angles are steeper than the front angle and the drill rig type.
Specifically, a drill rig equipped with a top hammer showed larger deviations.
4. Drilling successive rings with different rigs might cause undesired interaction effects. In
particular, this concerns a transition ring, drilled at a flatter ring front angle (70), which
approaches the preceding ring of the initial opening (80) according to plan with a minimum
burden of 1 m (ring no. 8 in Figure 22). It was observed that drilling deviations could cause it
to penetrate into the preceding ring of the opening. A separate blast then causes visible
damages to the charges of the transition ring which is charged with the rings before.
5. The best-fit plane further showed that the drilled ring front angle was on average 2.4 2.2
flatter than planned. Alignment deviations seem to be of stochastic character (Figure 53 in
Report 8). Several factors might be relevant for the observed alignment problems: i)
instabilities of the drill rig, ii) lack of precision in positioning the equipment, iii) misaligning
the feed beam, iv) topography at the collaring point, and, finally, v) structural geology.
53
Blast function
4.2
Burden instrumentation of SLC blast rings
4.2.1 Measurement problem

The conceptual layout for a future burden instrumentation of SLC rings is shown in Figure 31.
Figure 31. Future burden instrumentation of SLC blast rings (not to scale).
This layout comprises the drilling of boreholes from the longitudinal drifts (green in Figure 31) into
the burden of the SLC rings closest to the footwall and would focus on investigating the function of
the blast in the upper part:
1. Dynamics of burden movement (see section 4.2.2)
x
Swelling of the blasted burden and compaction of caving debris
Existence of voids prior to and after blasting
2. Breakage of burden (see section 4.2.3)

x
Verification of over- and underbreak
Quantification of fragmentation and mobilization
The implementation of the proposed measurement program in production mining is subject to some
complicating factors. The measurement holes have to pass at least 30 - 40 m of footwall before they
reach the nearest rings. Accurate surveying of the ring holes, careful planning and high-precision
drilling are necessary for the success of any future field test. Another factor is the correct placement
and installation of the measurement systems from the backside of the burden.
Studies of burden movement have been carried out previously with various systems ranging from very
simple to advanced ones, but with installation in front of the blast (Felice et al., 1991; Liu, 1994;
Chiappetta, 1998; Mishra & Gupta, 2002; Segarra et al., 2003). This study requires measurement of
burden movement from behind the blast. All measuring equipment must therefore be installed inside a
borehole or, alternatively, at the borehole collar. Many of the measurement systems and components
54
Blast function
available today could be used. However, their applicability, reliability and accuracy for this specific
task have to be assessed. This is the reason why controllable, simplified drift wall blasting experiments
were carried out, see Table 12.
Table 12. Overview of drift wall blasting experiments.
side
691
135
North
12
691
132
North
12
691
132
South
12
691
127
South
741
93
South
741
95
South
detonation failure; test abandoned
filled material
inappropriately filled, unconfined

blast
caved material
& free face
confined - unconfined blast
filled material
confined blast
reinforcement
no burden breakage; max. 0.1 m

movement
reinforcement
complete burden breakage; 0.5 - 1

m movement
x
x
Remarks
caved material
& free face
x
x
Type of
confinement
behind
drift
12
inside
level
in front
block
Test id
Blast site
behind
Instrumentation relative to
burden
Burden
Burden
dynamics
breakage
The first set of tests (test 1 4, Report 6) focused on the installation of redundant16 sensor systems on
or in burdens blasted against caved masses, see section 4.2.2.1. The second set of tests (test 5 and 6,
Report 7) aimed to further develop a specific measurement system (Report 7), see section 4.2.2.2.
4.2.2 Dynamics of burden movement
4.2.2.1
Confined drift wall blasting tests
Adjacent pillars in the completed production block 12 on the 691 m level were chosen for the first set
of tests (Report 6). Blasting at two sites (test 2, drift 132-130 and 4, drift 127-130) was carried out
against artificially filled drifts with a mixture of broken ore and waste loaded by an LHD machine. In
this way a confined blasting situation was achieved, which also allowed a later inspection of the blast
results. Two blastholes with diameter 115 mm were drilled for each test, parallel with the drift (see
Figure 32). The tests were extensively instrumented with different sensors, which were installed in
holes that where drilled from the drift behind into the burden area.
Table 13 summarizes the blast design.
16
By redundant is meant a sensor system that uses different gauges to measure the same quantities and that
can be used to assess the accuracy and repeatability of the signals.

55
Blast function
Table 13. Blast design, test 2 and 4 with blastholes B2 B5.
Parameters
Unit
Diameter
Hole length
Charge concentration
Stemming length
Effective burden
Effective spacing
Delay time
mm
m
kg/m
m
m
m
ms
Test 2
B2
B3
115
13.2
11.1
12.0
12.4
4.0
3.0
0.9
0.8
1.6
0
Test 4
B4
B5
115
13.0 14.0
11.4 11.6
2.5
2.5
1.5
1.7
2.0
0
The blast sites were documented with a face survey and structural mapping before and after blasting
with the photogrammetry system SMX3D (Gaich et al., 2009). All boreholes were surveyed with a
gyro based deviation probe Flexit SmartTool (Flexit; GeoDrilling Int., 2002). As an example, test
site 4 is shown in Figure 32.
Figure 32. Test site 4, 3D model (wireframe view) with blastholes (red), measurement holes (green) and
measurement points (blue cubes) within burden, (1) top-view and (2) isometric view.
The boreholes were charged with the standard emulsion explosive (KR0500) and primers (SLP-50),
see section 3.2. Programmable electronic detonators i-konTM (Orica) were used for initiation. The
boreholes were stemmed with sand/gravel. The velocity of detonation (VOD) and vibrations at
adjacent pillars were recorded to document detonation of the blastholes. A camera with high speed
filming capacity of 300 fps was used (Casio EX-F1) helped to verify the functionality of sensor
systems installed behind the burden and that the confinement was properly built up. Further
documentation of the situation after blasting was made in terms of damages to the sensor systems and
to determine final displacement.
56
Blast function
4.2.2.1.1
Instrumentation
The selection of a proper measurement method depends upon key factors such as:
x Measurement range of minimum 500 mm displacement with high sampling frequency
x Accurate and precise data
x Robustness in a mining environment i.e. low sensitivity to shock/vibration, dust, moisture, etc.
x Simple installation, i.e. inside a borehole or at the borehole collar
x Reasonably priced, either single-use or re-usable
Since only a limited number of experiments could be made a redundant instrumentation with various
different sensors was made. Table 14 provides an overview on different sensors that were tested.
Test 2 had three instrumented boreholes (M3-M5 with 76 mm) and test 4 had six holes (M10M16 with 76 mm and 115 mm; see Figure 32) respectively to measure burden movement.
In all measurement holes an anchor was grouted mid-way in the burden containing axially mounted 1axis PE (piezoelectric) and/or PR (piezoresistive) accelerometers (see section 4.2.2.2.1). For signal
transmission, steel cables each with three embedded electrical cables have been strain-relieved by a
conical termination filled with polyester resin mounted within the anchor. A cable magazine was
placed either behind the burden or at the borehole collar.
Figure 33. Anchor, single components and force-fit mounting with pulling rod. Test 6, hole M1.
57
Blast function
Figure 34. Incremental displacement sensor:

Coaxial cable spear, back side of the protection
casing. Test 4, hole M16.
Figure 35. Continuous displacement sensors: Drawwire and magnetostrictive sensor mounted at the
borehole collar. Test 2, hole M5.
The incremental displacement sensors, i.e. the coaxial cable spear (Olsson et al., 2009) and the fibreoptic zebra gauge (see section 4.2.2.2.2), could be installed closely behind the burden. To them the
movement of the anchor was transferred by a threaded hollow pulling tube, which also housed the
steel cables and was guided in a protection casing. A bolting of the latter to a mounting plate at the
collar in the adjacent drift was designed to keep the casing fixed in the boreholes so that a true relative
movement is measured. For the coaxial cable spear the pulling tube had a cutting knife and coaxial
cables were circumferentially mounted in the casing. For the fibre-optic zebra gauge a coded, blackand-white, adhesive foil was detected by 2 fibre photoelectric sensors mounted in the casing.
Due to their required space, some of the sensors, i.e. continuous displacement sensors (draw-wire and
magnetostrictive sensor) and the laser-light-barrier were mounted on a stand at the collar. The
movement of the anchor was transferred by the steel cable passing through a flexible teflon-coated
steel hose which gives both protection and a friction reduction. The spring system of the draw-wire
(rated for an acceleration of 1350 g) was used to tension and retract the cable assembly.
The instrumentation is in detail described in Report 6.
mounting stand
cable magazine
58
Type
Accelerometers
Continuous displacement
sensors
17
SPEAR
FIBRE
Coaxial cable
spear
Fibre-optic zebra
gauge
LASER
Laser distance
sensor
LASER1
MAGNET
Magnetostrictive
sensor
Laser light barrier
DRAW
PR1
Draw-wire sensor
Piezoresistive
accelerometer
PE
Piezoelectric
accelerometer
PR
Abbr.
Sensor system
GYcRS
GYFC2
gauge
controller
custom-made
permanent
magnet
WLL190T-2
IRF1800
reflex foil
Imos
Sick
Sick
LL3-DB02
amplifier
SigiCom
any
Sensor
Instruments
Sick
Santest
Firstmark
Controls
Endevco
Endevco
Endevco
PCB
M300
RG58, RG174
D-LAS-HS
transmitter,
receiver
coaxial cables
trigger
detector
fibre sensor
OD2-250W150
gauge
MG-R-2756
161-2145H-BSS
gauge
136
71-60K
gauge
amplifier
136
35011220KG
PCB
480C02
amplifier
gauge
amplifier
Endevco
Endevco
7255A-01
gauge
Producer
2793
Name
Components
Relative distances measured by

detection of black-and-white
coding, i.e. diffuse reflection
measured by a photoelectric system.
Position measured as time of travel

of a laser beam.
detection of a hole matrix, i.e. lightdark switches measured by a lasertransmitter system
detection of cable cuts, i.e. short
circuits measured as drop in
resistance.
Position of a permanent magnet

measured as time of travel of
mechanical pulses in a
magnetostrictive delay line.
Position measured by rolling up a

flexible cable from a spring-loaded
drum with angular potentiometer.
Acceleration measured as change in

electrical resistivity of a
semiconductor subjected to
mechanical stress.
Acceleration measured as
generation of electrical charge in
dielectric crystals subjected to
mechanical stress.
Principle
Table 14. Overview on different tested sensors.
Blast function
flexible
flexible
flexible
100-400
700
679
uncertain
uncertain
16667
infinite
300000
1330
3000
infinite
0-136000
0-10000
3-10000
Hz
mm
uncertain
Frequency
Range
59
++
+,o,-
Robustness17
Characteristics
Rating estimated based upon the factors shock, vibration, dust and moisture; explanation of symbols: +high, omedium and -low
Incremental displacement
sensors
--
++
++
+,o,-
Costs
4, 5, 6; lab
2, 4, 5; lab
5, 6
2, 4, 5
6; lab
4, 5, 6: lab
2, 4, 5, 6; lab
Test id
Blast function
4.2.2.1.2
Unconfined situation (test 2)
Filling of the drift for test 2 (see Figure 36) was incomplete, both in height and in length, which led to
a relatively unconfined blasting situation. This resulted in excessive material movement upwards and
outwards from the crosscut into the connecting longitudinal drift, and a wide trench was left in front of
the blast site, see Figure 37. Due to a staggered pillar layout only partial resistance from an opposite
drift wall was provided. Under this circumstance, the required filling degree in combination with a
small burden was assessed too conservatively.
18
Figure 36. Test site 2 with height isolines for filling.
Figure 37. Blast result with excessive material

movement, test site 2.
The collected data is a result of a combination of factors: the blasting situation, the sensors in use and
their mounting. As a consequence the time delays and the characteristics of the curves differ. The
horizontal velocity and displacement signals are summarized in Figure 38.
900
60
LASER (M3)
800
50
SPEAR (M4)
DRAW
(M5)
700
MAGNET (M5)
displacement [mm]
velocity [m/s]
40
DRAW (M3)
30
20
10
DRAW (M5)
DRAW (M3)
600
500
400
300
PE (M3)
LASER (M3)
200
PE (M3)
SPEAR (M4)
100
MAGNET (M5)
20
25
30
35
40
45
50
55
60
65
-10
20
time [ms]
25
30
35
40
Figure 38. Horizontal velocity and displacement, test 2.
18
45
time [ms]
Blastholes in red and measurement holes in green colour.

60
50
55
60
65
Blast function
With the blastholes detonating almost simultaneously at 20 ms, the anchor bearing the PE
accelerometer in hole M3 first moved after 1.9 ms. The instrumentation behind the burden in hole M4
(coaxial cable spear) came 6.5 ms after detonation and 1 ms later the instrumentation at the borehole
collars of holes M3 and M5 started to move. The comparatively long reaction time for the coaxial
cable spear could be caused by its incremental operating mode, i.e. an initial movement is required
until the first cable is cut, and by its comparatively high weight.
The velocity signals from the PE accelerometer and the coaxial cable spear clearly showed decay in
velocity, but the other gauges did not. Based on the blasting result with a massive cast of material, an
early retardation effect and an abrupt physical stop effect can be excluded. Hence an effect on the
initial measured burden movement and decrease in velocity is not expected within the measurement
range of the applied sensors. However, possible sources of error for the measurement could be found.
There was no linear guiding for the anchor with in-built 1-axis accelerometer. As there was no
resistance of masses either the anchor would most likely rotate during its trajectory at an early stage so
that the sensitivity in the direction of movement was reduced. For the coaxial cable spear, it was found
after retrieval that the protection casing moved unintentionally together with the spear because
concrete had penetrated into the casing during installation. The thick-walled casing moved about 1.5 m
from the wall and got stuck as it was severely bent. It was verified that the actual measuring tube with
installed cables did not move. However, a large frictional effect of the tight-fitting casing against the
borehole wall slowed down the movement and influenced the read out.
By contrast, the gauges linked to the cable assembly (draw-wire, magnetostrictive sensor and laserlight barrier) did not show any decrease in velocity for the designed measurement range of 700 mm.
Good agreement exists between the different measurement methods installed for the same borehole.
The mean velocity is approx. 42 m/s (M3) and 45 m/s (M5). There is a cyclic variation of the velocity
for both draw-wire sensors as they reach their maximum level. This might be related to the design of
the potentiometer and fast withdrawal of the steel cable. With the blast result in mind it is considered
possible that no retardation for the initial movement took place. According to this it is difficult to
assess if the spring system of the draw-wire could counteract the high inertia of the mass.
4.2.2.1.3
Confined situation (test 4)
A much better degree of fill was achieved for test 4 and a nearly confined blasting situation was
achieved. No movement on the surface of the muckpile was measured during blasting (high-speed
photography), or deduced from post-blast inspections. At about 100 ms after detonation, some gases
started to leak through the pile. Remote inspections through the instrument holes with a long-range
borescope and observations during mucking showed that there existed a void between the pillar and
the blasted material (Figure 40).
61
Blast function
19
Figure 39. Test site 4 with height isolines for filling.
Figure 40. Blast result with void between the

blasted burden and pillar, test 4.
The blasted material moved about 1.2 1.6 m in the upper part of the drift section. The areal
compaction is in the range of 4 5 % based on the void area in Figure 40. It is likely that the value of
compaction could be adjusted upwards as there was partial collapse of the void before or during the
mucking operation. The confinement, simultaneous initiation and stress situation in the present test
differed from regular SLC blast rings. Voids in operational SLC mining were identified to be related
to disturbed flow, i.e. the shallow-draw phenomenon (Selldn & Pierce, 2004; Power, 2004a-b;
Paper C, Report 8). Nevertheless, it raises the question of whether a bulking, or even voids closest to
the ring plane, are also present directly after blasting. This needs to be investigated, see section 7.2.
During mucking it was also observed that the burden has moved more or less as one unit during
blasting. It could be characterized as very coarsely fragmented, and was difficult to muck, with some
boulders that were as large as 1.5 - 2 m.
For test 4, all anchors with built-in accelerometers were linearly guided by a pulling tube within a
casing. The post-blast inspection revealed that the anchors were well fastened within the
concrete/borehole and moved with the masses towards the debris. Problematic, however, was that the
casings started to move together with the accelerating anchors and the surrounding material. The
casings likely became bent at the very beginning due to the shock waves arriving and causing the inner
pulling tube to become stuck inside the casing. A friction mechanism between casing and the throw of
material is also conceivable. The desired relative movement between the inner pulling tube and outer
casing, if any, only occurred at the very beginning of the movement. These mechanisms disturbed all
systems that were based upon a relative motion between an outer casing, with built-in measurement
19
Blastholes in red and measurement holes in green colour.

62
Blast function
system, and a pulling tube as the moving part. However, this problem can be solved (see section
4.2.2.2.2). Reasonably good measurement results for test 4 were obtained with the accelerometers and
draw-wire sensors. The horizontal velocity and displacement signals are summarized in Figure 41.
1300
35
30
1100
PE (M13)
PE (M15)
25
900
displacement [mm]
velocity [m/s]
PE (M13)
PE (M15)
20
15
PE (M16)
10
700
PE (M16)
500
DRAW (M13)
300
DRAW (M14)
100
0
15
25
35
45
55
65
75
-100 15
time [ms]
25
35
45
55
65
75
time [ms]
Figure 41. Horizontal velocity and displacement, test 4.
The burden movement during test 4, as measured with accelerometers, indicated a similar rapid oneor two-stage acceleration phase to a maximum velocity of 16 - 32 m/s, followed by a retardation
phase. In particular, the curves of the PE accelerometers for the very closely spaced holes M15 and
M16 exhibited very similar curve characteristics for the initial movement. The final stop and its
characteristics were not captured as the maximum displacement (necessary cable length > 1 m) of the
burden was underestimated. The total displacement measured corresponds to the maximum available
cable length of the accelerometer system. It is therefore reasonable to assume that the results obtained
up to this point were reliable.
As the entire system moved violently, i.e. anchor with pulling tube and casing, the bolting to the
neighbouring drift did not withstand the forces. Damage occurred to the cabling within the borehole
and at the collar to the mounting plates (bent or broken). This way, the draw-wire sensors were also
badly affected as they did not record more than about 300 mm of displacement. The draw-wire results
did not capture the initial acceleration phase, but after a delay of 3 - 4 ms the velocity signals swiftly
rise within 0.5 0.6 ms to the accelerometer levels. The initial jerks and the signal delays of the drawwire are either wave propagation phenomena or the response of the gauge related to excessive
acceleration (see section 4.2.2.1.1). The displacement records between 20 - 30 ms are quite reasonable
though compared with the integrated PE accelerometer signals.
63
Blast function
4.2.2.1.4
Numerical modeling
A numerical modeling investigation based on the results from the burden movement measurements
(see section 4.2.2.1.2 and 4.2.2.1.3) was undertaken using Blo-Up software, a component of the
Hybrid Stress Blast Model (HSBM). The objectives of the study (Paper D; Furtney, 2011) were
twofold: (i) to gain additional insight into the process of blasting in confined conditions and (ii) as a
verification study of Blo-Up. The Blo-Up software models non-ideal detonation, rock fracturing and
muckpile formation. The software is described in detail by Furtney et al. 2009. A so-called flexboundary condition models the behaviour at the boundary between the intact burden and the caved
material in an SLC ring. It shows compaction behaviour in which the modulus increases with
displacement and unloading occurs along a stiffer modulus.
Tests 2 and 4 described herein were analyzed using Blo-Up. Although test 2 was relatively unconfined
it allowed for a calibration of the rock mass properties of the burden. The objective was first to test the
model by exploring the simpler unconfined case before the more complex case of confined blasting
was studied.
Figure 42 is a cross-section of the confined model showing the boundary swell, fracturing and back
break. As in the field measurements a gap centered on the boreholes is observed.
20
Figure 42. Cross-sectional view of model predictions for burden swell.
20
Contours are for displacements of 0 m (blue) to 1.2 m (red), 60 ms after detonation. The point at which the
velocity was measured was on the intact-rock / broken-rock boundary adjacent to the lower hole.
64
Blast function
Figure 43 gives a summary of the results of the numerical investigation. In both the unconfined
conditions of test 2 and the confined conditions of test 4, Blo-Up reproduces the burden velocity and
displacement observed in the experiment. In the unconfined case the burden accelerates to a velocity
of about 45 m/s. In the confined case, the maximum velocity reached at the same point in the burden is
about 30 m/s.
Test 2: Unconfined situation
60
Test 4: Confined situation
40
Data
Blo-Up
35
50
velocity [m/s]
velocity [m/s]
30
40
30
20
25
20
15
10
Data
Blo-Up
10
5
0
0
25
30
35
40
45
15
50
time [ms]
25
35
45
55
65
75
time [ms]
Figure 43. Comparison of burden velocity measurements and Blo-Up model predictions.
4.2.2.2
Measurement system for burden movement
Based upon the experiences of the drift wall blasting blasts and parallel on-going laboratory tests
(Paper D, Report 6), a new measurement system installed closely behind the burden was developed
(Paper D, Report 7). The essential parts of the measurement system are illustrated schematically in
Figure 44 and in practice (test 6) in Figure 45.
Figure 44. Measurement system installed behind the burden (not to scale).
65
Blast function
Figure 45. Main components of the measurement system. Test 6, hole M1.
The aim is to have a system, which captures both the fast acceleration phase and the slower movement
until rest. The measurement system consists of a hollow pulling tube with an anchor incorporating an
accelerometer suited for long duration transients, which is grouted inside the burden. Behind the
burden, a displacement measuring system with fibre-optic sensors is mounted in a shock-absorbent
casing, which is kept in position. A coded tape is fastened to the pulling tube and detected by two fibre
photoelectric sensors. This allows detection of a change in the direction of movement.
4.2.2.2.1
Accelerometers
Both piezoelectric (PE) and piezoresistive (PR) accelerometers are widely used shock accelerometers.
However, their applicability for integration of the signal to velocity and displacement is often
troublesome and related to specific construction-conditioned problems (Chu, 1992).
Despite PR accelerometers having been improved considerably in their usable bandwidth (> 500 kHz)
and rigidity, the sensor can still be set into resonance as the gauge mechanism is practically undamped.
The result of this failure is complete loss of data and eventually permanent gauge damage. PE
accelerometers on the other hand are more robust under the same conditions, but have other
deficiencies:
x Inability to hold their charge output during a steady input (DC response).
x Zero-shift, i.e. failure of a sensors output returning to zero after a dynamic event.
The phenomenon of zero-shift arises during an explosively generated shock (pyroshock) as a
unidirectional shift of unpredictable polarity and amplitude. Measures to reduce the effects of zeroshift are the design of the accelerometer, i.e. avoidance of overstress of sensing element and sensor
part movements as well as care in the installation and set-up, i.e. avoidance of frictionally generated
cable noise (triboelectric effect) and base strain.
66
Blast function
For example, the design of the PE accelerometer, 7255A-01 (Endevco) is an attempt to eliminate the
zero-shift problem. It incorporates a built-in mechanical filter for sensor isolation and a matched
electronic low-pass filter in order to remove unwanted residual ringing superimposed on the
accelerometer output signals. Several independent test results have shown that the design virtually
eliminated the difficulty of zero-shift (Chu, 1988; Yiannakopoulos & Van der Schaaf, 1998).
4.2.2.2.2
Fibre-optic zebra gauge
The use of incremental displacement sensors to determine the time for initial face movement and the
rate of burden movement is not new within the field of blasting. Different systems, which yield timedistance data from the detection of cable breaks or cuts were successfully applied by Noren (1956),
Volchenko (1977) and Olsson et al. (2009).
The fibre-optic zebra gauge is an improvement in terms of resolution and ability to detect both forward
and backward motion (see section 4.2.2.2.4).
Two fibre photoelectric sensors detect the movement of the object by the detection of a coding printed
on adhesive foil. Detection occurs when the light beam, incident on the target, is reflected by the target
and received. The direction of movement is revealed from the sequence of the incoming signals. Close
spacing of the stripes gives an increased resolution to identify the acceleration and retardation phases
of the movement.
4.2.2.2.3
Laboratory tests
Simple laboratory tests with the aim of evaluating the performance of different accelerometers and
comparing their integration results with incremental displacement sensors (coaxial cable spear, fibreoptic zebra gauge) were carried out (Report 6). The measurement range was varied between 150 to
750 mm with different stoppage characteristics simulated. A constant velocity in the range of 3 6 m/s
could be achieved. The integrated accelerometer signals were compared to velocity and displacement
measured at the probe by the incremental displacement sensors. The coaxial cable spear proved to be
difficult as the cutting of coaxial cables induced ripples in the acceleration signal. By contrast, the
non-contact fibre-optic zebra gauge did not exert any influence on the acceleration signal. The results
from the PE unit and the PR unit looked very much alike, see Figure 46.
The integrated data coincided well with the velocity and displacement signals obtained with the fibreoptic zebra gauge. The stop was recognizable as a transient spike in the velocity signal and a kink in
displacement signal just before 2.10 s. For the time of movement (approx. 40 ms) zero-shift did not
prove significant. Continuing though to integrate the signal of the PE unit after the stop, revealed later
problems of zero-shift. This appeared as a ramp in velocity and a parabola in displacement. This
problem was also directly noticeable when comparing the signal levels before and after the shock
event. The mean velocity ratio just after and just before the stop event is 0.72 for the PE and 1.01 for
the PR unit (18 measurements). Even so, the PR unit did not perform better after shock.
67
Blast function
200
FIBRE
PE
PR
6
displacement [mm]
velocity [m/s]
150
4
100
50
FIBRE
-2
2.06
2.07
2.08
2.09
2.10
2.11
0
2.06
2.12
2.07
time [s]
2.08
2.09
2.10
PE
2.11
PR
2.12
time [s]
Figure 46. Velocity and displacement: PE and PR accelerometer compared with fibre-optic data; abrupt
stop at 150 mm.
The dynamics from a burden movement against confinement in terms of rock masses would
presumably look different with a) a much higher acceleration and initial velocity and b) a gradual
retardation effect as the confining masses get compacted. For the latter reason the stop used in the
form of hard plastic was replaced by a compression spring. A pronounced bounce effect and damping
500
400
300
200
-2
-4
1.85
displacement d [mm]
velocity v [m/s]
which reached a peak indicating the stoppage of the probe could be found, see Figure 47.
100
PE (v)
FIBRE (v)
PE (d)
FIBRE (d)
1.90
1.95
2.00
0
2.05
time t [s]
Figure 47. Velocity (v) and displacement (d): PE accelerometer compared with fibre-optic data, suspended
stop at 400 mm.
The use of PR accelerometers required care in cable selection to reduce triboelectric noise and lowpass filtering (10 kHz). By contrast, the PE accelerometer type is reliable due to a secure mounting
arrangement, ease of use, robustness and noise free, consistent results in the laboratory tests.
Comparing displacements and velocities derived by the PE unit with the fibre-optic measurements
pointed to reasonable agreement. On average, the velocity was increased by 7 37 % (165 meas.) and
the displacement decreased by 9 11 % (378 meas.) for the PE gauge.
68
Blast function
4.2.2.2.4
Field tests
For further development of the described measurement system (Report 7), collaboration was
undertaken with the project Rock support subjected to dynamic loading (Shirzadegan et al., 2011) in
which the effectiveness of various ground support systems under dynamic loading conditions are
tested. Rock burst loading is simulated using explosives in holes drilled behind differently reinforced
drift walls. These tests with an artificially generated initial movement and a possible rebound effect of
the drift wall were ideal for the development of a technique to measure face velocity and
displacement. Two tests were carried out at block 9, on the 741 m level at the Kiruna mine. Both were
single-hole shots with instrumentation either in front of the burden (test 5, drift 93, right side), or in
front and behind the burden (test 6, drift 95 98).
In test 5, a movement of at most about 0.1 m with a subsequent gentle rebound was recorded. In test 6,
the limit case of breakage was achieved as the burden for the section of lower charge concentration
moved slowly as a single block for about 0.5 1 m. The section with higher charge was unexpectedly
broken, but showed rather coarse fragmentation.
The fibre-optic zebra gauge gave an output that was comparable with other measuring techniques
12
100
FIBRE (d)
SPEAR (d)
80
60
40
velocity v [m/s]
SPEAR (v)
displacement d [mm]
velocity v [m/s]
FIBRE (v)
2.0
450
1.8
400
1.6
350
1.4
300
1.2
250
1.0
200
0.8
d = 1.24t + 3.68
2
R = 1.00
0.6
0
10
20
30
40
50
60
70
80
100
0.4
50
d = 1.70t - 7.31
R2 = 1.00
0.2
0
150
FIBRE (d)
LASER1 (d)
20
-3
FIBRE (v)
displacement d [mm]
when installed in front of the burden, see Figure 48 and Figure 49.
0.0
90
-50
0
time t [ms]
100
200
300
400
500
time t [ms]
Figure 48. Test 5, velocity (v) and displacement (d):

coaxial cable spear and fibre-optic data.
Figure 49. Test 6, velocity (v) and displacement (d):

laser distance sensor (Sick OD2-250W150) and
fibre-optic data.
Thus the gauge was a) fast in response, b) could detect changes in the direction of movement with a
high resolution and c) flexible in its measurement range. The limit of detection depends on the
frequency of the amplifier (response time of 60 s, i.e. 16.7 kHz), its settings and the width of coding
that is to be detected. During test 6, it was observed that for the current settings and a chosen blackwhite coding with stripes of 5 mm width, a velocity of 30 m/s is probably the upper detectable limit.
The field tests also showed that the double integrated accelerometer signals did not give an accurate
representation of the complete burden movement, but only of the first part. A splicing of the initial
69
Blast function
accelerometer data with the results of the fibre-optic zebra gauge is needed to record the whole
movement.
100
PE (v)
FIBRE (v)
PE (d)
FIBRE (d)
10
80
60
40
-5
20
-10
0
0
10
20
30
40
50
time t [ms]
60
70
80
displacement d [mm]
velocity v [m/s]
15
90
Figure 50. Test 5, velocity (v) and displacement (d): PE accelerometer and fibre-optic data.
Figure 50 compares velocity and displacement from the integrated accelerometer signal with the fibreoptic zebra gauge. The PE accelerometer picked up the initial wall velocity, its maximum
displacement and a rebound effect which thus gives a result comparable to that of the fibre-optic zebra
gauge. The observed double peak in velocity for the wall mounted accelerometer might have a
physical explanation as it represents the arrival of the shock wave followed by other reflections from
the drift face and boundaries of the pillar. A similar observation was made for a draw-wire sensor
mounted at the same section of the wall, but further down towards the floor. Possible explanations for
the later discrepancy between the integration result of the accelerometer signal and the fibre-optic
zebra gauge are i) a zero-shift problem occurring at a very early stage (after 5 - 10 ms), ii) mounting
issues for the fibre-optic zebra gauge, i.e. a lifting of the floor and subsequent low-frequency
oscillation of the mounting stand which was also observed from high-speed photography (at ~ 25 - 35
ms).
The excellent performance of the PE accelerometers achieved in the laboratory tests could not be
repeated during the field tests. It was observed that their failure is often related to specific spurious
pulses in the signal that occur during the movement and mask the actual signal in the low frequency
region. It remains uncertain if these are related to electrical disturbances, or to any real wave
reflections and mechanical impacts. The pulses result in a zero-shift and visible offset in the
integration result. The expected difference in behaviour of the PE and the PR accelerometers in the
field did not materialise. If such a difference exists it may become apparent during very slow motion
and as motion ceases.
70
Blast function
With the measurement system installed behind the burden in test 6, the initial face velocity was
measured. Then, as a result of the blasthole located nearby and penetrating gases, a blow-out of the
system occurred. The blow-out forced the fibre-sensing part of the zebra gauge backwards and
accelerated the apparent outward movement of the anchor. A curve splice of the initial integrated
accelerometer signal and the subsequent fibre-optic signal would yield the whole time history of the
movement, see Figure 51.
60
600
PE (v)
FIBRE (v)
PE (d)
FIBRE (d)
500
40
400
30
300
blow-out starts
20
200
10
100
displacement d [mm]
velocity v [m/s]
50
-10
-100
0
10
20
30
40
50
time t [ms]
Figure 51. Test 6, velocity (v) and displacement (d): PE accelerometer and fibre-optic data.
As the fibre-optic zebra gauge measured a relative movement between the anchor and the casing, the
starting position needs to better defined and the casing movement during measurement prevented.
Important constructive measures are therefore a grouted casing as well as an integrated stop ring on
the pulling tube (Figure 44). For the pulling tube either steel or aluminum was used as construction
material. An inertial effect was not measured though. With the small deformations observed and easier
handling the high-tension aluminum alloy Al 7075 is preferable.
71
Blast function
4.2.3 Breakage of burden

The breakage of burden was examined by time domain reflectometry (TDR) in coaxial cables grouted
in boreholes inside and behind the burden. Details of the technique and an overview on the
applications in geotechnical engineering can be found in O`Connor & Dowding (1999).
A TDR system responds to changes in impedance along the cable. Responses related to different
failure mechanisms will therefore be identified, like stretching or crushing without breaking, shortand open-circuiting. For low-f TDR systems the entire waveform signature is captured along the cable,
while for high-f TDR systems only the closest, severe deformation (short- or open-circuit) is registered
in distance and time. Its sensitivity to such deformations can be adjusted (discriminator setting).
TDR is used in caving mines to verify the cave initiation process and to serve as a long-term data
source on cave growth to be used for production schedules (Report 2). An investigation on the effect
of blasting and mucking on rock mass fracturing behind the blasted SLC rings was made by Krekula
(2004). Experiments using short time TDR to obtain rock motion data (crater formation, rock
fractures) are scarce (Schmitt & Dick, 1983; Dick & Edwards, 1985; Brent, 2010).
Inside the burden (see Table 12), cables were grouted in holes parallel with the blastholes and
monitored during blasting with fast-sampling (200 kHz) TDR (ShotTrack, SpeedVOD). The aim was
to find cable breaks that correspond to fracture propagation and whether the spacing and relative
occurrence of these could be related to the status of drift wall confinement.
From behind the burden (see Table 12), cables were grouted in holes penetrating between the
blastholes and monitored after blasting with static TDR (TDR 100). This monitoring determines the
effective burden and verifies that the planned breakage was complete and if overbreak occurred.
4.2.3.1
High-frequency (high-f) TDR measurements
4.2.3.1.1
Pretests
Crack propagation inside the burden was simulated in a simplified experiment based upon clear cable
cuts (Report 6). A detonating cord (Primacord 5 with 5.3 g/m) was tied around a measurement cable at
varying spacings, simulating a propagation of cracks at slow velocities, both 1000 and 2000 m/s. The
cable reflections measured with the fast-sampling TDR system (ShotTrack) showed excellent
agreement with the theoretical path and the signal stayed at the level of the actual cable length. Within
the same test, the fast-sampling TDR was also compared to a resistance wire system (DataTrap II). It
was found that the latter had problems in recognizing the exact point of the cable cut with the signal
subsequently either dropping or staying at this level.
A side-by-side comparison comprised the measuring of VOD of different explosive products (Report
6). The TDR gave reliable, repeatable and noise-free signals. Its wide range of application, to measure
72
Blast function
both high and low order explosive performance, deflagrations, malfunctioning explosives and
stemming behaviour was shown previously (Chiappetta, 1998).
4.2.3.1.2
Field tests
The fast-sampling TDR was repeatedly applied to monitor the breakage of coaxial cables grouted
parallel to the blastholes in the burden in the drift wall blasting test series (see section 4.2, Table 12).
It proved that a TDR cable initially could measure the velocity of detonation (VOD) from the blasthole
and continue recording pulses caused by cable breakage from the measurement holes in the burden.
For the measurement in the burden, it seemed that often the shockwave was registered within a certain
distance from the blasthole. A de-coupling of the cable did not make a difference.
Individual pulses registered and related to physical cable breaks seemed to exist only inside the
unconfined zone (test 2 and 3, see Table 12). Sparse data were recorded from the confined zones in
test 3 and 4 (test 3 and 4, see Table 12). An interpretation is that the movements against rock masses is
so limited, that the burden moved as one unit and hence cable breakages within the burden were
unlikely to occur. From test 4, large, undamaged cable pieces were retrieved while mucking, which
supports this interpretation.
4.2.3.1.3
Development of measuring method
It was concluded that it would be better to use a decoupled, sensitive cable with low crush threshold.
This cable would need to either withstand or recover from the initial shockwave and afterwards retain
sufficient sensitively so that small movements could be recorded in the form of a cut or short-circuit.
The foam-core coaxial cables tested did not seem to be appropriate for this task. The use of airdielectric coaxial cables (Chiappetta, 1998) was considered as an alternative. Pre-tests were made with
a modified bamboo construction, but failed due to impedance mismatches with the TDR units. Aircables with a spiral construction were unfortunately not available at this time.
Another approach could be to the development of a device with a very fast arm and pre-trigger
facilities (Valentim, 2010), which allows capturing of the raw data pulse and that hence would be
similar to present low-f TDR instrumentation. This way, even small deformations along the cable
could probably be analyzed during blasting and burden movement.
4.2.3.2
Low-frequency (low-f) TDR measurements
4.2.3.2.1
Laboratory tests
Laboratory shear tests comprising 9 cable samples were performed at Complab (LTU), see Report 2.
To study the influence of the shear angle on the TDR response, different coaxial cables were grouted
centrically into steel pipes. A servo-controlled testing machine (Dartec, 600 kN capacity) was used to
shear the samples along predefined cutting planes of the steel pipes, see Figure 52. The shear angles
(0, 30 and 45) were linked to the failure mode (see Figure 53) as an angle was defined between the
73
Blast function
centre line of the pipe and the shear plane at the sliding, with the sign depending on the direction of
movement (down- or upwards).
Figure 52. Testing machine with clamped

sample, Complab, LTU.
Figure 53. Different shear failure modes.
The tests focused on the following issues:

1. Correlation between TDR reflection magnitude21 and shear displacement
There existed a non-linear relationship between the reflection coefficient () and the actual shear
displacement (mm), see Figure 54. Especially in the beginning, the reflection amplitude to shear
displacement ratio (/mm) was quite small and the TDR system therefore insensitive. The sensitivity
at small deformities is dependent on the cable type, dimension and the response of the TDR.
2. Effect of different shear angles and failure modes
The shear displacement curves depended on the shear angles involved, see Figure 54. The larger the
shear angle, the smaller the reflection coefficient was at a given amount of displacement. The shape of
the reflection coefficient versus time, by contrast, did not reveal any information about the shear angle.
The mode of grout fracturing, cable deformation and failure affected by the shear zone width (i.e. gap
between pipe halves) effects the TDR signal. Localized shearing causes a distinct, large increase in the
reflection amplitude, whereas a longitudinal extension of the cable produces only a small, but
increasingly wide reflection with little increase in amplitude (Su, 1987; Dowding et al., 1988).
The width(m)/magnitude(-) ratio of the TDR reflections was suggested as a measure for the actual
shearing mechanism and failure of the cable since bending of the cable through a large shear zone
should give greater ratios than localized shear within small gaps (O`Connor & Dowding (1999).
However, a tendency towards larger ratios with increasing shear angle was not found (Figure 11 in
21
The reflection coefficient () is defined as voltage out (Vout) / voltage in (Vin).

74
Blast function
Report 2). Nor was there any observable increase in the ratios as the tests near completion at larger
shear displacements. It can thus be concluded that the decreased sensitivity (/mm) for increased shear
angles (Figure 54) is mainly attributable to an effect of the angle itself.
0.35
reflection coefficient [-]
0.30
0.25
0.20
0.15
shear angles: 0, 30, 45
0.10
campaigns I-III: cable type, failure mode

I: Smooth (S) Al-cable, shear+compression (hollow symbols)
II: Smooth (S) Al-cable, shear+tension (grey symbols)
III: Corrugated (C) Cu-cable, shear+tension (black symbols)
0.05
0.00
0
10
20
30
40
50
60
70
80
shear displacement d [mm]
Figure 54. Reflection coefficient versus shear displacement.
22
Different signal responses in the reflection coefficient versus time curves due to the actual shear
failure mode (see Figure 53), i.e. shear associated with compression or tension (down- or up-slip) may
exist. If so, they could at least within this test series not be identified.
3. Influence of cable length (attenuation and dispersion)
Cable length had a significant and quantifiable effect on TDR reflections (Figure 14 in Report 2).
Around 10 - 15 % attenuation in the reflection coefficient could be expected per 10 m cable extension.
This reduction in reflection magnitude results from an attenuation of the signal along the cables as
well as signal dispersion which increases the pulse width (Su, 1987). This way, longer cables also
affect the resolution to localize individual deformities as the width of the reflected signal also
increases. In addition, the width increases with progressive shear displacement on the cable, which
finally influences the possibility to resolve narrowly spaced cable deformities.
4. Influence of upstream crimps
With artificial upstream crimps placed in known positions and close to the expected deformation, the
waveform may be sampled only for a specific zone and hence resolution increased (Dowding et al.,
22
For campaign II, 45 shear angle, smooth Al-cable (II-S-45), a shear off did not occur. The outer smooth Al-
conductor began to slide relative to the jacket which caused the cable to be stretched in the shear zone.
75
Blast function
1988). A clear effect of upstream crimps on downstream deformations was not identifiable unless the
crimps are very large (Figure 15 17 in Report 2). This finding is consistent with that of Pierce et al.
(1994), which showed that multiple deformities upstream have a marginal effect. Thus, artificial
crimps (in the range of 10 - 20 m23) are considered to enhance the position accuracy of TDR
measurements for the identification of deformations on long cables.
5. Cable type
Two coaxial cable types, i.e. a smooth Al- and corrugated Cu cable ( 20 mm) with different
mechanical and electrical properties were tested. It was found that the corrugated Cu cable was
considerably more sensitive (m/mm shear displacement), but was associated with a smaller
maximum displacement before a final shear occurred. Both cable types needed a minimum shear
displacement (5 mm) before any movement was detected. For both, the level of noise is around 1 m.
4.2.3.2.2
Field test
Static TDR measurements were performed on cables grouted in holes intersecting the burden of blast
site 2 (see section 4.2.2.1.2) in the confined drift wall blasting tests (Report 6). The aim was to
ascertain if the burden breakage could be verified and distinguished from shear deformations as
simulated in the laboratory tests. A couple of parameters, such as cable type, diameter and application
of crimps which could possibly influence the length reading accuracy were varied.
The post blast monitoring showed that the blast induced dynamic failure, i.e. the burden breakage,
showed a clear difference to a more localized shear failure: The cables were stretched extensively
which yielded a different signal width(m)/magnitude(-) ratio RO for the respective reflection
coefficient curve. The blasting test gave RO >> 1 for the final cut-off of the cable and the laboratory
shear tests gave RO < 0.5 just before the cable was sheared off.
The remotely detected burden breakage from TDR measurements was as close as 0.1 m as compared
with the verified position after mucking of the test site was completed. In particular, the construction
manner of the corrugated Cu cable was sensitive to tension failure. The sensitivity (reflection
coefficient ) of this cable type was 4 times greater than for the smooth Al cable type of the same
diameter. A smaller absolute error was identified for thicker cables ( 20 mm vs. 15 mm). Using
crimps did not improve the location of cable breaks. The situation might change when considerably
longer cables are used (Pierce et al., 1994).
23
Artificial crimps were produced by squeezing the cable with an adjustable vise-grip wrench (grip head with 10
mm). For the used cables a 30 % diameter reduction was found to be optimal as a reference TDR reflection.
76
FRAGMENTATION
Indications existed that the Kiruna magnetite could have different fragmentation characteristics than
normal hard rock. Previous blast damage investigations had e.g. shown that the cracking behind halfcasts from perimeter blasting in drifts in magnetite looked quite different from that behind half-casts in
waste rock (Nyberg et al., 2000). In the hard waste rock the crack pattern is mainly radial, with many
short crushing cracks at the borehole and a smaller number of long distinct cracks. In ore the crack
pattern is more diffuse. Such a diffuse network of cracks would probably occur if the magnetite
showed plastic behaviour.
Quantitative assessment of the fragmentation of blasted rock on a larger scale is a difficult task. First
of all, obtaining representative sampling constitutes a major problem. Secondly, sieving would at
present appear to be the only reliable method of assessing fragmentation quantitatively in a production
environment.
The objective of this investigation was to find out whether the magnetite behaves like waste rock from
a blasting (more specifically fragmentation) point of view (Paper A, Report 1). Fragmentation
measurements from small-scale tests to run of the mine studies are presented. On a model scale, the
natural breakage characteristics (NBC; Steiner, 1991 & 1998) were determined and compared to
model-scale blasts. In full-scale, fragmentation was measured by sieving material loaded from one
specific SLC blast ring. In addition, results from 2D image analysis were compared with the actual
sieving data (Report 4).
77
78
Fragmentation
5.1
NBC approach and small-scale blasts
The NBC approach is based on the concept that a fragmented material exhibits a specific natural
breakage characteristic. The concept originates from mechanical comminution studies (Steiner, 1991
& 1998) and can be summarized as follows (Ouchterlony, 2003):
1.
When rock particles are broken in the steps or sub-circuits of an optimum comminution
sequence (OCS) in the laboratory, the resulting fragmentation curve is the steepest
possible. Both the comminution sequence and the ancillary conditions guarantee that, for
the OCS, a machinery independent fragmentation with focus on the material
characteristics takes place.
2.
When the product stream of each sub-circuit is classified, the resulting log-log
fragmentation curves are shifted parallel vertically upward as the comminution progresses.
3.
When the specific surface As (cm2/kg) created by an OCS is plotted versus the energy
consumed Es (J/g), the points fall more or less on a straight line. This line is termed the
energy register function of the material. It is material specific and its slope equals the
Rittinger coefficient R (cm2/J).
4.
All technical, i.e. non-optimal, comminution processes produce points As that for a given
Es value fall below the energy register line because they are less energy efficient.
Moser et al. (2000 & 2003) showed that the NBC approach may be applied to the blasting of model
cylinders and full-scale blasts in the same rock type with similar results. The OCS built up to analyze
the breakage behaviour of Kiruna magnetite was similar to that used within the Less Fines Project
(Moser, 2005) but differed in the size reduction ratios (Rohrmoser et al., 2007). The sequence,
designed on a succession of numerous comminution stages in closed circuits with pre-screening and
the setting of each apparatus, guaranteed a small size reduction ratio. The circuit design in the
laboratory was simulated by cyclic comminution tests at a high circulating load (minimum 100 %).
To determine the NBC of Kiruna magnetite, samples in the form of drill cores with 140 mm were
taken from two different mine sites at level 878, block 28, drift 287 (M3) and block 16, drift 148
(M4). The samples were residues from a previous small-scale blasting campaign (Johansson et al.,
2007).
Fragmentation curves from the respective mechanical comminution stages are compared to those from
model-scale blasts in Figure 55 and Figure 56. The first three mechanical fragmentation steps (streams
1, 4 and 7) refer to a jaw crusher whereas the subsequent two steps refer to a rod mill (streams 10 and
13). The form of the curves is in reasonable agreement for crushing, grinding and blasting of
79
Fragmentation
magnetite on model scale. The agreement for the magnetite quality M4 is exceptionally high. This
indicates that the magnetite reacts to blasting like typical hard waste rock.
mass passing [%]
100
10
NBC, stream 1
NBC, stream 4
NBC, stream 7
NBC, stream 10
NBC, stream 13
M3-2-05 confined
M3-4-05 confined
M-3-1-05 unconfined
M-3-3-05 unconfined
0.1
0.01
0.1
10
100
1000
mesh size [mm]

Figure 55. Fragment size distribution for NBC tests and model-scale blasts, M3 ore type.
mass passing [%]
100
10
NBC, stream 1
NBC, stream 4
NBC, stream 7
NBC, stream 10
NBC, stream 13
M4-1-05 confined
M4-2-05 confined
M4-3-05 unconfined
M4-4-05 unconfined
0.1
0.01
0.1
10
100
1000
mesh size [mm]

Figure 56. Fragment size distribution for NBC tests and model-scale blasts, M4 ore type.
Differences especially in the blasting results for the M3 quality might be related to internal variations
of material properties. With respect to the shape of the curves it is noticeable that the M3 ore type has
generally a more pronounced inflection point at the downwards slope. Differentiating the curves
mathematically shows that the minima for the M3 type are compared with the M4 type located at
slightly smaller mesh sizes (0.5 and 0.7 mm, respectively) which indicates a finer grain assembly
80
Fragmentation
(Grasedieck, 2007). The number of tests is not sufficient to tell whether the position of minima could
also depend on the method of comminution (mechanically or blasting in confined/unconfined
situation).
Plotting the energy register curves for the two different magnetite samples, see Figure 57, yields a
significant difference with respect to the breakage behaviour. The harder to blast M4 quality has a
smaller Rittinger coefficient, 30 cm2/J than the M3 quality, 46 cm2/J. This difference is exceptionally
large for one rock type and both samples are highly resistant to fragmentation due to a relatively small
Rittinger coefficient (Moser, 2007).
specific outer surface As [cm2/g]
350
300
As = 45.97Es + 22.55
250
R2 = 0.969
As = 30.01Es + 17.49
R2 = 0.991
200
150
100
50
M3
M4
0
0
10
12
cumulative net energy consumption Es [J/g]

Figure 57. Energy register functions for both ore types M3 and M4.
5.2
Previous full-scale fragmentation tests
Table 15 gives an overview of previous fragmentation measurements carried out on blasted rock in
Kiruna (Lundberg, 1961a-b; Maripuu, 1968). The present mine layout and operational conditions are
quite different so a direct comparison of their results may be misleading.
Maripuu`s (1968) investigation of fragmented caving debris is the most extensive one because of the
large number of samples studied at different extraction rates. For a few samples even information for
the fines (0.074 - 10 mm) range is available. Both, the three and five parameter Swebrec functions
(Ouchterlony, 2005) are a good fit with the data set, see Figure 58 and do not show any irregularities
in the breakage behaviour. From a historic data point of view the magnetite ore thus shows
fragmentation characteristics which are comparable to those of hard waste rock. The parameter x50
covers a wide range and its average value amounts to 86 mm. No trend of x 50 versus extraction rate is
directly noticeable.
81
Fragmentation
Maripuu also made an attempt to divide the caving debris masses into pure magnetite ore and waste
rock, wherein it was observed that the fragment size distribution for waste rock is considerably coarser
than that for ore with the average values x50 = 161 mm and x50 = 79 mm respectively. Further, the
percentage of fines for the waste rock was much higher than for the ore, i.e. the fragment size
distribution is much flatter.
Table 15. Previous fragmentation measurements for blasted rock in Kiruna.
Characteristics
Year
Unit
Lundberg
Lundberg
Maripuu
Wimmer
1961a
1961b
1968
Paper 1, Report
Caving debris
Drifting
Caving
Caving debris
Sieved material
Level
248
257
302
320
907
Coor
Y26 - 27
Y27 - 28
Y24
Y35 - 36
Y37
Drift
239 - 240
355 - 357
377, ring 7
Ore type
Sublevel interval
28.5
Blasts
Individual samples
41
Extraction rate
begin/mid/end
10 - 202 %
7.5 - 9.0 %
Total masses
kg
180 323
44 000
197 683
96 454
Screen sizes
mm
10 - 500
10 - 500
Mining
area
10-500 (41)
0.074-10 (4)
0.063-200
1.0-1.2
1.8
2.1
1.8
Hole diameter
mm
33.5-36
51
32
51
115
Holes / round
12
12
53
10
Drill metres / round
66
70
120
50.8 - 53.4
159.9
tonnes
361.4-
591.3
180.0
534.0 - 550.0
6 591
Explosive type
Dynalit
Dynalit
& ANFO
ANFO
KR0500 &
Kimulux 82
Specific charge
kg/tonne
0.12
0.19
0.33
0.15 - 0.17
0.23
Burden / advance
Tonnage / round
: 0.22 kg/tonne (ANFO)

: calculative break down in ore and waste rock
: visually classified as pure ore without any contamination
82
Fragmentation
Figure 58. Average fragment size distribution for caving debris, Maripuu`s data (1968).
5.3
24
Recent full-scale fragmentation tests
5.3.1 Test layout and working procedure

A total of 6 buckets and 96 tonnes of magnetite ore were taken from one ring (level 907, block 37,
drift 377) of the Kiruna mine (Paper A, Report 1). The buckets were loaded in series after an initial
extraction of 20 buckets corresponding to an extraction rate of ~ 7.5 % from a totally loaded mass of
6591 tonnes. Since the actual ring was situated close to the hanging wall it had not yet reached full
height and thus the standard layout of the SLC rings has not been used. An undisturbed previous
extraction as well as operational conditions led to the selection of this site for the test.
The local rock conditions were judged to be competent and undisturbed. The lower part of the ring
consisted of low phosphorous B1 and B2 ore (Fe ~ 67 %, P ~ 0.01 %) and the upper part to some
extent of high phosphorous D ore (Fe ~ 59 %, P > 0.3 %). B2 ore is, compared to B1 ore, slightly
more inhomogeneous and typically contains green actinolite and pyrite (Niiranen, 2008).
The ring comprised 7 boreholes with diameter 115 mm, a front angle of 80 and a burden of 3 m to the
next ring plane. A total of 159.9 drilling meters and a theoretical in-situ volume of 4462 m3 yielded a
specific drilling amount of 0.17 m/m3. Since water was present in some of the boreholes almost half of
the charging (47 %) required a substitution of the standard bulk emulsion (KR0500) by the packaged
24
5 blasts; 41 samples > 10 mm and 4 samples < 10 mm sieved; Extended (5 parameters) Swebrec function fit.
Data range 0.074 - 800 mm. Curve fit parameters: x50 = 86.1 mm, xmax = 1072 mm, a = 0.999, b = 2.893, c = 2.1
with r2 = 0.999.
83
Fragmentation
emulsion Kimulux 82. The actual specific charge thus amounted to only 0.23 kg/t instead of the
planned 0.35 kg/t. Apart from this, the charge columns and initiation were implemented according to
the operating standard (see section 3.2).
Figure 59. Working procedure for full-scale sieving test.
In order to ensure loading of pure magnetite the 6 individual samples were taken well before any
waste rock inflow. The whole test campaign is summarized by Figure 59.
Images to analyze fragmentation (see section 5.3.3) were taken both from the loader buckets filled
with material and piles spread out (un- and re-mixed) on a plane tarmac surface. Furthermore, 3D
models were computed by means of photogrammetry (SMX3D; Gaich et al., 2009) which for the
moment only provides a detailed documentation of the piles. The sieving was accomplished with a
mobile, track mounted, circular vibrating screen unit (Finlay Hydrascreens, Finlay 883). A moderate
feed rate at the apron conveyor was chosen to feed the 2-deck screen-box. The top deck of the mobile
Finlay 883 sieving unit had cross beams, which cover almost all available openings. Furthermore, only
one amongst four possible screen segments was used for the 200 and 100 mm screens. Thus the
accuracy of the corresponding values was considered questionable and they have therefore been
excluded in the analysis. The amount as well as the size of boulders has been recorded. Laboratory
samples were taken from the stream at the conveyor belt from the masses < 63 mm and sieved down to
0.063 mm.
84
Fragmentation
5.3.2 Sieving results

5.3.2.1
Sieving curves for 0 - 63 mm fraction samples
Figure 60 shows the cumulative fragment size distributions for all 6 lab samples from 0 - 63 mm. At
first sight a pronounced self-similarity can be observed since the curves are shifted upward in parallel
in the log-log diagram at least up to 5 mm and excluding sample 1 up to 10 mm. This is expected
when fine material follows the NBC concept. Inadequacies in the sampling method may explain the
slightly deviant curve for sample 1.
A curve fitting to the averaged lab sample data using the 3 parameter Swebrec function, see Figure 61,
gives a near perfect fit with an r2 value of 0.999 over a large interval from 0.063 - 63 mm.
Furthermore, the residuals are stochastically distributed. To adjust the fitting in the fines region a
weighting of the squared residuals by 1/(x) has been made.
mass passing [%]
100
10
1
0.01
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
Sample 6
Average
0.1
10
100
mesh size [mm]

Figure 60. Sieving curves for 0 - 63 mm fraction samples, log-log scales.
Figure 62 shows the continuous GGS (Gates-Gaudin-Schuhmann) exponent plots (Ouchterlony &
Moser, 2006) derived from the extended Swebrec function for both the present-day sieving data and
Maripuu`s data (caving debris). All curves fall on top of each other, which indicates that the sieving
curves from the lab samples are of NBC character (Ouchterlony & Moser, 2006). It is clearly visible
that the minima for the actual as well as the old data occur at x ~ 0.35 - 0.40 mm. Based on experience
the local position can directly be related to the petrographic character of the investigated ore
(Grasedieck, 2007). The minimum lies where the majority of grains change from polymineralic
assemblies to monomineralic pieces.
85
Fragmentation
25
Figure 61. Average fragment size distribution for 0 - 63 mm fraction samples with Swebrec function fit.
0.8
0.7
GGS exponent [-]
0.6
0.5
Sample 1
0.4
Sample 2
Sample 3
0.3
Sample 4
Sample 5
0.2
Sample 6
0.1
Average
x ~ 0.35 mm
0.0
0.01
0.1
Maripuu: caving debris

10
100
1000
mesh size [mm]

Figure 62. Continuous GGS exponent curves for actual full-scale sieving and Maripuu`s data.
Interestingly the described characteristic is not associated with the chemical condition since the
investigated magnetite ore types are disparate (low versus high phosphorous ore type). Thus the
chemical condition does not seem to have altered the characteristics of the grain assembly much. A
petrographic study of the different ore types would shed more light on this matter.
25
Basic (3 parameters) Swebrec function fit. Data range 0.063 - 63 mm. Parameters: x50 = 8.04 mm, xmax = 83
mm, b = 2.239 with r2 = 0.999.

86
Fragmentation
5.3.2.2
Evaluation of complete sieving curves
The construction of complete sieving curves is described in detail elsewhere (Paper A, Report 1).
Substantial differences between the fragment size distributions for six consecutively taken buckets can
be found, see Figure 63. The parameter x50 varies between 14.3 mm (sample 5) and 277.6 mm (sample
2). Similar large variations in the fragment size distribution for blasted magnetite ore were observed
earlier by Maripuu (1968). On average x50 amounts to 86.5 mm, however, and this is directly
comparable to the mean x50 value of 86 mm that Maripuu (1968) measured in earlier tests (see 5.2).
The shapes of the complete sieving curves do not entirely fit the Swebrec function. This is somewhat
surprising as a deviation from this behaviour is an exception (Ouchterlony, 2009). An increased
amount of fines < 40 mm causes a flattening and thus non-uniform continuation of the curve to coarser
fractions.
At first sight the curves are similar to those obtained by blasting near the critical burden, i.e. a limited
number of large fragments plus a fines tail that follows the Swebrec function (Ouchterlony, 2009).
This is a tempting explanation as the specific charge in the SLC ring was substantially lower than
normal, see section 5.3.1. However, a disturbance-free extraction for the investigated ring makes this
assumption rather doubtful.
mass passing [%]
100
10
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
Sample 6
Average
0.1
0.01
0.1
10
100
1000
mesh size [mm]

Figure 63. Constructed, complete sieving curves for all samples.
On the other hand, the flattening effect observed for the mid sizes may also be due to selective
breakage behaviour and it is comparable to the shape of fragment size distributions typically affected
by autogenous grinding (Lynch, 1977; Hahne et al., 2003). This is a direct analogy to the secondary
fragmentation in block caving (Brown, 2003; Eadie, 2002) as the blocks constituting the primary
87
Fragmentation
fragmentation move down through the draw column to the respective draw point. A similar occurrence
seems to be likely in the drawing of caving debris from a high SLC ring. Larsson (1998) found that the
fragmentation for caving debris was much finer than for the blasted ore and this phenomenon could
therefore be present at an early loading stage at the draw point.
Such a process may cause a depletion of medium size fractions, but still leave strong large fragments.
Several of the following mechanisms might be relevant for causing secondary fragmentation:
x Self-breakage depending on shape and strength of particles
x Abrasion effects between particles (autogenous grinding)
x Crushing of fragments under superimposed load
x Fracturing and crushing of larger blocks under different forms of load
Because of the low specific charge within the blasted ring (see section 5.3.1) more coarse material
than usual can be expected and would constitute an ideal feed for the described mechanisms related to
larger fragments.
Much self-breakage was observed during the sieving process as many of the fragments that dropped
from the conveyor belt split apart when hitting the truck tray. Previous investigations of ore processing
characteristics have in particular shown that a) a significant fraction has in fact the unusual tendency to
break very quickly within a batch mill, thus generating a large fraction of very fine material (Lichter,
2007) and b) the resistance of feed samples both to impact- and abrasion breakage increases markedly
as the distance of sampling from the mining face is increased (Hahne et al., 2003).
The draw height and consequently the stress regime as well as the corresponding draw rate may be
influencing the resulting secondary fragmentation. The sublevel height in the mining layout is now
28.5 m. It was 9 m when Maripuu (1968) made his study, in which the curves show no distinct effect
of secondary fragmentation. Effects from both confined blasting and longer flow path on
fragmentation are likely. On the other hand, the present buckets were taken before 10 % draw, which
indicates relatively short paths, unless the draw is extremely uneven, see section 6.2.2.
The presence and extent of secondary fragmentation effects is difficult to verify. However, two ways
to detect it are: a) the shape and surface properties of fine particles could reveal if the material has
been subjected to autogenous grinding (Forssberg & Zhai, 1985), and, b) the whole process could be
studied in full-scale with defined output parameters within an ore pass.
Segregation and thus preferential flow of certain fractions in the SLC extraction process could also
help to explain the observations.
88
Fragmentation
Finally, the details of the sieving process, with some openings covered underneath by beams of the
machinery and reduced number of sieve segments used, could have distorted the results. The possibly
tainted data was excluded, but is not certain that this was entirely successful.
Despite this flattening deviation from Swebrec distribution behaviour which is likely to be caused by
the internal flow mechanisms in the SLC process the main conclusion is still that the magnetite
behaves like waste rock from a fragmentation point of view.
5.3.3 2D image analysis (WipFrag software)
Sieving is the method that defines the true fragment size distribution if the process itself is done
without error. Methods based on digital imaging to determine fragment size distribution of rock are
widely used though. These have the major advantage of producing fast estimates of size distributions
or fragmentation characteristics without interfering with production (Ouchterlony, 2003). The
accuracy of fragment size distribution by image analysis is of continued interest (Maerz & Zhou,
1998; Sanchidrin et al., 2009). The available data set from full-sieving made it possible to study the
capability of the fragmentation analysis software WipFrag (version 2.6; Maerz et al., 1996) in more
detail. Fragmentation was assessed both from 2D images of buckets (underground) and piles spread
and re-mixed (on the surface). To quantify the results of the WipFrag evaluation the derived percentile
sizes (x25, x50, xc, x75 and x90) were compared with the percentiles from full-scale sieving for the
respective sample (Report 4). Three questions were of interest:
1. Manual versus automatic net delineation (640 480 resolution)
The automatic evaluation (Figure 5 in Report 4) based on LKAB`s standard edge detection parameters
(Lith, 2004) displayed a systematic error as small grains were fused and larger fragments had
disintegrated. Reasonable correlation curves (Figure 7 in Report 4) could be established for bucket and
pile images which allow correction of automatically analyzed images. However, linking old recordings
to these fragmentation calibration shots using the WipFrag camera on-site under real production
conditions would be required. The latest version (version 2010, build 6) of WipFrag incorporates reworked edge detections settings. Rapid first-trials using the built-in auto detection settings have given
better results compared with LKAB`s standard parameters to generate a fragmentation net.
The manual evaluation (Figure 6 in Report 4) showed that the deviations with a maximum of 0.1 m are
of more erratic nature and generally show a coarser result for the image analysis than the sieving data.
Deviations from the sieved data in the fines region (0 0.2 m) arose from the low image resolution
which makes identification of small particles difficult. By contrast, any percentile size in the range of
0.2 0.4 m gave a reliable estimate. Differences again occurring within the coarse region might be
explained by there being fewer particles within these size classes. This results in an image section
89
Fragmentation
which is not representative for the entire sample (segregation effect), and in the exposure of particles
which are not representative when compared with a quadratic sieve opening (flakiness effect)
2. Influence of image resolution, i.e. 640 480 versus 1600 1200 resolution (Figure 64)
For the 640 x 480 images, the deviations in the fines region (0 - 0.2 m) were within or above the
magnitude of the actual sizes of interest. Reliable estimates could be expected for the mid-range 0.2 0.4 m, but differences occurred again in the coarser fractions and might then be due to sampling error.
deviation (WipFragmanual - sieving) [m]
0.4
640 x 480
1600 x 1200
0.3
0.2
0.1
-0.1
-0.2
-0.3
0 - 0.05
0.05 - 0.1
0.1 - 0.2
0.2 - 0.3
0.3 - 0.4
> 0.4
percentile sizes (sieving) [m]
Figure 64. Influence of image resolution. Data count (24/12/18/15/12/9).
For the 1600 x 1200 images, the estimation of sizes in the fines region was considerably improved.
The effect measured as mean deviation from sizes determined by sieving amounted to 59.1, 48.9 and
30.5 % in the size classes 0 - 0.05, 0.05 - 0.1 and 0.1 - 0.2 m respectively. Almost no deviation from
the actual sieved sizes was recognizable for fractions 0.2 - 0.3 m. Again, deviations occurred at coarser
fractions than that and might be explained by sampling error.
3. Segregation effects (see Figure 65)
Severe segregation is likely to occur during the loading process and be present in the images taken
from buckets, which causes a wider spread of the measured sizes compared with pile images. This
effect increases in the large size fractions. Other factors may be: a) different lighting conditions
(underground with artificial light vs. surface with daylight), b) approx. 4 times larger image sections
(buckets 2.7 1.7 m and piles 5 m) with more particles being apparent (2.5 times at 1600 1200
and 1.9 times at 640 480) and c) autogenous grinding effects (see section 5.3.2.2).
90
Fragmentation
deviation (WipFragmanual - sieving) [m]
0.4
buckets
piles mix 1
piles mix 2
0.3
0.2
0.1
-0.1
-0.2
-0.3
0 - 0.15
0.15 - 0.3
> 0.3
percentile sizes (sieving) [m]
Figure 65. Segregation effects. Data count (16/7/7).
The larger image section (1600 x 1200) and more particles identifiable within the pile images do not
substantially improve the problematic deviations from the actual sieve sizes within the fine fractions.
The errors related to bucket images appear to be more erratic whereas comparing piles before (mix 1)
and after re-mixing (mix 2) a tendency for smaller sizes can be recognized after materials handling.
91
92
GRAVITY FLOW
Monitoring gravity flow was in the past typically done by markers (Brunton, 2009). They were
recovered either visually at the draw point or by magnetic separation during the later material handling
process. Recently, radio frequency identification (RFID) tags became available, which allow tracking
of ore flow from the blast to the downstream processes (Whiteman, 2010; Wortley et al., 2011).
Smart Markers were developed and used by Brunton (2009) to study gravity flow in SLC mining. It
was expected that the system could be made available during this project but its development was
delayed a long time. Instead, some unconventional approaches were taken to study gravity flow.
The feasibility of using different geophysical borehole probes to identify variations in caving debris
was evaluated with field test in drifts filled with ore and waste (Report 3). Besides some promising
results, clear limitations related to the inhomogeneous nature of the material studied were found. It
was concluded that some of the limitations probably could only be overcome by a successful
combination of existing methods and new techniques that would take too long to develop.
The 3D image acquisition system (see section 4.1) was successfully used to survey flow disturbances
(Paper C, Report 8). There could be several causes for flow disturbances. However, these situations
might be viewed as only a snap-shot of the gravity flow mechanisms related to the previous rings for
which production figures (extraction, ore recovery) showed a disturbance-free extraction. In addition,
the effects of confined blasting on swell and breakage become clear. Based on these full-scale
observations two conceptual models are proposed, one for confined blasting and one for gravity flow.
93
94
Gravity flow
6.1
SLC flow disturbances
6.1.1 Occurrence of hang-up situations

In practice, two different hang-up situations occur:
x Boulder blockages are caused by the trapping or wedging of one or usually several large
blocks in or above the draw point. Remedial action might be secondary blasting.
x Frozen hang-ups are caused by inadequate space for swelling of blasted ore, as only the
material closest to the blast plane is sufficiently broken to be mobilized. The material further
away is either incompletely broken (ore) or heavily compacted in previous flow zones (ore
and waste) and therefore temporarily builds up as an arch.
The latter type of hang-up is often not directly obvious as it may occur higher up and might be
misjudged as a remnant pillar, in which a break-through to the upper level and cave was not
established yet. However, the extraction figures for a remnant pillar situation are usually much lower
and hence enable a distinction. Remedial action for this type of hang-up might be the continued
mucking, water flushing with a special tank vehicle, and the blasting in adjacent drifts or of
subsequent SLC rings. There are furthermore many hang-ups which are not registered, but which are
taken down by skilled mucking operators.
Among all 519 hang-ups, registered in connection with the use of the tank vehicle, between the years
2002 2011, clear drops in yearly occurrence are first apparent for the year 2004 and then for 2007,
see Figure 66.
120
number of affected drifts
100
number
hang-ups / Mtonne SLC production

80
60
40
20
0
2002
2003
2004
2005
2006
2007
2008
2009
2010
year
Figure 66. Hang-up statistics, year 2002 - 2011.
95
2011
hang-ups / Mtonne SLCproduction
number of total hang-ups
Gravity flow
There are many possible causes for this decrease, both operational and human factors. The first drop in
the number of hang-ups (2004) coincides with an organizational change in the charging work that was
made within LKAB. A cause for the second drop might be the change to a silo-shaped ring design (see
section 3.1) in production. This layout was successively introduced at the mine level 907 m where
production started in 2006.
Around 1/3 of the hang-ups occurred in the same drift as an earlier hang-up, which would indicate a
systemic bias. A breakdown of the data with respect to depth showed that the 560 m and 600 m levels
(northern part, so-called Sjmalmen or ore under the lake ) had a much larger number of hang-ups
than the average (8.5 - 12 hang-ups / Mtonne versus on average 3.6 hang-ups / Mtonne). Also, the
levels 767 m 820 m had an increased occurrence of hang-ups (4.5 7.4 hang-ups / Mtonne). With
respect to the distribution of hang-ups along the strike of the orebody an accumulation is seen for
block 16 (4.2 hang-ups / Mtonne); no other problem area is directly recognizable.
During 2007 - 2011, the average number of hang-ups was below 35 occurrences per year and can
therefore not be regarded as a general problem for an annual SLC production of around 22.3 Mtonne
(excluding production of the year 2009).
6.1.2 Observations from 3D mapping
Six so-called frozen hang-up situations were mapped in 3D. Based upon the current mucking state at
the survey they can be grouped and are schematically depicted in Table 16.
Table 16. Surveyed hang-ups classified according to type (vertical section).
1: drift 121, ring 33+34, 767 m level
3: drift 125, ring 6, 792 m level
Hang-ups
Type
96
Gravity flow
A representative image and corresponding section is plotted in Figure 67 and Figure 68 respectively.
Figure 67. Hang-up observation (type B) with waste

lenses, drift 135, 792 m level.
Figure 68. Section at 776.8 m and projection of

waste lens from ring 7, drift 135, 792 m level.
A gap between the ring plane and the previously blasted material was observed. The gap narrowed
with height and its closing was caused either by a clogging of individual rock pieces or by insufficient
breakage. Thereby, the burden was separated from the ring plane and the burden material showed
hardly any disaggregation. If there was mobilization it occurred along holes with bridges in between.
Problems while charging resulting in a reduced amount of charged explosives (20 - 40 %) might have
been the initializing factor for most of the observed hang-ups (hang-ups 2 - 6). In a single case (hangup 5) an undetonated hole was found. However, there could be several causes for the flow disturbance.
Gravity flow for the SLC rings previous to the hang-up is assumed to have preferentially occurred
along the ring plane within the identified flow zones. No major flow disturbances were recorded.
Extraction rates varied as usual, 127 54 %, but were on average higher than normal with a
corresponding ore recovery of 104 55 %. For a long period of extraction also a very limited waste
rock inflow was observed.
The extraction rates in connection with the hang-ups could vary a lot dependent on the mobility of the
material, the loading procedure and if material was extracted through the gap. For two cases (hang-ups
2 and 3, Table 16) the dimension of the hang-up and the extraction rate at the time of survey revealed
that an extraction through the gap was likely until it finally got clogged by rock pieces. In particular a
cyclic inflow of waste rock was observed and this is most likely connected with the formation of the
hang-up (see section 6.2.2). Two flow disturbances (hang-up 2 and 5, Table 16) existed only
temporarily. Mucking continued after the survey with extraction rates of 180 % and 93 % respectively.
Comparative data is given in the following Table 17 for the surveyed hang-up and Table 18 for the
mapped flow zones from previous rings.
97
Gravity flow
26
Table 17. Data for surveyed hang-ups.
extraction
hang-up block level drift
1
2
12
40
767
964
3
4
5
12
792
ring
at survey
total
total
121
395
33+34
10
%
42.4
85.9x
%
42.4
180.0x
%
34.7
152.8
125
68.3
68.3
65.4
56.5
92.7x
79.7
132
135
recovery
14
44.3
37.5
179.8
66.1
190.6
172.1
27
Table 18. Data for mapped flow lenses of previously extracted rings.
hangring
up
1
2
3
4
5
628
32
9
5
4
13
12
7
6
7
6
zcoord.
m
754.0
948.0
769.0
775.0
775.5
776.5
775.8
777.7
756.4
771.2
area
m2
10.9 36
6.4 25
11.1 23
n.a. n.a.
7.6 24
9.5 34
12.3 39
6.1 24
12.8 28
11.6 31
size of flow zones

width
thickness
max,
max,
mean,
%
m
m
m
7.4
70
2.2
1.5
7.0
79
2.3
0.9
10.3
53
1.7
1.1
n.a. n.a.
n.a.
n.a.
5.4
49
2.2
1.4
5.3
54
2.8
1.8
7.2
68
2.0
1.7
6.4
72
2.0
n.a.
11.0
73
2.0
1.2
7.0
53
2.7
1.7
swell extraction recovery

%
12
30
21
12
17
14
12
8
12
12
206.7
186.6x
89.9
176.7x
103.2x
70.6x
90.0x
n.a.
90.0x
n.a.
199.7
162.3x
67.7
138.9x
52.8x
57.2x
76.3x
n.a.
76.3x
n.a.
Among the six investigated hang-ups situations, ten lens-formed flow zones (flow lenses) from the
extraction of previous rings were exposed for which several similarities were observed.
1. Flow zones
Dependent upon the size of the hang-up either one or two lenses with caving masses are related to the
flow zones in the preceding rings (Figure 67). A sharp contact existed between the blasted ore and the
altered, fine-fragmented caving masses (ochre coloured). Several mechanisms, such as crushing
26 x
bucket scale temporarily inactive. Dependent on LHD type a weight of 21 or 25 tonne/bucket is assumed.
27 x
bucket scale temporarily inactive. Dependent on LHD type a weight of 21 or 25 tonne/bucket is assumed.
28
For the preceding ring no. 8 a hang-up occurred and a flow lens was therefore, not present, see Table 16.
98
Gravity flow
through the blast loads, abrasion and fracturing under different forms of loading during gravity flow,
may have caused the fine fragmentation. The type of material seemed to be purely waste rock and
would reflect that mucking was stopped as waste rock inflow became too large for the previous rings.
The identified flow zones were curved and closed up before reaching the side holes. Relative to the
ring width at the specific height of observation (8 31 m) its extension was 63 11 % for the six
hang-ups observed (Figure 68). The thickness of the flow lenses was on average 1.4 0.3 m and hence
about of the burden.
2. Swell and compaction
The observed swell29 of the blasted burden was largest for the mid-part of the ring and decreased
towards the sides. Referred to the planned burden the maximum swell varied between 8 30 %, and
was on average 15 6 % based upon the six hang-up observations. Swell at the sides was
insignificant.
Both the compaction of the caving masses in the previous flow zone and the presence of a void may be
required for the blast to work satisfactorily.
Due to the occurrence of 2 hang-ups in the same drift (hang-ups 5 and 6, Table 16), the same waste
lenses were exposed twice, but at different heights. The appearance of the waste lenses varied, but the
swell of around 8 12 % seemed to be constant (Table 18). This implies that the depth for blast
compaction is rather limited and only concerns the material filling the flow zone of the previous ring.
The ore in-between the waste lenses seems to be incompressible and acts as a limiting wall for the
blast compaction. It appeared to be fractured, but entirely immobilized at the height of the observed
hang-up. It is competent in such a way that it built up the limiting wall towards the hanging wall in all
surveyed hang-up situations.
29
Maximum swell based upon a plot with the projection of waste lens computed as maximum perpendicular
distance between the fitted ring plane and the respective waste lens. Relative swell is based on the planned
burden of 3 m. If an overbreak of 0.5 m is assumed the relative swell would be increased by 20 %.
99
Gravity flow
6.2
Conceptual models
6.2.1 Blast dynamics

Based on 3D mapping results of blast rings (see section 4.1) and hang-ups (see section 6.1.2) a
conceptual model, i.e. plausible sequence of events, for SLC ring blasts is outlined, see Figure 69.
1. Pre-blast situation
For the pre-blast situation a flow zone of the previously blasted ring is assumed to be located between
holes B2 - B3 and hole B7. Its width is about of the burden and its shape curved to the sides. The
front part of the burden is fractured, but immobilized. Available swell volume to the sides did not exist
and therefore only marginal dilation and disaggregation of the burden occurred. One of the side holes
in the previous ring (hole B1) did detonate, but did not break and caused underbreak. Most likely
drilling deviations caused a too large burden/spacing at this height of the ring. Overbreak is constant
for the ring face and about 0.5 m. Some structurally caused overbreak exists as well. The overbreak
zone is included in the flow zone which is filled by caving debris.
2. Detonation of mid holes (B4 and B5); nominal delay 25 ms
Burden movement is initiated due to dynamic loading of the blast holes and dilation of the blasted
material which swells into the flow zone of the previous ring. As this available zone is limited in depth
and width, this implies both compaction and secondary fragmentation of the material filling it.
3. Detonation of holes B3 and B6; nominal delay 50 ms
Borehole B6 detonates as planned. Due to the curved shape of the previous flow zone, the possibility
for swell of the subsequent ring could be assumed to be largest towards the middle of the ring.
Borehole B3 does not denote. Several mechanisms may be the cause for this (see section 4.1.2.3).
4. Detonation of holes B2 and B7; nominal delay 75 ms
Borehole B2 detonates but is heavily confined due to the undetonated adjacent hole. This may lead to
fragmentation problems in this area. Borehole B7 detonates as planned. Swell of blasted material and
compaction of the caving debris increases towards the middle of the ring.
5. Detonation of side holes (B1 and B8); nominal delay 100 ms
The side holes are highly confined in this scenario, resulting probably in poor fragmentation and low
loosening in the vicinity of these holes.
6. Post-blast situation
The post-blast situation reflects a blasted burden with significant spatial differences in fragmentation
and porosity. This affects material mobility and the initial conditions for gravity flow (see 6.2.2).
100
101
Figure 69. Plausible sequence of events in the blasting of a SLC ring (section perpendicular to ring plane at 20 m height, not to scale).
Gravity flow
Gravity flow
6.2.2 Disturbed gravity flow

The full-scale observations from hang-ups (see section 6.1.2) support two main conceptual models: 1)
shallow draw (Figure 70); and, as a result of flow disruption 2) temporary hang-up (Figure 71).
Figure 70. Conceptual model, shallow draw

(vertical section).
Figure 71. Conceptual model, temporary hang-up

(vertical section).
1. Shallow draw model

Gravity flow preferentially occurs along the ring plane in a flow zone with a crescent shaped crosssection. It is limited both to the sides and in depth, which directly reflects the effects of confined
blasting. Swell is largest for the mid holes and marginal to the sides. Breakage is incomplete both at
the sides and for the part of burden furthest from the ring plane. The observed width of these flow
zones is about 50 70 % of the ring width and its thickness about of the burden. In terms of area, no
more than 25 40 % of the planned burden is mobilized. Extraction rates for the observed flow zones
could partly be very high. Inflow of caving debris might occur either from above, or as the LHD
machine penetrates into the flow zone of the previous ring. Supposedly, once the top of the delimited
flow zone is reached, flow might be unrestricted and the upper ring part act more like a block cave
draw point. This would, depending upon material type and characteristics still allow high extraction
rates.
2. Temporary hang-up model
If gravity flow along the ring plane is obstructed, a hang-up situation might evolve. A possible
obstruction is incomplete breakage for the upper ring part or a clogging of the flow zone by rock
pieces. For the latter situation, a gap between ring plane and frozen hang-up would be observed. As
a consequence of this disruption, the flow pattern is altered and material loaded from previous blasts.
With continued extraction, the cavity enlarges and finally become unstable, collapses and both
previously frozen ore and waste rock enter the draw point periodically. In operational mining many
hang-ups are not recorded, but are rectified by the loading procedure. It may thus be the case that the
cyclic nature of inflow of waste rock during operational mucking is caused by temporary hang-ups
that are not detected and hence not recorded.
102
Gravity flow
6.2.3 Similarities with, and differences, from other models

Previous discussions (see section 2.3.2, Table 6) emphasized irregular and asymmetrical shape of the
draw body for large-scale SLC geometries. It follows that waste rock inflow, which is measured by
weighing buckets during the mucking procedure, could also vary greatly. Waste rock inflow could
occur either at an early stage or be delayed but once it occurred it generally exhibits a pulsating
character. The two models suggested in section 6.2.2 would account for this. However, it needs to be
considered in the context of other measurements and observations. Both the observed shallow depth
and the limited extension to the sides for the observed flow zones agree well with previous marker
trials made at Kiruna mine (Larsson, 1998) and Ridgeway mine (Power, 2004a-b; Brunton, 2009). The
visual observations (Selldn & Pierce, 2004; Power, 2004a-b) made occasionally in the past are
similar to the present ones. However, the present visual observations are unique in that they were
made systematically and enabled a detailed study of the flow disruption in 3D.
The appearance of the flow zones varied with height, but they are clearly delimited to the sides and in
depth. The burden was separated from the ring plane and moved as a unit. Only the blasted material
closest to the ring plane was sufficiently fragmented to be extracted. The part of the burden furthest
from the ring plane seems to be fractured, but not disaggregated. It seems to be incompressible and
acts as a limiting wall for the compaction of further blasts (see 6.1.2). This is in contrast to the
proposed theory by Kvapil (2008) in which the formation of a compacted interface is assumed to be
caused by some of the ore being pressed into the gaps of caved waste. This interface temporarily
builds up a stabilized strip and hence allows an extraction of almost the entire burden.
A waste rock inflow between rings penetrating the compact zones of ore seems to be rather unlikely
for regular flow. Differences in material mobility existed in the horizontal depth rather than in the
height of the blasted ring (see 6.1.2). A formation of cavities due to immobility somewhat higher up in
the flow zone and subsequent bulging of waste from previous rings, as assumed in the cavity
formation theory (Hustrulid, 2000), is hence unlikely.
The overbreak observed for the confined blasts was similar to that for the unconfined blast rings (see
section 4.1.3). If no structural large overbreak occurs, the effect of material flow diverging to the
location of overbreak is assumed to be insignificant.
All six hang-up situations (10 observed flow lenses) support the hypothesis that gravity flow prior to
the flow disturbance occurred within the described flow lenses. Production figures showed a
disturbance-free extraction. Therefore, the argument seems valid that the observed consequences of
the disturbed gravity flow models might be an inherent part of production, see section 7.2.
103
104
CONCLUDING REMARKS
The results are discussed and conclusions drawn with respect to development of new measurement
techniques and understanding of breakage and flow. Thereupon, suggestions for future work and
research are made.
105
106
Concluding remarks
7.1
Discussion and conclusions
7.1.1 Development of new measurement techniques

7.1.1.1
3D mapping of SLC blast rings
Unique access that allows mapping of the geometry of the blast is created when a new draw point is
opened and when a hang-up occurs. A remotely operated 3D photogrammetry system was custom
built and used to reconstruct the ring front and/or cavity as a geo-referenced mesh model with colour
information to achieve this (see section 4.1.1.1). The 3D images provided an objective documentation
of the local geology and allowed mapping of the structures to be performed on rock faces, which are
inaccessible with conventional mapping. The blasting results were visualized and evaluated by
software specifically developed (see section 4.1.1.2).
Various blasting situations have been observed and this allowed a deeper study (see section 4.1.2 4.1.4) of: i) geometry of blasted cavity, ii) over- and underbreak and their effects on subsequently
blasted rings, iii) interaction effects between adjacent holes and adjacent rings, iv) mapping of
geological structures and their influence on the blast result, and v) drilling accuracy based on
identified boreholes. Monitoring under hang-up situations (see section 6.1.2), revealed the actual
effects of confined blasting and gravity flow. By providing and quantifying information hitherto
inaccessible, the system has proven to be a valuable tool in understanding SLC breakage and flow.
7.1.1.2
Burden instrumentation of SLC blast rings
The instrumentation of the burden is a crucial factor in studying the effect of confinement on the
dynamics of burden movement and breakage and in quantifying the blast function of an SLC ring. The
instrument development work was initiated so that SLC blasting may in the long term be better
understood (see section 4.2.1).
The main focus was therefore the development of measuring equipment that could be used in fullscale SLC blasting and installed behind the rings. This measurement problem is new and the methods
are untested. Controllable, simplified drift wall blasting tests were therefore carried out (see section
4.2.2.1). In the field tests, blastholes were drilled in a pillar, parallel to a crosscut. The burden
constraint was achieved either by filling the drift with rock masses or reinforcing the drift wall.
Burden movement was measured with different redundant sensors, which were installed in holes
drilled from a parallel crosscut. A system, which combines the initial movement recorded by an
accelerometer designed to minimize zero-shift with the later slower movement displacement recorded
by the fibre-optic zebra gauge, see section 4.2.2.2, proved useful and is recommended for future use.
Piezoelectric accelerometers with an in-built mechanical filter in combination with an electronic lowpass filter gave excellent performance in laboratory tests as the integrated data coincided for a long
time window (up to 150 ms) with conventional measurement techniques. However, the field tests
107
Concluding remarks
showed that the accelerometers tested were not suitable as a stand-alone technique designed to yield
information on complete burden movement, but are complementary. It was observed that their failure
is often related to spurious pulses in the signal that occured during the movement and masked the
actual signal in the low frequency region. The fibre-optic zebra gauge worked reliably and its output
was comparable with those of other measuring techniques when installed in-front of the burden.
Response was rapid and the gauge could detect changes in the direction of movement with high
resolution. The gauges could be built with various measurement ranges, for example 0.5 m, 1 m or
greater.
The maximum velocity varied between 16 32 m/s with an areal compaction in the range of 4 5 %
in the filled drift wall blasting tests. A void existed between the intact pillar and the blasted material.
The velocity for the unconfined situation yielded a considerably higher velocity (42 47 m/s).
Breakage of burden was examined by time-domain reflectometry (TDR) of coaxial cables grouted in
boreholes inside and behind the burden and monitored both during blasting and after blasting, see
section 4.2.3.
The high-f TDR initially measured the velocity of detonation (VOD) from the blasthole and continued
recording pulses caused by cable deformities from the measurement holes in the burden. The
shockwave often was registered from the nearby blasthole. Individual pulses registered and related to
cable breaks were found only inside the unconfined zone and close to the stemming limit. There was
little data recorded from the confined zones. One interpretation is that the movement of the confined
part of the burden is so limited that the burden moved as one unit and hence cable breakage inside it is
unlikely to occur. It is concluded that the successful application of this measurement method is
sensitive to proper cable selection and may require the analysis of raw data pulses with a modified
instrument.
The low-f TDR could be used in full-scale tests to verify burden breakage (over- and underbreak).
This method is especially important in the verification of breakage in the boundary areas and at the
uppermost top of the blast ring as these areas cannot be accessed in any other way. The laboratory
shear tests showed that recording the reflection coefficient with progressive shear displacement (i.e.
monitoring time) yields information about the shear angle of the rock movement. If this information is
combined with the local structural geology a 3D model of the damaged zone behind SLC rings can be
made. The field tests further showed that blast-induced dynamic failure, i.e. the burden breakage,
could be distinguished from other more localized shear failures occurring behind the burden.
108
Concluding remarks
7.1.2 Understanding breakage and flow

The basis for the following discussion of results is the findings from full-scale measurements in
production. This includes results from 3D mapping of both unconfined and confined SLC blast rings
and a large-scale fragmentation study of material loaded from a draw point.
7.1.2.1
Initial opening blasts
It was observed that the boxhole and the adjacent cut holes (2 x 2 m square) broke to their full height,
but problems evolved when widening the opening to the sides, see section 4.1.2.1.
Holes, which as planned, are to be longer than the length of the boxholes especially cause problems.
There is no realistic chance that they would break the burden. In terms of the blast result a
symmetrical opening blast, i.e. boreholes in a ring and in-between individual rings, should be
advantageous. If an asymmetric ring design is applied to the opening rings, the boxhole is preferably
placed adjacent to the longest boreholes. However, equally long box- and side holes also create a
problem related to spacing and initiation. An effect of timing was observed in that a shorter time delay
caused a smaller breakable burden and in that an increased burden required a longer delay. In the
planning of the adjacent holes to the boxhole and cut holes a burden of 2.0 - 2.5 m should not be
exceeded. Subsequent holes further widening the initial space could be spaced up to 3 m and a longer
time delay is recommended.
It was observed that with 12 surveyed initial opening blasts (method B and variety B1, see Table 7)
the hanging wall was not exposed more than twice. This suggests that either the boreholes did not
reach the modeled boundary or that the transition from orebody to hanging wall was more diffuse (e.g.
comprised breccia). The stringent adaption of the ring geometries to the predicted hanging wall that is
presently made is therefore brought into question. In terms of the blast result a symmetrical opening
blast layout, i.e. equally long boreholes (in a ring) and parallel holes in-between rings, should be
advantageous.
A well-defined hanging wall acts as a weakness zone. It was observed that once the hanging wall was
exposed, further SLC rings caused a fall of rock masses and thus a substantial increase in height.
The separate blasting of rings which are planned to be as close as 1 m causes problems. In addition,
drilling deviations could cause them to be even closer or drilled into one another, see section 4.1.4. It
is recommended that this problem be solved by combined drilling and blasting of such rings.
7.1.2.2
Unconfined ring blasts (scale-up of SLC rings)
The sequential scaling-up of SLC rings within the opening method B (see Table 7) implies a difficult
blasting situation, see section 4.1.2.2. In particular, the rapid height increase of consecutive rings
considerably reduced the broken area. On average 56 21 % of the planned ring area was broken.
Variations in blast data and the rock mass structure had no significant effect. Even if the breakage was
109
Concluding remarks
100 %, it did not necessarily lead to higher extraction rates as the stoppage criteria for loading are
dependent upon other factors than in standard SLC production, see section 4.1.2.2
The effective burden varied considerably along the broken ring face. It was on average 2.4 0.4 m,
but could be as large as 3.6 m. The study showed that a smaller effective burden tended to result in a
somewhat greater breakage depth. However, the burden at the maximum breakage depth does not
explain why the breakage stops at this point. The result depends more on incomplete breakage of the
preceding ring and the inability to scale-up the ring heights any further. At the same time, the breakage
angle became acute (< 70).
An attempt was made to identify the malfunctions and to categorize possible causes for the identified
unbroken boreholes, see section 4.1.2.3. About 60 % of all boreholes were incompletely broken. The
majority of the problems are probably related to the specific blasting situation in openings, but a
considerable proportion (10 %) was related to undetonated charges. Interaction effects between
boreholes and within the same ring might be an important problem. The identified problems with
initiation (i.e. post blast existence of the detonating cord) are unexpected and to determine their causes
would require a closer examination, see section 7.2.
The mean overbreak calculated for the entire area of the blast rings (36 surveys) was -0.5 0.2 m. The
mean overbreak occurring at the brow only, was near to 15 % larger (-0.6 0.4 m). The increased
standard deviation for the mean overbreak at the brow damage indicates that the actual amount could
vary greatly in particular cases. The largest overbreak was -1.5 0.7 m for the brow specific area, and
could in some cases even reach the subsequent ring 3 m away. The mid part of the ring is most often
affected.
An overbreak behind -0.5 m occurred for 29 21 % of the broken area. Excessive overbreak higher up
on the ring face was seldom observed, and was then structurally limited to small areas (< 10 m2 with a
maximum of 2/3 of the burden to the next ring).
Neither the specific charge nor the maximum charge per delay stage could explain the resulting overall
overbreak. The position of initiation seemed to have a near significant influence as brow damage
increased from |-0.5| 0.2 m to |-0.6| 0.4 m if the initiation took place closer to the collar.
The effects of unfavorable orientation of structures and of local fallouts were observed from individual
analysis of the blasting results. Possible predictors related to the rock mass structures were tested to
quantify this effect on the overall overbreak. It was found that the strike direction of structures had a
near significant effect as sub-parallel, orientated structures tended to give larger overbreak. The dip
angle and the dip direction did not have any explanatory effect though.
110
Concluding remarks
An analysis of the spatial position of overbreak did not show a correlation with nearby charges.
Overbreak is almost constant over large portions of the face (excluding brow damage). Large radial
cracks all over the face caused by blasting are the predominant cause of overbreak in addition to the
effects of natural jointing. The effect of the brow acting as a free face became apparent and increased
overbreak could be expected up to a ring height of about 12 m.
Drilling accuracy and errors in setting-out, collaring and alignment were studied based upon identified
borehole sockets and half casts, but this was work of preliminary nature.
7.1.2.3
Confined ring blasts (hang-ups)
The effects from confined ring blasting and gravity flow could be investigated in hang-up situations,
see section 6.1.2. Production figures (extraction, recovery) indicated a disturbance-free extraction
prior to the hang-up. Extraction rates varied as usual, 127 54 %, but were on average higher than
normal with a corresponding ore recovery of 104 55 %. A very limited waste rock inflow also was
observed for a long period of extraction. Problems while charging resulting in a reduced amount of
charged explosive (20 40 %) and hence breakage problems might have been the initializing factor
for some of the observed hang-ups. However, there could be several causes of the flow disturbance.
Among the six investigated hang-ups, ten flow zones from the extraction of previous rings were
exposed for which several similarities were observed.
Gravity flow for the SLC rings prior to the hang-up rings preferentially occurred along the ring plane
within easily identified flow zones, called the shallow draw phenomenon (Power, 2004 a-b; Selldn
& Pierce 2004). Due to inadequate space for swelling of blasted ore, only the material closest to the
blast plane and the mid part of the ring is sufficiently broken to be mobilized and the material further
away is fractured, but immobilized.
The flow zones in the hang-ups were sharply delimited and filled with altered, fine-fragmented caving
masses (ochre coloured). Several mechanisms, such as self-breakage and abrasion during material
flow and crushing effects through the blast loads might explain the fine fragmentation. Its extension
relative to the ring width at the specific height of observation (8 33 m) was 63 11 %. The thickness
of the lenses was on average 1.4 0.3 m and hence about of the burden.
The observed swell of the blasted burden was largest for the mid-part of the ring and decreased
towards the sides. Referred to the planned burden the maximum swell varied between 8 30 % and
was on average 15 6 %. Swell at the sides was insignificant.
Both compaction of the caving masses in the previous flow zone and the presence of a void may be
required to achieve a satisfactory blast. The ore in-between the flow lenses seems to be incompressible
and acts as a limiting wall against which compaction of the blast takes place. It appeared to be
111
Concluding remarks
fractured and entirely immobilized at the height of the observed hang-up. It is competent insofar as it
built up a limiting wall towards the hanging wall in all surveyed situations.
The overbreak observed for the confined blasts was similar to that for the unconfined blast rings. Very
seldom was overbreak so large that flow significantly diverged to the location of that overbreak.
Based upon these observations, conceptual models for confined blasting and disturbed gravity flow are
outlined, see section 6.2. It is suggested that waste inflow occurred from above or, as the LHD
machine penetrates into the previous ring filled with waste, from that ring. Alternatively, if flow along
the ring plane was obstructed, a hang-up situation evolved. As the extraction zone progressed
upwards, the cavity became unstable, collapsed and both temporarily frozen ore and waste rock
from the flow lenses of previous rings entered the draw point.
7.1.2.4
Fragmentation measurements
The fragmentation characteristics for blasted magnetite ore from a specific SLC ring were
investigated, see section 5.3. For comparison lab-scale data from crushing and grinding (section 5.1)
as well as blasting plus historical full-scale data (Maripuu, 1968) from the mine were analyzed
(section 5.2).
The present data confirms that the material basically follows the NBC criteria within the fines region
(< 10 mm). The chemical conditions of Maripuu`s samples differ from those of the study. However,
the minimum in the GGS plots coincides with the investigated samples and lies at x ~ 0.35 0.40 mm.
A petrographic study considering different ore types would probably give information about possible
interrelations. The complete fragmentation distribution for our buckets yields an extremely large
variation for the parameter x50, within 14.3 278 mm, with an average of 86 mm, the same as in
Maripuu`s data. However, the shapes of the present sieving curves do not entirely fit the Swebrec
function. This is surprising as a deviation from this behavior is an exception (Ouchterlony, 2009). The
details of our sieving process, with openings covered by beams of the machinery could have distorted
our results. Exclusion of critical data was an attempt to compensate for this distortion.
The relative flattening of our sieving curves may also be due to selective breakage in the mid-range,
which would increase the relative amount of fines. This resembles the behavior in autogenous
grinding, i.e. self-breakage of large fragments creates pebbles that then grind the mid-fractions finer,
and is also an effect attributable to secondary fragmentation in block caving. For SLC, also blast loads
also could be assumed to have a crushing effect on the compressible material. In addition segregation
effects, and thus preferential flow for certain fractions, could explain the observations.
112
Concluding remarks
The sublevel height was 28.5 m at the time of the study. It was 9.0 m when Maripuu made his study in
which altered fragmentation characteristics were not observed. From the larger SLC layout one could
expect more comminution to occur in the debris flow due to longer flow paths. On the other hand, the
present samples were taken before 10 % draw, which would indicate relatively short paths, unless the
draw was very uneven (e.g. shallow draw phenomenon).
The main conclusion is that magnetite behaves like normal hard rock from a fragmentation point of
view despite this flattening deviation from Swebrec distribution behaviour as this is likely to be caused
by the internal flow mechanisms in the SLC process.
Finally, the sieving data was compared to the results from 2D image analysis, see section 5.3.3. In
conclusion, several shortcomings of image based fragmentation analysis clearly are apparent and are
related to the automatic net delineation, an influence of image resolution, and segregation effects, see
section 5.3.3. A continuous fragmentation measurement for SLC production may not be needed in
general. There might, however, be situations where an image based fragmentation analysis could be
useful, e.g. as a comparative method for the verification of effects of operational changes.
7.2
Proposed future work and research
Detailed operative suggestions for an improved design of the initial opening blasts and scale-up of
SLC rings are described and discussed in Report 8. A group of designated persons was established
recently who will optimize the current drill and blast work and develop standards.
SLC has undergone major changes in the past (see section 2.1). The present silo-shaped layout of the
SLC blast rings was adapted to the findings of flow research from markers. Compared to the previous
fan-shaped layout an increased proportion of boreholes are confined. In addition, the scale consistently
increased and the technique was applied to depths for which comparative experience does not exist.
Thus higher in-situ stresses and cave pressures might affect the blast performance and subsequently
gravity flow of broken rock. Other studies (see section 2.3.2) and this work (see section 6) indicate
that gravity flow is of asymmetric and disturbed character. Operating figures for the extraction of these
flow zones did not indicate any flow disturbances and were higher than average. It remains to be
determined whether flow disturbances are an inherent part of SLC ring blasting and whether an
undisturbed gravity flow would occur if all holes and rings detonate and break as planned. Continued
full-scale research and analysis of results from different measurements are therefore recommended.
New measurement techniques were developed in this research. It is recommended that they be used in
combination with other, new and powerful technique. Areas of future research study are:
113
Concluding remarks
Smart marker tests

o
Development of the extraction zone during ore extraction,
Interaction effects between rings (extraction from previous rings, overbreak),
Correlation with blast design, mucking position, weighing data and fragmentation,
Identification of movement zones, i.e. tracking ore flow in time and space. This
requires further system development though.
Hang-up surveys
o
3D imaging and analysis,
Data analysis, i.e. occurrence of hang-ups related to blast design, mining sequence,
operational data, etc.
Instrumented, semi- and full-scale confined blasting tests

o
Measurement of burden movement, swell and compaction (fibre-optic zebra gauge),
Correlation with blast data, ore and caved rock properties prior to and after blasting
Identification of under- and overbreak (TDR technique).
Verification of detonation problems

o
Controlled, single-hole shots in open stope situations,
Regular production follow-ups,
Drilling campaign and optical inspection

o
Are there existing voids prior to blasting?
Does a gap at the ring face exist directly after blasting?
Verification of gaps and waste rock inflow during mucking from above,
Verification of the banding effect, i.e. alternate waste and ore stripes.
Despite the limitations of small-scale experiments, they have provided quantitative and qualitative
results for the design and operation of SLC throughout the years. The applicability of the combined
models incorporating the effects from confined blasting and material flow investigations remains.
Recent developments in numerical modeling are of great interest. Results in this thesis provide good
inputs for the calibration of numerical models. The use of HSBM (Blo-Up component; Furtney et al.,
2009) and REBOP (Pierce, 2004) software would advance clarification of the complex process of blast
dynamics, breakage and material flow in SLC.
According to that a series of starting points exist to further increase the understanding of the effects of
blasting on material flow behavior and recovery, which is essential for the long-term objective to
improve breakage and flow in sublevel caving.
114
REFERENCES
Abramov, V.F. & Gorbunov, V.A, (1978). Determination of the shift of stowed material during ore
breaking in a confined space. Soviet mining science, 14(2), 157-165.
Airey, L.D. (1965). The introduction of mechanized mining methods at Mufulira Copper Mines Ltd.
In J.M. Dew & R.T. Modigan (Eds.), 8th Commonwealth Mining and Metallurgical Congress (pp. 1626). Dunedin, New Zealand: University of Otaga.
Alford, C.G. (1978). Computer simulation models for the gravity flow of ore in sublevel caving
(Master thesis). University of Melbourne, Melbourne.
Andersson, P. (2004). Analys av data frn fjrrlastmaskin vid skopfyllning i skivras i KUJ [Analysis of
data from remote controlled LHD-machine during bucket filling in sublevel caving at Kiruna mine]
(Master thesis). Lule University of Technology, Lule.
Baiden, G.R., Bissiri, Y. & Saari, A.V. (2008). Real time sensing of rock flow in a block cave mine. In
H. Schunnesson & E. Nordlund (Eds.), 5th International Conference and Exhibition on Mass Mining
(pp. 993-1002). Lule, Sweden: University of Technology.
Belen`kii, E.V., Kovtun, I.N. & Fedorenko, P.I. (1969). Buffer properties of caved ore. Soviet Mining
Science, 5(4), 444-447.
Bergmark, J.E. (1975). Berkning av ortavstnd och frsttning vid skivrasbrytning [Calculation of
drift spacing and ring burden for sublevel caving] (Note RU 75-16). Malmberget, Sweden: LKAB.
Brent, G. (2010), Orica Australia Pty Ltd. Personal communication.
Brown, E.T. (2003). Block caving geomechanics. The international caving study, stage I 1997-2000.
Indooroopilly, Australia: Julius Kruttschnitt Mineral Research Centre.
Brunton, I.D. (2009). The impact of blasting on sublevel caving flow behaviour and recovery
(Doctoral thesis). University of Queensland, Brisbane.
Brunton, I.D. & Chitombo, G.P. (2009). Modeling the impact of SLC blast design & performance on
material recovery. In J.A. Sanchidrin (Ed.), 9th International Symposium on Rock Fragmentation by
Bull, G. & Page, C.H. (2000). Sublevel caving today`s dependable low cost ore factory. In G.
Chitombo (Ed.), 3rd International Conference and Exhibition on Mass Mining (pp. 537-556).
Melbourne, Australia: Australasian Institute of Mining and Metallurgy.
115
Chen, J.Y. & Boshkow, S. (1981). Recent developments and applications of bulk mining methods in
the Peoples Republic of China. In D.R. Stewart (Ed.), Design and operation of caving and sublevel
stoping mines (pp. 393-423). New York, USA: Society of Mining Engineers of the American Institute
of Mining, Metallurgical, and Petroleum Engineers.
Chiappetta, R.F. (1998). Blast monitoring instrumentation and analysis techniques, with an emphasis
on field applications. Fragblast International Journal of Blasting and Fragmentation, 2(1), 79-122.
Chitombo, G.P. (2010). Cave mining 16 years after Laubscher`s 1994 cave mining state of the
art In Y. Potvin (Ed.), 2nd International Symposium on block and sublevel caving (pp. 603-622).
Perth, Australia: Australian Centre for Geomechanics, The University of Western Australia.
Chu, A.S. (1988). Built-in mechanical filter in a shock accelerometer. In 59th Shock and Vibration
Symposium (vol.1, pp. 251-269). Washington, USA: Shock and Vibration Information Centre, U.S.
Naval Research Centre.
Chu, A.S. (1992). Problems in high-shock measurement (Technical paper TP 308). San Juan
Capistrano, USA: Endevco Corporation.
Clout, J. (2004). The continuation of full scale sublevel caving experiments Cadia Valley operation
underground (Ridgeway gold mine). (Unpublished report, ENGG 7240, Engineering project 4A).
Brisbane, Australia: University of Queensland.
Cokayne, E.W. (1982). Sublevel Caving. In W.A. Hustrulid, Underground mining methods handbook
(pp. 872-879). New York, USA: Society of Mining Engineers of the American Institute of Mining,
Metallurgical, and Petroleum Engineers.
Cullum, A.J. (1974). The effects of confined blasting on rock fragmentation and flow characteristics in
sublevel caving (Master thesis). University of Queensland, Brisbane.
Dick, R.D. & Edwards, C.L. (1985). Instrumentation and diagnostic techniques used by the Los
Alamos National laboratory in fragmentation experiments in oil shale. In Fourney W.L., Boade R.R. &
Costin L.S. (Eds.), Fragmentation by Blasting (pp. 30-40). Bethel, USA: Society for Experimental
Mechanics.
Dowding, C.H., Su, M.B. & O`Connor, K.M. (1988). Principles of time domain reflectometry applied
to measurement of rock mass deformation. International Journal of Rock Mechanics and Mining
Sciences, 25(5), 287-297.
Eadie, B.A. (2002). Modeling primary and secondary fragmentation for block caving (Doctoral
thesis). University of Queensland, Brisbane.
116
Fedorenko, P.I. & Kovtun, I.N. (1977). Influence of the shape and size of the compensation spaces on
the crushing quality in ore breaking by boreholes. Soviet mining science, 13(1), 31-34.
Felice, J.J., Klass, W. & Spathis, A.T. (1991). Burden velocity measurement in cast blasting
operations at Mount Thorley Mine using POWERWAVE face velocity radar. In 7th Annual Symposium
on Explosives and Blasting Research (pp. 71-77). Cleveland, USA: ISEE, International Society of
Explosives Engineers.
Finkel, M. & Skalare, H. (1963). Skivrasbrytning delrapport ver hllutningens inverkan vid
modellfrsk [Sublevel caving interim report concerning the influence of hole inclination for modelscale tests] (Report series C, no. 7). Stockholm, Sweden: Svenska Gruvfreningen, Stiftelsen
Gruvforskningen.
Fjellborg, S. (2002). The value of measuring VOD in large scale sublevel caving. In X. Wang (Ed.),
7th International Symposium on Rock Fragmentation by Blasting (pp. 717-724). Bejing, China:
Metallurgical Industry Press.
Furtney J., Cundall, P.A. & Chitombo, G. (2009). Developments in numerical modeling of blast
induced rock fragmentation: Updates from the HSBM project. In J.A. Sanchidrin (Ed.), 9th
International Symposium on Rock Fragmentation by Blasting (pp. 335-342). London, England: CRC
Press.
Furtney, J. (2011). HSBM modeling of the Kiruna confined drift wall blasting tests. (Unpublished
technical memorandum). Minneapolis, USA: Itasca.
Forssberg, E. & Zhai, H. (1985). Shape and surface properties of the particles liberated by autogenous
grinding. Scandinavian Journal of Metallurgy, 14(1), 25-32.
Free, G.D. (1970). Mathematical and model studies of the flow of material in the sublevel caving
mining method (Master thesis). University of Queensland, Brisbane.
Frstrm, J. (1970). Underskning av ekvivalenta modellmaterial fr utveckling och projektering av
skivrasbrytning [Examination of equivalent model material for development and design of sublevel
caving] (Master thesis). Royal Institute of Technology, Stockholm.
Gaich, A., Ptsch, M., Moser, P. & Schubert, W. (2009). How 3D images support bench face
profiling, blast planning and rock mass characterisation. In J.A. Sanchidrin (Ed.), 9th International
Symposium on Rock Fragmentation by Blasting (pp. 85-90). London, England: CRC Press.
GeoDrilling Int. (2002). Drilling equipment for construction quarrying. GeoDrilling International,
10(1), 4-10.
117
Gorham-Bergeron, E., Kuszmaul, J.S., Bickel, T.C. & Shirey, D.L. (1987). Confined volume blasting
experiments description and analysis. In W.L. Fourney & R.D. Dick (Eds.), 2nd International
Symposium on Rock Fragmentation by Blasting (pp. 159-171). Bethel, USA: Society of Experimental
Mechanics.
Grasedieck, A. (2007). Die Natrliche Bruchcharakteristik (NBC) von Gesteinen in der Sprengtechnik
[Natural breakage characteristics (NBC) applied for rock blasting] (Doctoral thesis). University of
Leoben, Leoben.
Gustafsson, P. (1998). Waste rock content variations during gravity flow in sublevel caving (Doctoral
thesis). Lule University of Technology, Lule.
Hahne, R., Plsson, B.I. & Samskog, P.O. (2003). Ore characterisation for-and simulation of-primary
autogenous grinding. Minerals Engineering, 16(1), 13-19.
Hanisch, J., Ergenzinger, P. & Bonte, M. (2003). Dumpling an intelligent boulder for studying
internal processes of debris flows. In D. Rickenmann & C. Chen (Eds.), 3rd International Conference
on Debris-Flow Hazards Mitigation: Mechanics, Prediction and Assessment (pp. 843-849).
Rotterdam, Netherlands: Millpress.
Hedn, H. (1976). Fysikalisk modell fr beskrivning av gravitationsfldet [Physical model for the
description of gravity flow]. In H. Hedn (Ed.), Kompendium frn Rasbrytningseminarium
[Proceedings of cave mining seminar] (pp. I1 - I7). Kiruna, Sweden: LKAB.
Hedstrm, O. (2000). Funktionskontroll av produktionssalvor i Kiirunavaara [Function control of
production blasts at the Kiruna mine] (Master thesis). Lule University of Technology, Lule.
Hollins, B. & Tucker, J. (2004) Draw point analysis using a marker trial at the Perseverance Nickel
Mine, Leinster, Western Australia. In A. Karzulovic & M.A. Alafaro (Eds.), 4th International
Conference and Exhibition on Mass Mining (pp. 498-502). Santiago, Chile: Instituto de Ingenieros de
Chile.
Hustrulid, W.A. (2000). Method selection for large scale underground mining. In G. Chitombo (Ed.),
3rd International Conference and Exhibition on Mass Mining (pp. 29-56). Melbourne, Australia:
Australasian Institute of Mining and Metallurgy.
Janelid, I. & Kvapil, R. (1966). Sublevel caving. International Journal of Rock Mechanics and Mining
Sciences, 3(2), 129-153.
Janelid, I. (1972). Study of the gravity flow process in sublevel caving. Proc. International sublevel
caving symposium (pp. 1-23). Stockholm, Sweden: Atlas Copco.
118
Jarlenfors, J. & Holmberg, R. (1980). Sprngningsfrsk mot kompakterade bergmassor.

Modellfrsk i plexiglas [Blasting against compressible rock masses. Model scale experiments in
plexiglass] (SveDeFo Report DS 1980:5). Stockholm, Sweden: Swedish Detonic Research
Foundation.
Johansson, D., Ouchterlony, F. & Nyberg, U. (2007). Blasting against aggregate confinement,
fragmentation and swelling in model scale. In P. Moser (Ed.), 4th EFEE World Conference on
Explosives and Blasting (pp. 13-26). England: EFEE, European Federation of Explosives Engineers.
Johansson, D. (2011). Effects of confinement and initiation delay on fragmentation and waste rock
compaction. (Doctoral thesis). Lule University of Technology, Lule.
Kanchibotla, S., Valery, W., & Morrell, S. (1999). Modelling fines in blast fragmentation and its
impact on crushing and grinding. In Exploring the role of breakage in mining and quarrying (pp. 137144). Melbourne, Australia: AusIMM, The Australasian Institute of Mining and Metallurgy.
Kirpichenko, V.M. (1982). Study of ore breaking in a compressed medium by sets of parallel adjacent
bores in the Tash Tagol Mine. Soviet Mining Science, 18(3), 233-237.
Kleine, T., Townson, P. & Riihioja, K. (1993). Assessment and computer automated blast design. In J.
Elbrond & X. Tang (Eds.), 24th International Symposium on the Application of Computers and
Operations Research in the Mineral Industries (vol. 3, pp. 353-360). Montreal, Canada: Canadian
Institute of Mining, Metallurgy and Petroleum.
Koppanyi, F. (1960). Optimering av skivrasbrytning med modellfrsk [Optimization of sublevel
caving with model-scale tests] (Master thesis). Royal Institute of Technology, Stockholm.
Kosowan, M.I. (1999). Design and operational issues for increasing sublevel cave intervals at Stobie
Mine (Master thesis). Laurentian University, Greater Sudbury.
Krekula, H. (2004). En underskning av sprngskador och rrelser i bergmassan nra
brytningsfronten TDR-mtningar i skivrasbrytningen [An investigation of blast damage close to the
mining front TDR measurements in sublevel caving] (Master thesis). Lule University of
Technology, Lule.
Kuchta, M.E. (2002). A revised form of the Bergmark-Roos equation for describing the gravity flow
of broken rock. Mineral Resources Engineering, 11(4), 349-360.
Kvapil, R. (1998). The mechanics and design of sublevel caving systems. In R.E. Gertsch & R.L.
Bullock, Techniques in underground mining. Selections from underground mining methods handbook
(pp. 621-653). Littleton, USA: Society for Mining. Metallurgy, and Exploration, Inc.
119
Kvapil, R. (2008). Gravity flow in sublevel and panel caving a common sense approach. Lule,
Sweden: Lule Technical University.
Larsson, L. (1998). Slutrapport Projekt Skivras 2000 [Final report project sublevel caving 2000]
(Unpublished report 98-765). Kiruna, Sweden: LKAB.
Lichter, J. (2007): Review of feed size distribution and mill operating parameters on the performance
of the KA2 comminution circuit (Unpublished report LK01M). York, USA: Metso Minerals Industries
Inc., Metso Minerals Optimization Services.
Lith, A. (2004). Prediction of fragmentation for ring blasting in large-scale sublevel caving (Master
thesis). Delft University of Technology, Delft.
Lith, A., Kuchta, M. & Quinteiro, C. (2004). Prediction of fragmentation for ring blasting in largescale sublevel caving. In M. Hardygra, G. Paszkowsk & M. Sikora (Eds.), 13th International
Symposium on Mine Planning and Equipment Selection (pp. 257-266). Rotterdam, Netherlands:
Balkema.
Lundberg, S. (1961a): Tappning, transport och krossning av skivrasberg p 320 m avv. i KUJ.
Delutredning: Siktanalys av skivrasberg [Drawing, transportation and crushing of caving material at
level 320 m in the Kiruna underground mine. Interim report: fragment size distribution of material
from caving] (Unpublished report 116-1071a). Kiruna, Sweden: LKAB.
Lundberg, S. (1961b): Tappning, transport och krossning av skivrasberg p 320 m avv. i KUJ.
Delutredning: Siktanalys av ortberg [Drawing, transportation and crushing of caving material at level
320 m in the Kiruna underground mine. Interim report: fragment size distribution of material from
drifting] (Unpublished report 116-1071b). Kiruna, Sweden: LKAB.
Liu, Q. (1994). The application of high speed video system for blasting research. In 20th ISEE
Conference on Explosives and Blasting Technique (pp. 359-375). Cleveland, USA: ISEE, International
Society of Explosives Engineers.
Lynch, A.J. (1977). Mineral crushing and grinding circuits their simulation, optimisation, design
and control (vol. 1). Amsterdam, Netherlands: Elsevier Scientific Publishing Company.
Maerz, N.H., Palangio, T.C. & Franklin, J.A. (1996). WipFrag image based granulometry system. In
B. Mohanty (Ed.), 5th International Symposium on Rock Fragmentation by Blasting, Workshop on
measurement of blast fragmentation (pp. 91-99). Montreal, Canada: Balkema.
Maerz, N.H. & Zhou, W. (1998). Optical digital fragmentation measuring systems inherent sources
of error. Fragblast, International Journal for Blasting and Fragmentation, 2(4), 415-431.
120
Malmgren, L. (2012), LKAB Kiruna. Personal communication.

Maripuu, R. (1968). Underskning av siktanalys och styckeform frn skivrasbrytningen vid LKAB,
Kiruna [Investigation of fragment size and shape from sublevel caving at LKAB, Kiruna] (Master
thesis). Royal Institute of Technology, Stockholm.
Markenzon, E.I. (1967). Mechanism of blasting without a compensation space. Journal of Mining
Science, 3(1), 45 - 47.
Marklund, I. (1976): Sprngningens inverkan p rasfrloppet [Influence of blasting on gravity flow].
In H. Hedn (Ed.), Kompendium frn Rasbrytningseminarium [Proceedings of cave mining seminar]
(pp. L1 L9). Kiruna, Sweden: LKAB.
McMurray, D.T. (1982). Sublevel caving practice at Shabanie mine, Rhodesia. In W.A. Hustrulid,
Underground mining methods handbook (pp. 945-952). New York, USA: Society of Mining
Engineers of the American Institute of Mining, Metallurgical, and Petroleum Engineers.
Mishra, A.K. & Gupta, R.N. (2002). Design of blast using high resolution camera. In X. Wang (Ed.),
7th International Symposium on Rock Fragmentation by Blasting (pp. 378-389). Beijing, China:
Moser, P., Cheimanoff, N., Ortiz, R. & Hochholdinger, R. (2000). Breakage characteristics in rock
blasting. In R. Holmberg (Ed.), 1st EFEE World Conference on Explosives and Blasting (pp. 165-170).
Rotterdam, Netherlands: Balkema.
Moser, P., Olsson, M., Ouchterlony, F. & Grasedieck, A. (2003). Comparison of the blast
fragmentation from lab-scale and full-scale tests at Brap. In R. Holmberg (Ed.), 2nd EFEE World
Conference on Explosives and Blasting (pp. 449-458). Rotterdam, Netherlands: Balkema.
Moser, P. (2005). Less fines in aggregate and industrial minerals industry. In R. Holmberg (Ed.), 3rd
EFEE World Conference on Explosives and Blasting (pp. 567-574). England: EFEE, European
Federation of Explosives Engineers.
Moser, P. (2007), University of Leoben. Personal communication.
Newman, T. (1997). Blasting of intact ore against caved waste rock at Kiruna (Senior Thesis).
University of South Australia, Adelaide.
Newman, T., Hustrulid, W. & Quinteiro, C. (2008). Sublevel caving trial monitoring effects from
blasting an ore slice against caved rock at LKAB`s Kiruna mine, Sweden. In H. Schunnesson & E.
121
Nordlund (Eds.), 5th International Conference and Exhibition on Mass Mining (pp. 705-713). Lule,
Sweden: University of Technology.
Niiranen, K. (2008), LKAB Kiruna. Personal communication.
Nordqvist, A. (2012), LKAB Kiruna. Personal communication.
Noren, C.H. (1956). Blasting experiments in granite rock. Colorado School of Mines Quarterly, 51(3),
213-225.
Nyberg, U., Fjellborg, S., Olsson, M. & Ouchterlony, F. (2000). Bedmning av sprngskador i
ortkontur; Vibrationsmtningar, skadeprognoser och sprickkartering i magnetitmalm och grberg
[Vibration measurements, damage predictions and crack mapping in magnetite ore and waste rock]
(SveBeFo Rapport 50). Stockholm, Sweden: Svensk Bergteknisk Forskning.
O`Connor, K.M. & Dowding, C.H. (1999). GeoMeasurement by pulsing TDR cables and probes. Boca
Raton, USA: CRC Press.
Olsson, M. (1987). Sprngning mot kompakterat berg [Blasting against compressible rock masses]
(SveDeFo Report DS 1987:5). Stockholm, Sweden: Swedish Detonic Research Foundation.
Olsson, M., Nyberg, U. & Fjellborg, S. (2009). Kontrollerad snderbrytning vid skivrassprngning
inledande frsk [Controlled fragmentation in sublevel caving first tests] (Swebrec Report 2009:2).
Lule, Sweden: Lule University of Technology.
Onederra, I.A. (2004). Modelling fragmentation in underground production blasting. In A. Karzulovic
& M.A. Alafaro (Eds.), 4th International Conference and Exhibition on Mass Mining (pp. 365-370).
Santiago, Chile: Instituto de Ingenieros de Chile.
Onederra, I., Chitombo, G.P., Cundall, P.A., & Furtney, J.K. (2009). Towards a complete validation of
the lattice scheme in the Hyrbid Stress Blasting Model (HSBM). In J.A. Sanchidrin (Ed.), 9th
International Symposisum on Rock Fragmentation by Blasting (pp. 343-351). London, England: CRC
Press.
Ouchterlony, F. (2003). Influence of blasting on the size distribution and properties of muckpile
fragments, a state-of-the-art review. (Report MinFo project P2000-10: Energy optimization during
comminution. Stockholm, Sweden: Swebrec, Swedish Blasting Research Centre at Lule University of
Technology.
Ouchterlony, F. (2005). The Swebrec function: linking fragmentation by blasting and crushing.
Transactions of the Institute of Mining and Metallurgy, Section A: Mining Technology, 114, A29-A44.
122
Ouchterlony, F. & Moser, P. (2006). Likenesses and differences in the fragmentation of full-scale and
model-scale blasts. In 8th International Symposium on Rock Fragmentation by Blasting (pp. 207-220).
Santiago, Chile: Editec.
Ouchterlony, F., Olsson, M., Nyberg, U., Andersson, P. & Gustavsson, L. (2006). Constructing the
fragment size distribution of a bench blasting round, using the new Swebrec function. In 8th
International Symposium on Rock Fragmentation by Blasting (pp. 332-344). Santiago, Chile: Editec.
Ouchterlony, F. (2009). Fragmentation characterization; the Swebrec function and its use in blast
engineering. In J.A. Sanchidrin (Ed.), 9th International Symposium on Rock Fragmentation by
Panczakiewicz, T. (1977). Optimization of the sublevel caving mining method investigated by physical
models (Master thesis). University of Melbourne, Melbourne.
Petropoulos, N. (2011). Influence of confinement on fragmentation and investigation of the burden
movement small scale tests. (Master thesis). Lule University of Technology, Lule.
Pierce, C.E., Bilaine, C., Huang, F.C. & Dowding, C.H. (1994). Effects of multiple crimps and cable
length on reflection signatures from long cables. In K.M. O`Connor, C.H. Dowding & C.C. Jones
(Eds.), Symposium on Time Domain Reflectometry in Environmental, Infrastructure and Mining
Applications (pp. 540-554). Minneapolis, USA: U.S. Bureau of Mines, Department of Interior.
Pierce, M. (2004). Development of REBOP as a practical tool for draw prediction in caving mines. In
E.T. Brown, The International Caving Study I (Unpublished final report). Brisbane, Australia:
JKMRC and Itasca Consulting Group, Inc.
Power, G. (2004a). Modeling granular flow in caving mines: large scale physical modeling and full
scale experiments. (Doctoral thesis). University of Queensland, Brisbane.
Power, G. (2004b). Full scale SLC draw trials at Ridgeway Gold Mine. In A. Karzulovic & M.A.
Alafaro (Eds.), 4th International Conference and Exhibition on Mass Mining (pp. 225-230). Santiago,
Chile: Instituto de Ingenieros de Chile.
Power, G. (2005). SLC marker trials project (3.2.1) (Unpublished presentation). MMT meeting in
Sudbury, Canada, October 2005.
Preston, C.J. (1995). 3D blast design for ring blasting in underground mines. In Exploring the role of
breakage in mining and quarrying (pp. 1-11, supplement). Melbourne, Australia: AusIMM, The
123
Quinteiro, C., Larsson, L. & Hustrulid, W.A. (2001). Theory and practice of very large scale sublevel
caving. In W.A. Hustrulid & R.L. Bullock (Eds.), Underground mining methods engineering
fundamentals and international case studies (pp. 381-384). Littleton, USA: Society for Mining,
Metallurgy, and Exploration, Inc. (SME).
Quinteiro, C. (2004). Final report: Orica-fragmentation project (Unpublished report 05-722). Kiruna,
Sweden: LKAB.
Quinteiro, C. & Fjellborg, S. (2008). Measurements of borehole deviation in sublevel caving fans at
Kiruna mine. In H. Schunnesson & E. Nordlund (Eds.), 5th International Conference and Exhibition
on Mass Mining (pp. 543-551). Lule, Sweden: Lule University of Technology.
RingCave (Version 2012-05-20) [Computer software]. Kiruna, Sweden: LKAB (Nordqvist, A.).
Rohrmoser, S., Hollerer, H. & Comoli, C. (2007). Magnetite ore samples from LKABs operation in
Kiruna/Sweden. Optimized comminution sequence, natural breakage characteristic, energy register
function (Swebrec Report 2007:U1). Lule, Sweden: Lule University of Technology.
Rustan, A. (1970): Volymviktmetodens teoretiska grunder fr bestmning av malmhalt hos en
blandning av sprngd malm och grberg. Kinematik, svllning, uppluckring, och styckefall i
frsttningen vid sprngning mot lst berg i pall- och skivrasmodeller. [Theoretical basis of the
volume-weight-method for the determination of the ore content of a mixture of blasted ore and waste
rock. Kinematics, swelling, loosening and fragment size in the burden for confined blasting in modelscale.] (Licentiate thesis). Royal Institute of Technology, Stockholm.
Rustan, A., qvist, J., Lin, N.S., Stenstrm, T. & Khler, O. (1988). Minsta avstnd mellan laddade
spngborrhl i skivrasbrytning [Minimum distance between charged boreholes in sublevel caving]
(Technical report 1988:28 T). Lule, Sweden: Lule University of Technology.
Rustan, A. (1990). The importance of using joints to achieve scaled fragmentation in magnetite
concrete used for sublevel caving blast models. Engineering Fracture Mechanics, 35(1-3), 425-438.
Rustan, A. (1993). Minimum distance between charged boreholes for safe detonation. In H.P.
Rossmanith (Ed.), 4th International Symposium on Rock Fragmentation by Blasting (pp. 127-135).
Rotterdam, Netherlands: Balkema.
Rustan, A. (2000). Gravity flow of broken rock what is known and unknown. In G. Chitombo (Ed.),
3rd International Conference and Exhibition on Mass Mining (pp. 557-567). Melbourne, Australia:
124
Rustan, A. (2012). The dynamics and fragmentation of blasted ore slices in scaled sublevel caving and
slab models followed by accuracy analysis of the Volume weight method used for determination of
ore content at loading. Submitted to 10th International Symposium on Rock Fragmentation by Blasting.
Sanchidrin, J.A., Segarra, P., Ouchterlony, F. & Lpez, L.M. (2009). On the accuracy of fragment
size measurement by image analysis in combination with some distribution functions. Rock Mechanics
and Rock Engineering, 42(1), 95-116.
Sandstrm, P.O. (1972). Application and optimization of sublevel caving techniques. Engineering and
Mining Journal, 173(6), 112 125.
Schmitt, G.G. & Dick, R.D. (1983). Use of CORRTEX to measure explosive performance and stem
behaviour in oil shale fragmentation tests (Unpublished report). Los Alamos, USA: Los Alamos
National Laboratory.
Segarra, P., Sanchidrin, J.A., Lpez, Pascual, J.A., Ortiz, R., Gmez, A. & Smoech, B. (2003).
Analysis of bench face movement in quarry blasting. In R. Holmberg (Ed.), 2nd EFEE World
Conference on Explosives and Blasting (pp. 485-495). Rotterdam, Netherlands: Balkema.
Selldn, H. & Pierce, M. (2004). PFC3D modeling of flow behaviour in sublevel caving. In A.
Karzulovic & M.A. Alafaro (Eds.), 4th International Conference and Exhibition on Mass Mining (pp.
201-214). Santiago, Chile: Instituto de Ingenieros de Chile.
Shirzadegan, S., Nordlund, E., Nyberg, U., Zhang, P. & Malmgren, L. (2011): Rock support subjected
to dynamic loading: Field testing of ground support using simulated rockburst. In Q. Quian & Y Zhou
(Eds.), Harmonising rock engineering and the environment, 12th International Society for Rock
Mechanics International Congress on Rock Mechanics (pp. 1269-1273), Leiden, Netherlands: CRC
Press/Balkema.
ShapeMetriX3D (version 3.0.2) [computer software]. Graz, Austria: 3G Software & Measurement.
Sjstrand, W. (1957). Skivrasbrytning i Kiirunavaara med modellfrsk [Sublevel caving in the
Kiirunavaara mine with model-scale tests] (Master thesis). Royal Institute of Technology, Stockholm.
Stazhevskii, S.B. (1996). Features of flow of broken rock in extraction of ores with sublevel caving.
Journal of Mining Science, 32(5), 403-416.
Steiner, H.J. (1991). The significance of the Rittinger equation in present-day comminution
technology. In H. Schubert (Ed.), 17th International Mineral processing Congress (vol. 1, pp. 177188). Freiberg, Germany: Polygraphischer Bereich Bergakademie Freiberg.
125
Steiner, H.J. (1998). Zerkleinerungstechnische Eigenschaften von Gesteinen [Comminution technical

properties of rocks]. Felsbau, 16(5), 320-325.
Su, M.B. (1987). Quantification of cable deformation with time domain reflectometry (Doctoral
thesis). Northwestern University, Evanston, USA.
Tessem, S. & Wennberg, S. (1981). Longitudinal sublevel caving at Fosdalens BergverksAktieselskab, Norway. In D.R. Stewart (Ed.), Design and operation of caving and sublevel stoping
mines (pp. 365-371-423). New York, USA: Society of Mining Engineers of the American Institute of
Mining, Metallurgical, and Petroleum Engineers.
Tonon, F. & Kottenstette, J.T. (2007). Laser and photogrammetric methods for rock face
characterization: A workshop. In F. Tonon & J.T. Kottenstette (Eds.), 41st U.S. Rock Mechanics
Symposium (Golden Rocks 2006 50 Years of Rock Mechanics), Report on a workshop held in
conjunction with the conference (pp. 5-12). Alexandria, USA: American Rock Mechanics Association.
Trumbull, C.J. (1967). Model testing of ring inclination in sublevel caving (Bachelor thesis).
University of Queensland, Brisbane.
Valentim, L. (2010), TLC Engineering Solutions Pty Ltd. Personal communication.
Volchenko, N.G. (1977). Influence of charge arrangement geometry and short-delay blasting on the
crushing indices in compression blasting. Soviet Mining Science, 13(5), 488-493.
WipFrag (version 2.6) [computer software]. North Bay, Canada: WipWare Inc.
Whiteman, D.S. (2010). The Smart Marker system A new tool for measuring underground ore body
flow in block and sub-level mines. In Y. Potvin (Ed.), 2nd International Symposium on block and
sublevel caving (pp. 603-622). Perth, Australia: Australian Centre for Geomechanics, The University
of Western Australia.
Wortley, M., Nozawa, E. & Riihioja, K.J. (2011). Metso SmartTag The next generation and beyond.
In Y. Baafi, R.J. Kininmonth & Porter I. (Eds.), 35th International Symposium of Computers and
Operations Research in the Minerals Industry (APCOM) (pp. 841-851). Melbourne, Australia:
AusIMM, The Australasian Institute of Mining and Metallurgy.
Wregg, C.B. (1966). Model testing of ore recovery in sublevel caving (Bachelor thesis). University of
Queensland, Brisbane.
126
Yenge, L.I. (1981). Analysis of bulk flow of materials under gravity caving process Part 2:
Theoretical and physical modelling of gravity flow of broken rock. Colorado School of Mines
Quarterly, 76(3), 1-67.
Yiannakopoulos, G. & Van der Schaaf, P.J. (1998). Evaluation of accelerometer mechanical filters on
submerged cylinders near an underwater explosion. Shock and Vibration, 5(4), 255-265.
Zhang, G. (2004). Behaviour of caved ore mass in sublevel caving and its effect on ore dilution. In A.
Karzulovic & M.A. Alafaro (Eds.), 4th International Conference and Exhibition on Mass Mining (pp.
238-242). Santiago, Chile: Instituto de Ingenieros de Chile.
Zhang, Z.X. (2005). Increasing ore extraction by changing detonator positions in LKAB Malmberget
mine. Fragblast The International Journal of Blasting and Fragmentation, 9(1), 9-46.
127
128
PART 2
APPENDIX
The papers A - D are attached in printed form. They were published at different conferences and are
attached as they were printed.
The reports 1 - 8 are attached on a CD.
129
PAPER A
The fragment size distribution of Kiruna magnetite, from model-scale to

run of the mine
Matthias Wimmer, Finn Ouchterlony and Peter Moser.

Published in:
Proc. 5th International Conference and Exhibition on Mass Mining 2008 (pp. 691703).
Lule, Sweden: University of Technology.
5th International Conference and Exhibition on Mass Mining, Lule Sweden 9-11 June 2008
The fragment size distribution of Kiruna magnetite, from model-scale

to run of the mine
M. Wimmer Swebrec at Lule University of Technology, Sweden
F. Ouchterlony Swebrec at Lule University of Technology, Sweden
P. Moser University of Leoben, Austria
Abstract
The objective of this investigation constitutes a detailed study of the fragmentation characteristics for blasted
magnetite ore from a specific ring in the sublevel caving (SLC) mine in Kiruna.
For comparison lab scale data from crushing and grinding as well as blasting plus historical full-scale data
from the mine were analysed. The present data confirms that the material basically follows the Natural
Breakage Characteristics criteria within the fines region. The complete fragmentation distributions yield an
extremely large variation for the parameter x50. The shapes of the complete sieving curves do not entirely fit
the Swebrec function since they are flatter, roughly over the range 25-75 mm. The relative flattening of the
sieving curves may be due to selective breakage in the mid-range, which would increase the amount of
fines. This resembles the behaviour in autogenous grinding, i.e. the self-breakage of large fragments creates
pebbles that grind the mid-fractions to fine, and also an effect attributable to secondary fragmentation in
block caving.
Several different mechanisms are conceivable to account for an altered fragmentation mechanism. From the
present large SLC layout one could expect more comminution occurring in the debris flow, due to longer
flow paths. On the other hand, the present samples were taken before 10 % draw, which speaks in favour of
relatively short paths, unless the draw has been extremely uneven. Despite this flattening deviation from
Swebrec distribution behaviour which is likely to be caused by the internal flow mechanisms in the SLC
process the main conclusion is still that magnetite itself behaves like waste rock from a blasting point of
view.
Research objective
The heart of the sublevel caving (SLC) process at LKAB consists of three steps a) blast function of caving
rounds, b) fragmentation of caving panels and c) caving flow. For the ore recovery to be as high as possible
and for the waste rock inflow to be as low as possible all three steps have to work as planned.
Fragmentation can be regarded as key element with its direct influence on the gravity flow and productivity.
Quantitative assessment of the fragmentation of blasted rock at a larger scale is a difficult task. Next to
screening there is still no known reliable method of assessing fragmentation quantitatively in a production
environment. In addition, obtaining a representative sampling constitutes a major problem.
The objective of this investigation is to find out whether the Kiruna magnetite behaves like waste rock from
a blasting point of view. In order to do this the fragmentation characteristics for blasted magnetite ore from
the SLC operation in Kiruna have been studied in full-scale (Wimmer et al., 2008).
Indications have existed that the magnetite could obey diverse fragmentation characteristics. Previous blast
damage investigations had e.g. shown that the cracking behind half-casts from perimeter blasting in drifts in
magnetite looked quite different from that behind half-casts in waste rock (Nyberg et al., 2000). In waste
rock the crack pattern is mainly radial, with many short crushing cracks at the borehole and a smaller number
of long distinct cracks. In ore the crack pattern is more diffuse, more like a spiders web. Such a web would
probably occur if the magnetite showed plastic behaviour. There have been other circumstances in Kiruna
too where the ore has been thought of as showing plastic deformation behaviour rather than brittle behaviour.
The results are further compared with the Natural Breakage Characteristics properties (NBC; Steiner, 1991
& 1998; Rohrmoser et al., 2007) and fragmentation results from model-scale blasts (Johansson et al., 2007ab). The study presented herein is part of an ongoing PhD project entitled Improved breakage and flow in
sublevel caving undertaken by the principal author and financed by the Hjalmar Lundbohm Reserch Centre.
NBC approach and model scale blasts
The NBC approach is based on the concept that a fragmented material exhibits a specific Natural Breakage
Characteristic (Steiner, 1991 & 1998). The concept originates from mechanical comminution studies and
can be summarized as follows (Ouchterlony, 2004).
1. When rock particles are broken in the steps or sub-circuits of an Optimum Comminution Sequence
(OCS) in the laboratory, the resulting fragmentation curve is the steepest possible. Both the
comminution sequence and the ancillary conditions guarantee for the OCS a machinery non-specific
fragmentation with focus on the actual material characteristics.
2. When the product stream of each sub-circuit is classified, the resulting log-log fragmentation curves
are shifted parallel vertically upward as the comminution progresses.
3. When the specific surface As (m2/kg) created by an OCS is plotted versus the energy consumed Es
(J/kg), the points fall more or less on a straight line. This line is termed the energy register function
of the material. It is material specific and its slope equals the Rittinger coefficient R (m2/J).
4. All technical, i.e. non-optimal, comminution processes produce points As that for a given Es-value
fall below the energy register line because they are less energy efficient.
Moser et al. (2000 & 2003) have shown that the NBC approach may be applied to the blasting of model
cylinders and that full scale blasts in the same rock type give similar results.
The OCS built up to analyse the breakage behaviour of Kiruna magnetite is similar to that used within the
Less Fines Project (Moser, 2005) but differs in the size reduction ratios (Rohrmoser et al., 2007). The
sequence, designed on a succession of numerous comminution stages in closed circuits with pre-screening
and the setting of each apparatus, guarantees a small size reduction ratio. The circuit design in the laboratory
is simulated by cyclic comminution tests at a high circulating load (minimum 100 %).
To determine the NBC of Kiruna magnetite, samples in form of drill cores with 140 mm were taken from
two different mine sites at level 878, block 28, drift o287 (termed M3) and block 16, drift o148 (termed M4).
The samples are residues from a previous small-scale blasting campaign (Johansson et al., 2007a-b). The
latter tests comprised the blasting of magnetite cylinders of size 140x280 mm with PETN cord as
explosive in a centre hole resulting in a specific charge of 1.3 kg/m3. To better understand the mechanisms of
rock breakage and fragmentation under confined conditions some of the cylinders have been blasted under
debris confinement within in a steel cylinder of size 309 mm.
Fragmentation curves derived within the respective mechanical comminution stages are compared to those
derived from model-scale blasts in Figures 1 and 2. The first three mechanical fragmentation steps (streams
1, 4 and 7) refer to a jaw crusher whereas the subsequent two steps refer to a rod mill (streams 10 and 13).
The form of the curves is in reasonable agreement for crushing, grinding and blasting of magnetite in model
scale, whereas the agreement for the magnetite quality M4 is exceptionally high. This indicates that the
magnetite reacts to blasting like typical waste rock.
Differences especially in the blasting results for the M3 quality might be related to internal variations of
material properties. With respect to the shape of the curves it is noticeable that the M3 ore type has generally
a more pronounced inflection point at the downwards slope. Differentiating the curves mathematically shows
that the minima for the M3 quality are compared with the M4 type slightly at smaller mesh sizes (0.5 and 0.7
mm, respectively) which indicates a finer grain assembly. A distinct tendency regarding the position of the
minima depending on the method of comminution (mechanically or blasting in confined/un-confined
situation) is - from the available number of tests - not directly inferable.
692
100.0
Mass passing [%]
10.0
NBC, stream 1
NBC, stream 4
NBC, stream 7
NBC, stream 10
NBC, stream 13
M3-2-05 confined
1.0
M3-4-05 confined
M-3-1-05 un-confined
M-3-3-05 un-confined
0.1
0.01
0.1
10
100
1000
Mesh size [mm]
Figure 1
Fragment size distribution for NBC tests and model-scale blasts, M3 ore type
100.0
Mass passing [%]
10.0
NBC, stream 1
NBC, stream 4
NBC, stream 7
NBC, stream 10
NBC, stream 13
M4-1-05 confined
M4-2-05 confined
M4-3-05 un-confined
M4-4-05 un-confined
1.0
0.1
0.01
0.1
10
100
1000
Mesh size [mm]
Figure 2
Fragment size distribution for NBC tests and model-scale blasts, M4 ore type
Plotting the energy register curves for the two different magnetite samples, see Figure 3, yields a significant
difference with respect to the breakage behaviour. The harder to blast M3 quality has a smaller Rittinger
coefficient, 30 cm2/J than the M4 quality, 46 cm2/J. This difference is exceptionally large for one rock type.
In general, both samples are highly resistant to fragmentation due to a relatively small Rittinger coefficient.
350
250
Specific Outer Surface As [cm /g]
300
As = 45.97Es + 22.55
As = 30.01Es + 17.49
R2 = 0.969
R2 = 0.991
200
150
100
50
M3
M4
0
0
10
12
Cumulative Net Energy Consumption Es [J/g]
Figure 3
Energy register functions for both ore types (Es for stream 1 is linearly extrapolated)
693
Previous full-scale fragmentation tests
Table 1 gives an overview of previous fragmentation measurements carried out on blasted rock in Kiruna
(Lundberg, 1961a-b; Maripuu, 1968). The present mining layout and operational conditions are quite
different so a direct comparison with their results may be misleading.
Table 1
List of previous fragmentation measurements for blasted rock in Kiruna
Characteristics
Year
Unit
Lundberg
Lundberg
Maripuu
1961a
1961b
1968
Sieved material
Caving debris
Drifting
Wimmer
2007
Caving debris
Caving debris
Level
248
257
302
320
907
Coord.
Y26-27
Y27-28
Y24
Y35-36
Y37
Drift
239-240
355-357
377, ring 7
Ore type
Sublevel interval
28.5
Blasts
Individual samples
41
Extraction rate
begin/mid/end
10-202%
7.5-9.0%
Total masses
kg
180 323
44 000
197 683
96 454
Screen sizes
mm
10-500
10-500
Mining
area
Burden / advance
10-500 (41x)
0.074-10 (4x)
0.063-200
1.0-1.2
1.8
2.1
1.8
Hole diameter
mm
33.5-36
51
32
51
115
Holes / round
12
12
53
10
Drillmeters / round
66
70
120
50.8-53.4
159.9
Tonnage / round
361.4-395.6
591.3
180.0
534.0-550.0
6 591
Explosive type
Dynalit
Dynalit/Anfo
Anfo
Kimulux R&82
0.33
0.15-0.17
0.23
Specific charge
kg/t
0.12
0.19
: 0.22 kg/t (Anfo)

: calculative break down in ore and waste rock
: visually classified as pure ore without any contamination
Maripuu`s (1968) investigation of fragmented caving debris is the most extensive one because of the large
number of samples studied at different extraction rates. For a few samples even information for the fines
(0.074-10 mm) range is available. Both, the three and five parameter Swebrec functions (Ouchterlony, 2005)
give a good fitting to the data set, see Figure 4 and do not show any irregularities in the breakage behaviour.
From a historic data point of view the magnetite ore thus shows fragmentation characteristics, which are
comparable to those of waste rock. The parameter x50 covers a wide range and its average value amounts to
86 mm. No explicit trend of x50 versus extraction rate is directly noticeable.
Maripuu has also made an attempt to divide the caving debris masses into pure magnetite ore and waste rock,
wherein it has been observed that the fragment size distribution for waste rock is considerably coarser than
that for ore with x50 = 161 mm and x50 = 79 mm respectively. Further it is worth mentioning that the
percentage of fines for the waste rock is much higher than for the ore, i.e. the fragment size distribution is
much flatter.
694
Residuals [6]
Mass passing, %
-2
-2
10
10
1
0.01
0.1
10
100
1
1000
Mesh size, mm
Figure 4
Fragment size distribution for caving debris (5 blasts; > 10 mm: 41 samples and < 10 mm:
4 samples) with extended Sewbrec function fit. Data range 0.074-500 mm. Curve fit
parameters: x50 = 86.1 mm, xmax = 1072 mm, b = 2.893, a = 0.999, c = 2.1 with r2 = 0.999.
Present-day full-scale fragmentation tests
4.1 Test layout

A total of 6 buckets and 96 ton of magnetite ore have been taken from a specific ring (level 907, block 37,
drift o377) of the Kiruna mine. All 6 buckets have been taken in series after an initial extraction of 20
buckets corresponding to an extraction rate of ~ 7.5 % from a totally loaded mass of 6591 t. Since the actual
ring was situated close to the hanging wall it had not yet reached full height and thus the standard layout of
the SLC rings has not been used. An undisturbed previous extraction as well as operational conditions led to
the selection of this site for the actual test.
The local rock conditions were judged to be competent and undisturbed. The lower part of the ring consisted
of low phosphorous B1 and B2 ore (Fe ~ 67 %, P ~ 0.01 %) and the upper part to some extent of high
phosphorous D-ore (Fe ~ 59 %, P > 0.3 %). B2 ore is, compared to B1 ore, slightly more inhomogeneous and
contains typically green actinolite and pyrite (Niiranen, 2008).
The actual ring comprised 7 boreholes with diameter 115 mm, a forward inclination of 80 and a burden of 3
m to the next ring plane. An overall of 159.9 drilling meters and a theoretical in-situ volume of 4462 m3 yield
a specific drilling amount of 0.17 m/m3. The drilling was done by an automated drill rig (Atlas Copco AB,
SIMBA W469). Since water was present in some of the boreholes almost half of the charging (47 %)
required a substitution of the standard bulk emulsion (Kimulux R) by the packaged emulsion Kimulux 82.
The actual specific charge thus amounted to only 0.23 kg/t instead of the planned 0.35 kg/t. Apart from this
the charge columns and the initiation were implemented according to the operating standard.
4.2 Working procedure

In order to ensure loading of pure magnetite the 6 individual samples have been taken well before any waste
rock inflow. The whole test campaign is summarized by Figure 5.
Pictures have been taken both from the loader buckets filled with material and piles spread out (un- and remixed) on a plane tarmac surface. Results from a calibration attempt of the on-site used granulometry
analysis software WipFrag (Carlsten, 2002; Lith, 2003) will be presented at a later date. Furthermore, 3D
models have been computed by means of photogrammetry (Gaich et al., 2006) which provides for the
moment only a detailed documentation of the piles.
695
Figure 5
Principles of sieving campaign
Table 2
Weighed fractions and cumulative passing values for sieved samples

Sample
Fraction (mm)
1
kg
2
%
kg
3
%
kg
> 200
6 867.5
100.0
10 535.6
100.0
5 503.7
100.0
150-200
1 137.5
58.1
505.6
33.9
593.7
70.2
100-150
2 408.1
51.2
1836.2
30.7
3 064.3
67.0
63-100
299.2
36.5
160.4
19.2
333.4
50.5
35-63
716.7
34.7
414.6
18.1
850.8
48.7
12-35
1 306.7
30.3
771.6
15.5
2 072.1
44.1
0-12
3 656.7
22.3
1703.8
10.7
6 073.5
32.8
Total
16 392.4
15 927.8
18 491.5
Sample
Fraction (mm)
4
kg
5
%
kg
6
%
kg
> 200
5211.8
100.0
1799.9
100.0
3728.0
100.0
150-200
941.8
63.5
409.9
88.3
378.0
76.4
100-150
2 952.4
56.9
2 560.5
85.6
1 768.6
74.0
63-100
169.0
36.2
233.8
68.9
373.8
62.8
35-63
655.0
35.1
695.1
67.3
763.2
60.4
12-35
1 370.8
30.5
2 065.1
62.8
1 853.2
55.6
0-12
2 982.4
20.9
7 555.1
49.3
6 923.2
43.9
Total
14283.2
15319.4
15788.0
696
The top deck of the mobile Finlay 883 sieving unit had cross beams, which close almost all available
openings. Furthermore, only one amongst four possible screen segments was used for the 200 and 100 mm
screens. Thus the accuracy of the corresponding values were considered questionable and they have therefore
been excluded in the analysis. The amount as well as the size of boulders has been recorded. Laboratory
samples have been taken from the stream at the conveyor belt from the masses < 63 mm and sieved down to
0.063 mm by LKAB`s laboratory.
The individual fractions from the full-scale test have been weighed using a calibrated scale (20 kg accuracy)
at the industrial site of LKAB`s subsidiary company KGS. A cross-check with the scale used in the
laboratory has not shown any significant variance in the accuracy.
4.3 Sieving curves for 0-63 mm fraction samples

Figure 6 shows the cumulative fragment size distributions for all 6 lab samples from 0-63 mm. At first view
a pronounced self similarity can be observed since the curves are parallel shifted upward in the log-log
diagram at least up to 10mm. This is expected when fine material follows the NBC concept. Inadequacies in
the sampling method may explain the slightly deviant curve for sample 1.
Mass passing [%]
100
10
Sample 1
Sample 2
Sample 3
Sample 4
Sample 5
Sample 6
Average
1
0.01
0.1
10
100
Mesh size [mm]
Sieving curves for 0-63 mm fraction samples, log-log scales

1.5
1.5
-0.5
-0.5
Mass passing, %
Residuals [9]
Figure 6
10
1
0.1
10
10
1
100
Mesh size, mm
Figure 7
Fragment size distribution for 0-63 mm fraction samples with basic Swebrec function fit.
Data range 0.125-40 mm. Parameters: x50 = 8.01 mm, xmax = 87 mm, b = 2.287, r2 = 0.999.
697
A curve fitting to the averaged lab sample data using the 3 parameter Swebrec function, see Figure 7, gives a
near perfect fit with r2 value of 0.999 over a large interval from 0.125-40 mm.
Furthermore, the residuals are stochastically distributed. To adjust the fitting in the fines region a weighting
of the squared residuals by 1/(x) has been made.
Figure 8 shows the continuous GGS exponent plots derived from the extended Swebrec function for both
the present-day sieving data and Maripuu`s data (caving debris). All curves basically fall on top of each
other, which supports that the sieving curves from the lab samples have NBC character (Ouchterlony and
Moser, 2006). It is clearly visible that the minima for the actual as well as the old data occur at x~0.35-0.4
mm. Based on experience the local position can directly be related to the petrographic character of the
investigated ore (Grasedieck, 2007). In fact the minimum lies where the majority of grains change from
polymineralic assemblies to monomineralic pieces.
0.8
0.7
GGS exponent [-]
0.6
0.5
0.4
Sample 1
0.3
Sample 2
Sample 3
0.2
Sample 4
Sample 5
Sample 6
0.1
Average
Maripuu: caving debris
0.0
0.0
0.1
1.0
10.0
100.0
1000.0
Mesh size [mm]
Figure 8
Continuous GGS exponent curves for actual full-scale sieving and Maripuu`s data
Interestingly the described characteristic is not associated with the chemical composition since the
investigated magnetite ore types are disparate (low versus high phosphorous ore type). Thus the chemical
composition does not seem to have altered the characteristics of the grain assembly much. A petrographic
study of the different ore types would shed more light on this matter.
4.4 Evaluation of complete sieving curves

The construction of complete sieving curves was accomplished in the following way (Ouchterlony et al.,
2006; Wimmer et al., 2008):
1. From the measurements taken from boulders a theoretical, imaginary grizzly size can be computed
(i.e. largest dimension from the largest block). Since a grizzly with a given gap can be passed by
larger fragments than a square mesh with the same dimensions, a flakiness factor has to account for
the difference in screen geometry. The flakiness value has been found to be 1.10. The same flakiness
factor has been considered for the rectangular screen opening 12x30 mm, which gives an effective
mesh size of 13.2mm.
2. The screening at 200, 150 and 100 mm involved hexagonal punch plates whereas the respective
quadratic mesh size can be obtained by a comparison of the available opening (decreased area by a
factor of (3)/2.
698
3. Practical circumstances (see section 4.2) finally impeded the use of the data points at 200 and 100
mm.
4. Since the Swebrec function is relatively linear within the range of 10-60 % passing and the
cumulative sieving data for 63, 35 and 12 mm mesh sizes lie well within this range, absent data
points in the full-scale test can be taken from a straight-line interpolation.
5. As 13.2 mm is missing in the lab screen series, the 13.2 mm data point has been interpolated from
the 12.5 and 16 mm data points.
6. The linear curve behavior within the range of 10-60 % mass passing rendered a splicing at 13.2 mm
possible by scaling the lab sample data for all smaller screen sizes. This makes the lab sample a
relatively smooth continuation of the interpolation line into the fines region.
Table 3 as well as Figure 9 clarifies the procedure of splicing for the construction of a complete
fragmentation distribution on the basis of sample 3.
Table 3
Splicing procedure for construction of a complete sieving curve, sample 3
Data type
Xmax
Finlay, hex
Finlay, quadratic
Lab sample
Finlay, quadratic
Lab sample
Lab sample
Lab sample
Lab sample
Finlay, 12x30mm
Lab sample
Lab sample
Lab sample
Lab sample
Lab sample
Lab sample
Lab sample
Lab sample
Lab sample
Lab sample
Lab sample
Lab sample
Mesh
mm
390
150
63
40
35
30
25
20
16
12
12.5
10
8
6.3
5
4
2
1
0.5
0.25
0.125
0.063
Mesh
mm
429
129.9
63
40
35
30
25
20
16
13.2
12.5
10
8
6.3
5
4
2
1
0.5
0.25
0.125
0.063
Lab sample
Passing %
100.00
91.11
87.90
85.84
82.14
74.07
66.91
63.03
62.06
58.85
53.83
48.40
44.94
41.12
29.56
21.07
16.02
12.70
9.56
5.76
Mass
Fraction kg
6097.4
3397.7
850.8
2072.1
6073.5
* mesh size converted; italic values are interpolated
699
Full scale
Passing %
100.00
67.03
48.65
44.87
44.05
41.61
39.18
36.74
34.79
32.84
Splice
13.2mm
52.10
47.47
45.80
44.73
42.80
38.59
34.86
32.84
32.33
30.66
28.05
25.22
23.41
21.43
15.40
10.98
8.35
6.62
4.98
3.00
Final curve
Passing %
100.00
67.03
48.65
44.87
44.05
41.61
39.18
36.74
34.79
32.84
32.33
30.66
28.05
25.22
23.41
21.43
15.40
10.98
8.35
6.62
4.98
3.00
100
Full-scale sieving
90
Grizzly (imaginary)
80
200/100 mm, hex corrected (excluded)
70
Full-scale sieving: flakiness + hex corrected
Mass passing [%]
Lab sample 0-63 mm

60
Interpolated data
50
40
30
20
10
0
0.01
0.10
1.00
10.00
100.00
1000.00
Mesh size [mm]
Figure 9
Construction of complete sieving curve for sample 3, log-lin scales
Analysis and discussion of results
The differences between the fragment size distributions for six consecutively taken buckets are substantial,
see Figure 10. The parameter x50 varies between 14.3 mm (sample 5) and 277.6 mm (sample 2). Similar large
variations in the fragment size distribution for blasted magnetite ore have also been observed earlier by
Maripuu (1968) and are also common for the raw material delivered by the primary crusher to the processing
plant (Hahne et al., 2003). On average x50 amounts to 86.5 mm however and this is directly comparable to
the mean x50 value that Maripuu (1968) measured in earlier tests (see section 3).
100.0
Mass passing [%]
10.0
Sample 1
1.0
Sample 2
Sample 3
Sample 4
Sample 5
Sample 6
Average
0.1
0.01
0.1
10
Mesh size [mm]
Figure 10 Constructed, complete sieving curves for all samples
700
100
1000
The shapes of the complete sieving curves do not entirely fit the Swebrec function. By and large this is
somewhat surprising as a deviation from this behaviour is an exception. An increased amount of fines < 40
mm causes a flattening and thus non-uniform continuation of the curve to coarser fractions.
At first view the curves remind of those obtained by blasting near the critical burden, i.e. limited number of
large fragments plus a fines tail that follows the Swebrec function (Ouchterlony, 2008). This is a tempting
explanation as the specific charge in the SLC ring was substantially lower than normal, see chapter 4.1.
However, a disturbance free extraction for the investigated ring makes this assumption rather doubtful.
On the other hand, the flattening effect observed for the mid sizes may also be due to selective breakage
behaviour and it is comparable to the shape of fragment size distributions typically affected by autogeneous
grinding (Lynch, 1977; Hahne et al., 2003). This bears a direct analogy to an effect attributable to secondary
fragmentation in block caving (Brown, 2003; Eadie, 2002) as the blocks constituting the primary
fragmentation move down through the draw column to the respective draw point. A similar occurrence
seems to be likely in the drawing of caving debris from a SLC ring and has been assumed earlier (Larsson,
1998). A process like this may cause a depletion of medium size fractions but still leave strong large
fragments.
Several of the following mechanisms might be relevant for causing secondary fragmentation:
x
Self-breakage depending on shape and strength of particles
Abrasion effects between particles (autogenous grinding)
Crushing of fragments under superimposed load
Fracturing & failure of larger blocks under different forms of load
Because of the low specific charge within the blasted ring (see section 4.1) more coarse material than usual
can be expected and would constitute an ideal feed for the described mechanisms related to larger fragments.
Self breakage effects have been observed to a great extent during the sieving process as many of the
fragments that dropped from the conveyor belt split apart when hitting the truck tray. Previous investigations
of ore processing characteristics have in particular shown that a) a significant fraction has in fact the unusual
tendency to break very quickly within a batch mill, thus generating a large fraction of very fine material
(Lichter, 2007) and b) the resistance of feed samples both to impact- and abrasion breakage increases
markedly as the distance of sampling from the mining face is increased (Hahne et al., 2003).
It can be assumed that the actual draw height and consequently the stress regime as well as the corresponding
draw rate are influencing the resulting secondary fragmentation substantially. The scale in mining layout has
changed tremendously during the years and resulted in a sublevel height of 28.5 m nowadays. It was 9 m
when Maripuu (1968) made his study, in which the curves show no distinct effect of secondary
fragmentation. From this follows that one could expect more comminution occurring in the debris flow, due
to longer flow paths. On the other hand, the present buckets were taken before 10 % draw, which speaks in
favour of relatively short paths, unless the draw is extremely uneven like as for a progressively upwards
developing shallow draw phenomenon (Power, 2004a-b; Selldn and Pierce, 2004).
The actual presence and extent of secondary fragmentation effects is difficult to verify. However, two main
strategies would exist as a) the shape and surface properties of fine particles could tell if the material has
been exerted to autogenous grinding (Forssberg and Zhai, 1985) and b) the whole process could be studied in
full-scale with defined output parameters within an ore pass.
In addition to possible secondary fragmentation effects segregation and thus preferential flow of certain
fractions could also help to explain the observations made.
Finally, the details of the actual sieving process, with some openings covered underneath by beams of the
machinery and reduced number of sieve segments used, could have distorted the results though. We tried to
compensate for this by excluding the tainted data but we do not know for certain that this has been entirely
successful.
701
Despite this flattening deviation from Swebrec distribution behaviour which is likely to be caused by the
internal flow mechanisms in the SLC process the main conclusion is still that magnetite itself behaves like
waste rock from a blasting point of view.
Acknowledgement
Hjalmar Lundbohm Research Centre (HLRC) is thanked for supporting the PhD project Improved breakage
and flow in sublevel caving and the financial support of this sieving campaign. LKAB is thanked for the
support of this project and the practical support on-site. Special thanks goes to Joel Kangas for all help
provided in the practical matters of the sieving campaign
References
Brown, E.T. (2003) Block Caving Geomechanics The International Caving Study Stage I 1997-2000. JKMRC
monograph series in mining and mineral processing, No. 3, Brisbane, ISBN 1-74112-000-4, 516 p.
Carlsten, J. (2002) Fragmenteringsmtning med bildanalys. Internal report, LKAB.
Eadie, B.A. (2002) Modelling primary and secondary fragmentation for block caving. PhD thesis, University of
Queensland, Brisbane, Australia.
Forssberg, E. and Zhai, H. (1985) Shape and Surface Properties of the Particles Liberated by Autogenous Grinding
Scandinavian Journal of Metallurgy 14, pp. 25-32.
Gaich, A., Ptsch, M. and Schubert, W. (2006) Acquisition and assessment of geometric rock mass features by true 3D
images. 41st U.S. Rock Mechanics Symposium Golden Rocks 2006, Golden, Colorado, eds. Yale, D.P., Holtz,
S.C., Breeds, C., Ozbay, U., Vol. 2, American Rock Mechanics Association, pp. 738-747.
Grasedieck, A. (2007) Die Natrliche Bruchcharakteristik (NBC) von Gesteinen in der Sprengtechnik. PhD thesis,
University of Leoben, Austria.
Hahne, R., Plsson, B.I. and Samskog, P.O. (2003) Ore characterization for-and simulation of- primary autogenous
grinding Minerals Engineering 16, pp. 13-19.
Johansson, D., Ouchterlony, F. and Nyberg, U. (2007a) Modellfrsk med sprngning mot rasmassor inverkan p
styckefall och svllning In Proc. Rock Blasting Committee, BK 2007, Stockholm, pp. 105-120.
Johansson, D., Ouchterlony, F. and Nyberg, U. (2007b) Blasting against confinement, fragmentation and swelling in
model scale In Proc. IV. EFEE World Conference on Explosives and Blasting, ed. Moser, P., Vienna, pp. 13-26.
Larsson, L. (1998) Slutrapport Projekt Skivras 2000. Internal report, LKAB.
Lichter, J. (2007) Review of feed size distribution and mill operating parameters on the performance of the KA2
comminution circuit. Metso Minerals Optimization Services, project number: LK01M, internal report, LKAB.
Lith, A. (2003) Study on the usage of WipFrag to describe blasted ore fragmentation in sublevel caving operations at
the Kiruna mine. Internal report, LKAB.
Lundberg, S. (1961a) Tappning, transport och krossning av skivrasberg p 320 m avv. i KuJ. Delutredning 1: Siktanalys
av skivrasberg. Internal report 116-1071, LKAB
Lundberg, S. (1961b) Tappning, transport och krossning av skivrasberg p 320 m avv. i KuJ. Delutredning 2: Siktanalys
av ortberg. Internal report 116-1071, LKAB
Lynch, A.J. (1977) Mineral Crushing and Grinding Circuits Their Simulation, Optimisation, Design and Control.
Volume 1, Elsevier Scientific Publishing Company, Amsterdam, ISBN 0-444-41528-9, 342 p.
Maripuu, R. (1968) Underskning av siktanalys och styckeform frn skivrasberg vid LKAB, Kiruna. Diploma thesis,
Kungl. Tekniska Hgskolan, Stockholm.
Moser, P., Cheimanoff, N., Ortiz, R. and Hochholdinger, R. (2000) Breakage characteristics in rock blasting In Proc.
I. EFEE World Conference on Explosives and Blasting, ed. Holmberg, R., Munich, pp. 165-170.
Moser, P., Olsson, M., Ouchterlony, F. and Grasedieck, A. (2003) Comparison of the blast fragmentation from labscale and full-scale tests at Brap In Proc. II. EFEE World Conference on Explosives and Blasting, ed.
Holmberg, R., pp. 449-458.
Moser, P. (2005) Less Fines in aggregate and industrial minerals industry In Proc. III. EFEE World Conference on
Explosives and Blasting, ed. Holmberg, R., pp. 567-574.
Niiranen, F. (2008) Personal communication.
Nyberg, U., Fjellborg, S., Olsson, M. and Ouchterlony, F. (2000) Bedmning av sprngskador i ortkontur;
Vibrationsmtningar, skadeprognoser och sprickkartering i magnetitmalm och grberg. SveBeFo Rapport 50,
Svensk Bergteknisk Forskning, Stockholm.
Ouchterlony, F. (2004) Influence of blasting on the size distribution and properties of muckpile fragments, a state-ofthe-art review. Report project P2000-10: Energy optimization during comminution (Swedish Mineral Research
Organisation, MinFO)
Ouchterlony, F. (2005) The Swebrec function: linking fragmentation by blasting and crushing Mining Technology
(Transactions of the Institute of Mining and Metallurgy A) 114, pp. A29-A44.
702
Ouchterlony, F. and Moser, P. (2006) Likenesses and differences in the fragmentation of full-scale and model-scale
blasts In Proc. VIII. International Symposium on Rock Fragmentation by Blasting, Editec, Santiago, Chile, pp.
207-220.
Ouchterlony, F., Olsson, M., Nyberg, U., Andersson, P. and Gustavsson, L. (2006) Constructing the fragment size
distribution of a bench blasting round, using the new Swebrec function In Proc. VIII. International Symposium
on Rock Fragmentation by Blasting, Editec, Santiago, Chile, pp. 332-344.
Ouchterlony, F. (2008) Personal communication.
Power, G. (2004a): Full scale SLC draw trials at Ridgeway Gold Mine In Proc. MassMin 2004, Santiago, Chile, pp.
225-230.
Power, G. (2004b): Modelling granular flow in caving mines: large scale physical modeling and full scale experiments.
PhD thesis, University of Queensland, Brisbane, Australia.
Rohrmoser, S., Hollerer, H. and Comoli, C. (2007) Magnetite ore samples from LKAB`s operation in Kiruna/Sweden
Optimized comminution sequence, natural breakage characteristic, energy register function. Internal report, Inst.
of Mineral Processing, University of Leoben, Austria.
Selldn, H. and Pierce, M. (2004) PFC3D modeling of flow behavior in sublevel caving In Proc. MassMin 2004,
Santiago, Chile, pp. 201-214.
Steiner, H.J. (1991) The significance of the Rittinger equation in present-day comminution technology. In Proc. XVII.
International Minerals Processing Congress, Dresden, Band I, pp. 170-188.
Steiner, H.J. (1998) Zerkleinerungstechnische Eigenschaften von Gesteinen. Felsbau 16, pp. 320-325.
Wimmer, M., Kangas, J. and Ouchterlony, F. (2008) The fragment size distribution of Kiruna magnetite loaded from a
draw point Evaluation and analysis of a full-scale test. Swebrec Report 2008:U2, ISSN 1653-5006.
703
PAPER B
Referenced 3D images from inside cavities and behind rings in sublevel

caving
Matthias Wimmer, Finn Ouchterlony, Peter Moser, Anders Nordqvist

and Gunther Lenz.
Published in:
Proc. 9th International Symposium on Rock Fragmentation by Blasting 2009 (pp. 91100).
London, England: CRC Press.
Rock Fragmentation by Blasting Sanchidrin (ed)

2010 Taylor & Francis Group, London, ISBN 978-0-415-48296-7
Referenced 3D images from inside cavities and behind rings

in sublevel caving
M. Wimmer & F. Ouchterlony
Swebrec, Swedish Blasting Research Centre at Lule University of Technology, Sweden
P. Moser
Department of Mineral Resources & Petroleum Engineering, University of Leoben, Austria
A. Nordqvist
LKAB Kiruna, Sweden
G. Lenz
3G Software & Measurement, Graz, Austria
ABSTRACT: Disturbance free fragmentation and flow in sublevel caving (SLC) are desirable as they
could lead to an increase in ore recovery and a decrease in waste rock loading and nitrates leakage. This
requires a reasonable understanding of the actual rock breakage in semi-confined ring blasting situations, where both the blasted geometry and the breakage itself are hidden. Sometimes there is a chance to
inspect the actual conditions, such as in the opening of a new drawpoint and in hang-ups in which cavities
exist. For these situations a lightweight mobile camera crane has been constructed for use in the LKAB
Kiruna mine. A 3D photogrammetry system has been constructed, which enables the reconstruction of
the ring front/cavity as a geo-referenced mesh model with color information. It is remotely operated from
a safe distance behind in the drift. The paper first reviews the current drill and blast practice. Then it
presents the design, construction and application of the 3D image acquisition system. A case study shows
the measurement application and also the evaluation possibilities. So far the system promises to be a valuable tool in increasing the understanding of SLC breakage and flow.
1
explosive properties and timing are commonly

based on site experience (Bull & Page 2000, Brown
2003). With the general trend towards larger blast
layouts in the past years and considering the fundamental importance of blasting to the success
of a SLC mining operation it is remarkable that
only a limited number of well documented small
and full scale experiments have been undertaken to
quantify the impact of altered blast design parameters on the resulting material flow characteristics
(Larsson 1998, Quinteiro et al. 2001, Quinteiro
2004, Zhang 2005). Both the actual fragmentation
of the blasted ore itself and the geometry of the
real breakage front will affect the material flow.
In ongoing work, the effect of packed aggregate
material on blast fragmentation and aggregate
compaction is studied in model scale (Johansson
et al. 2007 & 2008). In the present paper the focus
is on the actual breakage geometry in full scale.
Nitrate leakage measurements and function
control measurements that are made at regular
intervals in the LKAB Kiruna mine indicate that
1015% of the holes within a single blasted ring do
INTRODUCTION
Blasting in sublevel caving (SLC) has been identified

throughout the literature to have a significant impact
upon material flow characteristics and therefore
on the overall performance (Janelid 1968, Cullum
1974, Marklund 1976, Kvapil 1982, Stazhevskii
1996, Bull and Page 2000, Hustrulid 2000, Rustan
2000, Power 2004a & b, Selldn & Pierce 2004,
Minchinton & Dare-Bryan 2005, Zhang 2005 &
2008, DeGagn & McKinnon 2005).
In SLC blasting takes place in a semi-confined
situation, where the blasted material is allowed to
swell due to the compaction of the caved material
and to a minor extent into the void volume of the
production drift. Even though several analytical and
empirical models have been developed in the past,
the interaction of semi-confined blasting conditions, SLC blast design and rock mass characteristics on blast performance are not well understood.
Layout criteria for ring blasting concerning overall geometry (ring inclination, shoulder hole angles,
design powder factor), burden/spacing ratios,
91
led to a mining sequence with a diagonal stope face.

The elongated ore deposit with a strike length of
around 4 km is subdivided into individual production blocks of about 400 m and sublevel intervals
of 28.5 m. Staggered drift placing with a spacing of 25 m and a Bezier-shaped cross-section of
7.0 5.2 m is applied.
Caving in each production drift is facilitated by
connecting to the drifts and caving masses located
in the level above. In general, two different opening
methods are applied. In one a singular opening slot
is located in a drift along the hangingwall connecting two adjacent panels, in the other 12 openings
are placed within the crosscut. The former method
can be regarded as the standard method and has
technically seen good functionality. By contrast
the latter needs to be applied in situations of a flat
dipping orebody e.g. if the orebody gets progressively thicker with depth or if sudden changes in
the strike direction exist. Both methods have 8
blastholes arranged in a square pattern (2 2 m)
around a cut hole which is reamed from 254 to
700 mm by boxhole drilling.
The SLC ring layout has in the past years undergone some basic changes, see Figure 1. Marker
trials (Larsson 1998) have shown that the layout
with shorter and flatter side holes to initiate interactive draw zones does not work but that flow
rather occurs within a relatively slim vertical zone.
Therefore the layout has, with a transition level in
between changed to a silo-shaped layout.
The standard ring layout involves drilling 8 holes
with a diameter of 115 mm and a burden of 3 m. The
individual rings have a forward inclination of 80 and
shoulder angles of typically 73. The specific drilling
amounts to about 0.03 m/tonne for a full-sized ring,
yielding a tonnage of 10000 t of ore. All drilling is
performed with fully automated drill rigs having a
total capacity of 1 million m per year. The drilling
equipment itself consists of a Wassara W100 DTH
hammer powered by a high-pressure water pump.
Data communication to and from the rigs is accomplished with wireless LAN technique and sent to a
control room from which they are controlled.
not detonate as planned (Hedstrm 2000, Fjellborg

2002, Zhang 2005).
If the holes then fail to break out, the breakage
front becomes uneven. This provides an irregular burden for the subsequent ring blast. Uneven
breakage then probably causes an uneven fragmentation and in extreme cases maybe even fractured
but immobilized pillar remnants that obstructs a
regular caving flow. Possible causes for poor breakage are uncharged holes that become clogged,
charged holes that are sheared off because of excessive back-breakage, dead pressing from charges in
previously blasted rings, influence from side-holes
in the same ring, etc.
However, to verify the actual reasons a reasonable understanding of the actual rock breakage in
semi-confined blasting situations, where both the
blasted geometry and the breakage itself are hidden, is required. Visual observations are sparse and
difficult to perform. A dedicated test area in the
LKAB Malmberget mine with blasts into an open
stope allowed taking photographs in steps from
the bottom to the top of 3 ring faces (Rustan et al.
1988, Rustan 1993). The subsequent interpretation
of 2D images stitched together led to revised design
rules and to a recommendation for an increased
minimum distance between charged boreholes to
obtain undisturbed detonations.
Recently the idea to study SLC ring blasting effects
in a more controllable environment has been tried in
the abandoned open pit in Svappavaara where parts
of a rotated standard ring have been drilled behind
a bench face (Olsson et al. 2008). Unfortunately the
geological conditions made a meaningful interpretation of the blast results difficult.
At about the same time a HLRC (Hjalmar
Lundbohm Research Centre) sponsored project
with the name Improved breakage and flow in
sublevel caving was initiated and the project is now
going on its 3rd year. Previous fragmentation work
in the project was recently presented (Wimmer et al.
2008). One of the tasks in the HLRC project has
been to construct a system, which allows the acquisition of geo-referenced 3D images inside cavities
and behind rings in the LKAB Kiruna mine and to
use it in the mine. This work is described after the
current drill and blast practice.
2
2.1
2.2
Charging and blasting procedure
Charging of the holes is done by trucks with a

capacity of about 10 tonne. They are specifically
designed for pumping the emulsion explosive
KR0500 (Kimulux Repumpable 0500), which is
sensitized by glass micro-balloons and contains
aluminum. Normal pumping capacity is around
7580 kg explosives per minute. Charging starts at
the bottom of the hole and the hose is retracted automatically, yielding a specific charge concentration
of 1112 kg/m. The operation is carried out via
electrical and remote-operated hydraulic systems.
CURRENT DRILL AND BLAST

PRACTICE AT LKAB IN KIRUNA
Drilling layout
The minable thickness of the ore deposit has

an average thickness of 80100 m with a dip of
6070 towards east, which enables predominantly
a cross-cutting SLC layout. Rock mechanical considerations as well as quality control reasons have
92
Figure 2.
Figure 1. North-South section showing different ring

layouts at levels 849, 878 and 907 m.
Standard charging column.
up to one year for pre-charged rings might occur

and crystallization within the emulsion may occur.
To some extent this is counteracted by the multiple primer configuration, which with its additional
PETN primer serves as a safety precaution. When
a ring is prepared for blasting the detonating cords
are pulled down from the borehole collar, bundled
by a 5 g/m firing-cord, which is finally electrically
initiated.
The described assembly is both easy to apply and
makes individual placement of the multiple primer
system possible according to a-priori charging
plans. Usually rings are individually blasted yielding a specific charge of about 0.3 kg/tonne for the
standard layout.
Sometimes packaged emulsion (Kimulux 82) needs

to be used in the case of wet boreholes.
The initiation itself (Figure 2) consists of a system
of two sliding primers SLP-50 (Sliding Primer-50)
composed of cap sensitive emulsion packed in a hollow plastic casing. A 25 g primer made of PETN
is placed in the central hole and initiated by a nonelectric detonator with a short lead (0.6 m) of shock
tube. The SLP-50 casing also has a guide notch for
a 3.6 g/m detonating cord, which ensures a safe lateral initiation of the shock tube. Before charging a
support claw is fixed at the end of the detonating
cord, which is pushed to the bottom of the hole. At
designated lengths, the two SLP-50 primers are fixed
to the cord by cable straps. The charging hose is thin
enough to pass the SLP casings. The following general drilling and charging restrictions apply: (i) minimum distance between charged parts of adjacent
holes = 1.4 m to avoid negative interaction effects,
(ii) maximum distance between adjacent holes = 4 m
to ensure proper breakage, and (iii) uncharged collar
length = 2 m or more, exactly 2 m for holes 1 and 8
(the side holes).
The firing pattern is normally designed to open
the round symmetrically from the centre of the ring
with delay intervals of 25 ms and the same delay
time for both boosters. Safety issues related to
brow damage, ease of access in terms of the existing muckpile and non-clogged boreholes as well as
efficiency are the reasons that typically 23 rings
are charged at the same time. Therefore sleeping
times are typically one month but extreme periods
3 APPLICATION OF A 3D IMAGE
ACQUISITION SYSTEM
3.1 Basic considerations
Both the actual blasted geometry and the breakage itself are hidden in the controlled SLC mining
operation, which makes it very difficult to observe
the actual blasting result. However, under certain
conditions these restrictions do not apply such as in
the opening of a new drawpoint and in hang-ups in
which cavities exist. The opening case constitutes a
controllable roof-left situation before caving is initiated. The latter is an unpredictable occurrence during mucking with mostly unknown circumstances
93
Figure 3 shows the measurement procedure and

Figure 4 the present measuring equipment in practical use on-site. Taking a single image pair yields
a partial model of the breakage face. However, a
complete 3D image of the entire breakage front
requires merging of several overlapping models.
Therefore image pairs are gathered in vertical
columns from the bottom up, starting in the drift
direction. The vertical tilt angle and the horizontal
swing angle that sweep the target area are guided
by a mounted surveillance camera. A set of cube
corner prisms arranged around the digital camera
allow its positioning in the mines coordinate system from total station measurements made when
acquiring the first images (lowest one in the central column in Figure 3; marked in bold). Parallel
mounted point lasers give light spots on the cavity surface that provide the length scale in each
image pair. An entire measurement typically takes
15 minutes, well in agreement with the mines risk
analysis to minimize the exposure time.
and thus requires increased safety precautions in

case of a survey. In principle both situations grant
unique access possibilities to the actual breakage
geometry. This makes it possible to analyze or
map: (i) the structure of the ring face, (ii) the drilling accuracy (visible half-casts vs. planned trajectories), (iii) the back- and under-breakage as well
as brow damage, and (iv) the actual broken area,
roughness, orientation and height.
Hang-ups are of special interest for two reasons
a) the blasting was actually carried out in a semiconfined condition and b) they give direct access
to study gravity flow mechanisms. Besides the trapping or wedging of one or several large blocks, the
particular type of hang-up in which ore is temporarily frozen has generated substantial interest
(shallow draw phenomenon; Power 2004, Selldn &
Pierce 2004).
Gravity flow related issues of interest are:
Geometry of the hang-up (height, span width)
Which type of material builds up the hang-up?
What do the fragmentation and void ratio of ore
and caving masses look like?
Does a gap exist between the pillar and the compacted material?
3.2
Design and construction of the camera crane
The design and the construction of the measuring

equipment were made in an interactive process
together with the central workshop of Lule University of Technology (LTU).
The lightweight crane is mounted on a car
trailer. It is collapsible and has a maximum 50
inclinable 12 m long boom. In combination with
an additional extension rod between the car and
the trailer a safety distance for the car of up to
23 m can be reached. A rotating camera head with
Additionally a correlation with the actual production data (drilling, blasting, loading, etc.)
should contribute to a better understanding of the
reasons for hang-ups. This should in the long term
also contribute to avoiding these situations.
The task was to design a system that would
allow a quantitative study of openings and hangups along the lines mentioned above considering
the risks of sudden cavity or hang-up collapse.
Eventually a photogrammetry system mounted
on a mobile camera crane with the option to
geo-reference the 3D images was chosen. Digital
photogrammetry can be regarded as a mature
technology yielding reliable and reproducible
results with fast acquisition time and, compared
with laser scanner technology, low hardware costs
(Tonon & Kottenstette 2006).
Each measurement made has to be closely coordinated in time with the responsible production
personnel. If an on-site safety judgement of a cavity permits a measurement to be made sufficient
material is mucked to facilitate a proper placement
of the cranes head within the cavity. The camera
crane is then transported in a collapsed state to the
test site and mounted at a reasonably safe distance
from the mining front. This takes 12 hours.
The measurement after inserting the crane in the
cavity is safely remotely operated from a distance of
around 50 m behind in the drift. Production in adjacent drifts is stopped because it could trigger rockfalls or slope failure of the material not mucked.
Figure 3. Measurement procedure on-site (not to scale).
Figure 4. Camera crane at a drawpoint in SLC crosscut.
94
interest from slightly different positions. A dense

set of corresponding image points is derived automatically from these images utilizing techniques
from multidimensional statistics. From this set
the relative orientation of the 2 camera positions
is computed. Intersecting the rays of corresponding image points from these positions yields a 3D
point cloud. Finally these points are triangulated
to form a surface mesh, which is directly linked to
the original image data.
Prerequisites are the calibration data of the
imaging system to take the distortion of the lenses
and alignment of the optical axis into account.
The resulting 3D image already represents the true
shape of the surface but lacks the correct size and
orientation in 3D space. However, the use of photogrammetry in this unique measurement situation
has required the development of some customized
software extensions to the existing ShapeMetriX3D
(SMX3D) software (Gaich et al. 2006).
To achieve proper scaling a projection unit
made up of 4 laser pointers has been mounted on
the camera rig, see Figure 6. The laser modules,
with guaranteed collinear laser beams and casing
axes, are installed on a 50 50 cm frame and the
unit emits parallel beams normal to the imaging
plane of the cameras. The scaling factor (s) for the
3D image can then be derived by the ratio of the
known distance between the beams and the distance between the normal projection onto the yzplane of 2 observed laser points, e.g. s = (P4 P5)/
[nproject(P11 P12)].
To support geo-referencing of the 3D image,
reflecting prisms have been arranged in an equilateral triangle of 700 mm side length around the
imaging system. By surveying the locations P1 to
P3 of these prisms in the object coordinate system
(x, y, z) and their a priori known positions in the
imaging coordinate system (x, y, z) a transformation matrix M can be derived. M contains the translation and rotation between these 2 systems, so all
points from the correctly scaled 3D image can be
transferred to the global object coordinate system.
ancillary equipment (see Figure 5) is mounted at

the end of the boom. A custom-calibrated digital
camera (Canon Powershot G7 with wide-angle
conversion lens 0.7 ) is moveable along a 2.2 m
long traverse. It is used to take the stereographic
image pair. A laser distance measurement device
(Riegl LaserScout) is used to measure the distance
to the object which allows the estimation of the
required camera spacing for the two images. Pretests have shown that the ratio of spacing to the
distance of the object could be up to 1/12 and still
yield reproducible and correct results. The positioning of the camera carriage is controlled by a
surveillance camera and another camera with a
fisheye lens gives a general overview on the cavity
under investigation. Lights with 2 1000 W provide sufficient illumination for taking the images.
A combination of high precision and light weight
was the goal when mounting the 4 laser modules
(LT-series, Laser Components) and the 360 mini
prisms (Leica GRZ101) on the crane head, which are
both used for the photogrammetry system described
below. The need to remotely control all components
of the equipment from a safe distance (up to 100 m)
required special solutions in terms of signal converting. A diesel generator with an engine output of 3
kW and voltage stabilizer powers the equipment.
The design and construction of the camera
crane took 1.5 years and consisted of a) preparatory work (7 months), b) actual construction at
LTU (4 months) and c) testing on-site with successive improvements for usage in the underground
mine and verification tests (7 months).
3.3 Photogrammetry system
A 3D image is the union of a set of photographic
images and the three-dimensional geometry of the
observed objects. It is generated by means of classical photogrammetry and extensions from computer vision (Gaich et al. 2006). The creation process
starts with taking a pair of images of the object of
Figure 5. Rotating camera head with ancillary equipment.
Figure 6. Geometric arrangement of the laser modules.
95
Due to the limited field of view of the imaging

system and the arrangement of the cameras it is
not possible to create a full spherical 3D image
from only one pair of images and thus requires a
merging technique. This procedure thereby takes
structural information as well as topographic
information into account and does not need any
reference points. Thereby corresponding points
are determined either interactively or by an automatic search algorithm in the ModelMerger part
of SMX3D. To facilitate this, in the actual application, the capability had to be extended to support a
larger number of models, i.e. the merging of up to
15 instead of the default 6 individual 3D models.
Figure 7.
A MEASUREMENT APPLICATION
AND ITS EVALUATION POSSIBILITIES
Isometric view of ring outlines nos. 112.
4.1 Description of test site

An overview on the selected test site (level 716, block
9, drift 84) is given by Figure 7. The 3D view with
bordered contour lines of the blasted rings shows
the stepwise increase in height from the actual opening placed within the crosscut. The stope geometry
has been adapted to the local formation of the orebody and also explains the asymmetric rings.
Figure 8 shows the geometry and initiation timing
of the opening and its subsequent individual rings.
Numbering of boreholes within an individual ring is
made from left to right. The highly confined opening,
rings nos. 18, was made as a single blast with use of
LP detonators (square-marked). The function of the
small-sized ring no 9 is solely to transfer to a ring
inclination of 70. For ring no 9 and henceforward
SP detonators (encircled) were used. The absence of
geometrical contact with the caving masses above
as well as a small span width of the opening at this
stage made a reasonably safe survey of this cavity
possible. The measurement on-site was made after
blasting ring no 9. Thus the surveyed breakage front
comprises a combination of ring no 8 (upper part,
80) as well as ring no 9 (lower part, 70).
Figure 8. Section at level 706.8 m showing rings nos. 110

including initiation times (LP detonators square marked,
SP detonators encircled).
fracture planes. The stereographic projection in

Figure 10 shows the mean orientation of the whole
ring face (great circle) and the identified discontinuities subdivided into different joint sets (poles).
The steep viewing angles cause an increased
occurrence of glancing intersections. This makes the
identification of discontinuities in the upper regions
difficult and therefore the evaluation is based on the
lower parts of the ring face. A structure set (007/65;
set 1) stands out, which perpendicularly and steeply
inclined intersects the ring face (102/57). Another
distinctive joint set is almost parallel to the face
(122/55; set 2). In addition, minor joint sets exist
at partial areas intersecting the face at various
angles and with different dip angles.
4.2 Assessment of local rock conditions

In order to evaluate the influence of the local rock
conditions on the blasting result a detailed mapping of discontinuities has been made by means
of the photogrammetrically derived 3D model
using the program JMX Analyst (Gaich et al.
2006), see Figure 9. The surveyed face consists of
massive, homogenous and fine-grained (<1 mm)
low phosphorous magnetite ore (so-called B-ore
with 6869% Fe and 0.02% P; Niiranen 2006).
Calcite and/or hematite fillings (<5 mm thick) are
present and facilitate the recognition of natural
4.3
Evaluation of the blasting results
Figure 11 shows the generated geo-referenced and

merged low-resolution 3D model imported into the
96
digital mining map, including the drill holes of rings

nos. 810. For a convenient handling of data the
number of points has been reduced to 5000 from the
approx. 1.1 million reconstructed 3D points of the
original grid with a resolution from 0.0160.042 m.
The 3D model in color reveals that half-casts can
only be seen a short distance before the actual breakage of the roof stopped. Note firstly that it is not
adequate in this situation to refer to borehole measurements since collar positions are not measured.
However, the drill rigs in use aim for planned toe
positions. This allows comparisons of identifiable
borehole sockets at the roof for ring no 8 with the
planned position at this specific height, see Table 1.
The boreholes have a mean deviation from their
planned positions in the x-direction of 0.40 0.21 m
and 0.49 0.43 m in the y-direction respectively.
We observe that the deviations in the y-direction
(i.e. within the ring plane) decrease symmetrically
towards the middle hole when holes become more
vertical. All holes are drilled much closer than actually planned but quite large deviations can only be
identified for the side holes which results in much
narrower boundaries of the blasted ring face. A new
view on Figure 11 indicates that for these holes the

collar position has simply been not held which could
be due to various geometrical/technical reasons.
To study back- or rather under-break effects sections are drawn perpendicularly to the plane of the
actual blasted ring (section 707.95701.38 m ring
no 9, section 701.08697.15 m ring no 8) with the
lowest starting 1 m up from the crown and along the
ring plane (section 707.95 m). Figure 12 plots sections at different heights including planned drillhole
positions of the actual blasted rings nos. 8 and 9 and
subsequent ring no 10. Uncharged collar lengths are
encircled and unbroken holes of ring no 8 as identified from the 3D image are marked with a square.
Despite the uncharged collar length in the lower
part severe back-breakage can be recognized up to
around 4 m and amounts up to 2/3 of the burden
of the next ring. This might in a more general way
be explained by the reflected tensile stress from the
upper surface of the drift. In this specific situation
we rather believe that an unfavorable orientation
of the strata and its activation by gasses during
blasting could be the cause. The described situation implies that charges in the next ring may get
Figure 9. Evaluation of joint sets based on the lower

ring part.
Figure 10. Stereographic projection analysis, 70 entries.
Figure 11. Geo-referenced and merged 3D model imported to the digital mining map together with ring nos. 810.
(1) side view, viewing direction north, (2) front view, viewing direction west and (3) isometric view.
97
damaged and the gravity flow might be disturbed.

Exceeding back breakage is also a safety problem
for the charging personnel.
The drilling deviations of the side holes from
ring nos. 8 and 9 cause a considerably smaller size
of the ring than planned to break out. They fail
eventually close to the section at 702.32 m. Hereby
the identifiable and distinct edges coincide exactly
with the measured drilling deviations. They demonstrate that the energy available is nearly insufficient to break out the actual burden. This problem
builds up and the breakage area gets progressively
smaller when with section at 701.38 only 3 out of 7
holes succeed to break out. With increasing height
the actual breakage area decreases further and this
finally results in a cavity with a chimney shaped
form and a height of around 12 m.
Table 1.
For the blasting of the following ring no 10

this is an unfavorable starting geometry. It will be
very difficult for ring no 10 to reach the designated
breaking height. This concerns especially holes nos.
6 and 7 on the left side where up to 8 m (or 50%) of
their heights remain unblasted. The right side is less
affected from this problem as approx. only 34 m
are unbroken (holes 13) and the holes are shorter.
By contrast the middle hole no 4 has broken out
almost to its designated height. As seen in Figure 7
it is facing the raise opening with a burden of just
1 m. Holes nos. 6 and 7 have on the other side a
planned burden of 2 m and a rapidly increasing toe
confinement to break out. Observations based on
the 3D image suggest that the borehole charges have
detonated but rather incomplete breakage occurred
which might actually have several reasons.
The most likely reason is a poor breakage within
the preceding rings that are part of the opening
where the blasting situation is highly confined.
This agrees with the experience on-site that any
problems occurring in an opening of a new drift
are difficult to correct in the following blasts. As
breakage problems start to occur precisely at the
height where the ring no 8 changes from partly
uncharged collar lengths to fully charged columns
and at the same time rings nos. 8 and 9 come closer
to each other, interaction effects are also conceivable. The first interaction effect could occur in ring
no 8 because the drilling deviations in the ring
work towards smaller spacings. The second one is
dead-pressing effects in ring no 9 from charges in
ring no 8 when the rings get as close as the planned
minimum burden distance of 1 m.
Evaluation of measured borehole sockets.
Planned borehole
Target valuemeasurement
Position
Height
Nos.
Side
angle
Length x
m
m
y
m
r
m
Left
m
1
2
3
4
5
6
7
59.62
70.95
82.10
92.94
102.97
111.96
120.38
11.10
10.47
10.78
12.36
14.50
17.78
17.11
1.11
0.60
0.04
0.02
0.23
0.50
0.90
1.12
0.72
0.48
0.45
0.39
0.54
1.25
3.10
3.63
4.54
1.03
3.40
7.96
7.31
0.17
0.41
0.48
0.45
0.31
0.21
0.80
Figure 12. Sections perpendicular to the actual blasted ring plane at different heights with uncharged collar lengths
(encircled) and unbroken hole parts (marked by squares).
98
The HLRC project will increasingly focus on the

study of hang-ups occuring at irregular intervals.
They do however allow a direct analysis of gravity flow related issues such as; i) geometry of the
hang-up (height, span width), ii) material type (ore,
caving masses), iii) fragmentation and void ratio,
iv) verification of the existence of gaps between blast
front and v) conditions of the compacted material,
etc. The insight then gained will go more towards the
heart of the SLC process. The ring that resulted in
a hang-up malfunction is a deviation from working
breakage and flow and not related to the opening with
its special problems. This insight and understanding
of the reasons for hang-ups will then hopefully contribute to avoid such situations in the future.
At the same time final preparations are made to
use TDR (Time-Domain Reflectometry) technology to complement the visual observations of cavities. It will be used to verify the blast function in the
upper part of the ring (above 25 m height). Boreholes will be drilled from the longitudinal drifts in
the footwall, piercing the interfaces of the intact
pillar and the succeeding blast rings (Wimmer &
Ouchterlony 2008). Coaxial cables grouted into
the boreholes will then be long-term monitored to
determine both the actual completed breakage and
back-breakage for each ring and the deformations
behind the blasting front. The sum of all these measurements and results analysis will hopefully direct
appropriate innovative actions that contribute to
improved breakage and flow in sublevel caving.
This preliminary analysis shows the potential

that lies in an interpretation of the 3D images.
In the future a more systematic survey of several
cavities will be made in the ongoing HLRC project.
This will allow better conclusions about cause and
effect of imperfect ring blasts.
5
CONCLUDING REMARKS
Sublevel caving is a highly productive mining

method with the major disadvantages of high ore
loss and dilution. Several factors may be causing
this but the literature generally agrees that blasting
has a significant impact upon material flow characteristics and hence on the overall performance.
Both nitrate and function control measurements
have regularly shown that several holes in each
ring may not detonate as planned, which leads to a
number of negative results such as:
An unscheduled reduction of the planned specific charge
Fragmentation becomes coarser and the frequency of boulders increases
Irregularities in the material flow and problems
during loading operation
Immobilized but fractured pillar remnants causing ore losses
Increased costs in downstream operations such
as crushing/grinding
Nitrogen compounds contaminating surface
and ground water in the environment.
ACKNOWLEDGEMENT
However, to determine the causes behind different scenarios a reasonable understanding of

the actual rock breakage, where both the blasted
geometry and the breakage itself are hidden, is
required. Visual observations are sparse and difficult to perform.
The construction and use of the 3D image
acquisition system presented here is an important
step forward towards a better insight in the actual
process. The case study has demonstrated that an
evaluation of the following quantities is possible;
i) geological structure mapping, ii) drilling accuracy based on identified boreholes and iii) analysis
of back- under-break. A systematic survey, which
is in progress, will allow a deeper study of; i) interaction effects between adjacent holes, ii) effective
burden values for subsequent blasts, iii) a roughness description of the broken surface etc. From
such data various blasting situations and layouts
in areas with different rock conditions may be analyzed and also correlated with actual production
data (drilling, charging, loading) etc. Combining
and generalizing the results of these different observations will hopefully increase the understanding
of the SLC blasting and flow process considerably.
The Hjalmar Lundbohm Research Centre (HLRC), a

research foundation at LTU and endowed by LKAB
is thanked for its financial support throughout the
PhD project Improved breakage and flow in sublevel caving. LTUs central workshop, represented
by Tommy Nilsson, and the colleagues at Complab
are thanked for numerous and valuable discussions
in connection with the development of the measurement equipment. Our colleagues at the company
3G Software & Measurement in Graz, Austria are
thanked for their innovative support. Kari Niiranen
and Agneta Nordmark from LKAB have assisted
with technical matters. Special thanks finally go to
Joel Kangas, Ola Sllstrm and the surveyors team
represented by Anders Berg for all their help consistently provided in practical matters on-site.
REFERENCES
Brown, E.T. 2003. Block Caving GeomechanicsThe
International Caving Study I 19972000. JKMRC
Monograph Series in Mining and Mineral Processing
3: 226244. Indooroopilly, QLD: JKMRC.
99
Bull, G. & Page, C.H. 2000. Sublevel cavingtodays

dependable low cost ore factory. In G. Chitombo
(ed.), Proc. 3th Int. Conf. and Exhib. on Mass Mining:
537556, Brisbane, 29 October2 November. Brisbane,
QLD: AusIMM.
Cullum, A.J. 1974. The effects of confined blasting on
rock fragmentation and flow characteristics in sublevel
caving. MSc Thesis, 161 p. Brisbane: Univ. QLD.
DeGagn, D.O. & McKinnon, S.D. 2005. The influence
of blasting fragmentation on ore recovery in sublevel cave mines. In G. Chen et al. (eds), Proc. Alaska
Rocks 2005Rock Mechanics for Energy, Mineral and
Infrastructure Dev. in the Northern Region: ARMA/
USRMS 05-811, Anchorage, 2529 June. Alexandria,
VA: Am. Rock Mechs Ass.
Fjellborg, S. 2002. The value of measuring VOD in large scale
sublevel caving. In X. Wang (ed.), Proc. 7th Int. Symp. on
Rock Fragmentation by Blasting: 717724, Beijing, 1115
August. Beijing: Metallurgical Industry Press.
Gaich, A., Ptsch, M. & Schubert, W. 2006. Acquisition
and assessment of geometric rock mass features by true
3D images. In D.P. Yale et al. (eds), Proc. 41st U.S. Rock
Mech. Symp.Golden Rocks 2006: 738747, Golden,
1721 June. Alexandria, VA: Am. Rock Mechs Ass.
Hedstrm, O. 2000: Funktionskontroll av produktionssalvor i Kiirunavaara. MSc Thesis, 42 p. Lule: LTU.
In Swedish.
Hustrulid, W.A. 2000. Method selection for large scale
underground mining. In G. Chitombo (ed.), Proc. 3th
Int. Conf. and Exhib. on Mass Mining: 2956, Brisbane,
29 October-2 November. Brisbane, QLD: AusIMM.
Janelid, I. 1968. Sublevel caving: how to use it; what are
advantages, problems. World Mining 21(9): 7678.
Johansson, D., Ouchterlony, F. & Nyberg, U. 2007. Blasting against confinement, fragmentation and swelling
in model scale. In P. Moser et al. (eds), Proc. 4th EFEE
World Conference on Explosives and Blasting: 1326,
Vienna, 911 September. Vienna: European Federation of Explosives Engineers.
Johansson, D., Ouchterlony, F., Edin, J., Martinsson, L. &
Nyberg, U. 2008. Blasting against confinement, fragmentation and compaction in model scale. In H. Schunnesson & E. Nordlund (eds), Proc. 5th Int. Conf. and Exhib.
on Mass Mining: 681690, Lule, 911 June. Lule: Div.
on Mining and Geotechnical Engineering.
Kvapil, R. 1982. The mechanics and design of sublevel
caving systems. In W.A. Hustrulid (ed.), Underground
mining methods handbook: 880897. New York: SME.
Larsson, L. 1998. Slutrapport Projekt Skivras 2000.
Internal report, 37 p. Kiruna: LKAB. In Swedish.
Marklund, I. 1976. Sprngningens inverkan p rasfrloppet. In H. Hedn (ed.), Proc. RasbrytningsSeminarium: L1-L9, Kiruna, 1718 November.
Kiruna: LKAB. In Swedish.
Minchinton, A. & Dare-Bryan, P. 2005. The application of computer modeling for blasting and flow in
sublevel caving operations. In Proc. 9th Underground
Operators Conference: 6573, Perth, 79 March.
Perth, WA: AusIMM.
Niiranen, K. 2006. The deep parts of the Kiirunavaara
apatite-magnetite orebody in Kiruna, Northern
Sweden and the impact of the future mining on the
township of Kiruna. In S. Dominy (ed.), Proc. 6th Int.
Mining Geology Conference: 5767, Darwin, 2123

August. Darwin, NT: AusIMM.
Olsson, M. & Fjellborg, S. 2009. Lgesrapport Fullskalefrsk i Svappavaara. Internal report, in preparation.
Stockholm: Swebrec. In Swedish.
Power, G. 2004a. Modeling granular flow in caving mines:
large scale physical modeling and full scale experiments. PhD thesis, 265 p. Brisbane: Univ. QLD.
Power, G. 2004b. Full scale SLC draw trials at Ridgeway
Gold Mine. In A. Karzulovic & M.A. Alafaro (eds),
Proc. 4th Int. Conf. and Exhib. on Mass Mining:
225230, Santiago, 2225 August. Santiago: Instituto
de Ingenieros de Chile.
Quinteiro, C., Larsson, L. & Hustrulid, W.A. 2001. Theory and practice of very large scale sublevel caving.
In W.A. Hustrulid & R.L. Bullock (eds), Underground
mining methodsengineering fundamentals and international case studies: 381384. Littleton, CO: SME.
Quinteiro, C. 2004. Final Report: Orica-Fragmentation
Project. Internal report, 39 p. Kiruna: LKAB.
Rustan, A., qvist, J., Nie, S., Stenstrm, T. & Khler O.
1988. Minsta avstnd mellan laddade sprngborrhl
i skivrasbrytning. Technical report 1988:28T, 39 p.
Lule: LTU. In Swedish.
Rustan, A. 1993. Minimum Distance between charged
boreholes for safe detonation. In H.P. Rossmanith
(ed.), Proc. 4th Symp. on Rock Fragmentation by Blasting: 127135, Vienna, 58 July. Rotterdam: Balkema.
Rustan, A. 2000. Gravity flow of broken rockwhat is
known and unknown. In G. Chitombo (ed.), Proc. 3th
Int. Conf. and Exhib. on Mass Mining: 557567, Brisbane,
29 October-2 November. Brisbane, QLD: AusIMM.
Selldn, H. & Pierce, M. 2004. PFC3D modeling of flow
behaviour in sublevel caving. In A. Karzulovic & M.A.
Alafaro (eds), Proc. 4th Int. Conf. and Exhib. on Mass
Mining: 201214, Santiago, 2225 August. Santiago:
Instituto de Ingenieros de Chile.
Stazhevskii, S.B. 1996. Features of flow of broken rock
in extraction of ores with sublevel caving. Journal of
Mining Science 32(5): 403416.
Tonon, F. & Kottenstette, J.T. 2006. Laser and photogrammetric methods for rock face characterization.
Report on a workshop held in conjunction with the
41st U.S. Rock Mech. Symp.Golden Rocks 2006,
120 p. Alexandria, VA: Am. Rock Mechs Ass.
Wimmer, M. & Ouchterlony, F. 2008. Application of TimeDomain Reflectometry (TDR) for block and sublevel
caving minesState-of-the-art and preliminary laboratory shear tests. Internal Report, 30 p. Lule: Swebrec.
Wimmer, M., Ouchterlony, F. & Moser P. 2008. The
fragment size distribution of Kiruna magnetite, from
model-scale to run of the mine. In H. Schunnesson &
E. Nordlund (eds), Proc. 5th Int. Conf. and Exhib. on
Mass Mining: 691703, Lule, 911 June. Lule: Div.
on Mining and Geotechnical Engineering.
Zhang, Z.X. 2005. Increasing ore extraction by changing
detonator positions in LKAB Malmberget mine. FragblastInt. J. Blasting and Fragmentation 9(1): 946.
Zhang, Z.X. 2008. Impact of rock blasting on mining
engineering. In H. Schunnesson & E. Nordlund
(eds), Proc. 5th Int. Conf. and Exhib. on Mass Mining:
671680, Lule, 911 June. Lule: Div. on Mining and
Geotechnical Engineering.
100
PAPER C
3D mapping of sublevel caving (SLC) blast rings and ore flow

disturbances in the LKAB Kiruna mine
Matthias Wimmer, Anders Nordqvist, Finn Ouchterlony, Hkan Selldn

and Gunther Lenz.
Published in:
Proc. 6th International Conference and Exhibition on Mass Mining 2012 (on CD).
Sudbury, Canada: Laurentian University.
3D mapping of sublevel caving (SLC) blast rings and ore flow disturbances in the LKAB
Kiruna mine
Matthias Wimmer & Anders A. Nordqvist

LKAB Kiruna, Sweden
Finn Ouchterlony
Swebrec, Swedish Blasting Research Centre at Lule University of Technology, Sweden
Hkan Selldn
LKAB Kiruna, Sweden
Gunther Lenz
3G Software & Measurement, Austria
ABSTRACT
Sublevel caving (SLC) is a highly productive mining method with the major disadvantages of
irrepressible ore loss and dilution. The confined blasting situation of SLC rings is commonly regarded
to have a significant impact upon the material flow characteristics and hence on the overall
performance. The initial conditions for the ore flow after blasting are unknown, as both the blasted
geometry and the fragmentation itself are normally hidden. However, when opening a new drawpoint
and in hang-up situations an inspection of the actual conditions is feasible. For this purpose, a
remotely operated 3D photogrammetry system was custom built and used to reconstruct the ring front
and/or cavity as a geo-referenced mesh model with colour information. Various blasting situations
have been observed and this allowed a deeper study of the i) broken geometry and height, ii) over- and
underbreak and their effects on subsequently blasted rings, iii) interaction effects between adjacent
holes/rings, iv) drilling accuracy based on identified boreholes and v) mapping of geological structures
and their influence on the blast result. The capabilities of the 3D image acquisition system to evaluate
the blasting results are demonstrated with an example, in which a series of blasts were surveyed.
Monitoring under hang-up situations, has then revealed the actual effects from confined blasting and
gravity flow related issues. The 3D photogrammetry system is now increasingly being used to study
hang-ups. By revealing and quantifying hitherto inaccessible information, the system has proven to be
a valuable tool in increasing the understanding of both SLC breakage and flow.
BIOGRAPHY
Matthias Wimmer is a senior R&D engineer at LKAB Kiruna. His research interests
and specialties are focused on blasting, fragmentation and gravity flow in sublevel
caving. Matthias graduated at the University of Leoben in Austria. During his study
period he gained practical insight in various mining operations and at different tunnel
construction sites in Europe. He continued thereupon at the Swedish Blasting
Research Centre (Swebrec) to work on a PhD thesis entitled Improved breakage and
flow in sublevel caving which is expected to be completed during 2012.
1. INTRODUCTION
Blasting in sublevel caving (SLC) has been identified throughout the literature to have a significant
impact upon material flow characteristics and therefore on the overall performance of the SLC method
(Brunton, 2009). Blasting in SLC takes place in a semi-confined situation. Blasted material swells
while the caved material compacts, and also, to a lesser extent, fills the void volume of the production
drift. Several analytical and empirical models have been developed in the past. However,
understanding of the interaction of semi-confined blasting conditions, SLC blast design and rock mass
characteristics on rock breaking performance is rudimentary.
Nitrate leakage measurements and function control measurements in the LKAB Kiruna mine indicate
that 10-15 % of the holes within a single blasted ring do not detonate as planned (Fjellborg, 2002;
Hedstrm, 2000; Zhang, 2005). If individual holes then fail to break out, the breakage front becomes
uneven. This provides an irregular burden for the subsequent ring blast. Uneven breakage probably
then causes an uneven fragmentation, and in extreme cases maybe even fractured yet immobilized
pillar remnants that obstruct a regular caving flow. Possible causes for poor breakage are uncharged
holes that become clogged, charged holes that are sheared off because of excessive overbreak, dead
pressing from charges in previously blasted rings, influence from side holes in the same ring, etc.
However, to verify the reasons a reasonable understanding of the actual rock breakage in semiconfined blasting situations, where both the blasted geometry and the breakage itself are hidden, is
required. Visual observations are sparse and difficult to perform. A dedicated test area in the LKAB
Malmberget mine with blasts into an open stope allowed taking photographs in steps from the bottom
to the top of 3 ring faces (Rustan, 1993). The subsequent interpretation of 2D images stitched together
led to revised design rules and to a recommendation for an increased minimum distance between
charged boreholes to obtain undisturbed detonations. More recently, the idea to study SLC ring
blasting effects in a more controllable environment has been tried in the abandoned open pit in
Svappavaara where parts of a rotated standard ring have been drilled behind a bench face (Olsson,
Nyberg & Fjellborg, 2009). Unfortunately the geological conditions made a meaningful interpretation
of the blast results difficult. An alternative access to study rock breakage for SLC rings are
sporadically occurring hang-up situations with a temporarily stable arching of the caving material.
Observations are scarce though (Power, 2004; Selldn & Pierce, 2004).
The actual fragmentation and compaction of the blasted ore as well as the geometry of the real
breakage front will affect the material flow. With the objective to systematically map the blasting of
SLC rings and hang-up situations a system was constructed, which allows the acquisition of georeferenced 3D images inside cavities and behind rings (Wimmer, Ouchterlony, Moser, Nordqvist &
Lenz, 2009). In the present paper the focus is on the mapping of the actual breakage geometry of
blasted SLC rings and flow disturbances in full scale.
2. APPLICATION OF A 3D IMAGE ACQUISITION SYSTEM

2.1. Basic considerations
Both the actual blasted geometry and the breakage itself are hidden in the controlled SLC mining
operation, which makes it very difficult to observe the actual blasting result. However, under certain
conditions such as in the opening of a new drawpoint and in hang-ups, observations are possible. The
opening case constitutes a controllable blind open stope situation before caving is initiated. The latter
is an unpredictable occurrence during mucking with mostly unknown circumstances and thus requires
increased safety precautions in case of a survey. In principle both situations allow unique access
possibilities to the actual breakage geometry. This makes it possible to study, i) broken geometry and
height, ii) over- and underbreak and their effects on subsequently blasted rings, iii) interaction effects
between adjacent holes/rings, iv) drilling accuracy based on identified boreholes and v) mapping of
geological structures and their influence on the blast result. Frozen hang-up situations, in which a
temporarily stable arch was built up during gravity flow, are of special interest for the following two
reasons, a) the blasting was actually carried out in a semi-confined condition and b) they give direct
access to study gravity flow mechanisms. Some issues of interest are, i) geometry of the hang-up
(height, span width), ii) material type (ore, waste, caving masses), iii) fragmentation and void ratio, iv)
verification of the existence of gaps and v) conditions of the compacted material.
2.2. Measurement procedure on-site
The task was to design a system that would allow a quantitative study of openings and hang-ups along
the lines mentioned above considering the risks of sudden cavity or hang-up collapse. Eventually a
photogrammetry system mounted on a rotating camera head with the option to geo-reference the 3D
images was chosen. The system was mounted at the top of a 12 m long light-weight camera crane. For
the survey of openings the mobile crane was mounted on a car trailer. For the survey of hang-up
situations the camera crane was mounted with an adapter to the hammer of a scaling machine. To
enhance safety against a debris surge an additional pile of material was loaded in-front.
Figure 1: (Left) Survey at a drawpoint in SLC crosscut, hang-up situation.

Figure 2: (Right) Measurement procedure on-site, opening situation (not to scale).
An application of the equipment for the survey of a hang-up is shown in Figure 1 and the measurement
procedure in Figure 2. A single image pair yields a partial 3D model of the breakage face. However, a
complete 3D image of the entire breakage front requires merging of several overlapping models.
Therefore image pairs are gathered in vertical columns starting from the bottom. A set of mini prisms
arranged around the digital camera allow its positioning in the mines coordinate system from total
station measurements made when acquiring the first images (bold in Figure 2). Parallel mounted point
lasers provide light spots on the cavity surface giving the length scale in each image pair.
2.3. Photogrammetry system
The design and the construction of the measuring equipment were made together with the central
workshop of Lule University of Technology (LTU).
A rotating camera head with additional equipment (see Figure 3) is mounted at the end of the lightweight crane. A custom-calibrated digital camera with wide-angle conversion lens is moveable along a
2.2 m long beam. It is used to take the stereographic image pair. A laser distance measurement device
is used to measure the distance to the object, which allows the estimation of the required camera
spacing for the two images. The positioning of the camera carriage is controlled by a surveillance
camera and another camera with a fisheye lens gives a general overview on the cavity under
investigation. Two halogen lamps provide sufficient illumination for taking the images. A combination
of high precision and light weight was the goal when mounting the 4 laser modules and the 360 mini
prisms on the crane head, which are both used for the photogrammetry system described below. The
need to remotely control all components of the equipment from a safe distance (up to 100 m) required
special solutions in terms of signal converting.
3D images are generated by means of classical photogrammetry and extensions from computer vision
(Gaich, Ptsch, Moser & Schubert, 2009). The creation process starts with taking a pair of images of
the object of interest from slightly different positions. A dense set of corresponding image points is
derived automatically from these images utilizing techniques from multidimensional statistics. From
this set the relative orientation of the 2 camera positions is computed. Intersecting the rays of
corresponding image points from these positions yields a 3D point cloud. Finally these points are
triangulated to form a surface mesh, which is directly linked to the original image data. Prerequisites
are the calibration data of the imaging system to take the distortion of the lenses and alignment of the
optical axis into account. The resulting 3D image already represents the true shape of the surface but
lacks the correct scale, position and orientation in 3D space.
Figure 3: (Left) Measuring equipment mounted at a remote head.

Figure 4: (Right) Geometric arrangement of the laser modules.
However, the use of photogrammetry for the present task required the development of some
customized software extensions to the existing ShapeMetriX3D (SMX3D) software.
To achieve proper scaling a projection unit made up of 4 laser pointers was mounted on the camera
rig, see Figure 4. The laser modules are installed on a 5050 cm frame and the unit emits parallel
beams normal to the imaging plane of the camera positions. The scaling factor (s) for the 3D image
can then be derived by the ratio of the known distance between the beams and the distance between
the normal projection onto the y'z'-plane of 2 observed laser points, e.g. s = (P4-P5) / [nproject(P11P12)]. To support geo-referencing of the 3D image, reflecting prisms were arranged in an equilateral
triangle of 700 mm side length around the imaging system. By surveying the locations P1 to P3 of these
prisms in the object coordinate system (x,y,z) and their a priori known positions in the imaging
coordinate system (x',y',z') a transformation matrix M can be derived. M contains the translation and
rotation between these 2 systems, so all points from the correctly scaled 3D image can be transferred
to the global object coordinate system. Both the scaling and geo-referencing are implemented in the
software module LKAB Referencer. As the time spent surveying in a hang-up situation should be
minimized for safety reasons, it was decided that geo-referencing of the 3D image was not necessary.
With another custom-tailored extension of the SMX3D software, called the Interactive Referencer, a
3D image can be roughly aligned to a given coordinate system by interactively rotating the 3D image.
For instance, the 3D image is aligned based upon an identified half cast at the brow area and its
planned coordinates. In this way, the 3D images could be brought into the mine coordinate system and
the hang-up relative to adjacent SLC rings visualized.
Due to the limited field of view of the imaging system and the arrangement of the cameras it is not
possible to create a full spherical 3D image from only one pair of images and hence requires a merging
technique. This procedure takes structural information as well as topographic information into account
and does not need any reference points. Thereby corresponding points are determined either
interactively or by an automatic search algorithm in the Model Merger component of the SMX3D
software. To facilitate this, in the actual application, its capability had to be extended to support a
larger number of models, i.e. the merging of up to 15 instead of the default 6 individual 3D models.
Several tests were made to quantify the geometric accuracy of the measurement system. It was found
that the absolute error for geo-referencing and scaling with the described method was as small as 11
5 mm when compared with actual surveyed reference points. The relative error between a large
merged 3D image (740 m2, 12 individual 3D models) and another geo-referenced 3D model was
almost insignificant. At most it was in the range of 1-2 dm and can be related to the merging process.
2.4. Evaluation process

The 3D images provide an objective documentation of the local geology and allow structural mapping
to be performed on rock faces, which are essentially inaccessible for conventional mapping. From the
generated 3D images the visible rock mass structure was mapped using the component JMXAnalyst
within SMX3D, which enables to measure geometric data on the rock face. The mapped structures are
presented in stereographic projection plots, summarized for a certain drift and interpreted in
combination with the plotted orientation of the respective ring faces, see Figure 5. In addition, it was
emphasized to map all identifiable borehole objects (e.g. sockets, half casts, explosive remnants etc.).
The blasting results of the surveyed SLC rings were visualized and evaluated within the specifically
developed software RingCave (version 2012-03-21), see Figure 6. Programming was done by using
Delphi programming environment, which supports structured and object-orientated design. Data is
stored in a database management system (MS Access).
Figure 5: (Left) 3D image mapping at the roof inside an opening, KI-16-964-150-10.

Figure 6: (Right) Overview, RingCave software.
Starting point is the 3D image generated within the SMX3D software and imported as a CAD file into
RingCave. In addition, the following objects may be specifically identified within the mapping of the
3D image, separately imported into RingCave and linked to the respective model:
Point and line objects, model objects, i.e. borehole sockets, half casts etc. or
Polylines, 3D polygons, i.e. waste lenses, geological discontinuities, etc.
In a further step, the available drill, charge and blast data (rings) are imported from LKAB`s
database Giron for the actual surveyed SLC ring and its adjacent rings. Optionally, solely the ring
outlines from adjacent drifts could be imported (ring polygons).
In particular, RingCave assists with the calculation of various sections at a 3D model relative to a
defined plane or borehole which enables a detailed analysis of the blasting result. Information derived
from 3D mapping, i.e. model objects, is used in a versatile way to further compute:
Distances to a plane (over- and underbreak) relative to a best-fit plane
Static / dynamic explosive distribution (Kleine, Townson & Riihioja, 1993)
Breakage depths and spacing to adjacent holes
Minimum distances (effective burden) to the 3D model
Drilling deviations
Furthermore, quality measures, like the ratio area/minimum angle (m2/), can be calculated for all
triangles and hence areas of less accuracy within a 3D image might be located.
All computation results including relevant attributes can be visualized both in 2D and 3D.
3. SURVEY INSIDE OPENINGS

In the following, the capabilities of the 3D image acquisition system to evaluate the blasting results are
demonstrated with an example, in which a series of blasts were surveyed. A survey of SLC ring blasts
inside openings is possible as long as the blind open stope is not in contact with the caving masses
above and a maximum span width is not exceeded. In particular, the drifts with 2 openings along one
crosscut near the hanging wall are suitable for following up the blasting results of SLC rings inbetween. An overview of the selected test site (block 16, level 964, drift 150) is given by Figure 7. The
openings are marked in light green, the surveyed rings in dark green color and numbered. The 3D
view with bordered contour lines shows the stepwise increase in height and adaption of the stope
geometry to the local geometry of the orebody. The flattening of the ring inclination from 80 to 70
between ring nos. 8-13 reduces the high toe confinement to some extent. A completed breakage
though is unexpected for these boreholes and rather a loosening might be the actual effect. Even if ore
losses might sometimes be large for the opening, critical for the actual success of the SLC mine is the
second opening. Figure 8 shows the geometry and initiation timing for the smaller opening blast and
its subsequent individual rings. For the blasting of the opening LP (long period) and henceforward SP
(short period) detonators are used.
Figure 7: (Left) Isometric view of ring outlines, ring nos. 1-23 (right to left), KI-16-964-150.
Figure 8: (Right) Section at level 952.0 m, ring nos. 1-9, KI-16-964-150.
Evaluation of the blasting results
The structure mapping of the individual ring faces showed that there exist two main structure sets, see
Figure 5. Structure set 1 intersects the faces at an angle of 70 and dip angle 50 towards SSW and set
2 is roughly parallel with the face but dips in the opposite direction towards E with an angle of 70.
The surveyed faces consisted of massive, homogeneous and low/high phosphorous magnetite ore.
Absolute drilling deviations showed a systematic deviation towards positive x (in drift direction) and
a tendency for negative y. This problem became even clearer in ring no. 8 and further on when
another drill rig was used. Conceivable are a wrong set-up of the rig or ring planes and also alignment
errors in the front angle. In particular, this affected ring no. 8, as the planned minimum burden
distance of 1 m was not met anymore. The separate blasting of the ring no. 7 verifiably damaged
charges (hole nos. 3 and 4).
The blasting of the actual opening close to the hanging wall contact constitutes a highly confined
situation, while the blasting of the subsequent rings is more unconfined until contact with the caving
masses was created. Despite the consistent increase in area of 20 30 % for the subsequent ring blasts,
a reasonable breakage height was achieved, see Figure 9. In the opening, the boxhole and its adjacent
cut holes broke to its full lengths, see Figure 10. There were problems though with the actual widening
within the opening blast. As an after effect of the opening blast, the broken depths for the boreholes
within the subsequent rings are reduced towards the sides as well. In addition, extremely difficult
conditions exist for side holes within asymmetrically planned rings. An entirely unbroken side hole
could be found as well (ring 12, no 1) but the actual reason is difficult to identify. Borehole deviations
were not measured, nor could it be verified if the charge detonated. The extraordinarily long sleep time
for the charges (282 days) might be related to the breakage failure. Both the actual breakage depth and
broken area were nearly the same for all rings with 71 5 % and respectively 63 7 % relative to
their nominal values. This agrees also well with the experience on-site that any problems occurring in
an opening of a new drift are difficult to correct in the following blasts.
Figure 9: (Left) Isometric view of ring layouts with breakage heights, ring nos. 1-23, KI-16-964-150.
Figure 10: (Right) Geo-referenced and merged 3D model for ring no. 10, KI-16-964-150-10.
Although the actual hanging wall was not exposed in any of the 3D images, the parallel orientated
structure set 2 seemed to contribute to the rising in height. At the same time overbreak is caused by
this structure relative to subsequent rings, see Figure 11. In general the overbreak was limited to small
areas (< 10 m2) and less than 1 m, but could in exceptional cases (e.g. ring no 13) amount up to 2/3 of
the burden to the next ring. Though a considerably higher specific energy amount was present for the
blasting of the double ring (ring nos. 14 and 15), an increased brow damage or overbreak was not
verified. Severe overbreak could sometimes be observed to occur especially at the brow area despite
the uncharged collar lengths. This might be explained by the reflected tensile stress from the upper
surface of the drift but also be in connection with a unfavorable orientation of strata (set 1).
Figure 11: (Left) Over- and underbreak (m) for ring no. 12, distance to plane plot, KI-16-964-150-12.
Figure 12: (Right) Effective burden for ring no. 13, hole no. 5, KI-16-964-150-12.
Following several blasts in a row, an evaluation of the effective burden is possible. Assuming a burden
of 3 m realistic to break, ring no. 13 would break only about 40 % of its planned ring area (145 m2).
The blasting revealed that a much larger area (227 m2, based upon area with over- and underbreak
0.5 m) was broken as it could be expected. The effective burden and respective direction vector
towards the 3D model is exemplified for borehole no. 5 in Figure 12. With the post-blast information
of the actual breakage limit for this specific borehole the variations in effective burden and maximum
broken burden can be calculated. For this specific ring the actual broken burden varied between 2.2
2.6 m for the mid holes and was around 2.7 m for the side holes nos. 8 and 9. For the opposite side
hole no. 1 it was on average 3.9 m. Similarly, the maximum burden varied largely between 2.4 and 4.5
m and reflects the varying site geometry and feasible breakage angles. Again, the unconfined blast
situation and the structurally caused overbreak could have contributed to the result.
4. SURVEYS INSIDE HANG-UPS

Due to inadequate space for swelling of blasted ore, only the material closest to the blast plane is
sufficiently broken to be mobilized, and the material further away is heavily confined and temporarily
builds up an arch. A representative image and corresponding section of a so-called frozen hang-up
situation is plotted in Figure 13 and respectively Figure 14. A gap between the ring plane and the
blasted material was observed. The gap narrowed with height and its closing up was caused either by a
clogging of individual rock pieces or rather by insufficient breakage. Thereby, the burden was
separated and the material itself showed hardly any disaggregation at all and if there was mobilization
it occurred along holes with bridges in between. Problems while charging resulting in a reduced
amount of charged explosives (30 - 40 %) might have been the initializing factor for some of the
observed hang-ups. There could be several causes for the flow disturbance. However, the situation
itself might be viewed as a snap-shot to investigate gravity flow related to the previous rings for which
production figures (extraction, ore recovery) showed a disturbance-free extraction.
Figure 13: (Left) Hang-up observation with waste lenses, KI-12-792-135-8.

Figure 14: (Right) Section at 776.8 m and projection of waste lens 2, ring nos. 5-8, KI-12-792-135-8.
Among the 6 investigated hang-up situations several similarities were observed.
1. Dependent upon the size of the hang-up either one or two lenses with caving masses could be
assigned to the flow zones in the preceding rings.
2. A sharp contact existed between the blasted ore and the altered, fine-fragmented caving
masses (ochre coloured). Several mechanisms, such as self-breakage, abrasion and crushing
through the blast loads might be relevant for the fine fragmentation.
3. The type of material seemed to be purely waste rock and would reflect that mucking was
ceased as waste rock inflow became too large for the previous rings.
4. The identified flow zones were curved and ended before reaching the side holes. Relative to
the ring width at the specific height of observation their extension was 63 11 %. The
thickness was on average 1.4 0.3 m and hence about of the burden.
5. The actual height for the hang-up occurrence varied largely between 8 to 31 m. The ore in
between the waste lenses appeared to be fractured but entirely immobilized at the height of the
existent hang-up. A waste rock inflow in between rings penetrating these compact zones
seems to be nearly impossible during regular flow.
6. Swell was largest for the mid-part of the ring and decreased towards the sides. Referred to the
planned burden the maximum swell varied between 8 30 % but was on average 15 6 %.
Swell at the sides was insignificant.
Gravity flow for the SLC rings previous to the hang-up is assumed to have preferentially occurred
along the ring plane within the identified flow zones (shallow draw phenomenon; Power, 2004). No
major flow disturbances were recorded. Extraction rates varied as usual, 127 54 %, but lay on
average higher than normal with a corresponding ore recovery of 104 55 %. For a long period of
extraction also a very limited waste rock inflow was observed.
The extraction rates in connection with the occurrence of the hang-ups could vary a lot dependent on
the mobility of the material, the loading procedure and if material was discharged through the gap. In
particular a cyclic inflow of waste rock was observed and this is most likely connected with the
formation of the hang-up. Gravity flow progresses upwards, the cavity gets unstable, collapses and
both, temporarily frozen ore and waste rock from the flow lenses of previous rings enter the
drawpoint periodically.
Blasting, before it came to a hang-ups situation could be assumed to have taken place in a confined
situation. However, compared with the SLC rings studied inside openings no remarkable differences
for the blasting results in terms of over- and underbreak were found.
5. CONCLUDING REMARKS
The 3D image acquisition system has proven to be a valuable tool in increasing the understanding of
both SLC breakage and flow by revealing and quantifying hitherto inaccessible information. With
totally 30 surveyed SLC ring blasts inside openings a reasonable large data set is exists for a
comprehensive analysis of different influencing factors (Wimmer, Nordqvist, Ouchterlony, Selldn,
2012). Measurements are continued with the focus on standard production control and on testing new
layouts. Experiments with hole-by-hole initiation might be conducted inside openings to further study
interaction effects between holes. The potential of this new tool is large and new applications will
surely surface as people become aware of its existence.
With respect to the hang-up observations a pivotal question remained: Are the recent observations of
flow disruptions an inherent part of blasting SLC rings? An answer seems to be obtainable in the near
future with continued investigations and especially if the results are put in a broader context with other
potential new techniques, e.g. smart marker trials (Brunton, 2009), instrumented confined blasting
tests (Wimmer & Ouchterlony, 2011) etc.
ACKNOWLEDGEMENT
The Hjalmar Lundbohm Research Centre (HLRC), a research foundation at LTU and endowed by
LKAB is thanked for its financial support throughout the PhD project Improved breakage and flow in
sublevel caving. The colleagues from the central workshop and Complab at LTU are thanked for
numerous valuable discussions related to the development of the measurement equipment. Our
colleagues at the company 3G Software & Measurement in Graz, Austria are thanked for their
innovative support. Special thanks finally go to our technicians and the mine surveyors in the LKAB
Kiruna mine for all their help consistently provided in practical matters on-site.
REFERENCES
Brunton, I.A. (2009). The impact of blasting on sublevel caving flow behaviour and recovery
(Doctoral dissertation). University of Queensland, Brisbane, Australia.
Fjellborg, S. (2002). The value of measuring VOD in large scale sublevel caving. In X. Wang (Ed.),
7th International Symposium on Rock Fragmentation by Blasting (pp. 717-724). Bejing, China:
Gaich, A., Ptsch, M., Moser, P. & Schubert, W. (2009). How 3D images support bench face
profiling, blast planning and rock mass characterisation. In J.A. Sanchidrin (Ed.), 9th
International Symposium on Rock Fragmentation by Blasting (pp. 85-90). London, England:
CRC Press.
Hedstrm, O. (2000). Funktionskontroll av produktionssalvor i Kiirunavaara [Function control of
production blasts at the Kiruna mine] (Master thesis). Lule University of Technology, Lule,
Sweden.
Kleine, T., Townson, P. & Riihioja, K. (1993). Assessment and computer automated blast design. In J.
Elbrond & X. Tang (Eds.), 24th International Symposium on the Application of Computers and
Operations Research in the Mineral Industries (vol. 3, pp. 353-360). Montreal, Canada:
Canadian Institute of Mining, Metallurgy and Petroleum.
Olsson, M., Nyberg, U. & Fjellborg, S. (2009). Kontrollerad snderbrytning vid skivrassprngning
inledande frsk [Controlled fragmentation in sublevel caving first tests] (Swebrec Report
2009:2). Lule, Sweden: Lule University of Technology.
Power, G. (2004). Modeling granular flow in caving mines: large scale physical modeling and full
scale experiments. (Doctoral dissertation). University of Queensland, Brisbane, Australia.
RingCave (Version 2012-03-21) [Computer software]. Kiruna, Sweden: LKAB (Nordqvist, A.).
Rustan, A. (1993). Minimum distance between charged boreholes for safe detonation. In H.P.
Rossmanith (Ed.), 4th International Symposium on Rock Fragmentation by Blasting (pp. 127135). Rotterdam, Netherlands: Balkema.
Selldn, H. & Pierce, M. (2004). PFC3D modeling of flow behaviour in sublevel caving. In A.
Karzulovic & M.A. Alafaro (Eds.), 4th International Conference and Exhibition on Mass
Mining (pp. 201-214). Santiago, Chile: Instituto de Ingenieros de Chile.
ShapeMetriX3D (version 3.0.2) [computer software]. Graz, Austria: 3G Software & Measurement.
Wimmer, M., Ouchterlony, F., Moser, P., Nordqvist, A. & Lenz, G. (2009). Referenced 3D images
from inside cavities and behind rings in sublevel caving. In J.A. Sanchidrin (Ed.), 9th
International Symposium on Rock Fragmentation by Blasting (pp. 91-100). London, England:
CRC Press.
Wimmer, M. & Ouchterlony, F. (2011). Study of burden movement in confined drift wall blasting tests
in block 12, 691 m level, Kiruna mine (Swebrec Report 2011:U1). Lule, Sweden: Lule
University of Technology.
Wimmer, M., Nordqvist, A., Ouchterlony, F. & Selldn, H. (2012). 3D mapping of sublevel caving
(SLC) rings and flow disturbances in the LKAB Kiruna mine (Swebrec Report 2012:U1).
Lule, Sweden: Lule University of Technology.
Zhang, Z.X. (2005). Increasing ore extraction by changing detonator positions in LKAB Malmberget
mine. Fragblast International Journal of Blasting and Fragmentation, 9(1), 9-46.
PAPER D
Burden movement in confined drift wall blasting tests studies at the

LKAB Kiruna SLC mine
Matthias Wimmer, Anders Nordqvist, Finn Ouchterlony, Ulf Nyberg

and Jason Furtney.
Accepted for presentation at:
10th International Symposium on Rock Fragmentation by Blasting 2012.
New Delhi, India.
Burden movement in confined drift wall blasting tests studied at the

LKAB Kiruna SLC mine
M. Wimmer & A. Nordqvist
LKAB, Kiruna, Sweden
F. Ouchterlony & U. Nyberg

Swebrec at Lule University of Technology, Lule, Sweden
J. Furtney
Itasca Consulting Group, Minneapolis, USA
ABSTRACT: Blasting in SLC (sublevel caving) takes place under varying confinement. Blasted material
swells while the caved material compacts, and also, to a lesser extent, fills parts of the void volume of the
production drift. Several analytical and empirical models have been developed in the past. However, understanding of the interaction of confined blasting conditions, SLC blast design and rock mass characteristics on
rock breaking performance is rudimentary. Instrumentation of the blasted burden with various sensors and
study of the dynamics of burden movement against confinement is therefore important. Such measurements
are very scarce and thus different systems have been tested both in laboratory and in the field. Main focus was
thereby the development of measuring equipment that could be scaled up to full-scale SLC blasting and installed behind the rings. In the field tests, blastholes were drilled in a pillar, parallel to a cross cut drift in the
LKAB Kiruna SLC mine. The burden constraint was achieved either by filling the drift of 7.0 x 5.2 m (width
x height) with rock masses or reinforcing the drift wall. The tests were instrumented with different redundant
sensors, which were installed in holes drilled from a parallel cross cut. The most promising one was a concept, which combines the initial movement recorded by an accelerometer designed to minimize zero-shift
with the final displacement recorded by a fibre photoelectric sensor (fibre-optic zebra gauge). For the filled
drift wall blasting test the maximum velocity varied between 16 - 32 m/s with a compaction in the range of 4 5 %. A gap was verified to exist between the intact pillar and the blasted material. By comparison, the velocity for an unconfined situation yielded a considerably increased velocity (42 - 47 m/s). In addition, Blo-Up,
a component of the Hybrid Stress Blast Model (HSBM), was used to model the blasting results. The objective
was to show that reasonable predictions of fragmentation and burden movement under confined conditions
can be made. An unconfined reference experiment was used to calibrate Blo-Up and forward predictions of
the confined case were made. In both cases, Blo-Up accurately reproduced the burden velocity and displacement observed in the experiments. The suggested measurement concept could be used for further measurements of confined burden movement in filled drift wall blasting tests and under controlled situations. As mentioned the final objective is instrumentation of the burden in SLC production rings.
1 INTRODUCTION
Blasting in sublevel caving (SLC) has been identified throughout the literature to have a significant
impact upon material flow characteristics and therefore on the overall performance of the SLC method.
Brunton (2009) gives an excellent literature review.
Blasting in SLC takes place in a situation with varying confinement. Blasted material swells while the
caved material compacts, and also, to a lesser extent,
fills parts of the void volume of the production drift.
Its dynamic behaviour is dependent upon several parameters. Given factors are on the one hand material
properties of both the blasted and compressible ma-
terial. Controllable factors on the other side include

the ring layout and blast design.
A number of analytical models and physical experiments were deployed in the past to either quantify the fragmentation or understand the dynamic
process of blasting into a compressible material.
With a limited number of small-scale tests primarily
the final swell and compaction were investigated
(Belen`kii 1969, Rustan 1970, Cullum 1974, Volchenko 1977, Kirpichenko 1982, Zhang 2004 and
Johansson 2011). Measurements in full-scale were
mentioned by Kirpichenko (1982). However, no details of the measurement method were given. Newman et al. (2008) measured the final swell by physical means when blasting an ore slice against caved
material in a drift at Kiruna mine.
Measurements of the burden dynamics of blasting

into a compressible material are sparse. Small-scale
tests were carried out by Rustan (1970), Volchenko
1977, Kirpichenko 1982 and more recently by Johansson (2011) and Petropoulos (2011).
2 BURDEN INSTRUMENTATION OF SLC
BLAST RINGS
Understanding the interaction of confined blasting
conditions, blast design and rock mass characteristics on rock breaking performance is rudimentary.
Therefore the idea arose to instrument the burden
with various sensors to study the effect of confinement on dynamics of burden movement and breakage in full-scale. The conceptual layout for a future
test program is shown in Figure 1.
stallation of the measurement systems from the

backside of the burden.
Studies of burden movement have been carried
out previously with various systems ranging from
very simple to advanced ones but with installation
in-front of the blast. This study requires measurement of burden movement from the reverse side. All
measuring equipment must therefore be installed inside a borehole or alternatively at the borehole collar. In principle, a variety of measurement systems
and components available today could probably be
used. However, their applicability, reliability and
accuracy remain to be assessed for this specific task.
The fact, that the measurement problem is new
and the methods used are untested, are the reason
that controllable, simplified drift wall blasting experiments were carried out. Main focus was the development of measuring equipment that could be
scaled up to future full-scale SLC blasting and installed behind the rings. Due to different conditions
these tests to not intend to model full-scale blasts.
3 DRIFT WALL BLASTING TESTS
Figure 1. Future burden instrumentation of SLC blast rings.
The tests focus on investigating the functioning

of the blast in the upper part of the rings. These tests
comprise the drilling of boreholes from the longitudinal drifts into the burden area of the blast rings
closest to the footwall to study the following:
Dynamics of burden movement
Swelling of the blasted burden and compaction of
caving debris
Existence of voids prior to and after blasting
Breakage of burden
Verification of over- and underbreak
Quantification of fragmentation and mobilization
Breakage of burden was examined by static and
dynamic time domain reflectometry (TDR) of coaxial cables grouted in boreholes within the burden
(Wimmer & Ouchterlony 2008, Wimmer et al.
2011a). The present paper relates though to the
study of burden movement.
The implementation of the proposed measurement program in production mining is subject to
some complicating factors. The measurement holes
have to pass at least 30 - 40 m of footwall before
they reach the nearest rings. Accurate surveying of
the ring holes, careful planning and high-precision
drilling are necessary for a successful future field
test. Another factor is the correct placement and in-
Adjacent pillars in the completed production block

12 on the 691 m level were chosen for the tests
(Wimmer et al. 2011a). Blasting at two sites (test 2,
drift 132-130 and 4, drift 127-130) was carried out
against artificially filled drifts with a mixture of broken ore and waste loaded by a LHD machine. This
way a confined blasting situation should be achieved
which also allowed a later inspection of the blast results.
Two blastholes with diameter 115 mm were
drilled for each test parallel with the drift. The tests
were extensively instrumented with different sensors
which were installed in holes that where drilled from
the drift behind into the burden area. Table 1 summarizes the blast design.
TABLE 1. Blast design, test 2 and 4 with blastholes B2 B5.
Test 2
Test 4
Parameters
Unit
B2
B3
B4
B5
Diameter
mm
115
115
Hole length
m
13.2
11.1
13.0 14.0
Charge concentration kg/m 12.0
12.4
11.4 11.6
Stemming length
m
4.0
3.0
2.5
2.5
Effective burden
m
0.9
0.8
1.5
1.7
Effective spacing
m
1.6
2.0
Delay time
ms
0
0
The blast sites were well documented in terms of

a face survey and structure mapping before and after
blasting with the photogrammetry system ShapeMetriX3D (3G Software & Measurement). All
boreholes were surveyed with a gyro based deviation probe Flexit SmartTool (Flexit). As an example, blast site of test 4 is shown in Figure 2.
The boreholes were charged with the emulsion

explosive KR0500 (Kimulux Repumpable 0500),
which is sensitized by glass micro-balloons and contains aluminium. The initiation system consisted of
primers SLP-50 (Sliding Primer-50) composed of a
cap sensitive emulsion packed in a hollow plastic
casing. A 25 g primer made of PETN, wax and oil is
placed in the central hole. Programmable electronic
detonators i-konTM (Orica) were used for initiation.
The boreholes were stemmed with sand/gravel.
a redundant instrumentation with various different

sensors was made:
Accelerometers
Piezoelectric accelerometer (PE; Endevco
7255A-01) and piezoresistive accelerometer
(PR; PCB 35011220KG)
Continuous position sensors
Draw wire sensor (DRAW; Firstmark Controls
161-2145H-BSS)
Magnetostrictive position sensor (MAGNET;
Santest GYcRS with custom-made magnet)
Incremental relative distance sensors
Detection of a hole matrix with laser light barrier
(LASER; Sensor Instruments D-LAS-HS )
Detection of cable breakage as short circuits
(SPEAR; Swebrec`s coaxial cable spear)
Detection of coding with fibre photoelectric sensor (FIBRE; fibre-optic zebra gauge)
Figure 2. Measurement points (cubes) within burden for the

confined test 4, (1) top-view and (2) isometric view.
The velocity of detonation (VoD) and vibrations

at adjacent pillars were recorded to document detonation of the blastholes. For documentation purposes
a camera with high speed filming capacity of 300 fps
was used (Casio EX-F1). It helped to verify the
functionality of sensor systems installed behind the
burden and that the confinement was properly built
up. Special attention was given to further document
the situation after blasting in terms damages to the
sensor systems and to determine final displacement.
3.1 Measurement systems
The selection of a proper measurement method to
measure the dynamics of burden movement depends
upon key factors such as:
Measurement range of minimum 500 mm with
high sampling frequency
Simple installation, i.e. inside a borehole or at the
borehole collar
Robustness in a mining environment i.e. shock,
vibration, dust, moisture, etc.
Preferably comprehensive and highly reliable,
unbiased results
Reasonable priced, either single-use or re-usable
Measurement of the confined burden movement
in full-scale has been unresolved: the methods used
have therefore not previously been tested.
Several types of displacement transducers are described in the technical literature but their suitability
to reliably measure burden movement is rather uncertain. Because of this uncertainty and the fact that
only a limited number of experiments could be made
3.1.1 Instrumentation
Test 2 had three boreholes (M3-M5 with 76
mm) and test 4 six holes (M10-M16 with 76 mm
and 115 mm; Fig. 2) respectively which were instrumented to measure burden movement. In all
measurement holes an anchor was gouted mid-way
in the burden containing axially mounted 1-axis accelerometers. For signal transmission, steel cables
each with three embedded electrical cables have
been strain-relieved by a conical termination filled
with polyester resin mounted within the anchor. A
cable magazine was contained in a casing either behind the burden or at the borehole collar.
3.1.2 Installation behind the burden
The incremental relative distance sensors, coaxial
cable spear and fibre-optic zebra gauge, could be installed closely behind the burden. Thereby, the
movement of the anchor was transferred by a
threaded hollow pulling tube which also housed the
steel cables and was guided in a strong protection
casing. A bolting of the latter to a mounting plate at
the borehole collar in the neighbouring drift should
serve to keep the casing fixed in the holes so that a
true relative movement is measured. For the coaxial
cable spear the pulling tube had a cutting knife made
of hardened steel with an annular gap at the end to
gather the cut cable pieces. Coaxial cables were
circumferentially mounted within the casing and fed
through holes. The spacing of planned cable cuts
was successively increased, from fine to coarse resolution at the end of the movement. For the fibreoptic zebra gauge a coded, black-and-white, adhesive foil with strip widths either 5 or 10 mm was attached to the pulling tube. Movement was detected
by 2 fibre photoelectric sensors mounted inside the
3.1.3 Installation at the borehole collar

Due to their required space, some of the sensors, i.e.
continuous position sensors (draw-wire and magnetostrictive sensor) and the laser-light-barrier were
mounted on a stand at the collar.
The movement of the anchor was transferred by
the steel cable passing through a flexible teflon
coated steel hose which gives both protection and a
friction reduction. The sensors were mounted
asymmetrically on the stands with a cross-beam to
the element housing the cable termination within a
guide bar. The cable assembly was pre-tensioned by
clamping the cross-beam. The electrical cables from
the accelerometers leave the guided element at the
rear side to enter a cable magazine from which they
could be rapidly extracted. The spring system of the
draw-wire sensor is rated for a maximum cable acceleration of 1350 g which should be sufficient to
retract and to tension the cable assembly.
There was no linear guiding for the anchor with inbuilt 1-axis accelerometer. As there was no resistance of masses either most likely the anchor rotated
during its trajectory at an early stage so other components increasingly affected the reading. For the
coaxial cable spear, it was found that the spear
moved unintentionally together with a protection
casing because concrete penetrated during installation in this casing. The thick-walled casing moved
about 1.5 m from the wall and got stuck as it was
considerably bent. It was verified, that the actual
measuring tube with installed cables did not move.
However, a large frictional effect of the tight-fitting
casing within the borehole wall can be assumed
which affected the measurement result.
60
LASER (M3)
50
SPEAR (M4)
DRAW
(M5)
40
velocity [m/s]
casing. This would from the sequence of the arriving

signals reveal the direction of movement.
MAGNET (M5)
DRAW (M3)
30
20
10
3.2 Unconfined situation (test 2)

20
25
30
35
40
45
50
55
60
65
-10
time [ms]
900
800
700
displacement [mm]
Filling of the drift for test 2 was not complete, neither in height nor in length, which led to a relatively
unconfined blasting situation. This resulted in excessive material movement, upwards and outwards
from the drift and left a wide trench in front of the
blast site. It was caused by a too conservative assessment of the required filling degree, a small burden and a partly nonexistent opposite drift wall.
The collected data is a result of a combination of
factors: the blasting situation, the sensors used and
the construction and mounting of the systems. As a
consequence the time delays and the characteristics
of the curves differ. The velocity and displacement
signals are summarized in Figure 3.
With the blastholes detonating almost simultaneously at 20 ms, the anchor bearing the PE accelerometer in hole M3 first moved after 1.9 ms. The instrumentation behind the burden in hole M4 (coaxial
cable spear) came 6.5 ms after detonation and 1 ms
later the instrumentation at the borehole collar of
hole M3 and M5 started to move. The comparatively
long reaction time for the coaxial cable spear could
be based upon its incremental operating mode, i.e.
an initial movement is required until the first cable is
cut, and a comparatively high weight.
The velocity signals from the PE accelerometer
and the coaxial cable spear showed clearly decay in
velocity, but the other gauges did not. From the
blasting result with massive cast of material an early
retardation effect and rigorous physical stop effect
can be excluded. Hence an effect of the insufficient
filling on the initial measured burden movement and
decrease in velocity is doubtful but rather possible
error sources for the measurement should be found.
PE (M3)
DRAW (M5)
DRAW (M3)
600
500
400
300
PE (M3)
LASER (M3)
200
SPEAR (M4)
100
MAGNET (M5)
0
20
25
30
35
40
45
50
55
60
65
time [ms]
Figure 3. Velocity and displacement, test 2.
By contrast, the gauges linked to the cable assembly (draw-wire, magnetostrictive sensor and laser-light barrier) did not show any decrease in velocity for the designed measurement range of 700 mm.
Good agreement exists between the different measurement methods installed for the same borehole.
For hole M3 the mean velocity is around 42 m/s and
for hole M5 around 45 m/s. There is a cyclic variation of the velocity for both draw-wire sensors as
they reach their maximum level. This might be related to the design of the potentiometer and fast
withdrawal of the steel cable. With the blast result in
mind it is considered possible that an essential retardation for the initial movement did not exist. According to that it is difficult to assess if the spring
system of the draw-wire could on principle counteract the high inertia of the mass.
3.3 Confined situation (test 4)

A much better filling was achieved for test 4 and a
near confined blasting situation was achieved. A
movement on the surface of the muckpile was neither verified during blasting (high-speed filming)
nor from post-blast inspections. At about 100 ms after detonation, some gases started to leak through
the pile. Remote inspections through the instrument
holes with a long-range borescope and observations
during mucking showed that there existed a void between the pillar and the blasted material (Fig. 4).
very beginning of the movement. These mechanisms

disturbed all systems that were based upon a relative
motion between the outer casing, with built-in
measurement system, and the pulling tube as the
moving part. However, this problem can in the future constructively be solved as solely a bucklingresistant inner pulling tube reaches into the burden
and the actual measurement system within the casing
is placed well behind the burden.
Reliable measurement results for test 4 were obtained with the accelerometers and draw-wire sensors. The velocity and displacement signals are
summarized in Figure 5.
35
30
PE (M13)
velocity [m/s]
25
PE (M15)
20
15
PE (M16)
10
5
0
Figure 4. Blast result, test 4.
15
25
35
45
55
65
75
time [ms]
1300
1100
PE (M13)
PE (M15)
900
displacement [mm]
The blasted material moved about 1.2 1.6 m in

the upper part of the drift section. The compaction is
in the order of 4 5 % based on the void in Figure 4.
It is likely that the value of compaction could be adjusted upwards as there was partial collapse of the
gap before or during the mucking operation. Definitely, the degree of filling and compaction, the simultaneous initiation and the stress situation within
the present test differed from the SLC situation.
Voids in operational SLC mining were identified to
be related to disturbed flow, i.e. shallow-draw phenomenon (Selldn & Pierce 2004; Power 2004;
Wimmer et al. 2012). Nevertheless, it raises the
question if a bulking or even voids closest to the ring
plane possibly exist also directly after blasting.
During mucking it was also observed that the
burden has moved more or less as one unit during
blasting. It could be characterized as very coarsely
fragmented, and was difficult to muck, with some
boulders that were as large as 1.5 - 2 m.
For test 4, all anchors with built-in accelerometers were linearly guided by a pulling tube within a
casing. The post-blast inspection revealed that the
anchors were well fastened within the concrete/borehole and moved with the masses towards
the debris. Problematic, however, was that the casings started to move together with the accelerating
anchors and the surrounding material. The casing
became bent likely at the very beginning due to the
shock waves arriving and causing the inner pulling
tube to catch. A friction mechanism between casing
and the throw of material is also conceivable. The
desired relative movement between the inner pulling
tube and outer casing, if any, only occurred at the
700
PE (M16)
500
DRAW (M13)
300
DRAW (M14)
100
-100 15
25
35
45
55
65
75
time [ms]
Figure 5. Velocity and displacement, test 4.
The burden movement during test 4, as measured

with accelerometers, indicated a rapid one- or twostage acceleration phase to a maximum velocity of
16 - 32 m/s, followed by a retardation phase which
had a similar character for all gauges. In particular,
the curves of the PE accelerometers for the very
closely spaced holes M15 and M16 also exhibited
very similar curve characteristics for the initial
movement. The final stop and its characteristics
were not captured as the maximum displacement of
the burden was underestimated. The total displacement measured corresponds to the maximum available cable length of the accelerometer system. It is
therefore reasonable to assume that the results obtained up to this point are reliable.
As the entire systems moved violently, i.e. anchor
with pulling tube and casing, the bolting to the
neighbouring drift did not withstand the forces.
Damages occurred to the cabling within the borehole
and at the collar to the mounting plates (bent or bro-
ken in half). This way, the draw-wire sensors were

also badly affected as they did not record more than
about 300 mm of displacement. The draw-wire results did not capture the initial acceleration phase as
the signals after a delay of 3 - 4 ms rise within 0.5
0.6 ms to the accelerometer levels. The initial jerks
and the signal delay of the draw-wire are probably
wave propagation phenomena or the associated local
acceleration may have exceeded the manufacturer`s
limits. The displacement records between 20 - 30 ms
are quite reasonable though compared with the integrated PE accelerometer signals.
ditions of test 2 and the confined conditions of test

4, Blo-Up reproduces the burden velocity and displacement observed in the experiment. In the unconfined case the burden accelerates to a velocity of
about 45 m/s over a period of approximately 1 ms.
In the confined case, the maximum velocity reached
at the same point in the burden is about 30 m/s. The
expansion energy of the explosive product gasses is
consumed in overcoming interlocking in the burden
and in accelerating the burden mass. In the confined
case, the expansion energy is also consumed in
compacting the caved material, resulting in smaller
burden velocities and displacements.
60
To complement the understanding gained from the

burden movement measurements a numerical modeling investigation was also undertaken using the
HSBM Blo-Up software. The objectives of the
study were twofold: (i) to gain additional insight into
the process of blasting in confined conditions and
(ii) as a verification study of Blo-Up.
The Hybrid Stress Blasting Model (HSBM) is a
high level blast modeling research tool which provides results that can be used implicitly for practical
blast design. The code is being developed through
an international collaborative research project
funded by a consortium of companies which is comprised of explosive, equipment suppliers and major
mining companies. A key component of the HSBM
is the numerical code and user interface designated
as Blo-Up. The Blo-Up software uses a unique combination of continuous and discontinuous numerical
methods to represent the key processes occurring in
non-ideal detonation, rock fracturing and muckpile
formation. The model is described in detail by Furtney et al. 2009.
The flex-boundary condition was developed to
model the behaviour at the boundary between the intact burden and the caved material in an SLC ring.
This condition exists at the boundary of the model
so the caved material is not explicitly represented.
The broken rock material in the tests and in real SLC
rings are expected to show compaction behaviour in
which the modulus increases with displacement and
unloading occurs along a stiffer modulus. In order to
incorporate this compaction behaviour into the
model a non-linear hysteretic spring is parameterized to describe the waste material.
Tests 2 and 4 described herein are analyzed using
Blo-Up. Although test 2 was relatively unconfined it
provides excellent data on the burden movement.
This allows for a calibration of the material properties. The objective is to build confidence in the
model by exploring the simpler unconfined case before moving on to the less constrained and more
complex case of the confined blasting.
Figure 6 gives a summary of the results of the
numerical investigation. In both the unconfined con-
50
velocity [m/s]
3.4 Numerical modeling
Test 2: Unconfined situation
40
30
20
Data
Blo-Up
10
0
25
30
35
40
45
50
time [ms]
Test 4: Confined situation
40
Data
Blo-Up
35
velocity [m/s]
30
25
20
15
10
5
0
15
25
35
45
55
65
75
time [ms]
Figure 6. Comparison of burden velocity measurements and

Blo-Up model predictions.
Figure 7 is a cross section of the confined model

showing the boundary swell, fracturing and back
break. As in the field measurements a gap centered
on the boreholes is observed.
Figure 7. Model predictions of burden swell. Contours are for

displacements of 0 to 1.2 m, 60 ms after detonation.
4 DEVELOPMENT OF A MEASUREMENT
SYSTEM FOR BURDEN MOVEMENT
Based upon the experiences of the drift wall blasting
blasts and parallel on-going laboratory tests, an incremental relative measurement system installed
closely behind the burden was put into use. The
main idea is to have a system which captures both
the fast acceleration phase and the slower movement
until rest. The measurement system consists of a hollow pulling tube with an anchor incorporating an accelerometer suited for long duration transients,
which is grouted inside the burden. Behind the burden, a displacement measuring system with fibreoptic sensors is mounted in a shock-absorbent casing, which is kept in position. A coded tape is fastened to the pulling tube and detected by two fibre
photoelectric sensors. This allows detection of a
change in the direction of movement. Figure 8 illustrates the essential parts of the measurement system
as applied in the last field test.
Figure 8. Measurement system installed behind the burden.
tude. Measures to reduce the effects of zero-shift are

the design of the accelerometer, i.e. avoidance of
overstress of sensing element and sensor part
movements as well as care in the installation and
set-up, i.e. avoidance of frictionally generated cable
noise (triboelectric effect) and base strain.
For example, the design of the PE accelerometer,
7255A-01 (Endevco) is an attempt to eliminate the
zero-shift problem. It incorporates a built-in mechanical filter for sensor isolation and a matched
electronic low-pass filter in order to remove unwanted residual ringing superimposed on the accelerometer output signals. Several independent test results have shown that the design virtually eliminated
the difficulty of zero-shift (Chu 1988, Yiannakopoulos & Van der Schaaf 1998).
4.2 Fibre-optic zebra gauge
The usage of incremental relative distance sensors to
determine the time for initial face movement and the
rate of burden movement is not novel within the
field of blasting. Different systems which yield timedistance data as detection of cable breakage were
successfully applied by Noren (1956) and Olsson et
al. (2009).
The fibre-optic zebra gauge might be regarded as
a direct improvement. A fibre photoelectric sensor
detects the movement of the object by the detection
of a coding printed on adhesive foil. Detection occurs when the light beam, incident on the target, is
reflected by the target and received. The optical
plastic-fibre consists of a transmitter (core, 1 mm)
surrounded by 16 tiny ( 0.26 mm) receiver fibres
embedded in a flexible end tip (Sick LL3-DB02).
4.1 Accelerometers
Both, piezoelectric (PE) and piezoresistive (PR) accelerometers are widely used shock accelerometers.
However, their applicability for integration of the
signal to velocity and displacement is often troublesome and related to specific constructionconditioned problems (Chu 1992).
Despite that PR accelerometers were improved
considerably in their usable bandwidth (> 500 kHz)
and rigidity, the sensor can still be set into resonance
as the gauge mechanism is practically undamped.
The result of this failure is complete loss of data and
eventually permanent gauge damage. On the other
side, PE accelerometers are more robust under the
same conditions but have other deficiencies:
Inability to hold their charge output during a
steady input (DC response).
Zero-shift, i.e. failure of a sensors output returning to zero after a dynamic event.
The phenomenon of zero-shift arises during an
explosively generated shock (pyroshock), as a unidirectional shift of unpredictable polarity and ampli-
4.3 Laboratory tests

Laboratory tests with the aim to evaluate the performance of different accelerometers and compare
their integration results with incremental relative
displacement sensors (coaxial cable spear, fibreoptic system) were carried out.
The test set-up was intentionally kept simple and
robust with regard to its latter application in the mining environment. A probe that housed the accelerometer was guided within transparent PVC tubing
( 40 mm) mounted in a lathe.
The probe was pushed forward by a tube guiding
the cables and inserted through the hollow work
spindle. In this way a constant velocity in the range
of 3 6 m/s could be achieved for a certain distance.
The measurement range was varied between 150 and
750 mm and the characteristics of stoppage could be
differently simulated. The integrated accelerometer
signals were compared to velocity and displacement
measured at the probe by the incremental relative
distance sensors.
FIBRE
PE
PR
velocity [m/s]
800
6
PE (v)
FIBRE (v)
PE (d)
FIBRE (d)
700
4
600
velocity v [m/s]
tion effect as the confining masses get compacted.

These may also react as a spring system. For the latter reason the used stop in the form of hard plastic
was replaced by a compression spring. A pronounced bounce effect and damping which reached a
peak indicating the stoppage of the probe could be
found, see Figure 10. Further experiencing with
stopping systems of different rigidity indicated that a
harder stop was easier to identify as a clear stop.
500
400
0
1.80
1.85
1.90
1.95
2.00
2.05
300
2.10
displacement d [mm]
The incipient use of the cable breakage method

proved to be difficult as the cutting of coaxial cables
induced ripples in the acceleration signal. Consequently, a comparison with conventionally measured
velocity and displacement was largely incoherent.
By contrast, the non contact fibre-optic zebra
gauge did not exert any influence on the acceleration
signal. It also allowed detection of the direction of
movement with two systems mounted in a sequence
and gave an increased resolution to identify the acceleration and retardation phases.
The results from the PE unit compared with the
PR unit looked very much alike, see Figure 9.
200
-2
100
-4
0
2.04
0
time t [s]
2.06
2.08
2.10
2.12
2.14
-2
Figure 10. Velocity (v) and displacement (d), PE accelerometer

compared with fibre-optic gauge, suspended stop at 400 mm.
time [s]
200
displacement [mm]
160
120
80
40
FIBRE
0
2.04
2.06
2.08
2.10
PE
2.12
PR
2.14
time [s]
Figure 9. Velocity and displacement, PE and PR accelerometer

compared with fibre-optic gauge, abrupt stop at 150 mm.
The integrated data coincided well with the velocity and displacement as measured by the fibre optics. The actual stop was recognizable as a transient
spike in the velocity and a kink in displacement just
before 2.10 s. For the time window of interest (50 150 ms), zero-shift did not prove significant. Continuing though to integrate the signal of the PE unit
after the stop revealed later problems of zero-shift.
This appeared as a ramp in velocity and a parabola
in displacement. This problem was also directly noticeable when comparing the mean offset before (1)
to after the shock event (2). The mean 2/1 is 0.72 for
the PE and 1.01 respectively for the PR unit (18
measurements). Even so, the PR unit did not perform better after shock.
The dynamics from a burden movement against
confinement in terms of rock masses would presumably look different with a) a much higher acceleration and initial velocity and b) a gradual retarda-
The use of PR accelerometers required some special care. Attention was paid to the cable selection in
terms of special treatment to reduce triboelectric
noise. In particular, for the PR unit it was essential
to use low-pass filtering (10 kHz).
By contrast, the PE accelerometer type is reliable
due to a secure mounting arrangement, ease of use,
robustness and noise free, consistent results in the
laboratory tests. The excellent performance of the
PE unit is due to an effectively working in-built mechanical filter in combination with an electronic
low-pass filter. Comparing displacements and velocities derived by the PE unit with the fibre-optic
measurements pointed to a reasonable agreement.
On average, the velocity was increased by 7 37 %
(165 meas.) and the displacement decreased by 9
11 % (378 meas.) as the actual values.
4.4 Field tests
For further development of the described fibre-optic
system, collaboration with the project Rock support
subjected to dynamic loading (Shirzadegan 2011)
was valuable. In the test series the effectiveness of
various ground support systems under dynamic loading conditions are tested. Rock burst loading is
simulated using explosives in holes drilled behind
differently reinforced drift walls. The reinforcement
was of integrated character, i.e. rock bolts, shotcrete
and wire mesh. The equivalence of rock bursts and
blast-generated dynamic load is doubtful. However,
the tests with an artificially generated initial movement and a possible rebound effect were ideal for the
development of a reliable technique to measure face
2.0
400
1.8
350
1.6
300
velocity v [m/s]
1.4
250
1.2
200
1.0
0.8
150
FIBRE (v)
d = 1.24t + 3.68
R2 = 1.00
FIBRE (d)
0.6
100
LASER1 (d)
displacement d [mm]
Then, as a result of the blasthole located nearby and

penetrating gases, a blow-out of the system occurred. The blow-out forced the fibre-sensing backwards and accelerated the apparent outward movement of the anchor. A curve splice of the initial
integrated accelerometer signal and the later fibreoptic signal would yield the whole time history of
the movement, see Figure 13. As the fibre-optic
gauge measured a relative movement between the
anchor and the casing, two relevant mounting matters need to be solved in the future. The starting position needs to be well defined and casing movement
during measurement prevented. Important constructive measures are a grouted casing as well as an integrated stop ring on the pulling tube (Fig. 8).
50
0.4
d = 1.70t - 7.31
R2 = 1.00
0.2
0.0
0
100
200
300
time t [ms]
-50
500
400
Figure 11. Test 6, velocity (v) and displacement (d), laser distance sensor (Sick OD2-250W150) and fibre-optic gauge.
100
16
PE (v)
FIBRE (v)
PE (d)
FIBRE (d)
90
80
70
60
50
40
0
0
10
20
30
40
50
60
70
80
90
30
displacement d [mm]
velocity v [m/s]
12
20
-4
10
-8
0
time t [ms]
Figure 12. Test 5, velocity (v) and displacement (d), PE accelerometer and fibre-optic gauge.
50
600
PE (v)
FIBRE (v)
PE (d)
FIBRE (d)
400
30
300
20
blow-out starts
200
10
100
0
displacement d [mm]
500
40
velocity v [m/s]
velocity and displacement in a comparatively simple

and controllable way. Two tests were carried out at
block 9, on the 741 m level at the Kiruna mine
(Wimmer et al. 2011b). Both were single-hole shots
with instrumentation either in front of the burden
(test 5, drift 93, right side), or in front and behind the
burden (test 6, drift 95 98).
In test 5, a movement of at most about 0.1 m with a
subsequent gentle rebound was recorded. In test 6,
the limit case of breakage was achieved as the burden for the section of lower charge concentration
moved slowly as a single block for about 0.5 1 m.
The section with higher charge was unexpectedly
broken but showed rather coarse fragmentation.
The fibre-optic zebra gauge worked reliably and
its output was comparable with other measuring
techniques when installed in front of the burden. Velocity and displacement derived from it were compared to a laser distance sensor (Fig. 11).
The field tests showed also that the tested accelerometers were not suited as a stand-alone technique
to yield information on the complete burden movement but are rather complementary. A splicing of the
initial accelerometer data with the results of the fibre-optic gauge might be the best option.
Figure 12 compares velocity and displacement
from the integrated accelerometer signal with the fibre-optic gauge. The PE accelerometer picked up the
initial wall velocity as well as its maximum displacement and gives a result comparable to that of
the fibre-optic gauge. The observed double peak in
velocity for the wall mounted accelerometer might
have a physical explanation. It is reasonable to assume that this represents the arrival of shock wave at
the drift face followed by initial fracture propagation
and/or a reflection from the boundaries of the pillar.
A similar observation was made for a draw-wire
sensor mounted at the same section of the wall but
further down to the floor.
Possible explanations for the later discrepancy
between the integration result of the accelerometer
signal and the fibre-optic gauge are i) a zero-shift
problem occurring at a very early stage, ii) mounting
issues for the fibre-optic gauge, i.e. a lifting of the
floor and subsequent low-frequency oscillation of
the mounting stand which was also observed from
high-speed filming (at ~ 25- 35 ms).
The excellent performance of the PE accelerometers achieved in the laboratory tests could not be repeated for the field tests. It was observed that their
failure is often related to specific spurious pulses in
the signal that occur during the movement and mask
the actual signal in the low frequency region. The
expected difference in behaviour of the PE and the
PR accelerometers in the field did not materialise. If
such an effect exists it may be apparent during very
slow motion and as motion ceases.
With the measurement system installed behind
the burden, the initial face velocity was measured.
0
0
10
20
30
-10
40
50
-100
time t [ms]
Figure 13. Test 6, velocity (v) and displacement (d), PE accelerometer and fibre-optic gauge.
5 CONCLUDING REMARKS
The instrumentation of the burden to study the dynamic process of blasting into a compressible material is a crucial factor to quantify the blast function
within a SLC ring. This development work was initiated so that SLC blasting may in the long term be
better understood. Different redundant sensor systems were tested in both unconfined and confined
drift wall blasting tests. Blo-Up accurately reproduced the burden velocity and displacement observed in the experiments. The measurement methods deployed in the tests were further developed and
tested. The novel system, a PE accelerometer combined with the fibre-optic zebra gauge, will be further used to measure confined burden movement in
filled drift wall blasting tests under controlled situations and a full-scale SLC instrumentation is
planned. Moreover, the Blo-Up software might be
used in the near future to gain additional insight into
the complex process of blasting SLC rings.
ACKNOWLEDGEMENTS
The Hjalmar Lundbohm Research Centre (HLRC), a
research foundation at LTU and endowed by LKAB
is thanked for its financial support throughout the
PhD project Improved breakage and flow in sublevel caving. The colleagues from the central workshop and Complab at LTU are thanked for the support and numerous valuable discussions related to
the development of the measurement equipment.
Special thanks go to our technicians and the mine
surveyors in the LKAB Kiruna mine for all their
help consistently provided in practical matters.
REFERENCES
Belen`kii, E.V. Kovtun, I.N. & Fedorenko, P.I. 1969. Buffer
properties of caved ore. Sov Min Sci, 5(4): 444-447.
Brunton, I.A. 2009. The impact of blasting on sublevel caving
flow behaviour and recovery (Doctoral thesis). Univ of
Queensland, Brisbane.
Chu, A.S. 1988. Built-in mechanical filter in a shock accelerometer. In 59th Shock and Vibration Symp (1: 251-269).
Washington, USA: Shock and Vibration Information Centre, U.S. Naval Research Centre.
Chu, A.S. 1992. Problems in high-shock measurement (Paper
TP 308). San Juan Capistrano: Endevco Corporation.
Cullum, A.J. 1974. The effects of confined blasting on rock
fragmentation and flow characteristics in sublevel caving
(Master thesis). Univ of Queensland, Brisbane.
Furtney J., Cundall, P.A. & Chitombo, G. 2009. Developments
in numerical modeling of blast induced rock fragmentation:
Updates from the HSBM project. In J.A. Sanchidrin (ed.),
9th Int Symp on Rock Fragmentation by Blasting (335342). London: CRC Press.
Johansson, D. 2011. Effects of confinement and initiation delay
on fragmentation and waste rock compaction. (Doctoral
thesis). Lule Univ of Technology, Lule.
Kirpichenko, V.M. 1982. Study of ore breaking in a compressed medium by sets of parallel adjacent bores in the
Tash Tagol Mine. Sov Min Sci, 18(3): 233-237.
Newman, T., Hustrulid, W. & Quinteiro C. 2008. Sublevel
caving trial monitoring effects from blasting an ore slice
against caved rock at LKAB`s Kiruna mine, Sweden. In H.
Schunnesson & E. Nordlund (eds.), 5th Int Conf and Exhib
on Mass Mining (705-713). Lule: Univ of Technology.
Noren, C.H. 1956. Blasting experiments in granite rock. Colorado School of Mines Quarterly, 51(3): 213-225.
Olsson, M., Nyberg, U. & Fjellborg, S. 2009. Controlled fragmentation in sublevel caving first tests (Swebrec Report
2009:2). Lule, Sweden: Lule Univ of Technology. In
Swedish.
Petropoulos, N. 2011. Influence of confinement on fragmentation and investigation of the burden movement small
scale tests. (Master thesis). Lule Univ of Technology,
Lule.
Power, G. 2004. Modeling granular flow in caving mines:
large scale physical modeling and full scale experiments.
(Doctoral thesis). Univ of Queensland, Brisbane.
Rustan, A. 1970: Theoretical basics of the volume-weightmethod for the determination of the ore content of a mixture of blasted ore and waste rock. Kinematics, swelling,
loosening and fragment size in the burden for confined
blasting in model-scale. (Licentiate thesis). Royal Institute
of Technology, Stockholm. In Swedish.
Selldn, H. & Pierce, M. 2004. PFC3D modeling of flow behaviour in sublevel caving. In A. Karzulovic & M.A. Alafaro (eds.), 4th Int Conf and Exhib on Mass Mining (201214). Santiago: Instituto de Ingenieros de Chile.
Shirzadegan, S., Nordlund, E., Nyberg, U., Zhang, P. &
Malmgren, L. 2011: Rock support subjected to dynamic
loading: Field testing of ground support using simulated
rock burst. In Q. Quian & Y. Zhou (eds.), Harmonising
rock engineering and the environment, 12th International
Society for Rock Mechanics International Congress on
Rock Mechanics (1269-1273), Leiden, Netherlands: CRC
Press/Balkema.
Volchenko, N.G. 1977. Influence of charge arrangement geometry and short-delay blasting on the crushing indices in
compression blasting. Sov Min Sci, 13(5): 488-493.
Wimmer, M. & Ouchterlony, F. 2008. Application of time domain reflectometry (TDR) for block- and sublevel caving
mines State-of-the-art and preliminary laboratory shear
tests (Swebrec Report 2008:P3). Lule, Sweden: Lule
Univ of Technology.
Wimmer, M. & Ouchterlony, F. 2011a. Study of burden movement in confined drift wall blasting tests in block 12, 691 m
level, Kiruna mine (Swebrec Report 2011:P1). Lule, Sweden: Lule Univ of Technology.
Wimmer, M. & Nordqvist, A., Ouchterlony, F. 2011b. Burden
movement while blasting under constraints Tests with
new gauges in block 9, 741 m level, Kiruna mine (Swebrec
Report 2011:P2). Lule, Sweden: Lule Univ of Technology.
Wimmer, M., Nordqvist, A., Ouchterlony, F., Selldn, H. &
Lenz, G. 2012. 3D mapping of sublevel caving (SLC) blast
rings and ore flow disturbances in the LKAB Kiruna mine.
In G. Baiden & Y. Bissiri (eds.), 6th Int Conf and Exhib on
Mass Mining (CD). Sudbury: Laurentian Univ.
Yiannakopoulos, G. & van der Schaaf, P. J. 1998. Evaluation
of accelerometer mechanical filters on submerged cylinders
near an underwater explosion. Shock and Vibration, 5(4):
255-265.
Zhang, G. 2004. Behaviour of caved ore mass in sublevel caving and its effect on ore dilution. In A. Karzulovic & M.A.
Alafaro (eds.), 4th Int Conf and Exhib on Mass Mining
(238-242). Santiago, Chile: Instituto de Ingenieros de
Chile.

Towards Understanding Breakage and Flow in Sublevel Caving (SLC)

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Towards Understanding Breakage and Flow in Sublevel Caving (SLC)

Încărcat de

Drepturi de autor:

Formate disponibile

DOC TOR A L T H E S I S

Department of Civil, Environmental and Natural Resources Engineering

Lule University of Technology 2012

ISSN: 1402-1544 ISBN 978-91-7439-509-9

Towards Understanding Breakage and

Towards Understanding Breakage and Flow in Sublevel Caving (SLC)

Development of new measurement techniques

Division of Mining and Geotechnical Engineering

Printed by Universitetstryckeriet, Lule 2012

when you can measure what you are speaking about,

Sprngsalvors funktion, fragmentering och rasflde r viktiga delprocesser i skivrasbrytning. Fr att

Wimmer, M. & Ouchterlony, F. (2008). Application of time domain reflectometry

Wimmer, M. & Ouchterlony, F. (2008). Application of borehole geophysics to identify

Wimmer, M. & Ouchterlony, F. (2009): 2D image analysis using WipFrag software

Wimmer, M. (2010): Gravity flow of broken rock in sublevel caving State-of-the-art

Wimmer, M. & Ouchterlony, F. (2011). Study of burden movement in confined drift

Wimmer, M. & Nordqvist, A. & Ouchterlony, F. (2011). Burden movement while

Wimmer, M., Nordqvist, A., Ouchterlony, F. & Selldn, H. (2012). 3D mapping of

PART 1 EXTENDED SUMMARY ..................................................................................................................... 1

PROBLEM STATEMENT ............................................................................................................................ 5

SCIENTIFIC CONTRIBUTIONS ................................................................................................................... 7

BREAKAGE AND GRAVITY FLOW IN SLC ........................................................................................ 9

SUBLEVEL CAVING (SLC) .................................................................................................................... 11

BREAKAGE IN SLC ............................................................................................................................... 12

Small- and full-scale experiments .......................................................................................................... 13

Blasting into compressible materials .............................................................................................. 13

Ring blasting effects ........................................................................................................................ 19

Controllable factors and their effects on gravity flow ..................................................................... 21

Conceptual flow models .................................................................................................................. 24

SLC DRILL AND BLAST PRACTICE (KIRUNA MINE).................................................................... 27

SLC LAYOUT AND RING DESIGN ........................................................................................................... 29

CHARGING AND BLASTING ................................................................................................................... 32

BLAST FUNCTION .................................................................................................................................. 35

3D MAPPING OF SLC BLAST RINGS ....................................................................................................... 37

Measurement procedure on-site ...................................................................................................... 37

Photogrammetry system ......................................................................................................................... 37

Evaluation process .................................................................................................................................. 39

Blast results ..................................................................................................................................... 41

Initial opening blasts............................................................................................................................... 41

Scale-up of SLC rings ............................................................................................................................ 43

Categorization and discussion of breakage problems ............................................................................. 46

Drilling accuracy ............................................................................................................................ 52

BURDEN INSTRUMENTATION OF SLC BLAST RINGS .............................................................................. 54

Measurement problem ..................................................................................................................... 54

Dynamics of burden movement ....................................................................................................... 55

Confined drift wall blasting tests ............................................................................................................ 55

Unconfined situation (test 2).............................................................................................................. 60

Confined situation (test 4).................................................................................................................. 61

Numerical modeling .......................................................................................................................... 64

Fibre-optic zebra gauge ..................................................................................................................... 67

Laboratory tests ................................................................................................................................. 67

Field tests ........................................................................................................................................... 69

Breakage of burden ......................................................................................................................... 72

Field tests ........................................................................................................................................... 73

Development of measuring method ................................................................................................... 73

Laboratory tests ................................................................................................................................. 73

Field test ............................................................................................................................................ 76

NBC APPROACH AND SMALL-SCALE BLASTS ........................................................................................ 79

PREVIOUS FULL-SCALE FRAGMENTATION TESTS................................................................................... 81

RECENT FULL-SCALE FRAGMENTATION TESTS ...................................................................................... 83

Test layout and working procedure ................................................................................................. 83

Sieving results ................................................................................................................................. 85

Sieving curves for 0 - 63 mm fraction samples ...................................................................................... 85