International Journal of Rock Mechanics & Mining Sciences 127 (2020) 104237

Contents lists available at ScienceDirect

International Journal of Rock Mechanics and Mining Sciences

journal homepage: http://www.elsevier.com/locate/ijrmms

Experimental quantification of vertical stresses during gravity flow in

block caving
Raúl Castro a, b, *, Ren�e Go
�mez c, Matthew Pierce d, Juan Canales b
a
Laboratorio de Block Caving, Advanced Mining Technology Center, Universidad de Chile, Chile
b
Departamento de Ingeniería de Minas, Universidad de Chile, Chile
c
Facultad de Ingeniería, Universidad de Concepci�
on, Chile
d
Pierce Engineering, Minnesota, United States

A R T I C L E I N F O A B S T R A C T

Keywords: Usually in block and panel caving mines, when the caving connects to the surface, monitoring of vertical stresses
Caving mining over production drifts is minimal due to the low probability of seismic events. However, when the stresses over
Granular material the production level are related to an overburden by fragmented rock, the draw policies highly influence induced
Gravity flow
stress. In this context, we present an experimental set up using a laboratory-scale model to identify the role of
Induced stresses
Stress
draw strategy on induced vertical stresses in a block-caving layout. Here, three draw strategies are studied:
Rock isolated draw, panel caving draw, and block caving draw. Results show that induced vertical stresses can vary
between 0.3 and 2.8 times their initial vertical value and are highly influenced by both the distance from the
extraction front and the dimension of draw and non-draw areas. These findings provide useful information to
support the system design in block caving or panel caving methods to decrease induced vertical stresses related to
draw and define rules of extraction in an effort to avoid problems related to the maximum stress that can be
expected on production-level pillars.

1. Introduction walls, collapse and closure.

In granular material, such as caved rock ore, vertical stresses are
Current challenges in the mining environment include factors such as significantly lower than total column weight because of the arching ef­
deeper deposits, harder rock, higher in-situ stresses, greater production fect generated by shear stresses due to the rock friction. Janssen 21
demand and increased costs.1,2 One issue which can have consequences initially postulated Eq. (1) that provides the mean vertical stress, σv;0
at greater depths is that the rock mass already subjected to high stresses (Pa) in certain depth, z (m), considering overload, Q0, 22
suffers even higher induced stresses during mining activities. In partic­ 0 1 0 1
ular, induced stress occurs during the caving propagation stage in the Rh ρb g B μkz
C B μkz
C
advancing undercut front,3–5 where damage has been observed and re­ σ v;0 ¼ @1 e Rh
A þ Q0 @e Rh
A (1)
μk
ported in several block caving mines,6–11 mainly related to drift
convergence and pillar failure. Measurements of stresses, strain, crack
extension and seismicity have been carried out at this stage.12–15 How­ Where Rh is the hydraulic radius (area/perimeter; m) introduced by, 23
ever, in caving mines with greater depths, the column height of caved ρb is the bulk density (kg/m3), g is the gravity constant (m/s2), k is the
rock also increases, and consequently, there is an overload on the pro­ friction parameter that represents the horizontal and vertical stress
duction level. Usually in block or panel caving mines, when the caving ratio, σh =σv , z is the depth of caved rock (m), Q0 is the initial vertical
propagation connects to surface, vertical stress over the production overload (Pa) and μ is the friction between particles and model wall
drifts is not closely monitored because of the low frequency and expressed usually by tanðφw Þ, where φw is the friction angle of bin walls
magnitude of seismic events. Nevertheless, experience indicates that (degrees). This equation gives a reasonable result in static conditions.
production drift stability can also be affected during ore extraction in However, one of the common problems is determining the k parameter
productive areas16–20 causing drift convergence, concrete slabbing at the because of the high variability of stress in granular material. Different

* Corresponding author. Laboratorio de Block Caving, Advanced Mining Technology Center, Universidad de Chile, Chile.
E-mail address: [email protected] (R. Castro).

https://doi.org/10.1016/j.ijrmms.2020.104237
Received 13 March 2019; Received in revised form 29 January 2020; Accepted 30 January 2020
Available online 5 February 2020
1365-1609/© 2020 Published by Elsevier Ltd.
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

approaches have been used to define k such as, 24 operations are subjected to greater depths such as El Teniente and
Chuquicamata’s Underground Project in Chile, the Deep Ore Zone and
k¼1 senðφÞ (2)
Deep Mill Level Zone in Indonesia, Resolution Copper and the Hender­
Where φ is the material internal friction angle, the relation φ > φw has son mine in the United States and Cadia East in Australia.
been reported.23–25 Other approaches to define k are presented in Eqs.
(3) and (4), 2. Experimental methodology

k¼
1 senðφÞ
(3) Here a physical model for studying the vertical stress behaviour
1 þ senðφÞ under flow conditions was constructed representing different draw
strategies used in block and panel caving.
Eq. (3) is defined as an Active case, when σh < σv and Eq. (4) is
defined as a Passive case, when σ h > σv 26
2.1. Similitude analysis
1 þ senðφÞ
k¼ (4)
1 senðφÞ The focus of this work is to quantify stresses on granular material for
Walker 22
introduces in Eq. (5) a general relation for k using the block caving applications using laboratory-scale experiments. For this
Mohr-Coulomb circle considering that horizontal and vertical stresses reason, the geometric similitude is kept constant both in the physical
are not necessarily the principal stresses. model and the granular material with a length scale factor, λl , of 1:200.
Furthermore, following conditions of similitude, cohesionless material is
1 sen2 ðφÞ used to study stresses on granular material 35 to avoid adhesive forces
k¼ (5)
1 þ sen2 ðφÞ such as Van der Wall’s forces, liquid bridges and electrostatic effect at
laboratory scale. Then the material used must be dry and include frag­
The equations above were developed in physical models to quantify
ment sizes larger than 0.06 mm.40 On the other hand, the gravity, bulk
the magnitudes and distribution of stresses in static and dynamic con­
density and residual friction angle scale factors are the same in the
ditions.21–31 Eqs. (2), (3) and (5) have shown good correlation under
model as at mine scale, respectively λg ¼ λρ ¼ λφ ¼ 1. Then, the scale of
static conditions. In dynamic conditions, different results have been
stresses is equal to the length scale factor, λσ ¼ λl . This geometrical scale
reported for k and μ parameters in physical and numerical models,
24,32–34 considered (length scale, λl ¼ 1/200) ensures representation of the
mainly as a function of the granular material’s height. However,
induced stress phenomena. According to the scaling laws,35 the simili­
granular materials used in these studies (e.g. bean, corn, sand, wheat,
tude’s parameters are shown in Table 1. Here physical variables are
barley and coke) are different from material found in the mining envi­
based on the length scale used.
ronment, where wide particle size distributions, high strength, major
The following assumptions and simplifications were made to study
internal friction angles and angular fragment shapes are common. Few
the stress behaviour in block caving through physical modelling: (1) It is
physical models emulating mining conditions have been developed.35–37
assumed that the rock mass has caved to surface, forces in the system are
When granular material is under flow, at least two main zones the
due to the material’s weight, and in-situ stresses are not present. (2) The
movement zone and the stagnant zone – can be identified. In the
granular flow occurs in a 3D environment. (3) The rock breakage
movement zone (or draw zone), porosity increases and vertical stresses
mechanism related to secondary fragmentation is not considered. (4)
decrease due to flow, while in the stagnant zone (or non-draw zone)
The geometries of the flow zones in the physical model are comparable
vertical stresses increase. Experimental measures have shown that
as long as the materials have equal friction angles and particle shapes.
stresses can be transferred from the movement zone to the stagnant zone
The influence of friction in the model’s container walls is discussed in
during flow.35 Then, induced vertical stresses over the production drift
section 2.2.
depend on the movement of gravity flow zones. In the same way, Lorig 38
postulated that stresses in the isolated movement zone (IMZ) tend to be
lower than in surrounding stagnant material. This may occur because 2.2. Physical model
stresses are transferred through shear forces acting on boundaries be­
tween flow and stagnant material. Furthermore, it was concluded in The physical model represents a section of a mine operated by Block
Ref. 39 that the mean magnitude of forces acting inside the IMZ is in­ or Panel caving with the following dimensions: 140 m length, 46 m
dependent of the column height and weight. width and 480 m column height (see Fig. 1). The scale was chosen to
Pierce 39 proposed applying the tributary area theory to determine represent a high column height where vertical stress at the bottom
would be significant. In the production level, we selected El Teniente’s
the mean vertical stresses in the stagnant zone, σSZv , as:
30 m � 15 m layout.
Pn MZ
σv;0 AT i¼1 σ v;i Ai The mine section corresponds to three production drifts and nine
σ SZ
v ¼ (6)
Asz extraction drifts, which represent a total of 36 drawpoints (Fig. 2 and
Fig. 3-a). Fig. 2 shows the dimensions of the production level in the
Where σ v;0 is the mean vertical stress before draw (initial condition, Pa), physical model. Fine material was glued on the crown pillar over the
AT is total area (m2), σ MZ
v;i is the mean vertical stress of the movement
production drifts to represent the rock friction over pillars and the flow
zone i calculated using Eq. (1) and Eq. (3) (Pa), Ai is the area of the zones’ geometries, simulating 3D drawbells.
movement zone i (m2) and Asz is the area of the stagnant zone (m2).
Areas are where the caved material is placed in a plan view. Table 1
In caving mines, the draw is different in each case. For example, Scale factors based on length scale.
block caving extracts material in a different way than panel caving. Variable Factor Scale factor Value
Information on the influence of different kinds of draw on induced
Length λL λL 0.005
vertical overloads on the production level is, thus, generally lacking. Area λA λ2L 0.000025
Then, the objective of this investigation is to use laboratory-scale ex­ Volume λV λ3L 0.000000125
periments to quantify the role of draw strategy on induced vertical Time λT λ0.5
L 0.0707
Weight λ3L 0.000000125
stresses over pillars located in the production levels to define draw λW
Stress λS λL 0.005
policies that minimize induced stress in critical areas. In particular, Friction angle λø 1 1
nowadays a number of active or planned block and panel caving Density λρ 1 1

2
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

Fig. 1. Mine area represented in the physical model.

Fig. 2. Geometry of production level represented at the physical model scale.

The walls of the physical model as shown in Fig. 3-b were made of �0.039 kPa (8 g), a length of 11.5 cm and an area of 20 cm2. The po­
plexiglass. The plexiglass allows movement zones to be observed in the sition of load cells over the crown pillar depends on the purpose of the
model. This material also increased vertical stress compared with hori­ experiment. Thus, load cell location is shown with the results for each
zontal stress because of the low friction between plexiglass and granular test in Section 3 (Figs. 6, 9, 12 and 15).
material. For minor frictions, shear forces decrease between the model
wall and granular material, then more weight is transferred to the bot­ 2.3. Model media
tom (increasing vertical stress estimated there). For example, we can
calculate a vertical stress of 1.35 MPa applying Eq. (1), considering a Tests used crushed sulphide ore with a high aspect ratio (sphericity
depth of 500 m, an internal friction angle of 39� , a hydraulic radius of 0.58 and roundness 0.25, Cho et al.41 methodology), which is repre­
16.9 m, a bulk density of 1.4 t/m3 and a wall friction angle of 25� . If we sentative of caved rock. The particle size distribution was obtained by
considered a higher friction angle, like the internal friction angle of 39� , scaling (1:200) the primary fragmentation curve expected for Chuqui­
a hydraulic radius 74% higher gives us the same vertical stress. For this camata Underground Block caving project (see Fig. 4). The ore charac­
reason, in this study vertical stress results, σ v, will be analysed as a teristics are shown in Table 2.
function of their initial value, σv,0, using the ratio σv/σ v,0. Sulphide ore corresponds to a biotite and amphibole granitoid,
Up to six load cells were located within the model over the crown showing metallic mineralization of pyrite, chalcopyrite and bornite.
pillar for vertical stress measurements. Each load cell has a precision of Potasic and propilitic alteration can be seen in the rock samples and

3
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

Fig. 3. (a) Sector studied in experiments and main structures, plan view (b) Physical model of plexiglass used in laboratory.

Fig. 4. Fragment size distribution curve used in experiments scaled from a real Block-caving size distribution curve.

caving mines. Undercutting development, in Block and Panel caving, is

Table 2
the principal difference between them. Block-caving undercutting is
Ore characterization.
performed in just one stage, or block, at a time (Fig. 5-a), whereas panel-
Parameter Value Unit caving undercutting is performed over time by opening drawbells
Mean size, d50 4.0 mm monthly (Fig. 5-b). These differences usually imply that at the beginning
Coefficient of uniformity, Cu 2.0 of block caving extraction, all drawpoints of one block could be avail­
Point load index, IS50 6.2 � 1.6 MPa able. However, for different operational reasons, unmoved zones
Density 2.6 t/m3
Bulk density 1.42 t/m3
without extraction or overdrawn zones may also exist. On the other
Porosity 44 % hand, in panel caving, it is common to find open drawpoints, closed
Friction angle 39 �
drawpoints, and drawpoints with differences in their extraction
Repose angle 29 �
percentage.
Table 3 lists the four experimental set ups with their objectives, as
well as the draw strategy and the procedure chosen. To observe and
mafic minerals, respectively.
quantify result variability, two tests (A and B) were carried out with each
experimental set up (Tests 0, 1, 2 and 3) for a total of 8 experiments. We
2.4. Experimental set-up and procedure defined different draw strategies in each test then grouped results ac­
cording to the draw strategy used for analysing them: isolated draw
The main objective of this work is to study the stresses induced by ore (base case), panel caving and block caving draw strategies, respectively.
draw, considering common draw practices applied in Block and Panel

4
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

Fig. 5. Plan views (a) Block caving strategy: in blocks 1, 2 and 3 the extraction begins at different periods. Each block should be drawn uniformly. (b) Panel caving
strategy: in panels 1, 2, 3, 4 and 5 the extraction begins consecutively. Panels have different extraction percentages; greater percentages are in the first panels.

Table 3
Experimental set-up of gravity flow draw strategies.
Test Name Objective Procedure

0 Isolated draw To measure IMZ and Extraction was carried out

vertical pressures over the from one drawpoint (from
crown pillar for a single Extraction Drift (ED5))
drawpoint. next to the crown pillar.
Twenty kilograms of ore
drawn at 2.5 kg/h.
1 Panel caving To measure vertical Extraction was done from
pressures over the crown ED1 up to ED7. In total,
pillar considering a panel one kg were drawn per
caving draw scenario. drawpoint. During test
EDs 1 and 2 were closed
and EDs 6 and 7 opened.
Test draw rate ~5 kg/h.
2 Block caving – To measure vertical ED 1 to 3 and ED 7 to 9
60 m of pressures over the crown were drawn at the same
unmoved zone pillar with an intermediate rate until flow zones
non-draw area, following a reached the surface.
block-caving draw Afterwards, extraction
scenario. was done in ED 4 to 6. Test
draw rate ~6.5 kg/h.
3 Block caving – To measure vertical Extraction was conducted
30 m of pressures over the crown in all EDs except ED5. At
unmoved zone pillar with a smaller the end of the experiment,
intermediate non-draw extraction was conducted
area. in ED5. Test draw rate
~6.5 kg/h.

strategy, the extraction was carried out through a group of extraction

Fig. 6. (a) Isolated draw strategy configuration (front view) and (b) Load cell
drifts continuously until 1 kg per drawpoint was extracted; the draw­
location over crown pillar. points were then incorporated and/or closed depending on the period of
extraction following a panel caving scenario. In this work, the effect of
rate of draw was not studied because the stresses measured in the load
For each test, the load cells were located based on the point of in­
cells changed almost instantly in all extraction policies observed
terest defined by the draw strategy. The material described in the Model
regardless of extraction velocity.
Media section, previously homogenised, was then loaded in the physical
model to the top (2.5 m). The draw strategy defined for the test was then
programed through a servo-assisted motor. In block caving draw stra­ 2.5. Initial stress conditions
tegies, all drawpoints were drawn at the same time until near the end of
the tests when active drawpoints were stopped and drawpoints located Initial vertical stresses were measured in all tests as the following:
under the unmoved zone were drawn. For the panel caving draw 14.8 � 3.2 kPa in Test 0, 17.8 � 1.8 kPa in Test 1, 27.7 � 3.7 kPa in Test

5
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

2, 18.3 � 3.7 kPa in Test 3 and the initial vertical stresses based on Eq.
(1) is 18.7 kPa. Here the mean vertical stress is based on the material
characteristics defined in Table 2 and assuming k ¼ 0.37. The initial
vertical stress ranges are expected for physical measurement of granular
material due to arching, as reported by other studies.31,35–37

3. Draw test results

In this section vertical stress measurements are presented during

draw condition. In the first test run, the extraction from one drawpoint
defines the IMZ geometry of granular material used. Then, panel and
block caving draw strategies are tested showing the influence of the
extraction strategy on induced stress.

3.1. Test 0: isolated draw

Fig. 8. Induced vertical stress by mass drawn and IMZ radius evolution under
mass drawn. Isolated draw. Here, σv: vertical stress, σv,0: initial vertical stress,
In this test, ore draws from just one drawpoint located in the middle
RIMZ: IMZ radius (for tests A and B). The error bar shows the vertical stress
of the model (see Fig. 6). Fig. 6-b indicates the load cells’ location over variability (between tests A and B).
the crown pillar. The evolution of the IMZ geometry is according to the
pffiffiffi
kinematic theory normalized by d50, w ¼ 3:42 h 42 where w is the IMZ
width and h is the IMZ height presented against results in Fig. 7.
During this test, the IMZ radius reached 7.8 cm when 0.7 kg of ma­
terial were extracted (Fig. 8, right vertical axis). Here, load cells 2 and 3
were below the shadow of the flow zone. Then, the IMZ radius reached
16 cm when 14.6 kg of material were extracted, over load cell 1 (located
at 15.8 cm from the active drawpoint).
We observed that vertical stress decreased in load cells closer to the
drawpoint under extraction (load cells 2 and 3). Load cell 2 decreased
0.4 times its initial value, whereas load cell 3 decreased its initial value
0.6 times. Load cell 2 is closer than load cell 3 to the drawpoint under
extraction; then a distribution of vertical stress into the IMZ can be
assumed increasing from the IMZ’s centre. It is known that greater
(higher) porosity in granular material flow is generated next to the
extraction point. In draw zones, the bulk density is decreased due to
greater porosity. Greater porosity also implies fewer contact points to
transfer stresses in the granular material, which increases the stress
transfer on more compacted material, which in this case is the material
found in the non-draw zones.
On the other hand, load cell 1 showed no major variation from its
initial load condition. It can be seen that inside the IMZ, vertical stress
was lower than initial stress, as expected, and decreased closer to the
drawpoint. In the non-draw zone, overstress (on load cell 1) was not
observed. Possible reasons for this are its proximity to the flow zone, the
ratio between the draw and non-draw zones, and/or a shadow effect cast
by the IMZ. Further experiments could be done to identify the cause or
causes of this effect.
Fig. 9. (a) Panel caving draw strategy configuration (front view) and (b) Load
3.2. Test 1: panel caving draw strategy cell location over crown pillar.

In test 1, various drawpoints were opened based on the panel caving plan. Experiments here show a mass flow zone (interactive draw be­
strategy defined in section 2.4. Extraction drifts were drawn one by one tween drawpoints under extraction). The extraction drifts (ED) were
from right to left (Fig. 9-a), in accordance with a common production added from ED1 to ED7, between 27.1 and 47.7 kg ore was drawn from
ED1 to ED5, between 47.7 and 67.5 kg ore was drawn from ED3 to ED6,
and between 67.5 and 78.2 kg ore was drawn from ED3 to ED7.
Fig. 9-b shows the locations of the six load cells. This distribution
allows vertical stress measurements in draw and non-draw zones during
extraction. Here, the addition of extraction drifts caused the formation
of an unmoved zone in front of the extraction front. Fig. 10 shows the
test’s evolution, in which the width of the non-draw zone (defined by
Wr) continuously decreases while new EDs begin extraction. Induced
vertical stress in a non-draw zone will necessarily increase if its area is
decreased, as observed in Fig. 11.
Fig. 11 illustrates results for all load cells. The stress measurements
show overloads in the non-draw zone (also known as stagnant zone) and
Fig. 7. Isolated movement zone geometry evolution. under-loads in the draw zone (movement zone). Stresses recorded by

6
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

Fig. 10. Test 1 draw evolution during Panel caving draw strategy.

cell 1 (Fig. 11-a) decreased rapidly when extraction started because this
cell was below the flow zone, 0.30 σv,0. When 40–50 kg of ore was
extracted, drawpoints of ED 1 and 2 were closed, this increased the
induced vertical stress to 1.36 σv,0. Load cell 2 showed similar behaviour
with stress variations between 0.26 σv,0 below the flow, increasing to
0.54 σv,0 when EDs were closed. Load cell 3 was located in the stagnant
zone at the beginning of extraction. The vertical stresses showed no
change in value until drawpoints near (ED4) were extracted; at which
point, induced vertical stress decreased to 0.26 σv,0 (Fig. 11-a). Vertical
stresses on load cells 4, 5 and 6, located initially in the stagnant zone
away from drawpoints under extraction, continuously increased due to

Fig. 12. (a) Block caving draw strategy configuration with high non-draw zone
(front view) and (b) Load cell location over crown pillar.

the extraction (Fig. 11-b). In these cells, vertical stresses increased even
though the mass in the model was continuously decreasing. Then, ver­
tical stress on load cell 4 decreased to 0.36 σv,0 when the load cell was
Fig. 11. Induced vertical stresses by mass drawn (a) Results from Load cells 1, below the flow zone at the end of the test. In this test, load cell 6 reached
2 and 3 and (b) Results from load cells 4, 5 and 6. Here, σv: vertical stress, σv,0: the maximum value observed of 1.77 σ v,0.
initial vertical stress.

7
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

In the non-draw zone, a direct relation was observed in load cells 4, 5

and 6, between the vertical stress and the stagnant zone width (Wr),
such that when the stagnant zone decreased, the vertical stress
increased.

3.3. Test 2 and 3: block caving draw strategies

In tests 2 and 3, ore was extracted from the drawpoints in the outer
areas at the same time. For those drawpoints located in the middle of the
model, ore was not extracted until the end of the tests.

3.3.1. Test 2: non-ideal draw with 60 m of unmoved zone

The outer area extraction drifts (EDs) were drawn together while
three EDs were not drawn in the middle (see Fig. 12). The flow zone
evolution indicates that initially the flow zones interact among EDs 1
and 3 and, separately, among EDs 7 to 9, leaving an unmoved zone, Fig. 14. Induced vertical stresses by mass drawn load cells 1 to 6. Here, σv:
which has a width (Wr) observed in Fig. 13. vertical stress, σv,0: initial vertical stress.
When extraction progressed, the flow zones of each group of
extraction drifts interacted in height, leaving just a small portion of drawn, the induced vertical stress decreased to 0.35 σ v,0.
unmoved volume at the ore bottom. After approximately 95 kg extrac­
ted, draw strategy changed as EDs under draw (ED 1, 2, 3 and 7, 8, 9) 3.3.2. Test 3: non-ideal draw with 30 m of unmoved zone
were stopped and drawing was started from ED 4, 5 and 6. In this test, the same strategy as test 2 was used, but with a minor
Vertical stress measurements are shown in Fig. 14. The load behav­ unmoved zone to quantify higher induced vertical stress, and the rela­
iour observed was as follows: tion between the draw and non-draw areas as indicated in theory.39
Load cells 1, 2, 5 and 6 were placed within flow zones. When Here, only ED5 was not drawn (Fig. 15), while other EDs were under
extraction began, vertical stresses in these cells decreased to 0.31 σ v,0 draw. A small pillar was induced with draw strategy. One load cell was
and were maintained at that level until draw strategy changed. located within the non-draw zone (load cell 3 in Fig. 15-b, through ED5
For Load cells 3 and 4, located under the stagnant zone (Wr), vertical considering the El Teniente layout), while the other four load cells were
stresses were increased to 1.86 σ v,0 and 1.51 σ v,0, respectively. The dif­ located in draw zones over the crown pillar.
ference could have occurred because load cell 4 was closer to a move­ The evolution of flow zones showed that the no-flow zone rapidly
ment zone. Afterwards, as the moving front approached the location of disappeared as flow zones interacted leaving only a small, unmoved
the cells, the load decreased as much as 0.9 times from its original value. zone. In Fig. 16 it can be seen that the flow zones from both draw areas
At the end of the experiment, when drawing from extraction drifts 4 to 6 are interacting.
started, the vertical stress of these cells decreased 0.36 and 0.39 times From the measurements of load cells shown in Fig. 17, the vertical
from their original value, respectively. stress behaviour was as follows:
At the end of the experiment (around 95 kg), extraction from the EDs Load cell 3 was located in the permanent unmoved zone. In this case
1 to 3 and 7 to 9 was stopped and EDs 4 to 6 were drawn. A subsequent the induced vertical stress increased up to 2.81 σv,0. After 20 kg of
increase of vertical stresses on these cells was observed, reaching 0.6 to continuous draw, the movement zones overlapped over cell 3. The stress
1.1 of their initial value. here then decreased but was still higher than at its initial value (see
Load cell 3 registered the highest vertical stress on this test. It was Fig. 17). At the end of the experiment, after 100 kg of extraction, an
placed in the middle of the stagnant zone. In this case, induced vertical extraction drift below this cell was opened, and drawing started. In this
stresses increased up to 1.86 σv,0 and decreased continuously due to the case, the stress decreased as much as 45% from its original value.
mass removal from the system. Then, when extraction drifts 4 to 6 were

Fig. 13. Test 2 draw evolution in Block caving draw strategy (60 m).

8
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

stresses in granular material. These variables are the presence of un­

moved and moved areas, mass drawn, and distance to the extraction
front from any point in the unmoved and moved zones. Although pre­
vious studies have suggested a relationship among the variables studied,
few physical models have been used to actually measure the effects of
these variables on induced stresses on granular material related to draw
policies. Our study also took into consideration different draw strategies
commonly used in block and panel caving mines.
Fig. 18 shows the induced vertical stress measured in the stagnant
zone, σV (SZ – exp), and in the movement zone, σ V (MZ – exp), from Test
1: Panel caving draw strategy. Here, Eq. (6) is used to compare induced
stress estimated with vertical stress measured in the movement zone,
showing a good correlation with a standard deviation of 2.3. Addition­
ally, the vertical stresses in the movement zone are calculated with the
Janssen approach presented in Eq. (1).21 A good fit is obtained consid­
ering k as (1-senφ) and u as 2.6, i.e. vertical stress is higher than hori­
zontal stress on the movement zone of this experiment. In addition, a
good fit can also be obtained using Eq. (3). Fig. 18 shows results until 22
kg to avoid variation on stresses due to loss of weight.
To analyse the induced vertical stress in block caving strategies and
compare this with the results obtained in panel caving strategies,

Fig. 15. (a) Block caving draw strategy configuration with high non-draw zone
(front view) and (b) Load cell location over crown pillar.

Load cells 1, 2, 4 and 5 were placed below the flow zone and all these
cells showed a decrease in their stresses. The induced vertical stress of
these cells decreased by around 45% from their original values. When
extraction drift 5 was opened, the other extraction drifts were closed,
and induced vertical stress in these cells increased from 0.8 to 1.4 σ v,0.

4. Discussion

Fig. 17. Induced vertical stresses by mass drawn on load cells 1 to 5. Here, σv:
This study quantified the effect of key variables on induced vertical
vertical stress, σv,0: initial vertical stress.

Fig. 16. Test 3 draw evolution in Block caving draw strategy (30 m).

9
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

Fig. 18. Induced-vertical stresses in movement and stagnant zones.

induced vertical stress in the stagnant zone, σ SZ

v is calculated rewriting
stagnant zone.
Eq. (6) and normalizing it by the initial vertical stress, σv0 as, The results obtained from this experimental set up can be used to
� P MZ � define extraction policies to alleviate induced vertical stresses within
σ SZ σv AMZ 1 caved materials. Based on our results, one such strategy would be to
v
¼ AT (7)
σv0 σ v0 Asz increase the unmoved area proximal to drawpoints to allow vertical
stress to act on a larger area. Another strategy would be to draw from
Where σ v0 is the initial vertical stress, AT is the total caved area, σ MZ
v is unmoved zones to decrease stress.
the vertical stress in the movement zone, AMZ is the area of the move­ Many other interesting variables were not included in our study
ment zone and Asz is the area of the stagnant zone. It should be noted whose main objective was to quantify induced stress due to ore draw
that the induced vertical stress in the stagnant zone increases when the under controlled conditions. It would be useful in future studies to
area of the stagnant zone decreases. Additionally, in the stagnant zone observe the effects of variables such as humidity, draw rates and uni­
the induced vertical stress depends on the vertical stress in the move­ formity of draw on induced stress. Furthermore, findings from this study
ment zone. This relation between the induced vertical stress in the suggest other related studies. For example, future set ups could be
stagnant zone and its area is shown in the draw test in Fig. 19. constructed to more deeply analyse production pillar stability and the
Fig. 19 shows the vertical stress measured in the stagnant zone in influence of shear zones during ore draw.
Test 1, 2 and 3, and the induced vertical stress calculated using Eq. (7),
which used vertical stress measured in the movement zone of tests. As 5. Conclusions
observed previously in Fig. 18, induced stress correlates well with the
panel caving strategy in Fig. 19. However, in block caving tests, Eq. (7) In the flow zones of the different draw strategies, lower vertical
overestimates the induced vertical stress. The overestimation of induced stresses were measured, and consequently, as expected, there were
stress may occur because when both movement zones are in contact, the higher vertical stresses under no-flow zones. The size of the unmoved or
stress related to the initial density of the material is decreased in the moved areas influenced the magnitude of the induced vertical stresses.

Fig. 19. Induced-vertical stresses over stagnant zone relative to total area.

10
R. Castro et al. International Journal of Rock Mechanics and Mining Sciences 127 (2020) 104237

While the unmoved zone width decreased, the vertical stresses increased 14 Viegas G, Bosman K, Angus D, De Beer W, Urbancic T. Mapping cave front growth
utilising the collective behaviour of seismicity and velocity fields. In: Caving 2018,
here, and this was observed both in block caving and panel caving draw
Fourth International Symposium on Block and Sublevel Caving. Vancouver, Canada; 15-
strategies. 17. October 2018, 577-88.
In block caving strategies, it was found that vertical stress could in­ 15 Shen WL, Bai JB, Wang XY, Yu Y. Response and control technology for entry loaded
crease between 2 and 3 times its initial value when different unmoved by mining abutment stress of a thick hard roof. Int J Rock Mech Min Sci. 2016;100
(90):26–34.
zone widths were tested. On the other hand, with regard to panel caving, 16 Richardson MP. Area of draw influence and drawpoint spacing for block caving
the maximum vertical stress measured was found to be a function of the mines. In: DR Stewart DR, ed. Design and Operation of Caving and Sublevel Stoping
distance to the front of the draw zone and could reach values close to 2 Mines. 1981:149–156. New York.
17 Sahupala H, Brannon C, Annavarapu S, Osborne K. Recovery of extraction pillars in
times initial values. Knowing these magnitudes, not only can extraction the deep ore zone (DOZ) block cave, PT freeport Indonesia. In: Massmin 2008, 5th
policies be used to alleviate induced stress but also the maximum stress International Conference & Exhibition on Mass Mining. Lulea, Sweden; 9-11. June 2008,
that can be expected on the production-level pillars can be estimated in 191-02.
18 Bravo C. Recuperaci� on de reservas colapsadas en minas panel caving. Bachelor degree.
the beginning stages of a project. Then, support systems could be Santiago: Universidad de Chile; 2010 (in Spanish).
designed according to these requirements, thus avoiding problems like 19 Orellana M, Cifuentes C, Díaz J. Caving experiences in Esmeralda sector, El Teniente
those described by Sahupala et al., 17 Bravo, 18 Pierce, 20 and most likely mine. In: 3rd International Symposium on Block and Sublevel Caving. Santiago, Chile; 5-
6. June 2014:78–90.
in various other unpublished cases. 20 Pierce ME. Forecasting vulnerability of deep extraction level excavations to draw-
induced cave loads. Journal of Rock Mechanics and Geotechnical Engineering. 2019;11
Declaration of competing interest (3):527–534.
21 Janssen HA. Experiments regarding grain stress in soils. Zeitschrift Des Vereines
Deutscher Ingeniure. 1985;39(35):1045–1049. Translated from German by W.
Authors declare that we don’t have conflict of interest and all de Hustrulid and N. Krauland in Proceedings of the Massmin 2004. Santiago, Chile. p.
funding sources have been included. 293-00.
22 Walker D. An approximate theory for pressures and arches in hoppers. Chem Eng Sci.
1966;21:975–997.
Acknowledgment 23 Jenike AW, Johanson JR, Carson JW. Bin loads—Part 3: mass-flow bins. J Eng Ind.
1973;95(1):6–12.
This paper was (partially) funded by the CONICYT/PIA Project 24 Jaky J. Earth pressure – pressure in silos. In: Proceeding of the Second International
Conference on Soil Mechanics and Foundation Engineering. Rotterdam; 21-30. june 1948:
AFB180004. The authors gratefully acknowledge the support of the 103–107.
Block Caving Laboratory and the Advanced Mining Technology Centre 25 Pieper K. Investigation of silo loads in measuring models. J Eng Ind. 1969;91(2):
of the University of Chile. R. Go
�mez acknowledges the support of 365–372.
26 Nedderman RM. The method of differencial slices. In: Statics and Kinematics of
CONICYT PFCHA/DOCTORADO BECAS CHILE/2018 – 21180046. Granular Materials. Cambridge: University Press; 1992, 84-26.
27 Hoeg K. Stresses against underground structural cylinders. J Soil Mech Found Div.
References 1968;94:833–858.
28 Deutsch GP, Schmidt LC. Stresses on silo walls. J Eng Ind. 1969;91(2):450–457.
29 Munch-Andersen J, Ditlevsen O, Christensen C, Randrup-Thomsen S, Hoffmeyer P.
1 Flores G. Future challenges and why cave mining must change. In: 3rd International
Empirical stochastic silo load model. II: data Analysis. J Eng Mech. 1995;121(9):
Symposium on Block and Sublevel Caving. Santiago, Chile. June 2014:23–52, 5-6.
981–986.
2 Fairhurst C. Some challenges of deep mining. Engineering. 2017;3(4):527–537.
30 Vanel L, Claudin P, Bouchaud JP, Cates ME, Cl� ement E, Wittmer JP. Stresses in silos:
3 Laubscher DH. Cave mining-the state of the art. J S Afr Inst Min Metall. 1994;94(10):
comparison between theoretical models and new experiments. Phys Rev Lett. 2000;84
279–293.
(7):1439.
4 Trueman R, Pierce M, Wattimena R. Quantifying Stresses and Support Requirements in
31 Chou CS, Chuang YC, Smid J, Hsiau SS, Kuo JT. Flow patterns and stresses on the
the Undercut and Production Level Drifts of Block and Panel Caving Mines. Minneapolis,
wall in a moving granular bed with eccentric discharge. Adv Powder Technol. 2002;13
USA: JKMRC, The University of Queensland, Australia. Itasca Consulting Group;
(1):1–24.
2002.
32 Berntsen KN, Ditlevsen OD. Stochastic stresses in granular matter simulated by
5 Brown ET. Cave initation by undercutting. In: Block Caving Geomechanics, 2nd Edition,
dripping identical ellipses into plane silo. In: 8th International Conference on
the International Caving Study 1997 – 2004, JKMRC. The University of Queensland;
Applications of Probability and Statistics to Civil Engineering. 2000:57–64.
2007:229–287.
33 Colonnello C, Reyes LI, Cl� ement E, Guti�errez G. Behavior of grains in contact with the
6 Bartlett PJ, Croll A. Cave mining at premier mine. In: Massmin 2000 Proceedings.
wall of a silo during the initial instants of a discharge-driven collapse. Phys Stat Mech
Brisbane, Australia; 29 October – 2. November 2000:227–234.
Appl. 2014;398:35–42.
7 Callahan MF, Keskimaki KW, Fronapfel LC. Constructing and operating Henderson’s
34 Kobyłka R, Horabik J, Molenda M. Numerical simulation of the dynamic response
new 7210 production level. In: Massmin 2008, 5th International Conference &
due to discharge initiation of the grain silo. Int J Solid Struct. 2017;106:27–37.
Exhibition on Mass Mining. Lulea, Sweden; 9-11. June 2008:15–24.
35 Castro R. Study of the Mechanisms of Gravity Flow for Block Caving. PhD thesis.
8 Araneda O, Sougarret A. Lessons learned in cave mining at the El Teniente mine over
Australia: The University of Queensland; 2006.
the period 1997-2007. In: Massmin 2008, 5th International Conference & Exhibition on
36 Orellana LF. Evaluaci� on de variables de dise~
no del sistema de minería continua a partir de
Mass Mining. Lulea, Sweden; 9-11. June 2008:43–52.
experimentaci� on en laboratorio. MSc. Thesis. Santiago: Universidad de Chile; 2012 (in
9 Fernandez F, Evans P, Gelson R. Design and implementation of a damage assessment
Spanish).
system at Argyle Diamond’s block cave project. In: Caving 2010, Proceeding of the
37 Vergara P. Estudio experimental de flujo gravitacional en minería de panel caving.
Second International Symposium on Block and Sublevel Caving. Perth, Australia; 20-22.
Bachelor degree. Santiago: Universidad de Chile; 2016.
April 2010:65–81.
38 Lorig LJ. Relation between Caved Column Height and Vertical Stress at the Cave Base.
10 Lett JL, Brunton I, Capes GW, et al. Undercutting to surface breakthrough – Cadia
Final Report. International Caving Study, JKMRC and Itasca Consulting Group, Inc.;
East panel cave (stage 1). In: Seventh International Conference & Exhibition on Mass
2000.
Mining, Proceeding. Sydney, Australia; 9-11. May 2016:65–82.
39 Pierce M. A Model for Gravity Flow of Fragmented Rock in Block Caving Mines. PhD
11 Snyman L, Webster S, Samosir J. E48 cave extension at Northparkes. In: Seventh
Thesis. Australia: The University of Queensland; 2009.
International Conference & Exhibition on Mass Mining, Proceeding. Sydney, Australia; 9-
40 Knappett JA, Craig RF. Basic characteristics of soils. In: Carig’s Soil Mechanics. eighth
11. May 2016, 111-17.
ed. Spon press editorial; 2012:3–38.
12 Rojas E, Molina R, Bonani A, Constanzo H. The pre-undercut caving method at the El
41 Cho G, Dodds D, Santamarina JC. Particle shape effects on packing density, stiffness,
Teniente mine, Codelco Chile. In: Massmin 2000 Proceedings. Brisbane, Australia; 29
and strength: natural and crushed sands. J Geotech Geoenviron Eng. 2006;16(5):
October – 2. November 2000:261–266.
451–471.
13 Qui~nones L, Lagos C, Ortiz F, Farías E, Toro L, Villegas D. Undercut advance direction
42 Nedderman RM. The use of the kinematic model to predict the development of the
management at the North 3rd panel, Rio Blanco mine, Divisi� on Andina Codelco
stagnant zone boundary in the batch discharge of a bunker. Chem Eng Sci. 1995;50
Chile. In: 3rd International Symposium on Block and Sublevel Caving. Santiago, Chile; 5-
(6):959–965.
6. June 2014:91–97.

11