Professional Documents
Culture Documents
Pre-concentration
Mario Morin, Andrew Bamber, Malcolm Scoble
Department of Mining Engineering. University of British Columbia, Vancouver, Canada.
Abstract
The application of underground pre-concentration technology to narrow vein mining has the potential to change the
entire economics of bulk mining methods when used in narrow vein situations. The rejection of waste at or near the
face means that a higher-grade ore can be sent to surface. The waste rock can be reused directly as backfill. Swelling
of the muck through blasting means that only 60 to 65% can be used as fill. Planning and scheduling of the fill usage
is key to the success of this method. Through the use of simulation and system analysis, this paper examines the
planning and scheduling implications of underground pre-concentration when applied to traditional narrow vein
mining methods (cut-and-fill /drift-and-fill, longhole, AVOCA). It considers the logistical factors that represent
potential system bottlenecks that are key to understanding the correct implementation and control strategies.
Low
Medium
Medium
Medium
Medium
Medium
Medium
Medium
Shrinkage stoping
High
High
VCR
Shallow, thick or
irregular
High
High
Shallow, thick or
irregular
Surface, moderate to
steeply dipping
High
High
High
Low
Highest
Medium
Strip mining
Comments
Highly selective, typically not
mechanized, high cutoff grade
required
Selective, mixture of manual,
narrow seam equipment, high
cutoff grade required
Narrow thick seam
mechanized equipment, poor
extraction
Thick-seam mechanized
equipment, good extraction,
but requires fill
Thick-seam, mechanized,
moderate extraction, requires
fill
Bulk mining equipment,
requires good ground control,
unselective, dilution is a
problem
Bulk mining equipment,
requires good ground control,
unselective dilution is a
problem
Bulk mining, unselective,
low-grade
Surface blasting and
earthmoving equipment, no
selectivity, low grade.
Stripping ratio is a problem
Bulk surface earthmoving
equipment, (draglines), no
selectivity, low grade
The SADT model always begins with a context diagram that explains the purpose of the model as well as its
boundaries. Every SADT model is built from a single viewpoint and SADT models will look different depending on
the viewpoint. Figure 1 illustrates the context diagram for narrow vein mining with pre-concentration. Inputs, in this
case Production Requirements and Mining Resources, comes in from the left. Outputs, namely Hoisted Ore and
Hoisted Waste, are seen exiting to the right. What makes SADT different from other techniques is the consideration
of constraints (referred to as controls) and resources (referred to as mechanisms). Controls, seen coming in from the
top of the diagram, represent constraints on the process (shown in the box). M echanisms, seen coming in from the
bottom of the diagram, represent resources that will be used by the process in transforming the inputs into the
outputs.
production cycle? The addition of underground pre-concentration increases the complexity of the process and thus
increases the risk of a series failure, reducing the overall reliability of the process. These questions are best answered
through process simulation. Simulation can be used to model the overall process based on system variables such as
volumes, rates, capacities, speeds, distances and reliabilities.
Figure 3 illustrates the structured analysis for the Pre-concentrate Muck activity. Note that the pre-concentration
activity has three storage activities to store incoming Unsorted Muck and outgoing Concentrated Ore and Sorted
Waste Rock streams. These storage activities will act as surge bins to buffer the variability in pre-concentrating the
muck. The Pre-concentrate activity is under the control of several variables including Equipment Productivity &
Reliability, Mineralogical Characteristics of the muck, the Concentration Performance & Efficiency and the Preconcentrator Feedback.
pre-defined building blocks that are linked to create a model (Krahl, 2002; Imagine That!, 2004). The structured
analysis model created earlier is easily transferable to a simulation model.
To create a simulation model for narrow vein mining requires an understanding of the actual physical layout and
characteristics of the orebody. These characteristics include orebody depth, shape and spatial alignment as well as
mineralization grade and dispersion as well as rockmass strengths and behaviour. These characteristics will define
the mine layout and the mining method(s) that will be used to extract the resources. In the Canadian context, most
narrow vein orebodies tends to be vertical or sub-vertical deposits that are mined using cut-and-fill (and its many
variants, i.e. mechanized, cut-and-fill, drift-and-fill, and underhand cut-and-fill), longhole, and uppers (e.g. the
AVOCA method (Duke, 1981)). If the deposit is flat or near horizontal, potential mining methods could include
hardrock longwall and room-and-pillar. Based on the shape, grade, orientation and depth of the typical narrow vein
orebody, as well as accounting conservatively for possible factors such as weak ore and hangingwall strength, the
methods indicated for such an orebody using the Nicholas and modified Nicholas Procedure (Pakalnis et al) indicate
longwall, drift and fill or cut-and fill methods would be suitable.
Because waste rock is continuously produced, mining methods that provide an opportunity to quickly place the
waste rock or backfill back into the stope are better suited to underground pre-concentration. AVOCA and roomand-pillar have an advantage in this respect in that many small cuts or rooms are taken individually, thus reducing
the intermediate rock storage volume. However, both AVOCA and room-and-pillar operations would require the
construction of fill barricades for containing the fresh uncured cemented backfill.
Methods like cut-and-fill typically extract long cuts and the waste would have to be stored prior to being used. Of
course, if several cut-and-fill stopes were available and scheduled in such as way as to have at least one stope always
available for backfilling, then the need for intermediate storage could be considerably reduced perhaps even
eliminated. Similarly, the selection of initial stopes where the orezone is thinner will ensure the early availability of
a fill stope. However, having several active stopes at once means that the pre-development and mining equipment
requirements could be more extensive. As well, to minimize material travel times, development, production and
disposal areas should ideally be close together. This closeness could result in greater equipment congestion and
interference problems. These issued must be balanced in the pre-concentration scenario.
4. Simulation Model
For the purpose of this paper, the authors have chosen to simulate the extraction of a narrow-vein stringer orebody,
typical of a Sudbury Igneous Complex Footwall deposit. The hypothetical deposit is deep, of moderate grade, and is
associated with other deposits which have been mined previously to a depth of 1500m, thus principal access to the
orebody is already provided. Secondary access to the orebody is via a sub-vertical shaft from the 1500 2500m
level, with a haulage ramp for horizontal access to the deposit..
The cut-and-fill method has been used as a basis for the simulation. From a simulation perspective, it does not
matter what mining method is used provided ore can be produced at a certain rate. The method of access and layout
of the mining method however, is important. The orebody is deep and is reached via a sub-vertical shaft. A decline
is used to access the orebody from the bottom of the sub-shaft. Several production levels have been advanced into
the ore zone. The bottommost level is connected to an orepass, and a truck system will be used to haul the muck to
the pre-concentrator and backfill plant located close to the shaft. Cut-and-fill mining will be overhand (moving
upward) and backfill will be provided to the stope via a fill borehole connected to the uppermost level. Concentrated
Ore and Excess Waste Rock will be hoisted to surface (see Figure 2). Sorted Waste Rock will be sent to the backfill
plant to become Rockfill and later on Placed Backfill.
For the simulation, the production target was set as 1,000 tpd. Each stope will be advanced using a 4 metre wide, 3
metre high, 4 metre deep round. Each round produces nominally 48 m3 of muck or 148 tonnes. Operation was
assumed to be 12 hours per shift with two shifts per day. Blasting will take place every shift. We will also assume
that only four stopes plus one spare stope are all that is required to meet the production objective of 1000 tpd. The
model also resets once the fill-cycle is completed, allowing the simulation of eight complete mining and filling
cycles for each mining level.
It is important to define the operating characteristics of the pre-concentrator. The nominal target is to reject 60% of
the tons received at a recovery of 95% or better. Concentration Performance & Efficiency can be changed
depending what is happening within the stoping and rock storage bins. Muck storage capacity is available in the
orepass and in the pre-concentrator plant. If the mine is muck bound, the sorting efficiency of the pre-concentrator
can be increased. If backfill is needed, the efficiency can be reduced. This muck storage provides buffering between
the different processes and help smooth out the variations that naturally occur in a production setting. Figure 5
illustrates the basic design of the pre-concentrator. An important design criteria in the adoption of underground preconcentration is the provision of a bypass circuit should the pre-concentration facility become indefinitely
unavailable. The fundamentals of the underground pre-concentration system are described more fully in the paper on
Reducing Selectivity in Narrow-vein Mining Methods.
The basic principles of the mining simulation are described below:
A round is produced through the drilling and blasting of a stope. The ore is mucked by LHD in batches of 15t, and
dumped at the orepass over an average distance of 280m. Another round may not be blasted until the mucking cycle
is complete in each stope. A limit of one blast per 12h shift is also imposed. Stopes 1-4 are blasted on successive
shifts at the beginning of the simulation; the random delays built into the model means that this blasting order can
become staggered over the running of the simulation. A stope is mined out after 24 rounds have been blasted. 5% of
the ore blasted is rejected at the orepass by static grizzly. Rejects are transferred directly to the fill silo. Ore from the
orepass is collected by haul truck in 40t batches and delivered over a haul of 3km to the pre-concentration feed silo.
The pre-concentrator separates the ore into concentrate and waste. Concentrate is hoisted to surface via a hoisting
system with an appropriate reliability. Waste is delivered to the waste silo and hauled to the backfill silo over a
distance of 1km via 40t haul truck. Backfill is delivered to the stopes via a fill raise on each level. Fill is collected
from the fill raise on each level and returned to the stope by the production LHDs. Initial simulations showed that
due to the low utilization of the scoops, an additional LHD was not required for this task. Stopes are filled when fill
tons = 60% of the tons mined. A filled stope is reset by the simulation, and the mining cycle will commence for a
second iteration. The basic modeling parameters are listed in Table 2.
Table 2 Basic Parameters Used for Simulation
Parameter
Stope dimensions
Orebody thickness
Face advance
m3 / round
Ore density
t / round
Rounds per cut
Mucking
Mucking distance
Orepass capacity
Hauling
Haul distance
Pre-concentration
Feed storage capacity
Waste rejection
Metallurgical recovery
Concentrate production
Concentrate storage capacity
Backfill production
Backfill storage capacity
Value
4m x 3m x 4m
70m
4m
48
3.4 t / m3
144
24
4 x Wagner ST8, capacity 15 t, max speed 9 km/h
270 m
3000 t
2 x B40 haul trucks, capacity 40t, max speed 24 km/h
3 km
1000 tpd
1000 t
60%
95%
400 tpd
1000 t
600 tpd
1000 t
A feature of the model was the allocation of reliabilities to each production element in the model. Stopes, LHDs,
orepasses, haul trucks and the pre-concentrator were allocated realistic reliabilities. A base case of 85% reliability
for each step in the process (with a random distribution of mean-time-before-failure) was established. Simulation
was then used to explore the basic productivity for the system, the impact of pre-concentration and the continuous
generation of backfill on the mining cycle, and the impact of increasing and decreasing the reliability of each
production element in turn. The maximum backlog of ore at each point in the mining and processing system was
also recorded during the simulation.
The details of the mine model as described above have been captured in a discrete-event simulation model using
ExtendTM. Blocks from the generic and manufacturing modeling libraries have been used in constructing the
model. Figures 4 to 8 are block-level representations of the mining and pre-concentration processes in Extend. The
Extend model created is a steady-state discrete event model (Banks et al, 1999). It is steady-state because it does not
consider the initial development work to create the initial or start-up stopes or what happens to the model at the end
of the mine-life.
Figure 4 illustrates the ore generation process as an Extend simulation representation. An ore block containing
approximately 10,000t is generated on each mining level, based on the typical dimensions of the Onaping Depth
orebody and the mining width chosen. Four active stopes are generated in the ore block, containing between 1500
and 2500t depending on the thickness of the orebody; face dimensions are 4m x 3m with 3m rounds. Stope # 1 is
chosen to be the smallest in the ore block in order to make a fill stope available as soon as possible. Simulation starts
at time 0, a blast is initiated during a 12h shift and 144t of material is made available in the stope. Another blast may
not be initiated in the stope until the round has been mucked; a delay is allowed after each mucking cycle to allow
for drilling and blasting. Blasts in stopes 2, 3 and 4 are successively delayed by a shift. The stope availability block
generates a random distribution of MTBF around the percentage availability set for the stope. Availability in each
case can be adjusted between 0 100%. The backfill status block monitors the status of the ore block in order to in
initiate fill once the stope is exhausted, as well as signal the completion of backfill operations; stopes are regenerated
after the backfill cycle in order to simulate continuous mining operations.
Generate
stope tons
use
Initiate
Blasts
#
u
C
Count
Blasts
#
change
Stope
demand
use
Set V
#
u
reset3
L W
3a done
down
Stope3
reset3b
C
Count
STOP
change
blast
Status 3
Y
N
A
B
And
Y
A
A
B
a=0
LHD3
B
Y
Y
Full 3
Stope
Availability
Backfill
Status
Make
round
available
in stope
The ore generated in each blast is batched into 15t lots and made available to the LHD for collection. The mucking
route length and the haul speed is defined in the model; haul speed is on average 9km/h both ways, however this can
be varied. The LHD also has an availability based on a random distribution of MTBF. During the dump function,
ore is unbatched back into individual tons and delivered to the orepass. The LHD runs continuously until either a
failure occurs or the ore in the round is completely mucked. This process is shown in Figure 5.
Stope
Loads
No.3
C
Count
#
Var
C
Count
STOP
down
demand
out3
B
Y
Var
demand
A
a=1
Info
C
Count
Var
demand
demand
L W
C
Count
Var
n
LHD3
B
Y B
#3
144t round
available in
stope
reset3b
A
B
And
a<=14
out3
LHD haul to
orepass
Batch ore
to
15t LHD
Figure 5 Muck Haulage Simulation Model
Length
orepass
batch
Haul Route
Speed to Orepass:
To Orepass
batch
Info
#
F
L W
C
Count
Stope 1
?
0.05
reset4
Speed to Preconcentrator:
Length
0.95 a
b
select
To Backfill
Items in queue:
Var
Loads In:
demand
C
Count
C
Count
r
Throw
L W
Orepass:
autofill
Loads Out:
#
Waste Rock:
Status 4
reset4b
reset4
Ore Mined:
Orepass
Var
demand
C
Count
r
ore
n
demand
Status 4
Info
Var
Stope 2
reset4b
Ore from Stopes 1 4 is delivered by LHD to the orepass. A maximum of 5% oversize is scalped from the ore and
diverted to the rockfill bunker. The orepass has a maximum capacity of 500t. Ore is batched from the orepass into
40t lots for collection by the haul trucks. There are two haul trucks operating in tandem over a haul route of 2750m,
comprising 571m vertical gain over 12 inclination; the haul speed is 24 km/h on average both ways, and the trucks
also have an availability determined by a random distribution around the MTBF. Ore is dumped to the preconcentrator feed silo at the end of the haul route. Figure 7 illustrates the simulation model representation for preconcentration.
Hoist concentrated
ore to surface
stockpile
Reject 60%
waste on
average
ore
#
F
Preconcentrator
n
L W
Var
use
speed
demand
Length:
Var
F
demand
F
L W
Var
0.4
0.6?
L W
select
demand
Length:
F
Arrivals:
L W
Top:
Backfill Silo
Departures:
Bottom:
backfillout
Total:
backfillout
Pass rejects to
silo for
collection
#
Var
Info
n
demand
backfill
10
filledbstopes
Stope empty?
Catch
Full
Full 2
Full 3
Full 4
LHD's out:
No.1
No.2
No.3
Stope1a
Stope1b
Stope1c
No.4
Stope1d
Stope1e
Stope2a
filledbstopes
Preconcentrated
Var
F
n
L W
rockpass
Waste
delivered to
rockfill pass
filledbstopes
Var
demand
Stope2b
Stope2c
Stope2d
Stope3a
Stope3b
demand
Stope2e
Stope3c
Stope3d
Stope3e
filledbstopes
Stope4a
Stope4b
Kiruna #1
filled?
#1
#3
Stope4c
#2
Stope4d
Stope4e
#4
ASR
AGV
Info
speed X Y
Preconcentrated
backfill
Speed to Backfill Silo:
Speed to Rockpass:
Collect waste
and haul to
rockfill pass
11
% of Max
32%
100%
16%
46%
33.6%
32%
2.2%
24%
86%
72%
67%
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
40
50
60
70
80
Availability
90
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
100
Stope
LHD
Haul
Utilization
Productivity
Preconc
Sutil
LHDUtil
Hutil
Putil
6. Conclusions
From the simulation results a general trend can be seen. The basic simulation gives a productivity of 230 tpd for the
mining system as simulated. A drop in stope availability significantly drops the productivity of the system, and an
increase in availability of the stope substantially increases the productivity of the mining system. However, it should
be pointed out that if the stope is unavailable, it results in a lost blast. In a real mining situation, it is often possible to
make up for a lost blast but our simulator cannot account for this at this time.
LHD utilizations are moderate in all cases - utilization increases if stope availability increases and vice versa. Haul
truck utilization is low in all cases, indicating that two 40t haul trucks are overspecified for the duty, and could be
replaced by 20t units. Maximum orepass capacity varies with the combined productivity of the stopes and LHDs.
Rockfill capacity reaches a maximum of 2800 t when the overall system productivity is high, and is generally low
when the system is producing at a low capacity.
The impact of varying the performance of the LHD, haul truck, and pre-concentrator elements is insignificant when
compared to the impact of varying the productivity of the stope. The inherent 60% backfill generated by the preconcentrator on average requires a maximum surge of 3200t which is catered for between the waste silo of the preconcentration plant, the haul truck capacity and the capacity of the rock pass delivering fill to the stopes. The
12
continual requirement for fill is easily catered for by the LHD stope fleet even at the maximum production rate of
750 tpd. Backfill efficiency is high compared to stoping efficiency, and backfill rates are far higher than the basic
mining rate. The model also indicated that significant downtime can be incurred in the system due to lack of fill
during simulation runs where stope productivity was low, or haul truck or pre-concentrator productivity was low,
and thus the supply of backfill was interrupted.
The simulation demonstrates a number of critical issues in integrated underground mining and processing:
A continual supply of backfill generated in mining can be accommodated in the mining cycle
The generation of backfill is critical to the productivity of the system - interruptions to the mining cycle
occur when there is a shortage of backfill in the rock pass
The productivity of the mining systems is driven primarily by productivity at the stope, and in decreasing
proportions by the productivity of activities distant from the stoping activity
The low utilization of the haul trucks is indicative that only 2 such units would be required for the
production rate in the simulation
Reasonable variations in the availability of haul equipment result in increases in surge requirements at
various points in the system. However, the overall productivity of the system is not affected significantly
The simulation demonstrates clearly proof of concept of the underground mining and processing system, as well as
highlighting key productivity issues of stope and mucking system availability over downstream system component
availabilities.
Acknowledgements
The authors would like to acknowledge the contribution of Mr. Chris McCann of UBC Mining Engineering for his
support and commitment with the Extend simulation modeling.
References
BANKS, J., Carson, J.S., Nelson, B.L. (1999). Discrete-Event System Simulation. 2nd ed. Prentice-Hall. 548p.
BRUNNER, D., T., (2000). Simulation of Underground Mining Operations, Underground Mining Looks to the
Future, pp 705 711
CORKAL, T. and MACKENZIE, A. (1997). Simulation of Narrow Vein Mining. Proc. of the 13th Mine Operators
Conference Back to Basics Practical Use of Technology. Sudbury, Ontario. pp 130-140.
DUKE, J.D. (1981). AVOCA Mining Method at Texasgulf No. 2. Mine. Proc. of the 5th Annual Underground
Operators Conference. Canadian Institute of Mining, Feb 16-18. Sudbury, Ontario. 34 p.
FIPS PUBS (1993). Announcing the standard for integration definition for functional modeling (IDEF0). Federal
Information Processing Standards Publications, 183.
IMAGINE THAT! (2004). Extend Overview. Internet URL: http://www.imaginethatinc.com/prods_overview.html,
Web paged accessed June 2004.
KRAHL, D (2002). The Extend Simulation Environment. Proc. of the 2002 Winter Simulation Conference.
Yucesan, Chen, and Snowdon, eds. Internet URL: http://www.informs-cs.org/wsc02papers/026.pdf , San Diego,
California.
MARCA, D.A. and McGOWAN, C.L., 1993. IDEF/SADTtm Business Process and Enterprise Modeling. Pub. by
Eclectic Solutions Corp. San Diego, CA. 392 pp.
ROMER, W. and LAWLER, I. (2001). Integration of Mine2-4Dwith AutoMod on Onaping Depth Project.
Proceedings of the 15th Mine Operators Conference. Competitiveness through Operator Innovation. Sudbury,
Ontario.
13
ROSS, D.T., 1977. Structured Analysis (SA): A Language for Communicating Ideas, IEEE Transactions on
Software Engineering, Vol.3 No.1, pp. 16-34.
STURGUL, J.R., SMITH, M.L. (1993). Using GPSS/H to Simulate Complex Underground Mining Operations.
Proc. of the 2nd Int. Symp on Mine Mechanization and Automation, Lule, Sweden. pp. 801-810.
SWAN, G., and CAMPBELL, J. (2001). Onaping Depth: The Operational Challenges. Proceedings of the 15th Mine
Operators Conference. Competitiveness through Operator Innovation. Sudbury, Ontario.
VAGENAS, N. et al. (1995). Simulation for Design, Planning and Control in the Automated Mine. Proc. of the 1995
Mine Planning and Equipment Selection Conf., Singhal et al (eds.). Rotterdam: Balkema. pp. 271-276.
VAGENAS, N., et al. (1996). Simulation of Underground Hardrock Mining Using AutoMod. Proc. of the 1st Int.
Symp. of Mine Simulation via the Internet. Dec. 2-13. Rotterdam: Balkema.
VAGENAS, N., et al. (1998). Simulation of Teleremote Mining Systems. Proc. of the CIM98 Conf. Montreal,
Canada.
YAZICI, H.J., et al. (1999a). Simulation-Based Planning of Tele-operated Mining Processes. Proc. of the 5th Symp.
on Mine Mechanization and Automation Proc. of Telemin 1, June 14-17. Sudbury, Canada.
YAZICI, H.J., et al. (1999b). Simulation of Mining Method Alternatives in Underground Hardrock Mining. Proc. of
the CIM99 Conf. Calgary.
14