University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Colorado School of Mines
|
• Strebinger, C., Fig, M., Pardonner, D., Treffner, B., Bogin, Jr., G.E., and Brune,
J.F., “Investigation on the Overpressure Produced by High-Speed Methane Gas
Deflagrations in Confined Spaces”. SME Annual Conference and Exhibit.
February 2018.
Single Obstacle Experiments – Section 4.2.1, Section 4.2.2, and Section 4.2.3
➢ Purpose: Ignitions typically occur in or around the gob area, which has varying types of
rock rubble with different rock pile porosities, geometries, and void spacing.
➢ Outcomes:
• OEI: Cage obstacle induces turbulence in nearby unburned gases, resulting in an
increase in flame velocity by 18%.
• OEI: Decreasing the void spacing of an obstacle wall geometry from 73%, wall
H=3.8cm, to 13%, wall H=9.8cm, increases flame velocity by 17%.
• OEI: Increasing the obstacle wall length by 200% increases flame velocity by 12%.
• OEI: Increasing porosity from 67% to 77% increases flame speeds by 11%.
• OEI: Moving the obstacle location from 37cm to 62cm from the open end decreases
the relative velocity increase from 18% to 15%. Moving it to 87cm decreases the
relative velocity difference to 11% from the open case.
• CEI: A single obstacle wall with a void spacing of 73%, H=3.8cm, increases
downstream flame velocities by 27% and peak overpressures by 62%.
➢ Impact: These outcomes help determine which simulated gob parameters most affect the
flame – the amount of blockage has the greatest effect on both flame velocity and peak
overpressure. Additionally, location of obstacles relative to ignition location impacts
flame front propagation velocity significantly and depends on ignition location as well.
➢ Novelty: Other researchers have experimented with baffles, rings, and some solid
obstacle geometries. These experiments are novel in that the obstacles are made of solid
spheres, which more closely represents the turbulence induced by obstacles found in a
mine. For example, this research has shown that flame propagation trends over a solid
rectangle is fundamentally different than a flame passing over spheres or other obstacles.
This is because the spheres and obstacles induce more turbulence in the shear zone and
distort the flame front. Additionally, trends of increasing BR agrees with other
278
|
Colorado School of Mines
|
researchers (Chapman & Wheeler, 1926; Moen, Lee, Hjertager, Fuhre, & Eckhoff, 1982),
but Chapman and Wheeler (1926) found that increasing the wall length decreased
velocity. However, their experiment uses thick brass obstacles, which may be acting as a
heat sink and thereby decelerating the flame (Chapman & Wheeler, 1926). Another
novelty of these experiments is the exhaustion of obstacle parameters experimented in
this research – i.e. other researchers may only investigate blockage ratio, not porosity or
other factors. This allows for a more comprehensive validation of the on-going
combustion model.
➢ Presented at the 16th North American Mine Ventilation Symposium, Golden, CO, June
2017:
o Strebinger, C., Fig, M., Blackketter, K., Walz, L., Bogin, Jr., G.E., Brune, J.F.,
and Grubb, J.W. “A Fundamental Investigation of Simulated Gob Configurations
on Methane Flame Propagation”. 16th North American Mine Ventilation
Symposium, June 2017.
Simulated Gob Bed Experiments – Section 4.2.4
➢ Purpose: Longwall coal mines often have piles of rock rubble and the walls are made of
rock, which is why it is important to understand the different effects of a single obstacle
versus a pile of rubble on methane gas deflagrations.
➢ Outcomes:
• OEI: Simulated gob bed location has only a small impact on flame velocity and no
impact on overpressure for the heights and lengths gob beds investigated.
• CEI: Although a single obstacle wall of height 3.8cm increases flame velocity by
27% and peak overpressure by almost 62%, at the same location, a simulated gob bed
with H=2cm, L=30cm increases flame velocity by 32% and peak overpressure by
70%.
• CEI: Increasing the length and height of the simulated gob bed has less of an effect on
flame velocity (<5%) than peak overpressure. Peak overpressure increases 27% when
the height of the gob bed L=15cm was increased by 1cm (reducing the blockage ratio
from 96% to 89%).
279
|
Colorado School of Mines
|
• CEI: A simulated gob bed with height 2cm and length 15cm results in greater flame
velocities and peak overpressures than a single obstacle with height 3.8cm and length
6.35mm when located near the open end of the reactor.
• The process of flame acceleration between a single thin obstacle and simulated gob
bed is significantly different; a single obstacle results in a downstream vortex behind
the obstacle which can trap unburned gases which can help accelerate or decelerate
the flame. The simulated gob bed results in a turbulent boundary layer which interacts
with the bed similar to a porous media – resulting in a feedback loop, burning gases
in the porous media increase temperatures which increases combustion rates and
accelerates the main flame brush.
➢ Impact: These results help show how important it is to understand the impact of obstacle
geometry on methane acceleration mechanisms and not just focusing on the impact of
blockage ratio. The results also demonstrate how a longwall coal mine environment can
inherently exacerbate a methane gas deflagration and why it will be important to
discretely model the complex geometry of a mine; a wall has a significantly different
effect on flame acceleration than a pile of porous rock rubble.
➢ Novelty: There has been a significant amount of research of flame dynamics across a
single obstacle or through a porous media. We have only found one reference which
investigates flame acceleration (to DDT) across a porous media (Lee, Knystautas, &
Chan, 1985), but the porous media was across the entire length of the combustion reactor
because the researchers were mainly investigating DDT. This research is novel in that it
investigates the effects of the simulated gob bed height, length, and location on methane
flame dynamics and pressure generation.
➢ Presented at the 11th International Mine Ventilation Congress, Xi’an, China, September,
2018:
• Strebinger, C., Bogin, Jr., G.E., and Brune, J.F., “A Fundamental Study of High-
Speed Methane-Air Deflagrations Across Simulated Gob Walls and Sphere
Beds”. 11th International Mine Ventilation Congress, September 2018.
280
|
Colorado School of Mines
|
In-gob Ignition Experiments – Section 4.3
➢ Purpose: Methane gas explosions can occur near the gob and may travel to the longwall
face. Therefore, it is important to understand how an ignition between simulated gobs
may affect methane flame dynamics.
➢ Outcomes:
• The surface topology of granite rock results in immediate downstream flame
velocities 28% greater than smooth spheres.
• The granite rock slightly increases peak overpressure by 14% and sustains pressure
oscillations for 43% longer than the smooth glass spheres.
• All experiments resulted in hydrodynamic instabilities which inverted the flame front
(tulip flame), increasing combustion rates and pressure.
➢ Impact: These results show that the surface topology and roughness of the obstacle can
have a significant effect on methane gas deflagrations and can increase the peak
overpressure and duration of pressure oscillations. This is important because sustained
pressure oscillations can damage ventilation controls, harm human bodies, and stress
mine structures. The results also show that modeling rock rubble as smooth spheres is
insufficient and will require more detailed modeling of the gob area.
➢ Novelty: There has only been one other researcher we have found who has ignited a
combustible mixture between obstacles (van Wingerden & Zeeuwen, 1983). However, in
their experiments the obstacles were different and the flame could propagate freely in all
directions, whereas in a real longwall mine environment the flame would expand
preferentially in a horizontal direction. The experiments here are novel in that the flame
dynamics are different due to the preferential direction of flame propagation.
Additionally, we have found no experiments where researchers have used actual rock
rubble as an obstacle, which can have significant differences compared to smooth
obstacle especially in the wake zone behind the obstacle.
➢ Presented at the SME Annual Conference, Minneapolis, Minnesota, 2018:
• Strebinger, C., Fig, M., Pardonner, D., Treffner, B., Bogin, Jr., G.E., and Brune,
J.F., “Investigation on the Overpressure Produced by High-Speed Methane Gas
Deflagrations in Confined Spaces”. SME Annual Conference and Exhibit.
February 2018.
281
|
Colorado School of Mines
|
Ignition Location Experiments – Section 4.4
➢ Purpose: Since methane gas explosions can occur in a variety of different locations in a
mine, it is important to understand how methane flame propagation velocities and
overpressures change depending on ignition location.
➢ Outcomes:
• As ignition moved further from the open end, towards the middle of the reactor,
maximum flame velocities and overpressures increase.
• Ignition in the middle of the reactor results in pressures 270% greater than CEI
(ignition 11cm from closed end) and flame velocities only 45% less than CEI.
• The pressure trace of ignition in the middle of the reactor consists of high-frequency
oscillations with multiple modes.
• The resulting flame from ignition in the middle 2/3rds of the reactor is highly
turbulent and similar to a pulse jet.
➢ Impact: Ignition location impacts methane flame propagation and pressure generation the
greatest of all experiments performed. High frequency oscillations have a major impact
on the structural integrity of mine structures and can destroy ventilation controls.
Acoustics play an important role in distorting the flame front and accelerating
combustion rates.
➢ Novelty: Some researchers have ignited mixtures at the open and closed ends of their
combustion chambers, and some in the middle. But none have thoroughly investigated
the impact of varying the ignition along the length of the reactor, which was shown in
this manuscript to have differing impacts on propagation velocities and explosion
pressures. Additionally, no one has reported flame speeds, pressures, and shown images
of the propagating flame altogether.
➢ Presented at the SME Annual Conference, Minneapolis, Minnesota, 2018:
• Strebinger, C., Fig, M., Pardonner, D., Treffner, B., Bogin, Jr., G.E., and Brune,
J.F., “Investigation on the Overpressure Produced by High-Speed Methane Gas
Deflagrations in Confined Spaces”. SME Annual Conference and Exhibit.
February 2018
282
|
Colorado School of Mines
|
Confinement Experiments – Section 4.5
➢ Purpose: An explosion from the gob area towards the longwall face is extremely different
from an ignition on the longwall face, specifically the degree of confinement. These
experiments help further understand the impact of end conditions and confinement on
methane gas deflagrations.
➢ Outcomes:
• CEI: Relief holes have no difference on methane flame dynamics and overpressure.
• Port 2: The number of relief holes on the closed end of the reactor has less of an
effect on methane flame propagation velocities, but more of an effect on
overpressures.
• Port 2: A fully confined end condition results in overpressures at 97% greater than
with a small relief (D=1.2cm).
• Port 2: A fully confined ignition sustains pressure oscillations almost four times
longer (no relief holes).
➢ Impact: These results demonstrate how small changes in closed end conditions can lead
to large changes in methane flame dynamics. Similar to the ignition experiments, these
experiments further demonstrate the need to accurately predict overpressure and acoustic
impacts. For longwall coal mining, these results demonstrate the difficulty in
understanding severity of a methane gas explosion.
➢ Novelty: There has been a significant amount of research on the impact of relief on flame
propagation, though the majority of previous work on pressure relief is done with a single
opening (with the other end closed) and only one research group has looked into two
openings opposite each other (Guo, Wang, Liu, & Chen, 2017). As of yet, we have not
come across any researchers who have varied the relief and varied ignition location
together.
➢ Presented at the SME Annual Conference, Minneapolis, Minnesota, 2018:
• Strebinger, C., Fig, M., Pardonner, D., Treffner, B., Bogin, Jr., G.E., and Brune,
J.F., “Investigation on the Overpressure Produced by High-Speed Methane Gas
Deflagrations in Confined Spaces”. SME Annual Conference and Exhibit.
February 2018
283
|
Colorado School of Mines
|
Box Reactor Experiments – Section 4.6
➢ Purpose: EGZs can exist in many areas of a longwall coal mine, including the gob area
(Karacan, Ruiz, Cote, & Phipps, 2011). Ignitions may occur in the gob area, often due to
falling rock and rock-on-rock friction and the resulting methane gas explosion can pose a
serious risk to miners and cause serious damage to mine structures and equipment
(Brune, 2014). To better understand how a flame might propagate in a rectangular
reactor, an experiment box was setup and experiments were performed with and without
a porous media consisting of rock rubble.
➢ Outcomes:
• OEI v CEI: In the rectangular box, a confined ignition resulted in significantly faster
flame speeds than an unconfined ignition, further confirming previous experimental
results (Section 4.1).
• Flame propagation in the experimental box showed residual burning in the corners of
the enclosure, similar to other researchers (Solberg, Pappas, & Skramstad, 1981).
• When a porous media is present, the flame tends to move faster through the porous
media than through the open passageways. This is because the obstacles induce fluid
motion in the nearby gases, resulting in an increase in unburned gases to the flame
front, accelerating the flame.
➢ Impact: It is often through that an ignition in the gob will be quenched, however these
results show that depending on the void spacing and porosity of the gob, the flame can
actually accelerate through the gob before propagating in open corridors/tunnels.
➢ Novelty: Researchers investigating flame propagation through obstacles have
investigated blockage ratio, obstacle, spacing and other configurations as discussed in
Section 2.6. However, these researchers have only found one group that has
experimentally investigated flame propagation through an array of obstacles (van
Wingerden & Zeeuwen, 1983), but the setup had obstacles in the entire experimental
reactor. These experiments include both a porous medium and also open space for the
flame to propagate, providing more information on how a flame may propagate in a semi-
obstacle filled environment.
➢ Presented at the North American Mine Ventilation Congress, Montreal, Canada, 2019.
Paper was also nominated to be included in the CIM Journal:
284
|
Colorado School of Mines
|
• Strebinger, C., Bogin, Jr., G.E., Brune, J.F., “CFD Modeling of Methane Flame
Interaction with a Simulated Longwall Coal Mine Gob”. North American Mine
Ventilation Congress 2019.
8.2 Summary of Impactful Modeling Results
Modeling the Spark– Section 5.1.3
➢ Purpose: Modeling the initiation of combustion is difficult and this research has looked at
two main methods of modeling the initial spark kernel. The first method is a heat flux
model which models the electrodes as an aluminum spherical source with a constant heat
flux to the surroundings, referred to in this text as EM. The second method uses the
ANSYS Fluent Spark Model which was originally designed for spark ignition engines,
referred to in this text as SM.
➢ Outcomes:
• OEI: Both the EM and the SM fail to capture the flame front propagation velocity
trends from experiments. This is due to the model’s difficulty capturing buoyancy
which is a diffusion dominated process.
• OEI: Although the EM and SM do not capture the flame velocity trends, they both
match the maximum flame front propagation velocity within 5% for the lean and
stoichiometric mixtures.
• OEI/CEI: The SM takes significantly less simulation time than the EM; over 30%
savings for an OEI and over 80% savings for a CEI.
• A major benefit of the SM is that the spark can be located anywhere in the domain
without remeshing the entire domain.
➢ Impact: Current 2D models have less than half a million cells, but at the large-scale,
reducing simulation time is of the utmost importance and the SM should be implemented.
However, if researchers are interested more in accuracy, the EM may be more
appropriate, but requires the user to mesh the spark ahead of time. For applications to
longwall coal mining, researchers are more interested in moving the ignition location and
thus, constantly remaking and remeshing the domain is not ideal – therefore the SM
should be used.
285
|
Colorado School of Mines
|
➢ Novelty: Initiating combustion is difficult to model and this work shows the comparison
of two methods and discusses the advantages and disadvantages of both. Previous
researchers have not provided significant details on spark initiation and sensitivity
analysis along with reasonings behind using their methods.
Impact of Spark Model Inputs – Section 5.1.3
➢ Purpose: As discussed, researchers focused on initiating combustion using the ANSYS
Fluent Spark Model. An investigation was undertaken to understand the most influential
SM input parameters for this problem by investigating the impact of spark duration,
initial kernel diameter, and expansion model.
➢ Outcomes:
• The initial kernel diameter had the largest impact on flame propagation.
• Smaller initial kernel diameters led to slower FFPV, but the total pressure was not
as significantly impacted.
➢ Impact: These results are important because researchers must understand the limitations
and inputs to the ANSYS Fluent Spark Model which can lead to almost a 35% difference
downstream of the kernel.
➢ Novelty: This study gives a good overview of how small changes in model assumptions
can impact results, which not all researchers present when discussing model inputs.
Impact of Spark Ignition Energy – Figure 5.26, Figure 5.27, Figure 5.28, Table 5.7
➢ Purpose: In an underground longwall coal mine, a combustion event can be initiated by
rock-on-rock friction, rock-on-metal friction, hot streaks from metal-on-metal, or
spontaneous combustion among some. Methane air mixtures typically need 5mJ of
energy to initiate combustion, however some of the ignitions in an underground coal
mine can be much larger than 5mJ. Therefore, researchers investigated whether or not
ignition energy has an impact on methane flame propagation using the 2D 12cm diameter
model.
➢ Outcomes:
• Below 1J of ignition energy, the methane flame was unaffected by ignition
energy.
286
|
Colorado School of Mines
|
• An ignition energy of 1kJ increased flame speed and reduced flame time of
arrival.
➢ Impact/Novelty: The 2D model predicts ignition energy can impact flame propagation,
but not at ignition energies typically found in a longwall coal mine. For example, in a real
mine environment, explosions are typically ignited by rock-on-rock or rock-on-metal
friction which does not exceed 1J of energy. However, these outcomes show that
researchers investigating the detonability of methane flames in underground mines must
consider ignition energies greater than 1J.
Discrete Modeling of the Gob – Section 6.1.4, Section 6.3.1
➢ Purpose: EGZs can exist in many areas of a longwall coal mine, including the gob area
(Karacan, Ruiz, Cote, & Phipps, 2011). Ignitions may occur in the gob area, often due to
falling rock and rock-on-rock friction and the resulting methane gas explosion can pose a
serious risk to miners (Brune, 2014). Unfortunately, the got is an inaccessible area and
consists of varying levels of compacted rock rubble of different shapes and sizes.
Previous researchers modeling the movement and accumulation of EGZs in the gob have
often modeled the gob as a Darcy flow porous medium. However, this assumption must
be revisited when modeling a methane gas ignition in-and-around the gob.
➢ Outcomes:
• A methane-gas ignition cannot be modeled within a Darcy flow porous medium in
ANSYS Fluent, unless 100% porosity is assumed, which results in unrealistic flame
propagation. Thus, the gob must be modeled discretely to accurately model flame
interaction with the gob.
• The model presented in this work captures flame propagation velocity trends from
experiments.
• When modeling the gob as discrete objects, the shape of the object can impact flame
propagation velocities and local turbulence. Shapes that are not spherical predict
faster flame speeds, they increase upstream turbulence, and increase mixing of the
unburned gases.
287
|
Colorado School of Mines
|
➢ Novelty: To the knowledge of these researchers, no other modeling group using ANSYS
Fluent has shown that, when considering an ignition in the gob area, that the gob must be
modeled as discrete objects
➢ Presented at the North American Mine Ventilation Congress, Montreal, Canada, 2019.
Paper was also nominated to be included in the CIM Journal:
• Strebinger, C., Bogin, Jr., G.E., Brune, J.F., “CFD Modeling of Methane Flame
Interaction with a Simulated Longwall Coal Mine Gob”. North American Mine
Ventilation Congress 2019.
Full-Scale 3D Modeling –Chapter 7
➢ Purpose: Methane gas explosions in underground longwall coal mines can be extremely
devastating and often result in significant mine damage and loss of life as described in
Chapter 1. Although there are many researchers modeling the ventilation conditions in an
underground coal mine, there is no experimental or modeling research that demonstrates
how these explosions occur or how the flame and pressure waves tend to propagate. The
purpose of this research is to model a full-scale explosion to understand these phenomena
so that better prevention and mitigation strategies can be explored.
➢ Outcomes:
• Researchers have modeled a full-scale, 3D methane gas explosion at the longwall
face.
o Results show an expanding pressure wave at 350m/s ahead of the main
flame front traveling at 30-35m/s. The pressure wave preheats the nearby
gases by compressive heating. These preheated gases continue to
accelerate the main flame front.
o Results show that the expanding pressure wave can disturb the airflow
along the longwall face, redirecting the airflow into the gob area. This is
important because the redirected air can mix with the methane in the gob,
creating more EGZs with a potential for ignition. The redirected air also
allows for a lower pressure near the longwall face, potentially allowing
more methane to diffuse out from the face.
288
|
Colorado School of Mines
|
• Researchers have modeled full-scale, 3D subsections of a full-scale longwall
bleeder coal mine.
o For an ignition under a single shield, results show a flame expanding near
the shields at 30m/s, the same order of magnitude of those estimated near
ignition in the Upper Big Branch explosion (90m/s) (Page, et al., 2011).
These results are important because they show the potential for
incorporating dynamic meshing to continue to track the flame front
throughout the mine and reduce computational time.
o For ignition in the tailgate corner of the longwall face, results show an
expanding flame at 36m/s and expanding pressure wave at 350m/s,
reflecting off nearby walls/structures and diverting airflow.
• Researchers have modeling a full-scale, 2D longwall coal mine and ignition in the
hanging roof behind the tailgate shields.
o Results are similar to the 3D, full-scale mine explosion, showing a fast-
expanding pressure wave redirecting airflow along the longwall face.
o 2D simulations capture general flame and pressure wave propagation
trends observed in the high-fidelity, 3D models, but solves in less than 1
week.
➢ Impact: These results have enormous impact for the mining and combustion
communities. For the mining community, these simulations demonstrate how methane
gas explosions propagate and accelerate as well as impact nearby airflow patterns,
creating a hazard for future ignitions. These simulations can be used as a preventative
measure to help with mine design and the layout of seals, methane gas detectors, water
sprays, etc. for improved explosion prevention (Section 7.4.1). They can also be coupled
with ANSYS Mechanical to better understand the fatigue and stresses on nearby mine
equipment, seals, and pillars for future design (Section 7.4.1). Additionally, these
simulations can be used as an after-the-fact way of understand what led to an explosion.
➢ Novelty: These are the first full-scale, 3D CFD models of a methane-gas explosion in a
longwall coal mine.
➢ Researchers plan on drafting this as a journal article to be submitted by the end of
August, 2019.
289
|
Colorado School of Mines
|
CHAPTER 9
CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK
In conclusion, this research has presented both experimental and modeling results of
high-speed methane gas deflagrations in confined, obstacle filled spaces at a variety of scales for
the purpose of building a physically accurate, comprehensive CFD model capable of simulating a
full-scale methane gas explosion in an underground coal mine. Researchers have been successful
in modeling a methane gas explosion in a full-scale mine and have developed complimentary,
sub-section and 2D models as alternative, more time-effective methods of simulating these large-
scale explosions.
Three different full-scale models of methane gas explosions were created: 3D, full-scale
sub-section models, a 3D, full-scale model, and a 2D, full-scale model, each modeling different
methane gas explosion conditions. In the 3D, full-scale sub-section models, a methane gas
explosion was modeled 1) in the longwall face under a single shield, 2) in a discrete gob behind a
shield, and 3) in the tailgate corner of the longwall face. Results from the sub-section simulations
predict initial flame expansions at 30-36m/s and pressure wave expansion at 350m/s. Although
the UBB explosion estimated initial flame speeds near ignition to be 90m/s, these preliminary
results are of the same order of magnitude (Page, et al., 2011). Additionally, these sub-section
models show that the initial pressure wave can disturb the airflow in the longwall face, forcing
the air behind the shields and into the gob. This is important because that air can mix with
methane in the gob area, creating more EGZs with the potential for secondary ignitions. This
same phenomena was also captured when modeling a full-scale 3D explosion at the headgate
drum of the shearer, when the shearer is located near the tailgate entry. The initial pressure wave
expanded ahead of the flame front, diverting air away from the longwall face. In this case, the
pressure wave also preheated the air ahead of the flame, which may lead to flame acceleration
and eventual transition to a detonation. Although a transition to detonation was not modeled, it is
important to fundamentally understand the possible situations which could lead to a detonation.
Complimentary to the 3D models, a 2D mine model was set up and an ignition was
modeled in the gob behind the shields at the tailgate. Results agree with observations made from
the 3D models, showing an expanding pressure wave diverting airflow from the face. Although
290
|
Colorado School of Mines
|
more work needs to be done with this model and it is not as high-fidelity as the 3D models, it
shows the potential to model flame propagation throughout an entire mine in a short amount of
time, less than 1 week on an 8 core compute. Comparatively, the 3D sub-section models ran for
1.5-3 weeks on 8-12 cores and the 3D full-scale model ran for 2 weeks on almost 96 cores across
4 computes.
In summary, 3D full-scale methane gas explosions in a longwall coal mine have been
modeled and are the first ever simulations of these events. Based on results from these
simulations, this research recommends several different ways this information could be used to
help prevent and mitigate these disasters (Section 7.4.1). One of the major benefits of these CFD
models is the ability to simulate a variety of different explosion scenarios to estimate flame
speeds and overpressures. This type of information could be used to better distribute inert rock
dust in these areas or perhaps include additional water sprays, increasing the humidity thereby
decreasing the flame speed and pressure. These models could also be used to help dictate mine
designs; for example, ANSYS Mechanical can be coupled with Fluent to estimate the impact
forces on nearby structures. This could help in improving the layout of the mine in addition to
designing seals and pillars. ANSYS Fluent also has a Multiphase model which could be
investigated to 1) model water sprays for improved shearer/drum design, or could be used to 2)
model rock dust entrainment for an improved understanding of explosions which transition to
coal dust explosions (UBB 2010, (Page, et al., 2011)). Altogether, these models have the
potential to be incredibly useful tools to better understand the events which lead to an explosion
as well as designing ways to help mitigate these disasters.
Another major goal of this project was to perform experiments to understand the impact
of different mine conditions on methane flame propagation and overpressure. A variety of mine
conditions were explored including the impact of:
• Mixture stoichiometry
• Void spacing
• Obstacle porosity
• Obstacle location
• Obstacle geometry (wall, porous bed, etc)
• Ignition location
• Reactor end conditions (relief)
291
|
Colorado School of Mines
|
A summary of the impactful experimental results is discussed in Section 8.1 of Chapter 8
and shows that stoichiometry (presented in Section 4.1), void spacing (blockage ratio), ignition
location, and relief have some of the largest impact on high-speed methane gas deflagrations out
of all parameters investigated. However, in the exploration of the impact of these different mine
parameters, several interesting results were found. One of the most interesting discoveries during
the experiments was the role of pressure and acoustics in accelerating the flame. For example, in
many cases it is thought that the faster the flame the greater the overpressure. If one were to just
compare ignition at the open end versus ignition at the closed end, then this hypothesis would
hold. However, by exploring different ignition locations it was found that ignition in Port 2
resulted in higher pressures, but slower flame speeds compared to a closed-end ignition (Section
4.4). It was also found that high-frequency acoustics were excited in the tube at different modes,
resulting in a flame which traveled similarly to a pulsed jet (forward and backward, but in a
preferential direction). This was an extremely interesting discovery because there have not been
many researchers who have investigated this. Most researchers have ignited at the open end,
closed end, and some in the middle.
Another interesting discovery regarding pressure, was the tulip inversion which occurred
when ignition was between obstacles. Tulip inversions have been studied for a long time by
many researchers (Ellis & Wheeler, 1928; Guenoche & Jouy, 1953; Starke & Roth, 1989), but
this is the first time to the knowledge of these researchers, that a tulip inversion occurred from
ignition between obstacles. What is interesting about this discovery is that the overpressure
generated was almost half the maximum overpressure recorded during a closed-end ignition.
This is important because an ignition in the gob area could potentially generate significantly
more pressure (causing more damage) than ignition in a crosscut or along the longwall face.
Additionally, the pressure traces seem to show that different frequency pressure waves are
interacting (Figure 4.37, page 88). This is important because oscillating pressure waves could
entrain fresh air which could accelerate the flame and possibly lead to an autoignition event.
Alternatively, the pressure waves could continue to stretch the flame front, increasing
combustion rates and flame acceleration.
Finally, experiments performed in the box reactor were crucial to the understanding of
methane flame interaction with a porous medium, or gob. Before these experiments were
performed, these researchers hypothesized that the flame would tend to move faster in the open
292
|
Colorado School of Mines
|
spaces, thinking it would follow the path of least resistance. As shown in Section 4.6, this was
not the case; the flame tended to propagate faster through the rock gob than in the open spaces.
Unfortunately, there was not enough time to continue investigating many of these
interesting discoveries, which is why this research recommends the following future
experimental work:
Experiments for the 12cm diameter reactor
• Ventilation studies of methane near the longwall face show mixtures at the extremely
lean limit. Recommendations for future experimental work includes performing closed-
end ignition experiments in the 12cm diameter reactor for extremely lean mixtures.
• Research showed ignition location had one of the largest impacts on methane flame
propagation, pressure generation, and acoustics. Previous experiments have already been
performed, igniting stoichiometric mixtures at 11cm, 25cm, 50cm, 75cm, and 1.39m
from the open end. Researchers recommend repeating these experiments with ignition 1m
and 1.25m from the open end. Results from these experiments can be used to develop an
analytical model of the impact of ignition location on maximum flame front propagation
velocities and overpressures.
• Repeat in-gob ignition experiments with 1) multiple obstacles in both directions, 2)
obstacles of different porosities, 3) obstacles closer and further from ignition, and 4) at
different locations in the reactor (i.e. instead of just performing these experiments in Port
1, 25cm from the open end, repeat them in Port 2, 50cm from the open end). This will
help to better understand the role of pressure and acoustics on flame acceleration.
• Perform studies with different mixtures inside the reactor. For example, near the ignition
use a stoichiometric mixture, but further from ignition, fill the tube with 2% methane by
volume. This will be important for investigating whether an ignition in an EGZ can
continue to propagate into the longwall face because mine regulations do not allow
mixtures above 2% methane by volume anywhere in the bleeder entries. If above 2% is
detected, the active longwall panel operations are stopped.
• 2D modeling results showed that ignition energies above 1J may impact methane flame
propagation. Therefore, researchers recommend building an ignition system capable of
293
|
Colorado School of Mines
|
handling ignition energies between 1J-10kJ and capable of changing the ignition
duration.
Experiments for the Experimental Box
• Add ion sensors and pressure transducers to record flame speeds and overpressures and
repeat OEI and CEI experiments with and without a porous medium.
• Modeling results show that obstacle shape (circles, hexagons, squares) can impact local
turbulence and flame speeds. Additionally, other experimental evidence shows that small
changes in porosity can impact methane flame propagation. Therefore, researchers
recommend performing addition experiments in the box reactor using solid shapes such
as circles and squares and arranging them in different packing orientations and porosities.
This will help researchers to determine best practices for modeling the gob discretely in
the full-scale, 3D ventilation model.
• Researchers also recommend investigating the scalability of experiments in the
rectangular box by making larger and smaller boxes.
Experiments for the 71cm diameter reactor
• Researchers recommend investigating alternative methods of measuring the flame front
propagation velocity in this reactor including UV and IR sensors.
• Researchers recommend extending this reactor to 100m to investigate the transition of
methane gas deflagrations to detonations. From the results from this research, this is
currently being funded as a collaborative project between CSM, the University of
Maryland, and the Naval Research Lab.
As discussed, this project has successfully modeled a full-scale methane gas explosion in a
longwall coal mine and has recommended several ways to help build stronger mitigation and
prevention methods against such disasters. Additionally, this project has experimentally shown
which parameters of a mine have the most impact on methane flame and pressure wave
propagation. The main purpose of the experiments was to validate the CFD model at a variety of
scales under different conditions. In doing so, this research has several recommendations for
model improvement:
294
|
Colorado School of Mines
|
• Investigation of other turbulence models such as LES or k-ω, SST. Current models use
the standard k-ω, 2-equation turbulence model, but LES may be more appropriate,
especially at the large scale. An added benefit of the LES model is that the initial
turbulence parameters are estimated by the model.
• Investigation of more complex chemistry mechanisms. This research has shown that the
2-step mechanism grossly overpredicts the propagation speeds of rich flames and has
high rates of conversion of CO to CO . Improving this in future models will improve the
2
prediction of rich flames and it will also allow researchers to more accurately predict
leftover CO. This is important for longwall coal mine explosion mitigation because CO is
poisonous and can kill miners (Gates, et al., 2006).
• Investigation of advanced parallelization techniques. This research has shown that
modeling full-scale, 3D methane gas explosions requires a significant amount of
computational power and time. Further studies into parallelization of the model has the
potential to reduce computational times while maintaining model accuracy.
• Experimental studies have shown acoustics can play a large role on flame acceleration
and modeling results have also shown pressure waves can interact with mine structures.
Therefore, it will be important to capture these effects in future models. To do so will
require incorporation of an acoustic model.
295
|
Colorado School of Mines
|
APPENDIX B
DYNAMIC & OVERSET MESHING
One of the major challenges of this research is model scaling and more specifically,
balancing model accuracy versus simulation time as model domains become larger. This
research has investigated two main meshing techniques to help reduce the computational domain
while balancing accuracy: dynamic meshing and overset meshing. The main idea behind
dynamic meshing is creating a subsection of a larger domain and specifying the boundary(ies) to
move. For example, Figure B.1 shows a schematic of an active longwall model (Juganda, Brune,
Bogin, Grubb, & Lolon, 2017). As discussed, there is significant evidence that shows ignitions
can occur in or around the gob area and for the UBB explosion of 2010, the ignition occurred at
the tailgate corner as shown by the red box. As shown in Chapter 5, many of the models require
cells on the order of 5-1mm in order to accurately track the flame. However in the full-scale
model the cell sizes are on the order of centimeters. Thus, the idea behind dynamic meshing
would be to create a subsection with a fine mesh and move the boundaries, indicated with the red
arrows, according to a certain flow quantity.
Figure B.1 Schematic of an active panel of a longwall coal mine and a subsection of the tailgate
corner where an ignition may occur. Blue arrows represent airflow pattern. Red box indicates
possible tailgate subsection and red arrows show dynamic boundaries that move with the
resulting flame front. Figure modified from (Juganda, Brune, Bogin, Grubb, & Lolon, 2017).
307
|
Colorado School of Mines
|
To test the viability of dynamic meshing, 2D CFD combustion models of the 12cm
diameter reactor were set up to simulate a closed end ignition. A small subsection was also
created which was 25cm long at time, t=0s, and the ‘open’ end was made to be a dynamic,
moving zero gauge pressure outlet boundary as shown in Figure B.2. Figure B.3, Figure B.4, and
Figure B.5 show flame propagation towards the open end of the reactor for the base case and
dynamic mesh case. As can be seen in these figures, the dynamic mesh can be used with ANSYS
Fluent’s species model to track the flame front. However, as the moving boundary of the overset
mesh moves, the new domain is not initialized with methane which is why the flame eventually
consumes what little combustible mixture is in the reactor before subsiding. This can be
overcome by creating a user-defined function (UDF) that initializes the newly created moving
domain. Additionally, the UDF can include what direction and how much the moving boundary
will move dependent upon a flow variable such as total pressure, total temperature, reaction rate,
etc. Although in this small scenario the dynamic mesh did not give significant time savings, in
the full-scale, 3D model this method can save significant simulation time while not sacrificing
model accuracy.
The second meshing technique investigated is called overset meshing. The idea of overset
meshing is that the main domain and obstacles are meshed on a coarse base mesh as shown in
Figure B.6. A more refined grid, called the overset mesh, is then created and overlays the base
mesh and data is interpolated from the coarse base mesh to the refined overset mesh and solved.
The overset mesh can also be coupled with the dynamic mesh and move in any direction in the
domain. The main idea behind investigating this is to track the flame front or initial pressure
wave on the overset mesh to maintain accuracy on a refined grid without sacrificing simulation
time. Unfortunately implementing a dynamic overset mesh with combustion is not allowed while
using species transport or any other combustion model in ANSYS Fluent. However, this research
was able to show the viability of this method and investigate the advantages and disadvantages
of implementing a dynamic overset mesh with combustion in ANSYS Fluent.
To test the implementation of a dynamic overset mesh, a small computational domain
was setup as shown in Figure B.6. The domain has an inlet and outlet and two, parallel walls.
Simulations have been run with and without a small obstacle as shown, but for brevity, results
with the obstacle will be presented. To test the dynamic overset mesh method, the fluid flow was
designed to be laminar flow with a Re = 100. Thus the following conditions were used:
308
|
Colorado School of Mines
|
• Base mesh = 1mm quadrilaterals
• Overset mesh = 0.5mm quadrilaterals
• Fluid density = 1kg/m3
• Fluid dynamic viscosity = 1kg/m-s
• Laminar Flow
• Uniform inlet flow
• Re = 100
• Entry length = 6cm
• Boundary layer = 2-5mm
Results of the dynamic overset mesh are shown in Figure B.7, Figure B.8, and Figure
B.9. As shown in these figures, the dynamic overset mesh successfully moves throughout time.
However, when the overset mesh moves across the obstacle, the mesh does not know to not solve
this space and so it continues to interpolate inwards onto the obstacle. Although this does not
significantly change the flow patterns in this setup, it would change flame propagation in the
combustion model. Therefore, to navigate this problem, researchers have meshed the solid
obstacles and marked the cells as solids. When this is done, the dynamic overset mesh does not
interpolate onto the obstacle. This has been tested using the full-scale longwall ventilation model
with movement of the shearer, though not shown here.
Overall, the dynamic overset mesh can be used in ANSYS Fluent, but it cannot be
implemented simultaneously with species transport or other combustion models. However, this
research shows the viability of the method in Fluent when only considering fluid flow. To use
this simultaneously with combustion would require a complex UDF that solves species from one
mesh to the other. Implementation of this UDF could take several months, but is a realistic
method to tracking the flame front on a finer grid.
309
|
Colorado School of Mines
|
APPENDIX C
PRELIMINARY 3D MODELING OF THE 71CM DIAMETER REACTOR
A preliminary 3D model of the 71cm diameter reactor has been developed in ANSYS
Fluent v17.2. The model settings for this are similar to the 2D, 71cm diameter reactor model
presented in Section 5.3:
• Pressure-Based Solver
• Energy Equation
• Viscous Standard k-ω Turbulence Model
o Low Re Corrections
o Shear Flow Corrections
• Species Transport
o Volumetric Reactions
o Stiff Chemistry Solver
o Finite Rate Chemistry
▪ Density solved using ideal gas theory
▪ Diffusion solved using kinetic theory
▪ Metaghalchi and Keck laminar flame speed theory
• Spark Ignition Model
• PISO pressure-velocity coupling
• Continuity/energy residuals set to 10-4, Velocity/species residuals set to 10-3
• Second order in time and space
• Time step = 0.01ms
• k = 1.5 m2/s2 and ω = 25 1/s
• 3 levels of mesh adaption every time step
• Boundary Conditions:
o Walls – steel, 5mm roughness height, adiabatic
Figure C.1 presents results of a closed-end ignition (ignition 28.5cm from the closed end)
of stoichiometric methane-air mixture for a 5cm and 2.5cm mesh. As seen in this figure, results
show a large difference between the 5cm and the 2.5cm mesh, but due to computational
316
|
Colorado School of Mines
|
resources, these researchers were unable to solve this domain on smaller meshes. Despite these
large differences however, by taking the endpoints of the flame location as a function of time, the
5cm mesh predicts a flame speed of 206m/s and the 2.5cm mesh predicts a flame speed of
167m/s. Although the experiments resulted in a maximum flame speed of approximately 125m/s,
these preliminary 3D results show fair agreement with experiments and the flame shape shown in
Figures C.2 – Figure C.4 agree with previous findings.
In the future, researchers recommend investigating different meshing techniques such as
dynamic meshing in order to obtain mesh independence for this large, 3D model. Additionally,
future improvements can be made using Multigrid techniques, variable time-stepping, and/or
parallelization of the model across multiple computes.
Figure C.1 3D, 71cm diameter reactor model results for a closed-end ignition investigating the
impact of mesh size on flame front location versus time. Time step = 0.01ms. CH = 9.5%. 2D
4
Body mesh size = 5cm and 2.5cm. 3 levels of mesh adaption every time step. Temperature =
295K, Pressure = 76kPa. SM E = 60mJ.
ign
317
|
Colorado School of Mines
|
ER-4609
ABSTRACT
Over the years a number of excellent articles have been
written by a variety of authors on various aspects of mill
waste management/tailings dams. Unfortunately, when such
articles are for a technical audience, due to time and space
limitations, some topics are only briefly discussed.
All of the work presented in this thesis/report has
previously been published. This thesis/report presents a
compilation of the best sources on tailings dam and
impoundment design identified by the author during an
extensive literature search. The text of this report is taken
verbatim from these sources. This research was completed in
1994. Since then significant work has been done in tailings
management and impoundment design. This thesis/report is
intended for use as an instructional primer for mining
companies and regulating agencies.
An informational database has been developed including
many references for tailings dam design and construction. The
references used in the compilation of this report are
available in the database. The database (WASTE.EXE) is
included with the thesis on three 3.5" high density diskettes.
iii
|
Colorado School of Mines
|
ER-4609 1
CHAPTER 1
INTRODUCTION TO TAILINGS PONDS
The material presented in chapter one is compiled from
the Bureau of Mines information circular 8755, Design Guide
for Metal and Nonmetal Tailings Disposal, written by Roy L.
Soderburg and Richard A Busch.
1.1 Overview
Surface disposal of tailings uses dams and embankments
to form impoundments that retain both the tailings and mill
effluent. In the past, tailings were routinely discharged
into the nearest surface water course. As this practice fell
out of favor, it required only a modest advance in technology
to dam the water course, forming an impoundment in which the
tailing could settle from suspension. The prevalence of
surface disposal stems partly from the historical background
and also from the fact that a reasonably large surface
impoundment allows for clarification of discharged mill
effluent and its return to the mill for reuse. As used in this
guide, tailings ponds comprise embankments placed on ground
surface that are required to retain slurries of, waste and
water; they are constructed from tailings, borrow material, or
some of each. Some mines use deslimed tailings for
underground fill, leaving only the finer materials to be
impounded on the surface. The materials range from chemically
stable quartz to unstable feldspars which can alter to
micrometer-size clay.
|
Colorado School of Mines
|
ER-4609 2
1.2 Function of Tailings Ponds
The main function of a mine tailings pond is to store
solids permanently and to retain water temporarily. The
length of time that water must be retained ranges from a few
days to months, depending on gradation, mineralogy, etc. When
clarified, the water can be reclaimed for plant use or
discharged into the drainage.
When the water contains a serious pollutant, the
tailings dam must be designed to retain the water for longer
periods until the harmful chemicals have degraded or until the
water evaporates. A completely closed system is preferred in
all such cases, not only for conservation of water, but as a
necessity to prevent the pollutant from being discharged. The
seepage water from this type of dam must be controlled,
treated, and pumped back to the mill for reuse.
1.3 Basic Considerations
Economics continue to be of prime importance in the
design of tailings embankments, including site selection,
pumping requirements, length of pipe line, and capital versus
operating cost. The annual tonnage versus site acreage,
physical properties of tailings, type of embankment, method of
waste disposal, availability of construction materials,
climate, terrain, hydrology, geology, and nature of the
foundation at alternative sites are all important factors.
The consequences of failure should be fully considered in
establishing the factor of safety (FS) of the embankment
design. Embankments in seismically active areas should
undergo dynamic analysis to eliminate the possibility of
liquefaction from earthquake shock. Embankments in remote
|
Colorado School of Mines
|
ER-4609 3
areas can have a lower FS than needed in urban areas.
Operating costs for tailings disposal can be a big item in a
mining operation, and much thought should go into the study of
capital versus operating cost. In some cases, the plan with
the cheapest capital cost can be the most expensive when the
operating cost is added, and vice versa. Probably the
cheapest operation possible would be one where a few water-
type dams could be constructed to enclose a large area,
allowing the operator to merely dump the tailings; this would
completely eliminate operating labor except for pump operation
and periodic inspection.
1.4 Daily Tonnage
Operation of porphyry copper, taconite, and pebble
phosphate mines can more easily anticipate the ultimate area
needed for tailings disposal for the life of the deposit than
can operation of underground deep-vein mines. These surface
deposits are generally well defined with known ore reserves
for a given number of years. Knowing this and the anticipated
daily tonnage, definite plans for a tailings disposal area can
be made. Any planned expansion should be considered at.the
same time, keeping approximately 35 acres per 1,000 tons of
daily mill production for metal mines, preferably in two
separate areas. Taconite operations require about the same
acreage per 1,000 tons of waste produced. Under special
conditions, such as single-point discharge into large areas
where cheap land is available and other factors are favorable,
the area per 1, 000 tons of waste could go up two to three
times this, but observation of well-engineered taconite
tailings areas indicates that 35 acres per 1,000 tons is about
optimum where discharge pipelines surround the area.
|
Colorado School of Mines
|
ER-4609 4
Phosphate mines in flat terrain will require nearly an acre of
settling pond per acre of mined land until some improvement in
settling rate can be achieved. Because of the fineness of the
material and the low pulp density, it is deposited at a single
point at a time.
1.5 Size of Tailings Area
The size of the tailings embankment necessary for each 1,000
tons of milling capacity for a safe and efficient operation is
governed to some extent by the size of the grind, but mostly
by the terrain within the tailings area. A relatively level
area of a wide, open valley is an ideal site because of the
large volume of tailings placed per foot of elevation rise. A
starter dam constructed from borrow material is a very
important part of the entire impoundment. The purpose of this
dam is to contain the sand and provide a pond large enough to
insure sufficient water clarification at the start of
operations. The steeper the terrain within the embankment
area, the higher the starter dam must be to supply the storage
necessary for the sand and water until the embankment can be
raised with the beach sand. It is far better to make the
starter dam a bit higher than required because of the unknown
factors at startup of an impoundment. These unknowns are (1)
the efficiency of segregation of the sand and slime on the
beach, (2) the angle of the beach area, and (3) most
important, the retention time in the pond to get clean water.
A capacity curve plotting the volume against elevation should
be made, as well as a time-capacity curve to get the elevation
rise per year through the life of the impoundment (Fig. 1).
Where the maximum annual rise is limited to less than 8
feet per year, the active disposal area must be at least 20
|
Colorado School of Mines
|
ER-4609 5
acres per 1,000 tons of daily capacity. Operating at this
upper limit of rise per year for continuous operation might be
safe, but this depends on the grind, pulp density, and type of
material being impounded. From an operating and safety point
of view, a figure of 30 acres per 1,000 tons of daily capacity
is much better for the lower limit of a mature pond. If site
is on a hillside, the startup time is most critical because
the area of active storage is small. There is no established
rate that an embankment can be raised, but for a given
material, gradation, and pulp density there is a definite
MILLION TOMS
30 40 50 60
350
300
250
I
200
É Too rapid annual rise, unstable condition,
1 10.000 • 40,000 tpd in 2 years, steep terrain ( 30-percent1- slope)
150
100
2 ponds alternated,flat terrain (2 - percent slope), stable conditions,
50 25,000 tpd, 36 acres per thousand tpd production
Calendar Year
Figure 1. Capacity-elevation-time curve.
maximum rate of rise above which stability becomes a problem.
If the tailings cannot drain as fast as they are placed in
the pond, the phreatic surface rises and comes out the face
above the toe dam. When this occurs, seepage and piping take
place, lowering the safety factor to the danger point.
|
Colorado School of Mines
|
ER-4609 6
Possible solutions are to allow time for drainage and to place
a filter and rock surcharge on the toe. A rapid annual rise
is undesirable because the material does not have time to
properly drain, consolidate, and stabilize, nor is there time
to raise the peripheral dam.
The pond area required for clarification of the water
prior to reclamation of discharge into the local drainage is
difficult to determine by experimental or theoretical means.
The problem is to provide sufficient retention time to permit
the very fine fractions to settle before they reach the
decant. The settlement velocities of various grain sizes and
shapes can be determined theoretically; however, several
factors determine the effectiveness of settlement in the
field, such as grain size, percentage of slimes, pH of the
water, wave action, and depth of water.
The size of grind required to liberate the metal from
the waste can produce a material having 55 percent or more
minus 200 mesh so that the settling rate is quite slow.
Particles of 50-micrometer size have a settlement rate of 0.05
inch per second and will settle in a reasonable time even
though affected by wave action. The most difficult particles
to settle are those of 2 micrometers or less; these have a
theoretical settlement rate of 0.01 inch per second in still
water, but in fact may take days because of wave action.
The quality of the water returned to the mill or the
watershed will determine the retention time for any particular
mine. The time required may be as low as 2 days and as high
as 10 days, with an average of about 5.
|
Colorado School of Mines
|
ER-4609 7
CHAPTER 2
MATERIAL PROPERTIES OF TAILINGS
2.1 Introduction
The material presented in chapter two has been
extracted from two sources. Sections 2.2 and 2.3 are from
the Bureau of Mines information circular 8755, Design Guide
for Metal and Nonmetal Tailings Disposal, written by Roy L.
Soderburg and Richard A Busch. The rest of the chapter has
been extracted from Steven G. Vick’s book titled Planning,
Design, and Analysis of Tailings Dams.
2.2 Mill Tailings (after Soderberg and Bush, 1977)
Metal mine tailings include materials ranging from hard
quartz to mudstone with vast differences in physical
properties. Finely ground mill waste high in silica can
have a high shear angle at high densities with little or no
cohesion and still be very susceptible to erosion by wind
and water. Materials high in feldspar may have a high shear
strength when fresh, but can chemically change to clay with
time, reducing the strength. Relatively minor amounts of
sulfide can oxidize to form a crust and lower the pH enough
that vegetative growth is difficult or impossible without
adding topsoil or altering the material in some way. High-
sulfide tailings may ignite by spontaneous combustion or
produce acidic runoff, iron oxide, or hydroxide, which can
pollute large areas in a drainage basin. The sodium cyanide
|
Colorado School of Mines
|
ER-4609 8
from gold ore treatment plants requires retention time in
the tailings pond, and sometimes requires treatment with
chlorine or other oxidizing agents to neutralize the cyanide
to tolerable levels before release. The waste from uranium
mining and milling can be very dangerous for many years
owing to radioactive daughter products.
2.3 Classification of Tailings Types (after Soderberg and
Bush, 1977)
The types of tailings cover such a wide variety of
physical characteristics that generalization is difficult.
Not only do the types of tailings vary, but tailings within
any one ore type may differ substantially according to mill
process and the nature of the orebody.
Table 1 divides the various types of tailings into four
general categories according to both gradation and
plasticity. The first category, soft-rock tailings, are
those derived from shale ores, including fine coal refuse
and trôna insolubles. While these tailings ordinarily
contain some sand-sized materials, the clayey nature of the
slimes significantly influences the physical character.
Sands usually predominate for the second category,hard-
rock tailings, which includes the lead-zinc, copper,gold-
silver, molybdenum, and nickel types. Tailings are
primarily finely crushed silicate particles. Slimes, while
they may be present in substantial proportions, are derived
from the crushed host rock rather than clay and do not
usually exert an overwhelming influence on the behavior of
the tailings as a whole. Basic information on the
mineralogy of the ore, grinding operations, and
concentration procedures will usually permit reasonably
|
Colorado School of Mines
|
ER-4609 10
valid correlations with physical characteristics of the
hard-rock tailings reported herein.
Fine tailings, the third category, are those having
little or no sand and include phosphatic clays, bauxite red
muds, fine taconite tailings, and slimes from tar sands
tailings. The characteristics of slimes predominate for
these materials to the extent of rendering them largely
incompetent from a structural standpoint. These materials
may require long periods of time for sedimentation and
consolidation, and require large impoundment volumes.
Coarse tailings are those whose characteristics are
determined on the whole by the sizable coarse sand fraction
or, in the case of gypsum tailings, by non-plastic silt that
behaves more or less like a sand. This group includes the
coarse fraction of tar sands, uranium, gypsum, coarse
taconite, and phosphate sand tailings.
Because tailings in any one category share the same
broad physical characteristics, disposal problems are
usually somewhat similar. Thus, when dealing with tailings
from ores where there is little information available on
disposal practices, comparison with tailings in the same
general category may provide useful general guidelines. In
addition, changes in grinding at a particular mill may, for
example, produce considerably finer material, which may
change the category in which the tailings reside and
introduce new and different disposal problems. It is
important to recognize, however, that the above
classifications reflect only the broad physical
characteristics and engineering behavior of various tailings
types; chemical characteristics and environmental
considerations may be more important than physical behavior
in determining disposal practices in some cases.
|
Colorado School of Mines
|
ER-4609 11
2.4 Particle Sizes (after Vick, 1983)
The grind necessary to free the metallic minerals for
flotation ranges from about 30 percent to 80 percent minus
200 mesh (Fig. 2) . Sand-filling operations at some
underground mines remove the coarse sand, leaving an even
finer material to be impounded in tailings ponds.
Taconite plant waste products include a float product
of 3/8 to 1/2 inch size and a spiral and flotation reject
containing up to 70 percent minus 325 mesh, and there is a
possibility of even finer grind to 90 percent minus 325 mesh
to reduce the silica content in the pellets. Not all plants
have the same waste products, but all could have all or part
of those listed.
2.5 Depositional Characteristics (after Vick, 1983)
Central to an understanding of tailings behavior is the
nature of the depositional processes which tailings undergo.
Tailings are deposited hydraulically, usually by some form
of peripheral discharge method, either spigotting or
rotating single-point discharge. This results in an above
water tailings beach and a slimes zone associated with the
ponded decant water. For most types of tailings, the
beaches slope downward to the decant pond with an average
grade of 0.5-2.0% within the first several hundred feet.
Beaches on the steeper end of the range usually result from
higher pulp density and/or coarser gradation of the whole
tailings discharge. At distant points on exposed beaches,
the beach slope may flatten to as little as 0.1%. At these
distant locations, depositional processes may come to
|
Colorado School of Mines
|
ER-4609 13
resemble natural stream channel sedimentation, with shifting
braided flow channels and backwater regions. This
depositional process produces a highly heterogeneous beach
deposit. In the vertical direction, tailings beach deposits
are usually layered, with the percent fines varying as much
as 10-20% over several inches in thickness. If discharge
points or spigots are widely spaced, variations in fines
content of 50% or more can occur over short vertical
distances. Such extreme layering produced by thin slimes
layers within otherwise sandy beaches may result from
periodic encroachment of ponded water onto the beach where
thin layers of fines settle from suspension.
Horizontal variability is usually also significant,
with coarser particles settling from the slurry as it moves
over the beach, and finer suspended or colloidal particles
settling only when they reach the still water of the decant
pond to form the slimes zone.
Figure 3 summarizes measurements of fines content as a
function of distance for several tailings beach deposits.
The degree of grain-size segregation ranges from high to
almost nonexistent. The degree of sorting obviously depends
on the gradation characteristics of the whole tailings
discharge; slurry with a wide range of particle sizes is
more likely to exhibit beach grain-size segregation than
slurries containing poorly graded materials.
Deposition of slimes occurs by entirely processes than
those for tailings beaches. Sedimentation of slimes from
suspension in ponded water does not involve sorting by
saltation or particle rolling, but rather it is a relatively
straightforward process of vertical settling. The rate of
slimes sedimentation can have important effects on the size
of decant pond necessary for water clarification and on the
|
Colorado School of Mines
|
ER-4609 14
quantity of water available for mill recycle. Sedimentation
rates may be determined in the laboratory by pouring a
homogeneous slurry at the desired pulp density in a glass
cylinder. The advance of the interface between the water
and settled solids is recorded with time.
100
90 - *ev(^Ed<)e of decint cool _
8 BO
E
60 —|
3
40
6U
0 200 400 600 a00 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000
Distance from discharge point (ft)
Tailings
Curve Type S %-200 Pulp
1 GOLD -
2 COPPER 2.7 45 45
3 LEAD-ZINC 3.4 75
4 - 2.7 38 30
5 - 2.7 60 50
6 COPPER 3.0
Figure 3. Grain-size segregation along tailings beaches,
(after Vick, 1983)
Typical sedimentation test results are shown in figure
4, with the linear portion of the curve yielding the
sedimentation rate. Typical sedimentation rates for various
slimes are shown in Table 2.
In the absence of laboratory sedimentation tests, an
empirical rule for decant pond size is that it should allow
5 days of retention time and provide 10-25 acres of surface
pond area per 1,000 tons of tailings discharged per day.
|
Colorado School of Mines
|
ER-4609 18
values measured in actual impoundments for various types of
tailings.Slimes tailings generally show an average increase
of about 10 pcf per 100 ft of depth, with slightly smaller
rates of increase for the less compressible sands of about
5-10 pcf per 100 ft. High rates of density increase for
gypsum are not representative of tailings in general since
they are produced by long-term creep deformation of
individual grains.
2.7 Permeability
More than any other engineering property of tailings,
permeability is difficult to generalize. Average
permeability spans five or more orders of magnitude, from
10"2 cm/sec for clean, coarse sand tailings to as low as 10~7
cm/sec for well-consolidated slimes. Permeability varies as
a function of grain size and plasticity, depositional mode,
and depth within the deposit. General ranges of
permeability are shown in Table 4.
TABLE 4 Typical Tailings Permeability Ranges (after Vick,
1983)
Average Permeability
Type cm/sec
Clean, coarse, or cycloned sands with
less than 15% fines 1er2 to 10"3
Peripheral ^-discharged beach sands with
up to 30% fines 10'3 to 5 x 10'4
Non-plastic or low-plasticity slimes 1(T5 to 5 x icr7
High-plasticity slimes 10'4 to ICf8
|
Colorado School of Mines
|
ER-4609 19
2.8 Compressibility (after Vick, 1983)
Because of their loose depositional state, high
angularity, and grading characteristics, both sands and
slimes tailings are more compressible than most natural
soils of similar type. Compressibility is determined in the
one-dimensional compression (consolidation) test commonly
used to evaluate compressibility of clays in conventional
soil mechanics.
Interpretation of compressibility coefficients requires
specification of the stress range over which they apply.
Typical values for the compression index, Cc, determined in
one-dimensional compression tests are shown in Table 5
together with approximate stress ranges over which the
values were determined and corresponding initial void
ratios.
2.9 Consolidation (after Vick, 1983)
Primary consolidation governs the rate of pore pressure
dissipation under constant load, which can have important
implications for certain classes of stability and seepage
problems.
Primary consolidation for sand tailings occurs so
rapidly that it is difficult to measure in the laboratory.
The few available data suggest that the coefficient of
consolidation cv varies from about 5 x lO"1 to 102 cm2/sec
for beach sand deposits. For slimes tailings cv is
generally about 10~2 -10~4 cm2/sec, in the same range as
typically exhibited by natural clays. Reported data from
the literature for both sands and slimes tailings are.
|
Colorado School of Mines
|
ER-4609 21
summarized in Table 6.
TABLE 6 Typical Values of Coefficient of Consolidation, Cv
(after Vick, 1983)
Material Type cm 2 /Csvec
Copper beach sands 3.7 x 10"1
Copper slimes 1.5 x 10-1
Copper slimes 10"3 to 10'1
Molybdenum beach sands 102
Gold slimes 6.3 x 10"2
Lead-zinc slimes 10"2
Fine coal refuse 3 x 10"3 to
Bauxite slimes 10'3 to 5 x
Phosphate slimes 2 x icr4
2.10 Drained Shear Strength (after Vick, 1983)
Notwithstanding their generally loose depositional
state tailings have high drained (effective-stress) shear
strength owing primarily to their high degree of particle
angularity. It is not uncommon for tailings to show an
effective friction angle (a) 3 to 5 deg. higher than that of
similar natural soils at the same density and stress level.
With rare exceptions, tailings are cohesionless and show a
zero effective cohesion intercept C in properly performed
and interpreted laboratory tests.
Typical values of a for various materials, based on
laboratory tests of both undisturbed and remolded samples,
are shown in Table 7. In most cases, the tests were
performed on samples either at an initial density
-p 0 T—I o
|
Colorado School of Mines
|
ER-4609 29
Figure 7 shows the plan and construction details of a
tailings impoundment constructed in flat terrain in a low
rainfallf high evaporation environment. The bedrock is high
strength, moderately jointed norite. Soil cover is on
average about 1m of high plasticity silty clay of very stiff
consistency. Peak undrained strength is high (+ 150 kPa)
but residual effective strength is low (a = 15 ) . The
groundwater table is usually at the clay and bedrock
interface. In this example seismicity is insignificant.
The tailings grade from a medium sand to fine silt.
They contain potential contaminants, thus construction on
the clay to preserve groundwater quality is done. A closed
circuit water balance is possible because of the arid
climate.
In order to construct to a height of 50m, essentially
by upstream construction methods, hydraulic fill dams are
used in addition to conventional upstream discharge. In
this way the overall flat slopes required for stability in
the clays are created.
Construction of a surround dam starts with a 1m high
wall built of in situ clays. Tailings are discharged behind
the wall from spigots. Segregation occurs along the beach
which deposits coarser tailings near the perimeter, and
finer tailings near the penstocks or decant towers which are
used to remove water.
Once. the surround dams reach a suitable height,
discharge of tailings into the main impoundment begins.
This too is done by conventional spigotting.
In order to reduce as far as practicable seepage of
contaminants from tailings a dry disposal impoundment as
shown in Figure 8 is used. Underlying bedrock is sound and
a clay liner is installed over in situ sandy silts before
|
Colorado School of Mines
|
ER-4609 30
tailings deposition.
Figure 9 shows the plan and cross section of a tailings
impoundment designed for a steep narrow valley. Bedrock in
the area is a competent quartz monzonite with few fractures
or joints and an hydraulic conductivity of the order of 10"8
m/sec. Filling the base of the valley to a depth of 20m is
a deposit of medium dense alluvial sands with a hydraulic
conductivity of 10”5 m/sec. A tongue of low strength (c1 =
0, a1 = 2 0°) silty clay underlies a part of the sands. The
design earthquake acceleration at the site is O.lg.
The climate is dry : annual precipitation is 300mm and
evaporation is 750mm. Temperatures seldom fall below
freezing and winds are moderate due to shielding by
neighboring hills.
The tailings grade from a medium sand to a fine silt
and are suitable for cycloning. Chemically they do not give
rise to any potential ground or surface water contamination.
The toe embankment is constructed of mine waste rock.
Part of the alluvial material is removed both to construct
an upper starter dike and to remove the underlying low
strength clay. Dewatering of the sand before excavation is
required. A transition zone of screened rock is placed on
the upstream side of the rock embankment and a clay blanket
and liner on the upstream side of the starter dike.
Cyclones along the starter dike separate the tailings
into coarse and fine fractions. The coarser sands are used
to build the sand embankment by centerline techniques.
Drains beneath the cyclone underflow zone, the starter dike
and the rock embankment prevent the buildup of a phreatic
line. A series of diversion ditches plus the very small
catchment area of the impoundment control long-term buildup
of excess water on the impoundment. Sufficient freeboard
|
Colorado School of Mines
|
ER-4609 32
disposal is used. This method involves moving tailings as a
slurry and using placement techniques and impoundment design
features that cause the tailings to dry out once placed.
This drying out results either from sun drying of
unsubmerged tailings or controlled seepage of water from the
deposited mass. Such designs do not involve liners or
impermeable dikes or embankments ; but they do provide for
control of seepage and excess pool water. Potentially
contaminated water is treated during the mining period.
When deposition at the impoundment comes to an end the
volume of water that could seep from the tailings is small
and will continue to reduce as the tailings dries out.
The rest of section 3.2.2 was extracted from Steven G.
Vick's book Planning, Design, and Analysis of Tailings Dams.
This method appears to be best suited to disposal sites
located in relatively flat topography and where concentrated
runoff does not occur, at sites close to the mill where
pumping costs are minimized, and in low seismic risk areas.
In this regard, the thickened discharge method shares many
of the siting restrictions of upstream-type embankments.
Also, like upstream methods, thickened discharge disposal is
only applicable for tailings containing a reasonable sand
fraction and without a major proportion of clayey fines.
"Wet" disposal involves depositing and containing the
tailings in such a way that they are placed and are likely
to remain wet over extended periods. The design of a wet
tailings impoundment may involve liners and impermeable
cores in embankments or containment dikes. Impermeable
barriers are provided to contain and prevent movement to the
surrounding environment of water that may be contaminated.
Figure 9 describes an impoundment where wet disposal is
used.
|
Colorado School of Mines
|
ER-4609 33
Dry disposal is presently being used for disposal of
coal and uranium mine tailings. Disposal costs are high and
the method is best suited to small operations in dry
climates. Semi-dry disposal, too, is practical only where
evaporation exceeds precipitation. Thus in areas of high
precipitation wet disposal is the norm. If the tailings are
a potential pollutant, impermeable barriers will be
required.
3.3 Underground Disposal (after Vick, 1983)
The use of tailings backfill to increase ore recovery
places the most demanding requirements on the properties of
the material. In room-and-pillar mines, for example,
backfilling of stopes can allow subsequent remining and
recovery of ore in pillars, adding significantly to the
overall degree of recovery of the orebody. For backfill to
function effectively in this role, however, it must be free
standing and sufficiently stiff to accept appreciable load
transferred from the roof as pillars are mined. Achieving
these properties often requires addition of cement to the
sand tailings slurry, typically in proportions ranging from
about 1:20 to 1:30 by dry weight. In addition to cement
content, the gradation range, in-place density, and pulp
density of the slurry influence the strength and stiffness
of cemented tailings backfill. Sulfate-resistant cement may
be required for tailings derived from high-sulfide ores.
Disposal of tailings in underground mines purely for
storage purposes, outside of any mining-related function,
has not been routinely performed to date. However, many
room-and-pillar type mines in such materials as coal, trona.
|
Colorado School of Mines
|
ER-4609 34
and sometimes copper may produce large volumes of otherwise
unused underground space after mining in a certain area is
essentially completed. Use of this space for tailings
disposal can produce major advantages by reducing the area
and related disturbance required for surface impoundments.
This can also have a major cost advantage in cases where,
for example, impoundment lining or reclamation requirements
impose a severe cost penalty on surface tailings disposal.
Underground disposal below the water table may be of
particular advantage for tailings high in pyrite. By
keeping the tailings permanently saturated underground,
oxidation that could otherwise produce severe pH and heavy
metal contamination problems on the surface can be reduced.
3.3.1 Pit disposal (after Staub, 1978)
Disposal of tailings in mined-out pits is an appealing
alternative to other tailings management methods of the
recent past. In the 1950s and 60s, tailings were often
piled on the surface and left to dry. Wind erosion and
sheet runoff widely dispersed these tailings with their low
concentrations of radioactive components. Another common
method of containment was the construction of a ring dike
made from the sandy portion of the tailings. Clay slime and
contaminated water were impounded within the ring dikes.
Often the dikes were poorly designed and constructed, they
were located on the flood plains of major streams and their
reservoirs were unlined. Failure of several of these dikes
led to the uncontrolled surface discharge of tailings.
Although dikes have remained intact at most tailings
impoundments, groundwater contamination occurred by seepage
|
Colorado School of Mines
|
ER-4609 35
through the floor of the reservoir. More recently, high
earth fill embankments have been constructed across natural
drainage basins to impound slurried tailings within lined
reservoirs. While dams provide short-term (tens to perhaps
hundreds of years) protection against erosion of tailings,
the natural stream course will eventually breach the
embankment and cut intricate and progressively deepening
channels through the tailings. Even in the short-term there
is the risk of catastrophic failure of a poorly constructed
or earthquake damaged embankment. Disposal of tailings in
mined out pits reduces the impact of wind erosion,
eliminates the possibility of catastrophic failure and
reduces the possibility of stream erosion.
In an open pit that is being mined, in-pit tailings
disposal is not a practical alternative. Where multiple
pits are being mined, mined-out pits may be used for
tailings disposal.
3.4 Marine/Underwater Disposal (after Vick, 1983)
The effects of offshore disposal on water quality may
be limited if the chemical composition of the mill effluent
is relatively innocuous and if the tailings are relatively
coarse or sufficiently flocculated to settle rapidly without
excessive turbidity. It is also important that the point of
tailings discharge be in water sufficiently deep and far
from the shoreline to avoid the most biologically productive
and sensitive shallow-water and near-shore zones. Various
authors have reported results for water quality and
biological monitoring programs for various offshore disposal
schemes. The results of these studies generally indicate
|
Colorado School of Mines
|
ER-4609 36
minimal biological and water quality consequences for
offshore disposal, and suggest that offshore disposal areas
quickly become rehabilitated after discharge ceases. Other
studies of offshore disposal, however, indicate unexpectedly
large areas covered by the discharged tailings, as well as
turbidity problems.
Regardless of technical arguments for or against
offshore disposal, regulatory authorities view it with a
jaundiced eye. No other disposal method generates public
concern more rapidly and intensely, and mines using offshore
disposal can usually anticipate that control of their
tailings disposal, and therefore of their entire operation,
will ultimately reside in the political arena. For these
reasons, offshore disposal should realistically be
considered only as a last resort, after all other disposal
possibilities have been exhausted. Offshore methods,
however, may be the only possible option for tailings
disposal in some coastal areas where the combined effects of
extremely high precipitation, steep terrain, and high
seismicity make surface impoundments impossible from a
practical standpoint to safely design and construct.
3.5 Comparison of Disposal Options (after Caldwell, 1982)
In this section, the differences between alternative
tailings disposal methods are discussed. Emphasis is placed
on the differences between marine disposal and land disposal
in rugged, seismic, high rainfall terrain. For comparison,
reference is made to in-pit tailings disposal and
impoundments on flat terrain and disposal of "dry tailings"
in rolling country.
|
Colorado School of Mines
|
ER-4609 37
There are probably no mining projects where the
tailings engineer will have to choose between disposal in
the sea, in steep valleys, or on flat land, either by wet or
dry disposal techniques. However, there are projects where
the choice is between disposal in the sea or steep valleys;
and there are projects where the choice is between disposal
in a steep valley or on flat ground. When disposal is on
land, there is always a choice between dry or wet disposal.
Hence, although the comparisons made in this paper are
hypothetical, they are not entirely without practical
application as some of the examples described illustrate.
The method described here for use in comparing
alternative tailings disposal options involves a qualitative
evaluation of each method. For a list of defined factors,
five categories are defined ranging from very good to poor
(or very low to very high impact) . For the list of factors
in Tables 9 and 10 the impact of each disposal method may be
evaluated. The considerations leading to an evaluation of
poor, good, etc. are shown in Table 11.
There is room for argument about the definition of
ranking considerations and about the assignment of a
particular scale to each. Some may wish to change
definitions and scales and hence test the sensitivity of the
evaluation system to different opinions or judgment. As
done in tables 9 and 10 a system is established for
comparing marine tailings disposal to disposal in areas of
high precipitation and rugged topography, or any of the
other disposal options considered. Table 10 compares
operational and cost considerations.
In order to obtain a semi-quantitative evaluation,
hence ranking, a number is assigned to each factor; 1 for
very low impact through 5 for very high impact. This has
|
Colorado School of Mines
|
ER-4609 40
41M to u
£ £ tO
ofll
j u: uoi
TJ
4 u 31 0 01 T u«J
u t4 o1 u1 >4 E O<0 mu C to 0c 4 1‘ a Cl -i 0a H 1 S? s l 01i H-5 D • J 2 - m- ZO O Uu 3 kt 0 £®c » t «o u Q 3» toi
uto0 CUu TJ M = *> to Z u M to to
C
O’ TJ
£ U3 0 T -I cQ U t t4 4o oJD1 i Tf*— 3 C «4- J 0141 - ®t c t M U w3 op —0 WS * < O• W 3V - DU Q — «L u8 1I !U W41 l41 T < OwO J 0u 3 MVm M 3® ® t ®C o uO3 u 0e1 c 0u £ Ut u t4O 0 o1 ou Ot Vo l
to 0.
CM . 41 OM O OE u
G <TO i u « O5 J «4h • -£ £ H'C * fT -t U tt« Q O 4O Jo . £- f ® 3- 3-4 t T V 4 o« «CJ 1- . B °5t2 -.. T -U t« wO 1OJf — £ju - 34 4j MVc 4 U tO c o1l £ >W«k 4t T- M O E> W«4 J■ TT 0 M « w 3W a JJ •f a M O3U 4 Hu 1t #Vo 404 T S« ® J S o Wc ! T" Xi 04 O- J11 l -4 3k 4 mo1 t ) _ T - O4J tu« M3 U 3 tt u oo i u 4C ® aUtf 1l J® C « 0J M 0 £ U MO 0 0 V 0l U -«to 4
0>
>.J4 C1 » B 1« T 0J > U O « O14 -U 4 c
Ti—JI <e-• ti a4* uU cO i to 001 0
u e Oto' —-*« —i u C -4
ufd u4J Æ *mO Æ 4c o1i Isis c M 0 u10 0 $0 t « uo
u
0)
T XOJ *-CHO l —IC QI -u 3w t<U 0l >to inOI !cOI Ski —£ X01% 1®
14
K3 £u 1®
0
UOu £ Mu 10U C41 U -4 41
4-J
m
fd J! O
u Q4). O CO l—jQ• n T VJ T 4J 1 O Ol
fd — XC < - W 1OU 0 Æ C O I4C 1 —O >i J D 3O3 > -£4 10 <1101 £ ecu u Ca - MU 4 C 11
-H - XO HI 'HC 10 •U - 3< UL Wl T C «J J w1 34 >w j3 j u u 0*0 > I 0- *-T ® 1J 0 <Q e 0 . *t£ *o4 C u O £ 01 - at ®4 o j U M m: -J H w4 J1
<u > 3 U 44 Q,
4->
-rH o w
M ••4 O "4
-u -oc H U: ■ EOU O Q1 o 40 iS i • n -l UO O• 4*S * T5 «O CIC Jdl t £ uO U3Oi I > £5 £U o $H — jW n4 j* i -« —J >0W - 3 41«* • •U u1 OC O H J0 -“ U << *> » ®U0p H £^» H « u8 u H ^« u58 U 0 3' .u #O i0* d4 4 j QW I d Q * 1 -■ • T m“ u C9H * J8 —5 Ok < eD C tW•t 3 o0 t i u o V«> J -0, 1 4Q o 40 u 4tB - U1 iH> - ih f £u 3 u -4 1 «4 U f4G L- -4 1 i O e —- ijJ> W c®C4 a 3jJ 1* T c t 3W t„ oJ o. »—4 OOX> uE 4* < i* -" — J t®E hI J o• —T 5 MC OJ J l U2 QOC « M ’ JUw O J "T qU< 1 ®f C O 4J l 00l i £ UI <0 r0O M t» o4 4 i U 4» —kO - n1»* l4 t u uu C O Oec tW«o "U4 o41 14 £*0 M4 1 to0 U o4 1 01 - >t34 o41
fd
I—dI
to « u
»0
-t
u« q1*i c 0 »
uf >d 5 <8 2 tf$cj WS
3
T U3 C <J
0 T
" U
J
c4o 1
J U
101 M
J
00
-t
u « ® M0
U 41
t
1 ®
Mo
0
XOO 1M 0 £ U4
t
M« o1 J 1 1J 0
0 r4
UM Ou
3
eV
nri •
f1
u
u0 0 ® l4>« u5 1
fd • 06 ® 0 Ol • X > * < 01
ü
-H
&
Eh
tti
id
§
H
|
Colorado School of Mines
|
ER-4609 42
CHAPTER 4
SITE SELECTION
In this chapter, information on the site selection of
tailings dams is presented. Sections 4.1, 4.4.3, 4.4.4,
4.4.8, and 4.4.9 were compiled from information contained in
the Bureau of Mines information circular 8755, Design Guide
for Metal and Nonmetal Tailings Disposal, written by Roy L.
Soderburg and Richard A Busch. Sections 4.2, 4.3, 4.4,
4.4.1, 4.4.2, 4.4.5, 4.4.6, 4.4.7, 4.4.10, 4.4.11, 4.4.12,
and 4.4.13 were extracted from an article written by Earle
J. Klohn titled Geotechnical Investigations for Siting
Tailings Dams. Sections 4.5 through 4.7 were extracted from
Steven G. Vick's book titled Planning, Design, and Analysis
of Tailings Dams.
4.1 Overview of Site Selection (after Soderberg and bush,
1977)
The selection of a site for tailings disposal has to be
made when the plant and mine sites are selected. In the
feasibility study of a new property, a tentative tailings
site must be picked. It should be as close to the mill as
possible, and downstream from the mill for gravity flow of
the tailings. It must be of adequate size to accommodate
the annual tonnage of tailings without too rapid a rise in
the height of the embankment each year.
In a new area and early in the mine exploration period
data should be gathered in the area. All climatic data
should be gathered, and onsite measurements should be made
|
Colorado School of Mines
|
ER-4609 43
of stream flow and evaporation Sedimentation
characteristics, turbidity, pH, metallic ion count, etc., on
the proposed waste should be determined. In the United
States, U.S. Geological Survey (USGS) topographical maps are
usually available. Detailed contour maps of the impoundment
area are necessary for the planning and design of mine waste
embankments. Aerial photographs are useful for locating
geological features that may not be discernible by surface
reconnaissance and mapping and for locating potential
sources of construction materials.
The USGS maps are valuable for reconnaissance surveys,
for choosing a site, for measuring area and volume, and for
general geology, drainage area, creeks, etc. Major faults
should be avoided in the tailings area and especially in the
dam area. By the time of site selection, there should be
enough geological information available to eliminate
tailings sites on any mineralized areas, vein extensions,
potential shaft sites, pit access, or pit extensions. The
site should be far enough from projected mining to preclude
seepage, spills, or runs into the mine through faults,
shafts, or fractures from mining operation.
Habitation downstream from a potential tailings dam
would affect the design in that a higher factor of safety
would be necessary than in a remote area.
4.2 Siting Considerations (after Klohn, 1980)
Design of the tailings storage facility is a site
specific operation. A design considered suitable for one
site might be completely unacceptable at another.
Regulatory guidelines normally require that alternative
|
Colorado School of Mines
|
ER-4609 44
tailings storage areas be identified and the preferred site
selected on the basis of preliminary site investigations.
Once the preferred tailings storage site is selected,
detailed site investigation, leading to ultimate design of
the facility, should be carried out.
The geotechnical site investigations must provide the
designers sufficient data in the preliminary stage to select
the preferred site, and in the final stage to develop safe
and economical designs that satisfy all regulatory
requirements. To achieve these ends the geotechnical site
investigations must cover such items as: topography,
climate, hydrology, geology, hydrogeology, seismicity, site
stratigraphy (soil and bedrock), soil properties
(permeability, strength, compressibility, etc.),
availability of suitable borrow materials for dam
construction, and clay mineralogy and physiochemical
properties of potential soil liners. Not all of these items
normally would be classified as geotechnical. However, as
they are all interrelated and as they all should be carried
out as part of the engineering site investigations, they
will be treated collectively as "geotechnical site studies".
The team required to carry out the necessary site
investigations will be interdisciplinary in nature, and
preferably should be led by a geotechnical engineer having
broad experience and/or training in some of the related
disciplines (hydrogeology, hydrology, seismicity, etc.).
As previously indicated, the geotechnical site studies
should be carried out in stages. The first stage usually
referred to as the preliminary site investigations, should
be designed to provide an overall assessment of site
conditions. Sufficient work should be carried out to define
general site conditions and to identify problem areas for
|
Colorado School of Mines
|
ER-4609 45
each of the alternative sites considered. This permits
comparisons to be made of alternative sites so that the most
preferred site can be selected for detailed study.
The second state, usually referred to as the detailed
geotechnical site investigations, must cover all of the
previously outlined items in sufficient detail that the
designers of the tailings storage facility can produce safe
and environmentally acceptable designs that satisfy all of
the regulatory requirements.
4.3 Preliminary Site Investigations (after Klohn, 1980)
The preliminary site investigations must be
sufficiently comprehensive that meaningful comparisons can
be made between alternative tailings pond areas. The first
step in the preliminary site investigations should involve
the collection of all available data for the area. This
includes such items as:
Topographic maps - usually available from
governmental agencies. (In Canada, topographic
maps are available from the Department of Energy.
In U.S.A., maps are available from U.S. Department
of the Interior - Geological Survey).
Aerial Stereo Photographs - usually available from
Federal governmental sources and private air
survey companies. In the U.S.A., photos are
available from the Photographic Library, U.S.
Geological Survey.
|
Colorado School of Mines
|
ER-4609 47
Resources, Victoria. In the U.S.A., data is
available from the Seismic Records and Earthquake
Data File - National Oceanic and Atmospheric
Administration, Boulder, Colorado.
Supplemental Data - maps and reports of both the
federal and provincial or state departments of
agriculture and forestry.
In conjunction with the collection and study of the
above data, including an airphoto interpretation study to
determine site geology and sources of construction
materials, a thorough on-site examination of the proposed
tailings storage areas should also be made. This
examination should be conducted by experienced geotechnical
personnel. Of prime importance is the surficial geology
(the geology of all soil deposits overlying bedrock). This
is particularly true in areas where many of the valleys have
been infilled with hundreds of feet of soil deposits. Other
geological factors that should also be assessed, include:
evidence of landslide movements, evidence of weak planes
within the rock, evidence of faulting, probable permeability
of the bedrock in mining and the possibility of deep buried
channels. The on-site examination in combination with the
airphoto interpretation normally provides a good preliminary
assessment of the site geology.
The small scale topographic maps provide a means of
studying the general area surrounding the proposed mining
development to locate possible alternative tailings storage
areas. They are also useful for making preliminary
estimates of tailings pond storage volumes, the size of the
runoff area contributing to the tailings pond area, the
|
Colorado School of Mines
|
ER-4609 48
probable direction of groundwater flows, and possible
effects of the tailings pond on nearby developments.
The climatic and streamflow data, when combined with
the topographic data, enable a preliminary assessment of the
runoff characteristics to be made. This provides the
designer with information concerning average runoff volumes
and storm runoff volumes. Estimates of both peak flows and
total runoff volumes are required to assess such items as:
diversion structures, spillways, flood storage surcharge
required on top of the tailings pond, etc.
The combined information obtained from the collection
of published data and the on-site examinations may not be
adequate to permit an assessment of the merits of
alternative sites. In some instances, obvious factors, such
as highly permeable foundations and/or large catchment areas
contributing to the tailings pond, may eliminate a site. In
other instances, surficial vegetation and/or surficial
deposits may mask site conditions, so that further field
work is required before alternative sites can be
realistically assessed. Where this is the case, the next
step in the preliminary site investigations normally
involves digging test pits or test trenches, drilling a few
test holes and obtaining soil and/or rock samples, and
perhaps carrying out some preliminary geophysical work such
as seismic surveys or electrical resistivity surveys. If
test holes are drilled, in situ permeability test should be
run to assess the permeability of the soil and/or rock
underlying the waste impoundment area.
On the basis of the above outlined preliminary
geotechnical site investigations, the designers should be
able to assess the relative desirability, from a
geotechnical point of view, of the various alternatives
|
Colorado School of Mines
|
ER-4609 49
examined. However, it should be realized that because of
the several factors which impact the selection of the
tailings storage area, the most desirable area from a purely
geotechnical point of view may be unacceptable for other
reasons (i.e., too close to inhabited areas, conflict with
other uses, environmentally sensitive issues, etc.).
Consequently, the importance of considering more than one
possible area for tailings storage and assessing the
relative desirability of each area considered, is obvious.
Once the preliminary studies are completed and the most
suitable tailings pond area, compatible with all the
regulatory requirements, has been selected, detailed
geotechnical site investigations must be carried out.
4.4 Detailed Site Investigations (after Klohn, 1980)
Planning detailed geotechnical site investigations
tends to be a site specific operation. Once the preliminary
site investigations have been completed, the general site
conditions are defined and potential problem areas are
identified. Detailed geotechnical site investigations are
designed to answer the problems posed by the specific site.
4.4.1 Topography (after Klohn, 1980)
The scale and contour interval required for the site
maps varies with the function the mapping is intended to
perform. The government published contour maps having
1:50,000 scale and 100 foot contours are adequate for
assessing such items as: contributing watershed for the
tailings pond area; location of the tailings pond with
|
Colorado School of Mines
|
ER-4609 50
respect to existing facilities (towns, farms, highways,
major streams or lakes, etc); possible effects of seepage
losses on the surrounding area; etc.
Once the detailed site investigations are underway,
accurate mapping of the tailings storage area is required.
This is usually achieved by using photogrammetric mapping.
This mapping is fast, economical, and suitably accurate,
provided satisfactory. horizontal and vertical ground
control. Moreover, once the contour interval is selected
and the mapping completed to some standard scale, perhaps
1:2500, it is a simple matter to produce maps having the
same contour interval to any desired scale. A contour
interval commonly used for such mapping is 5 foot. These
maps may be used for computing storage volume for the
tailings pond, for laying out any required diversion ditches
and for laying out the tailings dam and dikes. Prior to
producing final design, surveys are required to produce
accurate layouts and grades.
4.4.2 Climate and Hydrology (after Klohn, 1980)
Climatic data, including such items as air temperature,
precipitation, humidity, wind solar radiation, and
evaporation, and stream flow data are used in assessing the
hydrology of a site.
The hydrology of a tailings storage area is critically
important for assessing the runoff volumes and flows that
may enter the tailings pond. There will always be some
catchment area contributing runoff to the tailings pond.
This area may vary from that of the tailings pond itself, to
a much larger drainage area in the case of an impoundment
|
Colorado School of Mines
|
ER-4609 51
formed by damming a valley into which several streams enter.
In the latter case, it is most desirable to divert all
possible streams and natural runoff around the tailings pond
area.
Substantial runoff volumes and flows can result from
heavy precipitation or snowmelt over relatively small
catchment areas, making the design of suitable diversion
structures a major undertaking. At the end of the mining
operation, unless the diversions are maintained in
perpetuity, the risk exists that they will become
inoperative and that tailings ponds will be subjected to
very large flood flows at some future date. For this case,
the tailings dam must be protected from possible overtopping
which would cause serious erosion and result in tailings
being washed downstream. The required protection would
involve the construction of large, expensive, permanent
spillways to safely handle the maximum possible flood flows.
To minimize these problems, the tailings pond area should be
located such that the contributing catchment area is a
minimum and not much larger than the tailings pond area
itself (See Figure 11) . This should minimize the size of
the permanent spillway required after abandonment and in
cases where the tailings pond has no contributing catchment,
it may be possible to eliminate the spillway by providing
sufficient freeboard to store the predicted flood volume.
4.4.3 Evaporation (after Soderberg and Bush, 1977)
In the arid Southwest United States as much as 84
inches of water per year may be lost by evaporation from a
tailings pond, and this is one of the major water losses.
|
Colorado School of Mines
|
ER-4609 53
depth of 8 inches for 12 to 18 months and evaporation is
monitored. Additional instruments can be installed near the
evaporating pan to relate the measured evaporation in the
pan to meteorological factors. Some of these instruments
are—
1. Wet-and dry-bulb thermometers for air and
precipitation temperatures, vapor pressures, and dew points.
2. Anemometer for wind.
3. Precipitation gages— one non-recording and one
weighing-type recording gage.
Pan coefficients (ratio of lake evaporation to pan
evaporation) are used to estimate the evaporation from lakes
and reservoirs. The evaporation from natural lakes and
reservoirs is 0.6 to 0.8 as much as from the Class A pan, a
coefficient of 0.7 is a good average figure.
4.4.4 Runoff (after Soderberg and Bush, 1977)
Runoff must be considered in designing a mine tailings
pond. The annual spring runoff can best be assessed by even
a few years of records for a given watershed. Where this
information is not available and the watershed is small,
hydraulic handbooks have simple equations to calculate
runoff flow rates. The National Weather Service has maximum
probable precipitation for a general area which can be used,
and assuming a saturated watershed, a runoff hydrograph can
be drawn. The design must be made to handle the 100-year
|
Colorado School of Mines
|
ER-4609 54
flood whether it is by spillway, diversion ditch, decant
tower with discharge lines, or pipe beneath the embankment.
In areas of high snowfall the maximum rain could occur
in the winter on deep snow pack with above-freezing
temperatures. The runoff could be increased by melting of a
large portion of the snow, so that the total runoff could be
even greater than the total rainfall.
A reliable method for estimating runoff volume and flow
requires three steps. Step 1 is the estimation of the
amount of precipitation in the form of rain or snow for a
duration equal to the time of concentration for the area.
This information is available from the National Weather
Service as are the maximum storm and the probable frequency
of occurrences. Step 2 is the assessment of the runoff
losses in the catchment area by vegetation, evaporation,
infiltration, and storage in lakes, etc., all depending on
the characteristics of the area. This step can be
eliminated by being conservative and assuming a saturated
watershed, which often happens when the main storm is
preceded by many days of rain.
Step 3 then assumes that all the precipitation is
runoff and the timing and quantity of the maximum flow are
the only problems. Precipitation and streamflow data from
previous years in the drainage will show the shape of the
hydrograph, which should be more accurate than a synthetic
streamflow hydrograph. Synthetic hydrographs are drawn from
generalized data available on published climatic maps and
records from adjacent areas. To obtain onsite information,
recording and non-recording rain gages, a snow storage gage,
and a recording streamflow measurement gage are necessary.
Streamflow measurements are also required to determine the
state-discharge relationship of the stream gage.
|
Colorado School of Mines
|
ER-4609 55
The National Weather Service has records of
precipitation and the USGS has streamflow records and
hydrographs which can supply information for a specific
watershed not directly covered by their streamflow gages.
4.4.5 Freeboard (after Klohn, 1980)
Regulatory requirements for tailings dams in many areas
stipulate that the design be essentially "closed circuit",
with no water allowed to leave the system unless treated to
meet regulatory water quality standards. This means that
tailings dams and dikes should be designed with sufficient
freeboard to store the maximum design flood without allowing
any uncontrolled discharge of effluent. The size of the
maximum design flood that should be used is an item on which
there is not universal agreement. In the United States, the
Nuclear Regulatory Commission (NRC) requires that for
uranium tailings dams, where the flood runoff is to be
stored by surcharging the tailings pond, the surcharge
capacity should be adequate to store a probable maximum
flood series, preceded or followed by a 100 year flood,
assuming a pond elevation equivalent to the average annual
runoff. The probable maximum flood series is defined as the
probable maximum flood (P.M.F.) preceded by three to five
days by a flood equal to 40% of the P.M.F. Where an
emergency spillway is provided, it should be capable of
passing the P.M.F.. The P.M.F. is defined as the largest
flood that may be expected from the most severe combination
of critical météorologie and hydrology parameters.
The U.S. Corp of Engineers has proposed a set of
guidelines for selecting spillway design floods. Table 12
|
Colorado School of Mines
|
ER-4609
56
0
u
c 0
© 'O ©
e © 2J © > >1
a a 3 k
0 0 JQ 3 CTto to
u to © S 2
© « 3 ©3 c k a. to
> > L, © 0-1 Ol to to
A|o © w 2 X c H — <N 2 <M 2 2
toQ TJto 3 © to m ou k ^ to
o © © © © Q >1'-*
i—1 i—1 33 k Q to I O O O
O © 3 a ©>,3 o o i_> O U
u C U > >, O ti v •
e c (0o U U 3 3 W c 3 3ro T O3 k 2 k 2to t 2o
o c © e ffl D rel © M -HU M©3 U »H O XO J >Ii to >1 to to
v a|a| tuo ot —X o 32u oo uo <a a u c «t m o tX o 0 V O ra» (a 0 ir> «-t .-t 0 ^2 < rHx
to
TJ
O
O
c
—I 2 c
I[y o c
H 0 73
C o< «
to 0
- (H 0 t tt o oo c E i Æo u C© E Ak ©
Û( >U
1 A
o
Oo «<
d
© E
0
©a 2k Va
0
3Ca e -û O >©©k <^C3E
-4>
Uk 3 3k©
3
- C tQ3 OH 7*
>
i ' oo oO c <n V0 T o ooc 1v vD 0 |o o o i n
A|
2H U H g 0Q <<5 J ■© O m 01 30a t> © c0 X© o 7 2© a ©© CO©3 t 3 Awk O k© k3 mJ X 2 tC < k 3 30 ©o0 I x kE t C( k© OEo 0 : X X ■3kk ©t ( C©o H30I i ■ 2© ©1 c © k OC0 0E
1
7 M© © C©3 ,© ©kD 3l r 0Ek
1
7 H© © EE C3 |O k ©© 3l UE
l
7 M© ©© C3 3©© D k 1l
C
-H
-P
ÜQ)
.—.
rH O
0) 00
co cn
M
o
c
w X oi 0)
(UI—I
c « C T 03 ) d
-iH 0 e 0
I—Ip D 0) •—t m © C ©r
<DQ)
10 EC©
•d4-> u ^ J U
•h m
3 to
O '—'
CN
i—I
H
PI
<
|
Colorado School of Mines
|
ER-4609 57
summarizes these guidelines. From Table 12 it can be seen
that the recommended design floods are based on the height
of the dam, the volume of water stored, and the potential
hazards to downstream life and property, and they range
between a 50 year recurrence frequency and the P.M.F..
Current good engineering practice for both major
conventional water storage dams and tailings dams requires
the use of the P.M.F. for design of the spillway. Where
flood storage is to be achieved by surcharging the tailings
pond the design normally involves some combination of wet
year runoffs and the P.M.F. The method proposed by the NRC
represents one such procedure.
Tailings ponds are structures that either should have
spillways designed to safely pass the P.M.F., or provide
sufficient freeboard to safely store a combination of wet
year runoffs and the P.M.F. This requirement should not
place an unreasonable burden on the tailings pond design
provided it has been sited such that it has a small
contributory watershed. If this is not possible, large
diversion and/or spillway structures, with all the problems
of perpetual maintenance, will likely be required.
4.4.6 Geotechnical Investigations (after Klohn, 1980)
Geological and subsurface investigations are closely
interrelated and are usually carried out simultaneously
under the collective name of "geotechnical investigations".
The main objectives of the detailed geotechnical
investigations should include determining:
The detailed soil and bedrock stratigraphy.
|
Colorado School of Mines
|
ER-4609 58
including depth, thickness, continuity, and
composition of each significant stratum.
The site geology, both surficial and bedrock.
This study should include a history of deposition
and erosion, including glaciation, and should
cover such items as: buried channels; collapsing
structures ; solution cavities ; tectonic movements
and faulting; weak formations such as shear
planes, bentonite layers, mylonite seams, etc.,
and site seismicity. Geologic maps should be
produced showing both the surficial and subsurface
geologic features of the site.
The site hydrogeology, including : definition of
all aquifers and aquicludes, determination of
bedrock topography and thickness of unconsolidated
sediments; determination of the piezometric
pressures in all aquifers; determination of
hydraulic conductivity; determination of local
and regional groundwater flow systems.
The geotechnical properties of the soil and rock
strata that may affect design of the waste
impoundment structure. For soils these should
include: water content, grain size determinations,
Atterberg limit tests, consolidation tests,
triaxial compression and/or direct shear tests,
permeability tests, and ion exchange capacity of
days proposed as impervious liners. For rock
these should include : shear strength along weak
layers and permeability of various strata.
|
Colorado School of Mines
|
ER-4609 59
The availability of suitable construction
materials for building dams or dikes and
impervious linings.
4.4.7 Geophysical Procedures (after Klohn, 1977)
The principal method used for determining detailed soil
and bedrock stratigraphy is drilling and sampling. Soil
samples are usually taken at 5 foot intervals with the type
of sample taken depending on both the nature of the soil and
the purposes of the sample. Samples of bedrock are obtained
by continuous coring using diamond drills and double or
triple core barrels. The types of drills and methods of
sampling for various soil conditions are indicated in Table
13.
Geophysical methods such as seismic and electrical
resistivity are often used to complement the drilling and
sampling programs. Under certain conditions these methods
can prove invaluable for quickly determining bedrock and
water table profiles. However, they require careful
interpretation in conjunction with geological information,
and work best where geologic .conditions are relatively
simple. Downhole geophysical methods such as seismic
electrical resistivity and neutron and gamma logging are
often used to correlate similar rock or soil strata between
different drill holes. These methods can be very effective.
Drilling records should be carefully maintained for all
exploration holes. They should contain a record of all
water losses or gains and the location of soft or shattered
zones of rock, as well as a detailed description of all soil
land rock strata encountered. The borehole water level
|
Colorado School of Mines
|
ER-4 609 61
d 4J
E 2 ^ eu o.
2 Z
5 1 E* <u « 5 S a* o-S* w o. O O Dû — 2 c OJ W -o
i ” « S 2 u) 3 ? o o •° i > n u w -~t c - Û.M-. • X V? O (0
« «9 " "O Er 12- ^U CO SU ” u SO IS3 o1
j
2 i*- » 3Û -*
2
2YV
!
:
1"
- eg
u">*
^^1- oeu °2 (O) " -Co
2
2<C 9L >L z*S J zW - àC "o
2
jeL — 51 1en See nn tC
u g- ™ % 2 o ” 2 3 o % g-« ? J o! S o 1 s z 2 5 OJ*o. "o 2 2
^ 3 01 2 ( o U <0 O 5 >; g-12 ° 2 V- X
^ 1 jC û. AJ «U JD
00 '° QO < °
o c 5 —1c t 1
^ s .
- c 2 C ^ 2 | 2 2 E - OJ 2
o *o ^ c 5 c 5 S U4 3 f o"
u Æ 2 5 rt « u 3 —1 en —4 g- 2 2 E 66 X 4J OJ1
u 5 'o. o ^ « E o oô S £ i ü Ü 1 ; Û. o y S' - c ë ë
CL '*-« e o. ^ e 2 “ -S "S. o “ c “ o c L- £ 5 2 3 Cp i -o
0 > d
« C *0 m
C5
•— |*o
°
®
<0
üS —g 1) ^« 7 uj " o2 S ai. g
<uc
>
2. U
T>
3
s
nC
^um
■O2
i
*o ° 5
U
2
i OJ. i
3
• ü2 f« lî «E " -W 2 ^” C ^C ' U u -O 2l ^3 E O Cf ë ■ C« ? « Su î ° 2u £m o Sae £ c £ ?? 2c 2 < UU 1 g)22 2“ X 33 2S Hui J £OJ 5 ^ 2 AJ 2U « 3M o-, zA sJ -
5 ’ o -i < ° S ^ ° Q en F "° <a < •° « AJ•o
>> c 5 e—1c , 1
• O 2 -, ^ 2 _ e eS5 2 E • OJ “ g . 2
O -O^ 3 C > î 15 u *r4 U —e en c <u S ü 3 i u - E o*
S 3
—C
*
—3
4
..i
U*
■* O2 u(D U 3 ^"q. W ™ ! 0Û. O
O -
u o-, S
>
° coX2 g 00 X
-
A CJ OJg q.2
^-4 (ft E d u ij C U-, C 2i -% g " 1 “ 1 ■u c U O <02 T gL u Ts•o c 3
jn u ^ E o.2ii 5 jQ > OJ c x "2 > 4J3 C £ V 2 <D« 2 2
" I w « _ | ^ 0 -y 2 " s i E" ° 1 2 5 : 3 c Z S Z o S s-l
r «o | . «iï " 1 3w 3 'o «3. 1 jCU• c «o
Û..:
Og- ) c „
1
«/J - -a ”O
o
U
O.
^ £3
c
-°
2
«
£" ua, -- £I 25 , £1 2E£c ga *2a
c
?s a 2i cI A 22J
5
S
X 23 2V «C aL ü2g i<
2 Z 2 z
2AJ iU >> U 3°A z -A
C
SJ
L
l <
y ff) c* c ^ >
c
7
i 1
-4 « i o. s 5 UI | i u u|
o o
c 5 > « « § ! —‘ | > 2 T*
o o 2 o. « 2 *x 1 f ü5 a •—1 "w ° 2 OJ 1<u jS j "i •£
^ 3 <U « ^ «A E 5 1 U ü«—* o
01 ^ ° (S " -° i2 5 ^ <
g> C >s U s 7
O § 10 *o o u d 5 "E -iic I _
en. o O -C O0 O
^ 1 S 3
g 1 C O 01 > -»-( -H
v) - SS 5
oj
3 2r • C2
/l
-r§ 4s 2° •2 ■2tû
-*
-c
o
5S" ^3 z
u
O0# 'r S-C
- T
2 E O. ^ U C 1 jj 2 « g° - 1 2 ë 3 I ■"Es
X ^ 3
al ° al < -° ^ °
g : « % >> 2 C £5
« 1 ^ ’ ü >! o o 2 u O oo *4-, ^ 3 ai A aJ i S m
o m ° ° c c c c1
* w ï C ”
E: - *- — 21 m «2 U w E. ? ^ g 3 V — i O 3
H a,~ S ° o 55 i•E « (U ^ U en p s
.deunitnoc(
31
ELBAT
|
Colorado School of Mines
|
ER-4609 62
should be measured at regular stages during drilling. Rock
cores should be carefully logged with particular attention
to such features as: core recovery, weathering, fractures,
joints, faulted zones, solution cavities, etc..
Tailings pond facilities are considered to be hydraulic
structures as they retain water as well as tailings. Three
basic properties that must be determined for the foundations
and abutments of all hydraulic structures are : the
compressibility, the shear strength, and the permeability.
In addition, the permeability of the tailings pond area
proper must be determined to ensure that unacceptably large
seepage losses do not occur. Seepage losses that can be
safely accepted for a conventional water storage dam may not
be acceptable for some tailings ponds because of the
contaminants contained in these losses.
4.4.8 Compaction (after Soderberg and Bush, 1977)
The moisture-density relationship for compacting soil
is obtained by the Standard Proctor method or the Modified
Proctor method. In the early days of compaction, when
construction equipment was small and gave relatively low
densities, the Standard Proctor density was the expected
value to be attained in the field. As construction
equipment and procedures were developed which gave higher
densities, the Modified Proctor method with over 4-1/2 times
the compactive effort of the Standard was adopted.
A definite relationship exists between the water
content of a soil at the time of placement and the amount of
compactive effort required to achieve a given density. If
silts and clays are too wet or too dry, the maximum density
|
Colorado School of Mines
|
ER-4609 63
will not be attained with a given compactive effort. The
objective of the laboratory procedure is to determine the
optimum water content and maximum density for the specified
compaction effort.
Sands are not as moisture dependent but should be
compacted either saturated or completely dry to avoid the
effect of "bulking".
4.4.9 Shear Strength (after Soderberg and Bush, 1977)
The shear strength of a soil may be measured by
triaxial compression tests or direct shear tests. Triaxial
tests measure the shear strengths under both drained and
undrained conditions with the sample maintained as near
field conditions as possible. Direct shear tests can define
shear strengths under limited conditions of moisture and
confinement.
Using the shear strength of a soil is common practice
for the design of an earth starter dam for tailings
disposal. In this case, the soil can be mechanically
compacted to a density where the shear angle, cohesion,
permeability, etc., needed for design can all be determined
by laboratory testing. To determine the same physical
properties for the tailings is a bit more difficult because
it is generally deposited hydraulically with no compaction.
If an old tailings pond is available for undisturbed
sampling and testing, these figures can be obtained and used
to determine the physical properties and geometry.
If an old tailings pond is not available, such as in
the case of an entirely new property, some assumptions must
be made as to the density of the deposited material, from
|
Colorado School of Mines
|
ER-4609 64
either laboratory tests or information from another property
with the same grind and rock types. The screen analysis,
mineralogy, pulp density of deposition, and cyclones, if
any, affect the material characteristics that determine the
shear angle, cohesion, drainage, etc. Soil shear strength
is also affected by many test factors including such items
as rate and method of loading, principal stress ratios,
degree of saturation, drainage, rate of specimen strain, and
total specimen strain. In selecting the shear strength
parameters that are to be used for specific analyses, an
estimate must be made of the probable strains and rates of
pore pressure dissipation under field conditions, and a
decision must be made as to whether "peak" or "residual"
shear strength values should be used to determine the angle
of internal friction o. Generally, if the void ratio value
(e) is small, the peak o is used. If e is high, residual a
is used. More testing is required to determine the shear
strength characteristics of soft soil than firm soil.
Triaxial shear testing is a very exacting process that
requires good equipment and much training and experience,
and it is best left to specialists in this field. The
sample must be taken with care in the field, prepared for
transport, and transported with minimum disturbance.
Extreme care must be used. The unconsolidated, undrained
test is described in ASTM D2850-70. This is an important
test because of its use in stability analysis.
Data obtained from shear strength tests are normally
presented in terms of effective stresses. During triaxial
testing, both total stresses and pore water pressures are
measured. The effective stress is the total stress minus
the pore water pressure. Plots of effective stress at three
different confining pressures for soil specimens at failure
|
Colorado School of Mines
|
ER-4609 66
permit assessment of probable seepage losses. In soils, the
permeability is influenced by particle size and thus pore
space, and the distribution of particle sizes or thedegree
of homogeneity. Coarse, granular soils exhibit the highest
permeabilities and fine-grained soils the lowest. Figure 13
presents a summation of the range of permeability for soils.
Also indicated in Figure 13 are types of permeability tests
applicable to each class of soil.
The permeability of intact rock is usually low. In
most rock masses, groundwater flows largely through
discontinuities such as joints and fissures. The nature,
orientation, and continuity of the joints and fissures
determine the permeability of the rock mass. Highly jointed
rocks, with open joints can be more pervious than clean
coarse gravels. Table 14 presents a range of typical
permeability values for soils and rocks.
The permeability of soils can be determined by
laboratory testing, or in the case of granular soils,
estimated from grain size curves. However, laboratory tests
may not accurately reflect the in situ permeability of the
soil as this depends on such factors as degree of
stratification and continuity of individual stratum.
Consequently, although laboratory permeability values may be
used for estimating the seepage flows through soils, these
are usually checked by in situ tests. Rock permeabilities,
because they depend on discontinuities in the rock mass,
cannot be determined by laboratory tests on intact rock and
consequently must be determined by in situ field testing.
Methods which are commonly used for direct in situ
measurement of permeability are summarized in Table 15.
Borehole permeability tests are the most commonly used in
situ permeability tests because of their ease of performance
|
Colorado School of Mines
|
ER-4609 71
and relatively low costs. However, they have the
disadvantage of testing a relatively small zone and large
numbers of carefully run tests are required.
Large scale pumping tests are usually performed by
pumping water out of a screened well in an aquifer. As
pumping proceeds the resulting lowering of groundwater
levels is measured in observation wells. Large scale
pumping tests are currently the preferred method for
determining the in situ permeability of soils and rock.
However, they are costly to perform, and the successful
location, execution and interpretation of well pumping tests
requires experienced contractors and hydrogeologists.
4.4.11 Groundwater Conditions (after Klohn, 1980)
To establish the existing groundwater regime,
piezometric pressures in the underlying soils and rock must
be determined. This is done by installing piezometers which
are instruments designed to measure water pressure at a
specific depth. Piezometers may vary from a simple
standpipe type, which measures the water level at a given
depth, to remote monitoring, electrical strain transducers.
The type of piezometer selected for a given installation
depends on the expected permeability of the soil or rock
into which it is installed and the required response time.
The most important characteristics required of any type of
piezometer are ruggedness, adequate accuracy, and long-term
reliability. Piezometer locations are selected to provide
groundwater pressures relevant to the design or monitoring
requirements of the particular site.
Detailed water monitoring programs should be set up
|
Colorado School of Mines
|
ER-4609 72
well before the start of operations to establish natural
background levels of contaminants. The required water
sampling program should be combined with the groundwater
regime studies, the in situ drill hole permeability tests,
and the pumping tests so that maximum use can be made of
these drill holes, wells, and piezometers in the water
sampling studies.
4.4.12 Seismicity (after Klohn, 1980)
In areas where seismic disturbances may occur, analyses
are required to determine the effects of the seismic forces
on the proposed tailings dams. Before such seismic analyses
can be made, it is necessary to select the magnitude and
location of the design earthquake for the site. Selection
of the magnitude of the design earthquake normally involves
obtaining the records for all recorded earthquakes in the
area together with statistical analyses predicting
magnitudes for various periods. Also required is a detailed
geological assessment of the structural geology of the area,
with particular attention to existing faults and their
history of movement. The exact procedures used to estimate
the design earthquake vary, but all methods involve a large
degree of judgment. In any event, as earthquake records are
only available over a relatively short period of time, the
most severe earthquake recorded for an area cannot be
assumed to be the largest that could occur in that area. A
procedure sometimes used to determine the hypothetical
earthquake that should be selected as the design earthquake
is to take the largest recorded earthquake for the area and
increase it by 1 Richter magnitude (10 times).
|
Colorado School of Mines
|
ER-4609 73
A method commonly used to determine the effects of the
selected design earthquake on a particular site, is to
assume that the earthquake occurs on the closest known,
possibly active fault. The fault is selected on the basis
of the geological studies that have been made for the area.
Attenuation tables are then used to estimate the magnitude
of the earthquake forces reaching the site as a result of
the design earthquake occurring on the selected fault.
If the geotechnical site investigations indicate that
the embankment is underlain by loose, saturated sand or
sensitive silt, or if the embankment itself is constructed
of such materials, the possibility of liquefaction occurring
must be considered. Liquefaction occurs when the dynamic
earthquake forces cause the pore water pressures within the
loose sand to rise to such high values that the deposit
loses its strength and liquefies. To make a liquefaction
analysis requires dynamic laboratory shear strength tests
on "undisturbed" samples of the loose, saturated soil to
determine its dynamic shear strength parameters of the soil
and incorporate the predicted earthquake forces, are then
used to determine the factor of safety of the foundation
against liquefaction.
Embankments and foundations composed of cohesive soils
or dense granular soils normally are not subject to
liquefaction under earthquake forces. These materials
usually exhibit little strength loss or build-up of pore
pressures during earthquake shaking. Dynamic stability
analyses, both simplified and rigorous may be used to assess
the stability of the embankment. Experience has shown that
provided the embankment and its foundations are not subject
to liquefaction failure, earthfill embankments can safely
withstand large earthquake forces without suffering
|
Colorado School of Mines
|
ER-4609 74
appreciable damage. Tailings dams, constructed from select
borrow materials to the same general standards as water
retention dams should be expected to resist earthquake
damage in a similar manner.
4.4.13 Instrumentation and monitoring prior to
construction. (after Klohn, 1980)
Piezometer installations made for the prime purpose of
measuring water pressures in both the overburden and bedrock
strata are one of the major items of preconstruction
instrumentation. All piezometers should be read on a
regular basis throughout the entire year. This permits
seasonal variations and trends in piezometric pressures to
be observed and recorded.
Water quality determinations should also be made on
samples taken from the various identified aquifers. The
piezometer installations, whenever possible should be
designed in such a manner that they may also be used as
water sampling stations and thus reduce or eliminate drill
holes required for water sampling programs.
Early in the site investigations, a climate-hydrology
package of observation should be formulated. These data are
scarce to non-existent at many mining sites and the
designers are usually forced to extrapolate from the nearest
site having such data when making their preliminary designs.
Climate-hydrology observations are therefore needed as soon
as possible to provide a check on the extrapolated data and
to start an accumulation of climate-hydrology data that
applies to the specific site. The climatic data required
include : precipitation, evaporation, air temperatures,
winds, and humidity. (Remote-reading weather stations.
|
Colorado School of Mines
|
ER-4609 75
which perform all of the required climate measurements are
available at relatively small cost). The hydrology data
required include streamflow measurements and snowpack
measurements (thickness and water content).
Performance monitoring of the operating tailings pond
likely would include such items as: piezometers, settlement
gauges, alignment gauges, inclinometers, and water quality
measurements. Post-operative monitoring likely would be
primarily concerned with water quality wind and water
erosion, and radon gas emissions for uranium tailings ponds.
4.5 Impoundment: layout (after Vick, 1983)
Impoundment layout is an integral part of the siting
process, since the suitability of a particular site cannot
be fully established without confirming that the site will
accept a particular impoundment configuration. Like
impoundment sites, impoundment layouts exist in infinite
variety. Nonetheless, several categories of impoundment
layouts can be defined that are generally compatible with
various topographic settings. Impoundment layout types
considered in this chapter include :
Ring dikes.
Cross-valley impoundments.
Sidehill impoundments.
Valley-bottom impoundments.
4.5.1 Ring Dikes (after Vick, 1983)
The ring dike impoundment layout is shown schematically
in Figure 14. Best suited for flat terrain in the absence
|
Colorado School of Mines
|
ER-4609 77
4.5.2 Gross-valley impoundments (after Vick, 1983)
Cross-valley impoundments, illustrated in Figure 15,
differ little in layout from a conventional water-storage
reservoir. As the name implies, the cross-valley
impoundment is confined by a dam extending from one valley
wall to another. Cross-valley type layouts can be nearly
universally applied to almost any natural topographic
depression, in either single- or multiple-impoundment form,
as shown in Figure 15, thus accounting for the prevalence of
this layout for tailings disposal. Paramount in the use of
the cross-valley layout is that the impoundment be located
near the head of the drainage basin to minimize flood
inflows. While sidehill diversion ditches can be used to
reduce normal runoff accumulation in cross-valley
impoundments, larger diversion channels to pass peak flood
flows around the impoundment are often not feasible because
of steep valley sidewalls. Flood runoff from large drainage
catchment areas can often be handled for cross-valley
impoundments only by storage, spillways, or separate water-
control dams upstream from the tailings impoundment.
4.5.3 Sidehill impoundments (after Vick, 1983)
The sidehill impoundment layout is shown in Figure 16.
This layout type encloses the impoundment by embankments on
three sides and therefore generally requires more fill than
the cross-valley option. This type of impoundment, however,
can be used where no incised drainages suitable for cross
valley impoundments are available— for example, on mountain-
front alluvial pediment deposits or where the available
|
Colorado School of Mines
|
ER-4609 80
incised drainages would have an excessive catchment area.
This type of layout is best suited for sidehill slopes of
less than about 10% grade; on steeper slopes, fill volumes
may become excessive in relation to storage volume achieved.
For downstream-type embankments, the upstream portion of the
embankment itself may occupy a significant proportion of
what would otherwise be impoundment storage volume.
4.5.4 Valley-bottom impoundments (after Vick, 1983)
Valley-bottom impoundments, depicted in Figure 17,
represent a compromise between cross-valley and sidehill
layouts. The valley-bottom option is well suited for cases
where the drainage catchment area would be too large for
cross-valley layouts, but hillside slopes are too steep for
practical application of the sidehill option. Since the
impoundment is enclosed by embankments on two sides, fill
requirements are generally intermediate between those for
cross-valley and sidehill layouts. Valley-bottom
impoundments are often laid out in multiple form, as shown
on Figure 17, in order to "stack" the impoundments one above
the other as the valley floor rises.
Central to the use of the valley-bottom layout is a
diversion channel to carry the full peak flood flow around
the impoundment. Diversion is usually necessary since these
impoundments, commonly located in relatively narrow valleys,
are often constructed across the stream channel. The
diversion channel usually corresponds to the gradient of the
original stream channel but is constructed tight against the
opposing valley wall. During initial layout, if sufficient
space is not allocated for the diversion channel, costly
|
Colorado School of Mines
|
ER-4609 82
excavation in valley sidewall rock may be required to
achieve necessary channel widths. Because peak flows under
PMF or similar flood conditions are usually large, widths
for diversion channels associated with valley-bottom
impoundments are often considerable. Excavated material can
often be conveniently used as starter dike fill. In
addition, it is frequently the case that high-velocity flow
will occur against the outer embankment face under design
flood conditions, requiring that lower portions of the
embankment be protected by riprap. This can make the use of
centerline or downstream embankment rising methods awkward
because of the need to continually replace the riprap as the
embankment face moves outward with progressive raises.
4.6 Single versus multiple impoundments (after Vick, 1983)
All four impoundment layout options described can be
implemented in either single-or multiple-impoundment form.
While the best choice depends on specific site conditions,
some general advantages and disadvantages apply.
Multiple impoundments usually require a greater total
quantity of embankment fill. In the extreme case for ring
dikes illustrated in Figure 14, 1.5 times as much fill is
required as for a single impoundment to achieve slightly
less total storage volume. In other cases, however,
particularly as illustrated in Figure 17 for valley-bottom
impoundments, the fill penalty is not so severe and multiple
impoundments may significantly aid in achieving the required
storage volume in a limited available space. Also, for
multiple cross-valley and sidehill impoundments shown in
Figures 15 and 16, the uppermost impoundment bears the full
|
Colorado School of Mines
|
ER-4609 83
burden of flood runoff inflows. Since the size of the
individual impoundment segment is much less than for a
single large impoundment, excessive flood storage
requirements for the uppermost segment may result, and
careful planning for control of surface water is required.
These disadvantages notwithstanding, the benefits from
multiple impoundments can be considerable, again depending
on individual site conditions. In general, multiple
impoundments are constructed sequentially, allowing for
smaller initial capital expenditures and producing cash-flow
benefits much the same as those realized for raised
embankments. Multiple impoundments also offer considerable
operational flexibility. Impoundment segments can be
constructed either strictly on an as-needed basis or in
advance of actual tailings storage requirements as fill
material or construction equipment become available. When
more than one segment has been constructed, discharge of
tailings can be alternated between the impoundments to
provide beneficial flexibility in impoundment operation.
Environmental benefits for multiple impoundments
compared to single impoundments of equivalent capacity can
be major. Generally, multiple impoundments are constructed
and filled sequentially. Thus, only a small portion of the
eventual total impoundment area is covered with water at any
given time. To the extent that seepage is directly
proportional to the area over which flow occurs, seepage
rates may be considerably reduced. At least as significant
is the fact that reclamation can proceed concurrently with
ongoing tailings disposal. Following filling of one
multiple-impoundment segment, reclamation can begin as
discharge is shifted to the next segment, thus minimizing
the area disturbed at any one time and reducing problems
|
Colorado School of Mines
|
ER-4609 84
related to blowing dust.
4.7 Optimization of impoundment layout (after Vick, 1983)
For a given impoundment layout, site, and embankment
type, there is often an optimum combination of embankment
height and impoundment area that will give the lowest fill
volume for the required storage capacity. The concept of
fill efficiency ratio is useful in this regard. The fill
efficiency ratio is defined as the ratio of impoundment
tailings storage volume to the volume of fill required to
achieve that storage. Since fill volume is usually closely
related to the cost of the impoundment, the fill efficiency
ratio provides an indirect indicator of relative impoundment
cost and is useful, not only in optimizing embankment height
and impoundment area for a given storage volume, but also in
comparing the relative costs of different impoundments with
dissimilar capacities. When applied to compare impoundments
at different sites, the fill efficiency ratio properly
penalizes those sites where higher embankments are required
for storage of flood runoff, since the ratio is defined in
terms of available tailings storage volume rather than total
reservoir volume.
To illustrate the use of the fill efficiency concept,
consider the simplified example shown in Figure 18. The
assumed embankment and impoundment configurations are shown
in Figure 18a. A sidehill-type layout is being planned on
ground sloping at a uniform 5%. The impoundment width is
fixed at 2,000ft by site boundaries, but impoundment length
and height can vary. The problem thus becomes to select the
most efficient height and length of the embankment.
|
Colorado School of Mines
|
ER-4609 86
Figure 18b shows the fill efficiency ratio plotted
against embankment height and impoundment length. Small
impoundments of low height produce less storage volume per
unit fill volume, as do large impoundments with high
embankments. For this particular example, the maximum fill
efficiency ratio yields an optimum embankment height of
about 50ft and a corresponding length of about 1,200ft. The
resulting storage volume might or might not be compatible
with volume requirements dictated by the mill tailings
output, but for large storage requirements this example
would suggest that a series of multiple sidehill
impoundments of the optimum dimensions would be efficient.
While the geometry of natural drainage basins is much
more complex than indicated by this example, it is often the
case that fill efficiency decreases for very high
embankments. Minimum embankment height is dictated by
tailings and flood storage requirements, but there often
comes a point of diminishing returns where new impoundments
at different sites would be more efficient than excessive
raises of a single large impoundment.
Another example illustrating fill efficiency is shown
in Figure 19. In this case, suppose that a 15-yr tailings
storage volume is required for a mill output of 2,000T/day.
The tailings will have an in-place dry density of 90pcf,
resulting in a tailings storage volume requirement of about
2.4 x 108f t3. The assumed impoundment is a ring dike
layout, as shown in Figure 19a, with square sides of length
L. The ground surface is assumed to be flat, and runoff
water inflow is negligible.
In this example, the problem becomes to select a
combination of impoundment height and area to achieve the
given storage volume. This can be accomplished by either a
|
Colorado School of Mines
|
ER-4609 88
very low, large impoundment or a high small one. The graph
in Figure 19b illustrates the diminishing returnsfor high
embankments in terms of their large fill volume and
indicates that the given storage volume can be achieved with
minimum embankment fill quantity for low dikes and a large
impoundment.
Both of the preceding examples have defined fill
quantity in terms of embankment fill only. In some cases
this can produce misleading results. For instance, the
example in Figure 19 is repeated in Figure 20, only this
time assuming that a 3-ft thick compacted clay liner is
required over the impoundment bottom to minimize seepage. A
similar situation might result from requirements for the
impoundment topsoil stripping or replacing topsoil over the
impoundment surface as a part of reclamation. Comparison of
Figure 19 with Figure 20 shows that total fill volumes are
substantially increased by the clay liner. More significant
is that the liner fill, being a function of impoundment
area, penalizes larger impoundments. For the example in
Figure 20, storage requirements can be met with a minimum
total fill volume for an embankment height of about 30ft and
impoundment width of 3,000ft. This serves to emphasize the
point that optimizing impoundment layout must often account
for earthwork requirements in the larger sense, including
reclamation and seepage-control measures, rather than being
based strictly on embankment fill requirements. Although
the above examples are simplified, the same general
principles apply to more realistic impoundment topography
and to all types of impoundment layouts. Establishing the
optimum layout is a trial-and-error procedure. While
experience is of considerable assistance, an optimum layout
can be arrived at using the concept of fill efficiency.
|
Colorado School of Mines
|
ER-4609 90
CHAPTER 5
TAILINGS DAM CONSTRUCTION METHODS
The material presented in chapter five has been
extracted from two sources. Sections 5.1, 5.3, 5.4, and 5.6
have been extracted from The Development of Current Tailings
Dam Design and Construction Methods, written by Earle J.
Klohn. Sections 5.2 and 5.5 have been compiled from Steven
G. Vick's book titled Planning, Design, and Analysis of
Tailings Dams.
5.1 Upstream Construction Using Tailings (after Klohn, 1980)
This is the oldest method of tailings dam construction
and is a natural development of the procedure of disposing
of the tailings as cheaply as possible. The dam is normally
constructed by spigotting in an upstream direction off a low
starter dike. Various methods are used to raise the dam
when the pond level nears the top of the starter dike. The
most common method of upstream construction is to raise the
dyke by dragging up material from the previously deposited
tailings as shown on Figure 21. Another procedure often
used in the past and still in use at a few sites, involves
using vertical timber forms to raise the dam in 2 to 3 foot
increments. Figure 22 is a sketch illustrating how the
procedure works. There are, of course, many other
variations of the upstream method, but perhaps the most
interesting is that shown on Figure 23 which illustrates the
use of cyclones to raise the dike.
|
Colorado School of Mines
|
ER-4609 94
The chief advantages of the upstream method of tailings
dam building are the low cost of construction, and the speed
with which the dam can be raised by each successive
increment of the dike. By using cyclones, the speed of
construction is increased, and a lead can be developed
between the crest of the dam and the top of the pond.
All upstream methods of tailings dam construction
suffer the disadvantage of being built on top of previously
deposited, unconsolidated tailings. Under static loading
conditions, there is a limiting height to which such a dam
can be built without danger of a shear failure. This height
will depend on the strength of the tailings within the zone
of shearing, the downstream slope of the tailings dam, and
the location of the phreatic line within the dam. A rise in
the phreatic line due to heavy rainfall, or blockage of
seepage outlets could cause failures. Under earthquake
loading, this type of dam can fail by liquefaction.
Figure 24 compares the water storage dam with the old,
upstream construction type of tailings dam. An examination
of the figure shows that this type of tailings dam does not
meet conventional dam requirements for slope stability,
seepage control (internal drainage),and resistance to
earthquake shocks. Any change in conditions that would
result in saturation of the outer sand dike could quickly
lead to failure by piping or sliding. Potential causes of
saturation include such items as: a rise in water levels in
the pond, freezing of seepage outlets on the downstream face
of the dam, and torrential rainfall. The upstream methods
of dam building are unsuitable for earthquake prone areas.
|
Colorado School of Mines
|
ER-4609 97
In general, downstream raising methods are well suited
to conditions where significant storage of water along with
the tailings is necessary. Because the phreatic surface can
be maintained at low levels within the embankment and
because the entire body of the fill can be compacted,
downstream raising methods are liquefaction resistant and
can be used in areas of high seismicity. Unlike upstream
embankments, raising rates are essentially unrestricted
because the downstream raises are structurally independent
of the spigotted tailings deposit. Downstream raising
methods are essentially equivalent in structural soundness
and behavior to water-retention type dams.
Downstream raising methods, however, require careful
advance planning. Because the toe of the dam progresses
outward as its height increases, sufficient space must be
left during layout of the starter dike to prevent
encroachment of the dam toe on property lines, roads,
utilities, diversion ditches, or topographic constraints.
The ultimate height of the embankment is often determined by
such restrictions at the toe.
The major disadvantage of the downstream raising method
is the comparatively large volume of embankment fill
required and the corresponding high cost. The availability
of fill for various raises of the dam may also impose
constraints on construction. In particular, if mine waste
or sand tailings are used for embankment construction, these
materials will be produced at a more or less constant rate.
The volume of fill required for each successive downstream
raise often increases exponentially as the embankment
increases in height. Advance planning is required to ensure
that fill material production rates will be sufficiently at
all times during the life of the embankment.
|
Colorado School of Mines
|
ER-4609 98
This problem is illustrated in Figure 26. Figure 26a
shows the elevation-volume curve for the impoundment, which
is strictly a function of topography. For a given rate of
mill tailings discharge, the impoundment elevation versus
time curve in Figure 2 6b can be derived. The elevation of
the tailings surface starts at zero and increases with time,
but typically at a decreasing rate. In addition, an
impoundment depth increment sufficient for storage of storm
runoff inflow must also be accounted for, as shown by the-
higher curve in Figure 2 6b. Storm runoff volume is usually
constant over time, and the impoundment depth allowance at
time zero represents the starter dike flood storage
capacity. Although the volume remains constant, the depth
required to retain this volume decreases with time because
of the increase in impoundment area at higher elevations.
Figure 2 6c shows the volume of dam fill required as a
function of dam crest elevation, a relationship determined
by the cross-section of the dam and topography along the dam
alignment. For the downstream method, each raise of
constant height requires increasingly greater fill volumes
to construct. Figure 2 6d shows the dam fill volume required
as a function of time, derived from Figures 26b and 26c.
The volume of fill required to keep the dam crest above the
elevation of the tailings (plus flood storage allowance)
increases exponentially with time. Superimposed on Figure
2 6d is a constant rate of fill production, such as mine
waste, assuming that the starter dike is constructed of
natural soil borrow. Although curves must be established
for each individual case, in the example shown fill
production is adequate initially, but then becomes
insufficient at higher dam elevations after longer periods
of time. This problem can be resolved by constructing a
|
Colorado School of Mines
|
ER-4609 100
higher initial starter dike of native soil borrow, shifting
the fill production curve in Figure 2 6d upward. Other
downstream construction methods are illustrated on Figures
27, 28, and 29.
5.3 Centerline construction (after Klohn, 1980)
The centerline method of tailings dam construction is
actually a variation of the downstream method. The only
difference being that instead of the crest of the dam moving
downstream as the dam is built, the crest is raised
vertically. This procedure allows the dam to be raised
faster, as less sand is required. Figure 30 illustrates one
type of centerline construction. The major advantages of
downstream methods of dam building are :
None of the embankment is built on previously
deposited, loose, fine tailings.
Placement and compaction control can be exercised
over the fill operation.
Underdrainage systems can be installed as
required, as the dam is built. The underdrainage
permits control of the line of saturation through
the dam and, hence, increases its stability.
The dam can be designed and subsequently
constructed to whatever degree of competency that
may be required, including resistance to
earthquake.
|
Colorado School of Mines
|
ER-4609 105
Usually, the dam can be raised above its original
ultimate design height with a minimum of problems
and design modifications. This is critically
important for most mining operations where the
original life of the mine might be extended by new
ore discoveries, higher metal prices, new methods
of metal extraction, etc.
The major disadvantage of all methods of downstream dam
building is the large volume of sand required to raise the
dam. In the early stages of operation it may not be
possible to produce sufficient volumes of sand to maintain
the crest of the tailings dam above the rising pond levels.
If this is the case, then either a higher starter dam is
required or the sand supply must be augmented with borrow
fill. Both procedures add to the cost of the initial
tailings facility.
Figure 31 presents a comparison between a water storage
dam and a tailings dam built using one of the downstream
methods of construction. The similarity of the two dams is
obvious. Both have substantial cross-sections and extensive
internal drainage. This type of tailings dam can be
designed to be stable under both static and seismic
loadings.
5.4 Conventional dam construction using open pit waste
(after Klohn, 1980)
Waste rock and overburden materials from the open-pit
stripping operation where economically available, can, in
most instances, be utilized to provide very stable tailings
|
Colorado School of Mines
|
ER-4609 107
dams, particularly under seismic loading. Unfortunately,
the availability of waste stripping from the open pit
operations does not always coincide with the construction
scheduling required to keep the dam crest above the tailings
pond. However, it may be possible to combine waste rock and
tailings sand to produce a safe economical tailings dam.
Figures 32, 33, and 34 illustrate designs utilizing wastes.
In recent years there has been a definite trend on
major projects to move closer to using conventional dam
design and construction procedures. This trend has
developed mainly because of regulatory requirements that
will not permit downstream discharge of tailings effluent
unless treated to meet water quality standards, combined
with the requirement that the tailings pond be able to
safely store the Probable Maximum Flood. For the
conventional tailings dam, the rise in pond water levels
associated with storing such floods can be quite
appreciable. Large rises in pond levels will drown the
slimes beach and can lead to serious seepage and piping
problems through the sand dam. Under these design
conditions it becomes necessary to incorporate an impervious
membrane in the tailings dam, and continue to extend this
membrane as the dam is raised. The materials used for
construction of the main portion of the tailings dam may be
open pit waste, borrow materials, cycloned tailings sand, or
a combination of these materials. Figure 33 illustrates
this type of tailings dam design.
|
Colorado School of Mines
|
ER-4609 111
5.5 Comparison of impoundment options (after Vick, 1983)
The selection of an appropriate surface impoundment
option for a particular tailings disposal problem requires
that the compatibility of the method to specific site
conditions, mill tailings and effluent production, and mine
production be carefully addressed. Suitability of various
surface impoundment options to different conditions is
summarized in Table 16.
Of particular interest in many cases are comparisons of
different embankment types on the basis of cost. To the
extent that embankment cost is proportional to total fill
volume, comparison of the various embankment types in Figure
35 is instructive. For equivalent embankment heights, and
for the particular configurations shown, downstream or
water-retention type embankments require roughly three times
more fill than an upstream embankment on the basis of
comparative cross-sectional area. A centerline embankment
would require about twice as much as an upstream embankment
of similar height. The divergence between fill volumes for
the embankment types become greater with increasing height.
Embankment fill costs are a significant item in many
cases, especially for high embankments and large tailings
production rates. However, the contribution of embankment
fill costs to the total cost of tailings disposal varies
widely. In some cases, costs for impoundment area topsoil
stripping, impoundment lining, or reclamation may far
outweigh embankment fill costs, making comparisons between
embankment types on the sole basis of fill cost misleading.
|
Colorado School of Mines
|
ER-4609 115
CHAPTER 6
TAILINGS DAM DESIGN
The material on tailings dam design has been compiled
from several sources. Sections 6.1, 6.5, 6.11, and 6.14
Controls, (except where noted) were extracted from Steven G.
Vick's book titled Planning, Design, and Analysis of
Tailings Dams. Sections 6.2 and 6.12 were compiled from The
Development of Current Tailings Dam Design and Construction
Methods, written by Earle J. Klohn. Sections 6.3, 6.4,
6.4.1, 6.4.2, 6.6, 6.6.1, 6.6.2, 6.6.3, 6.6.4, 6.7, 6.8,
6.9, 6.10, 6.13, 6.16 and 6.17 were compiled from the Bureau
of Mines information circular 8755, Design Guide for Metal
and Nonmetal Tailings Disposal, written by Roy L. Soderburg
and Richard A Busch. Section 6.14 Pond Inflow, was
extracted from Environmental Aspects and Surface Water
Control, written by Ernest A. Portfors.
6.1 Introduction (after Vick, 1983)
The type of tailings dam under consideration may be of
the water-retention variety or one of the raised
embankments, including upstream, downstream, or centerline
types. Determination of embankment type incorporates
consideration of the following key issues :
Mill-related factors.
Type of tailings and their engineering
characteristics. Mill output of tailings
|
Colorado School of Mines
|
ER-4609 117
state, depending on their fineness, their age, and
the location of the water table. However, under
severe seismic shock all saturated tailings are
likely to liquefy, becoming a fluid of high unit
weight.
A large part of the dam is usually constructed
using the coarser fraction of the tailings.
Most of the dam construction is carried out by the
mining operators, as part of the tailings disposal
operation, with the dam being raised as required
to stay ahead of the rising tailings pond.
The first factor affects the forces assumed to act on
the dam, especially under seismic loading. The second two
factors strongly influence the design section finally
selected for the dam. Because tailings dams usually are
constructed slowly over a period of many years, the designer
is able to select a design and then check its performance,
making modifications as required throughout the construction
period. This is a critically important aspect of tailings
dam design as it allows far more flexibility than is
available for design of conventional water retention dams.
With the above outlined qualifications, the basic
design requirements for tailings dams are very similar to
those for water storage dams. Although tailings are far
from being ideal dam-building materials, they are utilized
in most tailings dam designs for the obvious reason that
they are the cheapest available material. Some of the
disadvantages of tailings as a dam-building material are:
they are highly susceptible to internal piping, they present
|
Colorado School of Mines
|
ER-4609 118
highly erodible surfaces, the fine tailings are very
susceptible to frost action, and loose and saturated
tailings are subject to liquefaction under earthquake
shocks. Obviously, if tailings are to be used as the main
dam-building material, the tailings dam design must take
into account the undesirable physical properties of the
tailings. This usually is accomplished by incorporating
into the design such considerations as:
Separation of the tailings into sands and slimes,
with only the sands being used for dam building.
Control of the sand separation procedures to
ensure that the sand produced meets specified
gradation and permeability requirements.
Installation of internal filters and drains to
prevent piping and lower the phreatic surface
within the sand dam.
Compaction of the tailings sand to increase its
density. This increases the sands resistance to
liquefaction under earthquake shocks and permits
the safe use of steeper fill slopes. An
alternative solution to the liquefaction problem
is to accept a lower degree of compaction, use
flatter design slopes and install a positive
internal drainage system that prevents saturation.
Protection of erodible surfaces with vegetation,
coarse gravel or waste rock. (The tailings sands
are subject to wind and water erosion).
|
Colorado School of Mines
|
ER-4609 122
HYDROMETER ANALYSIS
SCREEN ANALYSIS
Hydrometer analysis reading -
US Standard Sieve Sizes
Screen analysis reading
30 40 50 7080100 140 200 325 400
Segregation by spigoting
Percent -200 mesh-38 percent
Pulp density 30 percent
Material-Granitic
Mill toiling
" 1973
Sample 100 feet Sample 900 feet from
and 200 feet from discharge (near clear water
discharge composite) pool)
Dike material
J 987 6 5 4 3 2 0198 7 6 5 4
Grain Size (mm)
Figure 36 Gradation of metal mine tailings-coarse grind,
low pulp density (after Soderberg and Bush,
1977)
HYDROMETER ANALYSIS
SCREEN ANALYSIS
Hydrometer analysis reading =
U S Standard Sieve Sizes
Screen analysis reading -
30 40 50 7080100 140 200 325 400
100 "4 t r t i i " "i i i 0
> E nX S Ge sg -r 2eg .7ation by sp gating^ 10 < X
1—
Percent-200 mesh- 60 percent
80 \ ruip aensny 20 t X—
Somple Material - Por o
5 70 (100 feet frorl scLh. orge? 'k \ \ - 3So ,0m 0pi 0e f2 e et from dU is c1 h ar1 ge 30 LU
o 60 X (of cecont towe) 40 > C- O
5 50 50 cr
Adill toilng UJ
1 CO
40 ■^1 60 cr
<
Ou, o
30 70 o
t—
20 80 z
UJ
- - o
10 90 cr
UJ
Q_
0 -100
5 4 1987 6 5 4 3 2 0198 7 6 5 4 .001
Grain Size (mm)
Figure 37. Gradation of metal mine tailings-fine grind,
high pulp density (after Soderberg and Bush,
1977)
ii.
|
Colorado School of Mines
|
ER-4609 124
flow net should be drawn to estimate the
pore water pressure resulting from
steady seepage within the embankment and
pervious foundation.
If the foundation contains compresible
strata, foundation pore pressures
estimated on the basis of consolidation
time theory should be taken into account
in the analysis and should be checked by
field measurements during and after
construction. (Proper blanket drains
should eliminate all pore water pressure
from the foundation).
Repeat the stability analyses until a
section has been found that has the
required factor of safety.
Again it must be emphasized that the parameters used in
the analysis are of paramount importance for accuracy and
are the most difficult .to obtain for a new property.
Probably the most difficult and the most important is the
phreatic line, which can be found by making a model of the
tailings pond and getting horizontal and vertical
permeabilities and using the finite-element method of
determining the flow through the embankment. In making the
model, the material must be discharged at the same grind and
pulp density as it will be in actual operation. The
permeabilities obtained in the samples of unconsolidated
material will not be the same as that of the same material
after consolidation, but the relative horizontal to vertical
|
Colorado School of Mines
|
ER-4609 125
permeability will be approximately the same. The vertical
sample could then be loaded in the consolidometer and
permeabilities measured at various loadings to simulate
various embankment heights, and from this the horizontal
permeabilities could be calculated at the same heights.
Small-volume, low-height embankments are quite simple
to design and operate. If borrow material is readily
available, an all-borrow dike may be desirable, or a small
starter dam with spigoting around the periphery and the
embankments raised with tailings sand the most economical.
Generally when the ultimate height of the dam is to be
only 50 to 100 feet, stability is not a problem, provided
good operating procedures and reasonable slopes are used and
the annual rise is kept low. The site selection sampling
and testing are still necessary to be sure that the
foundation is strong to accommodate the weight placed on it.
Embankments need a starter dam to provide sufficient
freeboard to prevent overtopping at the start and to provide
water storage for clarification and reclaim. Large
operations in a narrow valley may require an extra-large
starter dam because of the small initial acreage, or they
should have two or more sites in use at the start of
operations. Annual rise is very critical and can cause
trouble if it is too rapid. As much as 10 years leadtime
may be required before a single site could handle the total
tonnage in a large operation.
6.4 Starter dam design (after Soderberg and Bush, 1977)
The site investigation, including trenching, drilling,
sampling, and laboratory testing, should indicate the type.
|
Colorado School of Mines
|
ER-4609 126
quantities, and physical properties of the foundation
material in the dam area. From this information and the
properties of the tailings, the starter dam can be designed.
The material available for construction of the dam is most
frequently borrow material from within the disposal area.
If this is mostly clean sands and gravels with high
permeability, a pervious starter dam can be built. If it is
predominantly clay mixed with silts, sand, and gravels, an
impervious starter dam with filters and drains should be
built. Overburden or waste rock from open pit or
underground operations can also be used. The materials used
should be those that give adequate stability at least cost.
6.4.1 Pervious Starter Dam (after Soderberg and Bush,
1977)
Excavation for the base of the starter dam should be
down to a competent soil that will withstand the weight
contemplated. All the organic soil, trees, and brush should
be removed. On a smooth rock foundation with a 5- to 10-
percent slope, a trench cut into bedrock may be needed to
key the dam to the rock. Foundation defects such as open
cracks in the bedrock, clay seams, buried coarse talus
deposits, or pervious foundation soils should all be
remedied. Loose and pipable material should be excavated,
and open cracks filled to prevent piping under the dam.
All the possible problems and conditions for all
situations cannot be contemplated. Actual treatment of the
foundation depends on conditions exposed in the field and
must be solved there. Seepage through or beneath the
starter dam in this case is not bad except that it must be
controlled so that it does not lead to piping. On deep
|
Colorado School of Mines
|
ER-4609 127
alluvium most of the seepage would go out the bottom of the
pond with part of it flowing under the dam.
A pervious starter dam should have a permeability of
10~2 to 10~3 centimeters per second, but the main criterion
is that it have a higher permeability than the sands it is
retaining. It is necessary that the starter dam not retain
water so that the phreatic surface hits as low as possible
on the upstream face and does not emerge on the downstream
face. All the water that reaches the starter dam must go
freely through it to a collection pond below the downstream
toe. The sand-gravel mix must be placed in this layers and
compacted to 95 percent of Proctor to insure stability while
allowing flow through the dam. The borrow pits in the
material used for construction of the dam should be tested
for permeability in the laboratory at Standard Proctor
density and the material should be placed in the dam so that
the permeability increases downstream and the overall
permeability is greater than that of the sand.
6.4.2 Impervious Starter Dam (after Soderberg and
Bush, 1977)
If all or most of the borrow available for construction
within economical hauling distance of the site is a
relatively impervious material, or if the "downstream
method" of placing tailings is to be used, an impervious
starter dam should be built.
The method of construction for the impervious starter
dam is the same as for the pervious dam. Compaction should
be 95 percent of Standard Proctor, and the foundation
excavation and preparation should be the same. For the
ordinary upstream method of placing sands, the starter dam
|
Colorado School of Mines
|
ER-4609 128
should have drains to catch all the seepage water and let it
pass freely under the starter dam in pipe or blanket drains.
Under no condition should the starter dam retain water
against its upstream face because it would become saturated
and unstable. Under these conditions the seepage could
emerge high on the sand face above the top of the starter
dam, and remedial measures would be necessary. These
remedial measures are described elsewhere but are ' no
substitute for proper drainage, design, and construction.
The ultimate height that the dam could be built is
materially reduced if a high phreatic line is generated.
With the downstream method, the starter dam is at the
upstream toe of the completed dam. It can and should be
impervious relative to the sand and retain water as much as
possible. The seepage that eventually goes through and over
the top of the starter dam will move down through the more
pervious sand and into the drains between the starter and
downstream toe dam (Figure 40) . The stability of this
starter dam is not a problem because it eventually is
completely surrounded by tailings and sand on its top and
downstream and by slimes upstream.
The area between the upstream starter dam and the
downstream toe dam must have blanket or strip drains to
catch all the seepage and drain it out to a holding pond
where it can be recycled or discharged. These drains would
not be necessary if the cyclone sand were >100 times the
permeability of the starter dam.
The steeper the terrain and narrower the valley, the
higher the starter dam must be. If many years of leadtime
are available and the area can be raised slowly, a small
starter dam can be used, allowing the sand to build up to an
elevation where the area is large enough to take the entire
|
Colorado School of Mines
|
ER-4609 130
output continuously.
Two complete and separate areas are desirable in order
to allow de-activation of one area to drain, build sand
dams, raise pipes, etc., while the other area is activated.
At startup, with large tonnage, the fill time is quite
short even with two areas, and drainage may be the limiting
factor. They may not drain fast enough to allow the
necessary time to raise the dams with the sand. When the
original ground is steep (5% to 10% percent slope), the
drains may not be able to handle the water because the pond
area is small and the rate of rise is fast. In this case,
standby areas should be provided to take care of
emergencies.
6.5 Control of phreatic surface (after Vick, 1983)
The location of the phreatic surface, or internal water
level, within an embankment exerts a fundamental influence
on its behavior, and hence control of the phreatic surface
is of primary importance in embankment design. The phreatic
level governs to a large degree the overall stability of the
embankment under both static and seismic loading conditions,
in addition to influencing the susceptibility of the
embankment to seepage-induced failure.
The objective of prime importance is to keep the
phreatic surface as low as possible in the vicinity of the
embankment face. To the extent that the arrangement of
materials of differing permeability within the embankment
governs internal seepage patterns, control of the phreatic
surface dictates the types of materials required for
construction and their configuration in internal zones. A
|
Colorado School of Mines
|
ER-4609 131
general principle that guides embankment design in relation
to phreatic surface control is that permeability of various
internal zones should increase in the direction of seepage
flow. As permeability increases, the phreatic surface is
progressively lowered, and ideally the most pervious
available material should be located at or beneath the
embankment face.
This principle is illustrated in Figure 41. Figure 4la
shows an idealized upstream embankment in which permeability
increases in successive zones in the direction of seepage
flow, from low-permeability slimes near the decant pond to
high-permeability sands at the embankment face. In this
case, the phreatic surface is reasonably low near the face,
and seepage breakout on the face itself, which could induce
dangerous erosion and slumping, is avoided.
Figure 41b shows the same case, except with a low-
permeability zone at the face, such as might result from
perimeter dikes constructed of clayey natural soils. Here
the low-permeability zone impedes drainage and results in an
elevated phreatic surface that breaks out high on the
embankment face, producing conditions conducive to both mass
instability and such seepage-related problems as piping and
erosional sloughing. The principle of increasing
permeability in the direction of seepage flow applies in a
strict sense only to materials near the embankment face.
Figure 41c shows a downstream or water-retention type
embankment with an upstream core and a pervious downstream
shell. In this case, the permeability of the retained
tailings can be higher than that of the core with no
appreciable effect on the phreatic surface. It is possible
for a properly designed downstream or water-retention dam to
function entirely independently from the nature of the
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.