University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Colorado School of Mines
|
CHAPTER 4
MODELING ENVIRONMENT
This chapter describes the modeling environment of the mine ventilation layout, geomet-
ric representation, and flow characteristics used in the simulations. The model construction
creates a complete bleeder-ventilated longwall panel from startup room to the active face lo-
cation 3,109 m (10,200.0 ft) in length. As shown in Figure 1.4 the active panel has regulated
flow into the bleeder entries leading to the bleeder shaft and fan. This ventilation layout
is further discussed in Section 4.1. The active mining at the face is assumed stopped for a
sufficiently long period of time to consider steady-state simulation a good approximation.
The construction of a CFD model of the ventilation layout begins with the sketching of
individual sections approximating the geometry that represents the mine entries, face, gob,
etc. This is done in the ANSYS native software suite using ANSYSfi DesignModelerTM
software in order to facilitate geometry changes. All models were create in feet with the
exact dimension noted throughout this chapter. A discussion of each geometry section is
in Section 4.2. The unknown flow characteristics of the gob are then estimated using the
geomechanical simulation data from FLAC3D (Marts et al., 2014b) and fitted to the length,
width, and the mine’s stratigraphic type. The resulting porosity and viscous resistance,
equation fits and Fluent implementation are discussed in Section 4.3.
4.1 Ventilation Layout
The CFD simulation models the mixing of ventilation air and methane released from the
overlaying strata. Figure 4.1(a) shows the ventilation plan map of a bleeder-ventilated gob
with labeled Points A through F, and regulators at Point 1 and 2. Figure 4.1(b) shows the
ventilation network system that supplies 47 m3/s (100,000 cfm) to the panel at Point A.
This air is split to provide 33 m3/s (75,000 cfm) across the face to Point B and 12 m3/s
(25,000 cfm) inby the headgate entry to Point E. The headgate (HG) is on the right, and the
70
|
Colorado School of Mines
|
The model methane liberation rates are set to 2% of the total incoming ventilation air
and must be diluted to 2% at both the tailgate and headgate regulator. These regulators are
generally considered the examination points for bleeder effectiveness as defined in CFR 30.
However, the regulators at Point 2, as shown in Figure 4.1(a), are generally constructed to
restrict accidental personnel access to the startup room, and not considered as an essential
ventilation control. This project includes a study of the EGZs that form in response to the
ventilation controls at Point 1 and 2.
4.2 Geometry
The geometry of the mine dimensions are simplified to common dimensions of mine
entries. For example, the width of the development entries is assumed to be 6.1 m (20.0 ft).
The geometry models are sketched and dimensioned in the units of feet using constraining
relationships where possible in order that a simple parameter can be changed to create the
next model. An origin point is chosen for each part as noted in the following sections.
4.2.1 Overview of Geometry Sections
The mine geometry is broken down into repeatable sections to represent the fluid domain
of the ventilation system. Figure 4.2 shows the plan view of a panel with a three-dimensional
exploded view zooming in on the tailgate side of the panel. The geometry pieces consist of
the longwall face, tailgate and headgate, gob, void or gob-fringe, crosscuts and entries. The
gob height is 12.8 m (42.0 ft) and slanted at the headgate and tailgate sides to match the
observed angle and the formation of the gob-fringe as discussed by Worrall, 2012.
4.2.2 Void or Gob-fringe
The void or gob-fringe is formed as the roof collapses to form the gob and incompletely
fills the gateroads on either side of the longwall panel. This behavior is observed in 3 out of
4 of the Western United States coal mines in visited by Grubb (2008) and again observed
by the research group behind the headgate in Mine C and E. Figure 4.3 shows a picture
72
|
Colorado School of Mines
|
Figure 4.2: Three-dimensional Geometry Sections
of a gob-fringe, and the estimated geometry that might form. The back return ventilation
pattern shown in Figure 1.4 requires the construction of a large trapezoidal opening. This
was originally modeled by Marts et al. (2014a) in a paper on U-type ventilation and the
effects of a back return, and is adapted into the bleeder-ventilated longwall panel modeled
for this research.
The continuous void surrounding the gob is initially modeled by Worrall (2012). Fig-
ure 4.4 shows the sketch dimensions and origin point. The origin shown in the figure has
x-zero at the void interface to the gob, y-zero at the void interface to the headgate and z-zero
at the mine floor. The void is 0.9 m (3.0 ft) at the base, 0.3 m (1.0 ft) at the top, and 12.8 m
(42.0 ft) in height, with an offset of 4.7 m (16.0 ft) to tilt the gob-fringe. The length can be
adjusted to any panel length and is 3,109 m (10,200.0 ft).
Figure 4.5 shows the back return modeling the tailgate as it remains open to the first
crosscut inby. The origin is located at the center of the panel, at the gob interface to the
face and at the mine floor. This is the same origin as the final model. The panel width is
305 m (1,000.0 ft) and the tailgate end is located at minus 152 m (–500.0 ft) from the origin.
73
|
Colorado School of Mines
|
Figure 4.5: Back Return Gob-fringe Geometry
The bottom of the trapezoid is 8.8 m (29.0 ft), and the top is 3 m (10.0 ft). The height and
the inby shape remain the same as the continuous void shape.
4.2.3 Longwall Shields and Coal Face
The ventilated longwall face, as shown in Figure 1.3, allows leakage through the shield
into the gob area. This is modeled using the approach developed by Worrall, 2012 where it
is assumed that a gap area of 0.093 m2 (2.5 ft2) accumulates for every 5 shields. Figure 4.6
shows the geometry of a 6 m (19.5 ft) face depth in a 3.4 m (11.0 ft) tall coal seam offset by
the shield leakage gap depth of 0.15 m (0.5 ft). The gaps are modeled by a window placed in
the center of the coal seam approximately every 9.1 m (30.0 ft). The gap is 0.76 m (2.5 ft)
tall and 0.3 m (1.0 ft) wide. The longwall face width is 293 m (960.0 ft); and when including
the headgate and tailgate entries, the total width is 305 m (1,000.0 ft). The origin and global
axis shown in Figure 4.6 has x-zero at the center of the panel, y-zero at the face geometry
interface to the gob and z-zero at the mine floor.
The face model is improved by modeling individual shields, but is not included in this
research due to the computational complexity of the physics needed to define the flow and
75
|
Colorado School of Mines
|
Figure 4.6: Face Geometry with Approximate Shield Gaps
the number of cells required to capture the geometry. This work is the subject of a study
by Gilmore et al., 2015a conducted with the use of GPGPU.
Figure 4.7 shows the geometry of the headgate entry. The headgate side longwall face
entry is modeled with a single leakage gap connecting the face ventilation to the gob-fringe
located next to the gob. The origin has coordinates of x-zero at the outer edge of the gob
(or 500.0 ft from the panel center), y-zero at the gob to face interface and z-zero at the mine
floor. The same offset of 0.15 m (0.50 ft) is used from the shield gap model, which connects
the headgate to the gob-fringe. The entry length is 28.7 m (94.0 ft) from the origin point.
This geometry can be used for the tailgate entry by rotating it on the y-axis.
The headgate geometry used in this project is a simplification of the air pathway of an
actual mine headgate. The actual mine headgate entry would include following: the air
split to the belt of neutral air shown in green in Figure 4.1(a), the conveyor belt, crusher,
stage loader and headgate drive for the armored face conveyor. These details would require
extensive mesh refinements to resolve and would result in localized effects not within the
scope of this research.
76
|
Colorado School of Mines
|
4.2.5 Mine Entries
The mine entries’ geometry has a constant cross-section of the coal seam height of 3.4 m
(11.0 ft) with a width of 6.1 m (20.0 ft). Figure 4.9(a) shows a 18 m (60.0 ft) mine entry
extending in the positive y-direction, while Figure 4.9(b) shows a 67 m (220.0 ft) mine entry
extending in the negative y-direction. The origin shown in the figures has coordinates of
x-zero where the geometry extends in the positive x-direction and z-zero at the mine floor.
The entries connect with the crosscuts to make the airway network of the bleeder system.
This process is further discussed by the modular mesh approach in Chapter 6. Also, a third
tailgate entry, with a length of 34 m (110.0 ft), interfaces with the back return void section
and the face, thus modeling the open tailgate entry inby the shields.
(a) Mine Development Entry – 60 ft Segment
(b) Mine Development Entry – 220 ft Segment
Figure 4.9: Mine Development Entry Geometry
78
|
Colorado School of Mines
|
4.2.6 Gob Geometry
The gob geometry represents the fluid zone of the rubblized, collapsed roof rock. The
Fluent porous media model (see Section 3.7) is applied to this zone using the developed
equation fit in Section 4.3. Figure 4.10 shows the geometry of the gob, including the ge-
ometries applied as mesh refinements. Figure 4.10(a) shows a 240 m (800.0 ft) long gob. A
1-foot refinement geometry is shown in the figure applied at 168 m (550.0 ft) inby the face
with a 45 degree sloping to 46 m (150.0 ft) from the tailgate and headgate sides and is 46 m
(150.0 ft) inby the face. The 1-foot refinement geometry height is equal to that of the coal
seam, 3.4 m (11.0 ft). A 2-foot refinement geometry zone is centered in the middle of the
gob extending 61 m (200.0 ft) inby the 1-foot refinement zone. The gob is mirrored across
the panel center and is 152 m (497.0 ft) from the center to the tailgate side totaling 300 m
(994.0 ft) when completed. This width, when added to the gob-fringes of 0.9 m (3.0 ft) each,
is 305 m (1,000.0 ft) at the widest points from headgate to tailgate. The origin and global
coordinates are set for use for the porous media equation fit of x-zero at the center of the
panel, y-zero at the face and z-zero at the mine floor.
Figure 4.10(b) shows a 1,500 m (5,000.0 ft) gob using the same refinement regions, but
with the 1-foot region extending the depth of the panel. These geometry pieces are mirrored
across the panel center. Figure 4.11 shows a close-up of the tailgate with the specially shaped
gob geometry with the included refinement regions and the back return gob-fringe cut out
in red. This gob-fringe models the open tailgate gateroad that remains open to the first
crosscut inby.
The total modeled panel length is 3,100 m (10,200.0 ft), which includes the specially
shaped gob for the back return, 61 m (200.0 ft), and two 1500 m (5,000.0 ft) gob sections.
This is a short panel as compared to some actual mines in the Western United States that
use 4,830 m to 8,050 m (3-mile to 5-mile) long panels, although the flow and trends for each
should be similar.
79
|
Colorado School of Mines
|
4.2.7 Startup Room and Entries
The startup room section spans the panel width of 305 m (1,000.0 ft). This is modeled
using an open entry interfaced with the back of the gob. This is similar to the face section
without the leakage window gaps for the shields. Under normal conditions in a mine, the
startup room collapses into the gob and may have a similar shape of the gob-fringe or void
that forms when the roof collapses into the gateroads, however, the details are simplified to
a single open entry. The startup room is interconnected with one additional entry inby with
connecting crosscuts.
The crosscuts into the startup room at Point 2 in Figure 4.1(a) are separate geometry
sections that include a regulatory geometry restriction. This regulatory restriction is added
by using the Fluent “mark region” tool to add a new zone that becomes a separated, interior
fluid zone, which then can be modeled as a porous jump region or as a solid zone leaving a
window.
4.3 Model Porosity & Permeability – Equation Fitting
The gob porosity and permeability is calculated from the VSI output from the FLAC3D
modeling, as described by Marts et al., 2014b. Before the equation fitting process is applied,
the panel is broken up into scalable parts in order to fit any size longwall gob. The equation
fitting tool box in Matlab, cftool, is used to match the primary curve features of the data.
The 3D data set for position x, y, and VSI is fit using a least-squares approach to determine
the coefficients of a custom equation. The goodness of fit is determined by comparing
the R-square values, the number of coefficients, and by visually matching the shape. The
custom equations combine a series of polynomials and exponentials to reduce the number
of coefficients in each fitted section of the panel. The Knothe subsidence Equation 2.11 is
used to expand exponentials and polynomials to find an optimal number of coefficients to
recreate the given data.
80
|
Colorado School of Mines
|
in the center of the gob. The steep gradients near the gateroads suggest that this fit may
represent a mine that quickly caves into the gateroads and has a weak roof material that
compacts tightly.
4.3.3 Equation Fit for Mine E
The Mine E VSI uses a starting porosity of 50% and an initial porosity of 25% for the
host rock. Figure 4.14(b) shows that the viscous resistance is significantly less than Mine C.
Starting at the same value of 1.45 × 1051/m2, but only compacting to 5 × 1051/m2. The
porosity range is smaller from 50% to 32%. The gradients near the gateroads are shallow
and the final value of viscous resistance is an order of magnitude less than Mine C.
4.3.4 Volumetric Strain Increment Equations and Coefficients
The VSI equation and the corresponding coefficients are defined in the following section.
The following subscripts i, j, k, l are used to notate the powers of x and y, where i and j
correspond to the pre-exponential powers of x and y, and k and l correspond to the powers
of x and y inside the exponential function as given in following Equation 4.1
(cid:88)
(xy)f b xiyje−c ijklxkyl (4.1)
ijkl
where b is the pre-exponential coefficient, c is the coefficient inside the exponential
ijkl ijkl
function, and f is a fractional exponent or zero.
The permutations from 0 to 2 of ijkl are used in fitting the equations, and then through
a process of elimination the terms with the largest exponential coefficient are removed. This
process is repeated until the equation is no longer capable of a reasonable match to the
contours of the data. An term elimination threshold value of about 3,200 was observed
during this iterative process. Further improvements are made to the equation fit by fixing
values of previously unknown coefficients, thereby, simplifying the overall equation that
the Matlab tool box must solve. This process is used on three FLAC3D data sets with
reproducible success. The Mine E and Mine C equation fits are expressed in Equation 4.2
with coefficients given in Table 4.1.
82
|
Colorado School of Mines
|
CHAPTER 5
ANSYS FLUENT
This chapter presents the details of the Fluent model setup, solver settings, methane
source considerations, near-wall treatment and treatment of gravity. It also discusses the
boundary conditions, EGZ algorithm for post-processing results and development of a su-
percomputer interface script.
5.1 General Solver Settings
The Fluent software settings for the solver are shown in Table 5.1. The pressure-based
solver is used when the flow is incompressible, and the velocity formulation can be absolute
for slow flowing fluids. The time formulation is set to steady-state, however, transient cases
are possible; and the Fluent gravity model is turned off (see Section 5.4).
Table 5.1: Fluent Solver Settings
ANSYS Fluent General Settings
Type Pressure-Based
Velocity Formulation Absolute
Time Steady State
Gravity Off
The transport equation model settings are shown in Table 5.2. They were parametrically
determined by Worrall, 2012 to produce the best results. Worrall studied the effect of
different k −ε turbulent sub-model settings on the solution results and concluded that the
choice of the RNG k − ε turbulent model made little difference over the standard k − ε
model, but is reputed to be more accurate in a wider range of applications (ANSYS, 2014b).
The standard wall function was selected in order to simplify modeling of the entries and
the surface roughness of the gob-fringe. Choosing the differential viscosity model enables
the solution of the Equation 3.30 for effective viscosity to account for low-Reynolds number
90
|
Colorado School of Mines
|
effects. The porous media model is applied to the gob zone using the superficial velocity
formulation, which is based on the volumetric flow rate through the cell. The variable
viscous resistance (inverse of permeability) and porosity are applied via a Fluent UDF using
the equation fitting of FLAC3D data, as discussed in Section 4.3. The additional formulation
for porous media inertial resistance term in Equation 3.45 is set to zero for the formulation
of an initial solution and trend analysis.
Table 5.2: Fluent Model Settings
ANSYS Fluent Transport Models
Energy On
RNG k −ε, Standard Wall Function,
Viscous
Differential Viscosity Model
Zone: Gob
Porous Formulation: Superficial Velocity
Porous Media Model
Viscous Resistance & Porosity from UDF
Fluent material settings for species are shown in Table 5.3. The methane-air mixture
species formulation includes five species. For initial modeling of EGZ development, carbon
dioxide and water are not included in the model. This is the equivalent of modeling dry-air
with no additional sources of carbon dioxide during the mining process. However, sources
could exist in a mine from spontaneous combustion, as discussed in Section 2.4, or as a
seam gas similar to methane occurs. The three modeled species are methane, oxygen and
nitrogen, which requires only the solution for two species transport equations (see Section
3.5). Although, anairspeciesisavailableinFluentthatwouldsimplifythebleeder-ventilated
gob case, the code was built for use with three species.
The species density formulation is an incompressible ideal gas, which models the change
in density according to Equation 5.1
p
op
ρ = (5.1)
RT (cid:80) Yi
i Mw,i
where p is the operating pressure and R is the universal gas constant (8.134 J/K-mol).
op
91
|
Colorado School of Mines
|
The thermal conductivity is set as a constant 0.0454 W/m-K. The specific heat is formu-
lated using the mixing law, which is defined as the mass fraction average of the pure species
heat capacities, c , calculated by the following Equation 5.2
p
(cid:88)
c = Y c (5.2)
p i p,i
i
The viscosity formulation in Fluent is set using the ideal gas mixing law. The solver
computes the mixture viscosity based on kinetic theory given by the following Equation 5.3
(cid:88) X µ
i i
µ = (5.3)
(cid:80)
X φ
i j j ij
where
(cid:20)
(cid:16) (cid:17)1/2(cid:16)
(cid:17)1/4(cid:21)2
1+ µi Mw,j
µj Mw,i
φ = (5.4)
ij (cid:104) (cid:16) (cid:17)(cid:105)1/2
8 1+ Mw,i
Mw,j
The mass diffusivity uses the kinetic theory formulation from a modification on the
Chapman-Enskog formula to compute the diffusion coefficient as given by Equation 5.5
(cid:20) (cid:18) (cid:19)(cid:21)1/2
T3 1 + 1
D = 0.00188
Mw,i Mw,j
(5.5)
ij p absσ i2 jΩD
where p is the absolute pressure and Ω is the diffusion collision integral. The diffusion
abs D
collision integral is a measure of the interaction of the molecules, which is a function of
T/(ε/k ) where k is the Boltzmann constant (1.3806×10−23m2kgs−2K−1) and ε is the
B ij B
chemical species energy well depth defined from quantum mechanics. The quantity ε/k is
B
defined as the Lennard-Jones energy parameter for each species. The term (ε/k ) for a
B ij
(cid:113)
mixture is the geometric average given by the square root of the product, (ε/k ) (ε/k ) .
B i B j
The Lennard-Jones characteristic length scale, σ, is given in units of Angstroms for each
species and for a binary mixture σ is the arithmetic average of the species.
ij
5.2 Fluent Discretization and Solver Settings
The choice of gradient interpolation between the cell center and an adjacent face, trans-
port equation derivatives discretization, and segregated or coupled solver are all important
92
|
Colorado School of Mines
|
Table 5.3: Fluent Materials Settings
ANSYS Fluent Materials – Methane-air Mixture
Names Methane, Oxygen, Nitrogen
Density Incompressible-ideal-gas
Specific Heat Mixing-law
Thermal Conductivity 0.0454 W/m-K
Viscosity Ideal-gas-mixing-law
Mass Diffusivity Kinetic-theory
for the accuracy and convergence of the solution. The CFD general modeling approach rec-
ommends thatonce a solutionis achieved, then a process ofcell refinementand an increasein
discretization order is followed to quantify the accuracy of a solution. This is conducted by
analyzing the percent change in a variable of interest and reducing it to an acceptable value,
or by direct comparison with experimental data. The following sections outline the chosen
settings to calculate the initial solution. Further discussion on validation of the results is
presented in Chapter 7.
5.2.1 General Scalar for Transportation Equation Discretization
The control volume based technique in Fluent is applied to a general scalar transport
equation, yielding an algebraic equation solved numerically. The transport equation is inte-
grated over each control volume producing a discrete equation expressing the conservation
law on a control volume basis. Considering a general scalar transport equation for an un-
steady conservation law with the scalar quantity φ in integral form for an arbitrary control
volume V is given by the following Equation 5.6
ˆ ˛ ˛ ˆ
∂ρφ
dV + ρφV·dA = Γ ∇φ·dA+ S dV (5.6)
φ φ
∂t
V V
where ρ is the density, V is the velocity vector (equal to uˆi + vˆj in 2D), A is the surface
area vector, Γ is the diffusion coefficient for scalar φ, ∇φ is the gradient of φ (equal to
φ
(∂φ/∂x)ˆi+(∂φ/∂y)ˆj in 2D) and S is the source term per unit volume.
φ
93
|
Colorado School of Mines
|
Applying Equation 5.6 to each control volume in the computational domain of meshed
cells leads to a two-dimensional discretization of the scalar transport equation. Using the
triangular cell shown in Figure 5.1 gives the following discrete Equation 5.7
N N
∂ρφ (cid:88)faces (cid:88)faces
V + ρ V φ ·A = Γ ∇φ ·A +S V (5.7)
f f f f φ f f φ
∂t
f f
whereN isthenumberoffacesenclosingthecells,φ isthevalueofthescalarφconvected
faces f
through the face f, the quantity ρ V ·A is the mass flux through the face, A is the area
f f f f
of the face, ∇φ is the gradient of the scalar quantity at the face, V is the cell volume and
f
∂ρφV is the temporal discretization.
∂t
Figure 5.1: Control Volume Discretization
The discretized form of general scalar transport, Equation 5.7, represents the value of φ
at the cell center and its relationship to the surrounding neighboring cells. The system of
equations is, generally, a non-linear problem, it can be represented in linear form as written
in Equation 5.8
(cid:88)
a φ = a φ +b (5.8)
p nb nb
nb
where a is the coefficient at the cell center for the variable φ, the subscript nb refers to the
p
neighboring cell for coefficients a , and b for the variable φ . The number of coefficients
nb nb
dependsonthemeshtypeandisequaltothenumberoffacessharedwithinteriorneighboring
94
|
Colorado School of Mines
|
cells or exterior cells representing boundary conditions. A complete system results in a set
of algebraic equations with a sparse coefficient matrix solved using a point implicit (Gauss-
Seidel) linear equation solver with an imbedded algebraic multi-grid (AMG) method.
The AMG solver settings and speed-up in Fluent using the coupled velocity-pressure
solution method are the subject of a recent publication by Gilmore et al. (2015a). This
paper explores the limitations and benefits of using GPGPUs for mine ventilation research
aided by CFD simulations.
The solution variables φ are computed and stored at the cell centers, c and c , as shown
0 1
in Figure 5.1, and the face values φ must be interpolated. By default in the porous medium,
f
the solver calculates the solution at every cell face. The upwind scheme interpolation options
available in Fluent are first-order, second-order, power-law, Quadratic Upstream Interpola-
tion for Convection Kinematics (QUICK) and third-order Monotonic Upstream-Centered
Scheme for Conservation Laws (MUSCL). It is also necessary to compute the gradients and
derivatives in the solution process in the convection and diffusion terms in the flow con-
servation equations. Fluent offers three schemes: Green-Gauss, cell-based; Green-Gauss,
node-based and least squares, cell-based.
5.2.2 Fluent Solution Settings
The Fluent solution method settings are given in Table 5.4. The Semi-Implicit Method
for Pressure-Linkage Equations (SIMPLE) is used to couple the pressure-velocity field. This
sets the solver to use the iterative solution process outlined for the pressure-based segregated
algorithm as shown in Figure 3.1.
The default gradient setting in Fluent, least squares cell based, selection is kept. This
method assumes the solution varies linearly and the relative accuracy is comparable to the
node-basedgradient, butlessexpensive, computationally. Thechangein acellvaluebetween
c and c (see Figure 5.1) along the position vector between the two cell centroids can be
0 1
expressed as Equation 5.9
(∇φ) ·(cid:52)r = (φ −φ ) (5.9)
c0 i ci c0
95
|
Colorado School of Mines
|
Table 5.4: Fluent Solution Method
ANSYS Fluent Solution Methods
Pressure-Velocity Coupling SIMPLE
Spatial Discretization
Gradient Least Squares Cell Based
Pressure PRESTO!
Momentum 1st Order Upwind
Turbulent Kinetic Energy 1st Order Upwind
Turbulent Dissipation Rate 1st Order Upwind
Species 2nd Order Upwind
Energy 1st Order Upwind
Under-Relaxation Factors Default
Thus, the gradient becomes a function of mesh geometry. The various setting effects on the
final solution were determined by Worrall (2012) to have little impact.
The pressure spatial discretization choice is governed by the domination of the domain
by a porous media. The recommended setting is PRESTO! (PREssure STaggering Option),
which computes the face pressure as well as the cell pressure. PRESTO! is more computa-
tionally intensive, but is required for an accurate solution in porous media flows (Patankar,
1980). The PRESTO! scheme uses the discrete continuity balance for a control volume about
the face to compute the pressure. Note that for triangular and tetrahedral meshes, compa-
rable accuracy is obtained using a similar algorithm, and therefore PRESTO! is available for
all mesh types.
In Fluent Version 15.0 and greater, PRESTO! is intrinsically used by the solver for all
porous media cell zones allowing the user to select a discretization scheme for the remaining
fluid zones. Therefore, a first-order, second-order or QUICK scheme can be used in the
mine entries to more accurately resolve the solution. PRESTO! was the scheme used for this
research as most solutions were obtained using Version 14.5 of the software.
The remaining transport equations are first-order discretization except species. A notice-
able difference is observed in the species contour plots between first-order and second-order.
A higher order discretization is recommend by Fluent, however, solution stability and con-
96
|
Colorado School of Mines
|
vergence issues remain a challenge for future work.
The energy equation may be de-selected as there is no significant heat transfer in the
model with thetemperature boundary conditionsset to 300 K. Asobserved in final solutions,
the temperature ranges from 299.8 K to 300.6 K with no noticeable impact on the other
transport equations.
5.2.3 First-order and Second-order Accuracy
The first-order scheme assigns the field variable at the cell faces by assuming the field
variable at the cell centroid represents a cell average value. Thus, in a first-order upwind
scheme, the face value φ is equal to the value of φ from the upwind cell centroid.
f
The second-order scheme computes the cell face values using a multidimensional linear
reconstruction based on the selected gradient evaluation scheme. Through a Taylor series
expansion about the cell centroid, the second-order upwind scheme computes the face value
φ as given in Equation 5.10
f
φ = φ+∇φ·r (5.10)
f
where ∇φ is calculated using Equation 5.9,and r is the vector from the upwind cell centroid
to the face centroid. For example, r as shown in Figure 5.1 when c is upwind of c .
1 1 0
The solution order of accuracy for first-order and second-order is given in the following
Equations 5.11–5.12
(cid:18) ∂φ(cid:19) φ −φ O((cid:52)x)
i+1 i
= + (cid:124) (cid:123)(cid:122) (cid:125) (5.11)
∂x (cid:52)x Truncation error
(cid:0) (cid:1)
(cid:18) ∂φ(cid:19) −φ +4φ −3φ O (cid:52)x2
i+2 i+1 i
= + (cid:124) (cid:123)(cid:122) (cid:125) (5.12)
∂x 2(cid:52)x Truncation error
The modeling solution accuracy of mine gas mixtures in large underground bleeder-
ventilated gobs varies linearly with the spatial component of the cell size when using the
first-order upwind discretization scheme. The scale of the mine is on the order of 4 to 5
miles and the smallest geometric shape is approximated by the shield longwall face model
97
|
Colorado School of Mines
|
shown in Figure 4.6. The mesh face sizing control used to model the shield gaps is one-
tenth of a foot and the largest face element size in the center of the gob is 16-feet. As a
result, initial solutions have two orders of magnitude linear error from an unknown actual
solution. However, as discussed in Chapter 7, this error is significantly reduced through
mesh refinement and increases to the second-order scheme.
5.2.4 Solution Approach and Convergence
A steady-state CFD simulation implies that the solution no longer significantly changes
withfurtheriterations. Fluentrecommendscheckingthefollowingtodetermineconvergence:
(cid:136) Discrete conservation equations are solved to a specific tolerance
(cid:136) Overall mass, momentum, energy and scalar balances
(cid:136) Decrease in residuals by at least three orders of magnitude
(cid:136) Energy residual decrease by six orders of magnitude
(cid:136) Species residual decrease by five orders of magnitude
(cid:136) Monitor relevant key variables
The specific tolerance to the governing equations is determined by examining the key
relevant variable, which in this project is the normalized EGZ size (see Section 5.5). The
overall mass balance is reported for all inlets and outlets of the model as 5×10−5kg/s, which
is four orders of magnitude less than the smallest mass inlet. The momentum and energy
balance are not of importance in this simulation as the temperature boundary conditions
are uniform and the momentum transfers to a fixed geometry
Fluent reports the root mean squared average of the global normalized residual to the
linearized governing equations in 5.8, where a residual is defined in Equation 5.13
(cid:88)
R a φ − a φ −b (5.13)
P = P P NB NB P
NB
98
|
Colorado School of Mines
|
which is a measure of how well the linearized system of equations is converged to the solution
of the represented non-linear system.
This research solved a steady-state solution using the following process, and then used
this solution to re-initialize all following cases. The transport equations are turned on one-
by-one and iterated until the residuals convergence criteria are reached. The initial solution
for the laminar case is partially solved followed by the turbulent and temperature equations,
and then the species transport equations. Figure 5.2 shows this process for 1,500 iterations
where the jump in residuals at 300, 700, and 1,000 correspond to each step in the solution
approach. The residuals appear steady for species and energy, but further reduction in
the velocity and turbulent residuals are needed before a final solution is reached, which is
determined by calculating the percent change in EGZ size.
Figure 5.2: Initial CFD Simulation Residuals
Figure 5.3 shows the residuals starting from a previously solved data set solved for 5,000
iterations more after a change in boundary conditions. These residuals suggest that the
convergencecriteriacanbesetto1×10−4 forcontinuity,1×10−5 forthevelocities,1×10−6 for
theenergy,0.001forturbulenceand1.3×10−4 forthespeciesequations. Thesolutionappears
to converge with 1,300 iterations although the EGZ is still changing. Figure 5.4 shows the
normalized EGZ and the percent change in EGZ. Using the EGZ as the determining variable
for convergence the solution can be considered converged at approximately 3,200 iterations
when the percent change in EGZ drops below 0.5%. This process is used for all final results
99
|
Colorado School of Mines
|
5.3.1 Boundary Conditions
The boundary conditions of the models at the inlets and outlets are match to regulatory
conditions when using common ventilation quantities. The model boundaries are at common
measurement points for flow quantity and gas concentrations. Therefore, the predicted flow
within the model would represent an operational mine to the extent that the geometry and
mesh represent the ventilation system.
The inlets are defined as velocity-inlets in Fluent, which sets the flow rates to the mea-
sured quantities typically found in bleeder-ventilated gobs. The inlets of the model are the
headgate entry supplying air to the face (Point A), the entry on the headgate side supplying
air inby the face (air flowing past Point A), tailgate entry supplying air to the face (Point C),
and tailgate center entry supplying air inby the face (Point D) as shown in Figure 4.1. Each
inlet is set with a velocity magnitude normal to the boundary as listed in Table 5.5. The hy-
draulic diameter of 4.3 m (14 ft) and turbulent intensity of 3% are calculated from Equations
3.26 and 3.39. The species mole fraction is set for the composition of air as 20.95% oxygen,
with the remainder as nitrogen and ignoring the 1% other gases normally contained in air.
The outlet of the model is set to a pressure outlet with zero gauge pressure and backflow
conditions same as the inlets.
Table 5.5: Inlet Boundary Conditions
Name Velocity, m/s Air Quantity. m3/s (cfm)
inlet hg 1.732 35.4 (75,000)
inlet hg-entry60ft y+ 0.577 11.8 (25,000)
inlet tg 0.231 4.72 (10,000)
inlet tg-gentry60ft y+.1 0.346 7.07 (15,000)
5.3.2 Methane Sources
Methane may come from three main locations: upper or lower coal seams and the seam
being mined. This project neglects seam gases produced during active mining; therefore, the
101
|
Colorado School of Mines
|
model is simplified to a longwall panel that is complete, or is inactive. The methane source
is modeled as an upper rider seam well above the caved zone. The methane must enter the
mine ventilation system through the partially intact strata layers above the gob. The rider
seam supplies methane from an infinitely larger reservoir that readily refills the cleated coal
structure. This is evident as the permeability of the coal seam is orders of magnitude higher
than the surrounding strata. As a result, this produces an evenly distributed methane source
at the bottom of the rider coal seam.
The methane source may further be simplified by allowing the methane inlet to be moved
to the top of the gob, when neglecting the operation of GVBs located in the strata. The
choice to locate the methane inlet boundary condition at the top of the gob is further
supported when considering the orders of magnitude difference in permeability between the
intact strata and the gob. Further consideration of methane sources is discussed by Worrall
(2012).
The volume of the methane inlet was calibrated to supply a concentration of 2% methane
at the bleeder outlets (Point 1 and Point F), which is the concentration limit in a bleeder
entry. The top of the gob has an area of 911,600 m2 (9,812,000 ft2) on a 3,109 m (10,200.0 ft)
long panel, with a constant velocity inlet of 1.31×10−6m/s producing 119 m3/s (2,530 cfm)
of methane. The turbulent conditions are set with the intensity of 0.1% and length scale
of 0.35 m, however, the solution is insensitive to the choice of these turbulent boundary
conditions.
5.3.3 Turbulent Near-Wall Treatment
The ANSYS Fluent Theory Guide 2014b discusses the equations and use of near-wall
treatment, and is summarized in the following section. The highest gradients in the flow
solution are often near walls where the no-slip boundary condition must be imposed. Also,
the walls serve as the main source of vorticity and turbulent production in the formation
of the momentum boundary layer. The geometry modeling, mesh refinement and numerical
expense required to fully resolve the solution gradients near walls is beyond most CFD
102
|
Colorado School of Mines
|
modeling needs. It is important to accurately predict the pressure drop and turbulent effects
produced by modeling fluid flow across walls.
Figure5.5showsthatthenear-wallregionflowcanbesubdividedintothreelayersplotted
onsemi-logcoordinateswhereUisthefreestreamvelocity, U oru isthefrictionvelocityas
T τ
(cid:112)
defined as τ /ρ, where τ is the wall shear stress and y+ ≡ ρu /µ a dimensionless distance
w w τ
from the nearest wall. The three regions are the inner viscous sublayer where the flow is
almost laminar, which is dominated by the molecular viscosity; the outer fully-turbulent
layer where turbulence is the dominating flow factor; and a buffer layer, or blending region,
where molecular viscosity and turbulence are equally important. The use of a semi-empirical
formula or wall function to calculate the effects of the viscous sublayer and the buffer layer
eliminated the need to fully resolve the flow near the wall. The use of a wall function requires
thatnogridrefinementsnearthewallalloway+ valuebelow15, exceptwhenusingascalable
wall function.
Figure 5.5: Subdivisions of the Near-Wall Region (redrawn for clarity) (ANSYS, 2014b)
Thisresearchusesthestandardwallfunctiontocalculatenearwalleffects. Thissimplified
the modeling of mine entry wall roughness and gob-fringe roughness. The use of a wall
roughness of 0.15 m (0.5 ft) and a wall constant of 0.6 for the gob-fringe effectively models
the surface of the gob and surface of the gob-fringe with the intact strata, while the mine
entries remained the default values of a smooth duct. The momentum equations and species
103
|
Colorado School of Mines
|
equations defining the law-of-the-wall in the standard wall function are given in Equations
5.14–5.16. For the mean velocity
1
U∗ = ln(Ey∗)−(cid:52)B (5.14)
κ
where,
U C1/4k1/2
U∗ ≡ P µ P (5.15)
τ /ρ
w
is the dimensionless velocity and
ρC1/4k1/2
y∗ ≡ µ P y (5.16)
P
µ
is the dimensionless distance from the wall, κ is the Carman constant of 0.4187, E is the
empirical constant of 9.793, where the subscript P refers to the wall-adjacent cell, U is
P
the mean velocity of the fluid at the wall-adjacent cell centroid, k is the turbulence kinetic
P
energy at the wall-adjacent cell centroid, y is the distance from the centroid of the wall-
P
adjacent cell to the wall and µ is the dynamic viscosity of the fluid.
The recommended range of y∗ values depends on the overall Reynolds number. The lower
limit is always in the order of y∗ ∼ 15. The upper limit for high Reynolds number flows can
be several thousand, while for low Reynolds number flows it may be as small as one hundred.
Therefore, ANSYS recommends avoiding the application of wall functions in low Reynolds
number flows, as it limits the overall number of nodes, which may be placed in the boundary
layer.
Themathematicalformoftheslopeinterceptterm,(cid:52)B inEquation5.14hasthefollowing
form, 1/κlnf where f is the roughness function. The form of f is determined by the value
r r r
of the non-dimensional roughness height given in Equation 5.17
ρC1/4k1/2
K+ = µ P K (5.17)
s µ s
where K is the physical roughness height.
s
There are three flow regimes: hydrodynamically smooth, transitional and fully rough,
modeled by Equation 5.18
104
|
Colorado School of Mines
|
energy and the dissipation rate are calculated with the local equilibrium hypothesis with the
assumption that they are equal in the wall-adjacent cell. The kinetic energy production is
calculated by Equation 5.22
∂U τ
w
G ≈ τ = τ (5.22)
k w ∂y w κρC1/4k1/2y
µ P P
and the dissipation rate is computed by Equation 5.23
C3/4k3/2
ε = µ P (5.23)
P
κy
P
The wall boundary conditions used are summarized in Table 5.6. The application of the
roughness coefficients to the gob-fringe wall requires that the interface between the crosscuts
and the gob-fringe be created, then the remaining wall not in contact with a crosscut surface
is assigned the wall boundary conditions. The surface roughness on the interface between
the gob, a porous media, and the gob-fringe, a fluid zone, is accounted for in the roughness
applied to the gob-fringe wall. The mine entries’ walls are considered smooth. Although to
properly model the actual pressure drop across a mine entry, a more complete geometry and
roughness would have to be considered. The simplified modeling of the longwall face (as
discussed in Section 4.2.3) requires a generalization for the mine entries.
Table 5.6: Wall Boundary Conditions
Wall Roughness Constant Roughness
Entries 0.5 0
Gob-fringe Walls 0.6 0.15
5.4 Gravity
TheeffectsofbuoyancyofthesourcetermG intheturbulentproductionequationandin
b
turbulentdissipationrateequation(seeEquations3.28and3.29)formingaspeciesseparation
gradient within a porous media are not fully understood. Attempts to activate gravity in
the simulations resulted in un-converged cases. A simple two-dimensional flow simulation
suggests that a flow separation will be produced with gravity enabled, but when disabled,
106
|
Colorado School of Mines
|
the species gradient becomes dispersed. A U-type ventilation simulation with gravity one-
thousandth of a G, results in the expected flow separation between methane and air. The
methane gradients in the upper layers are most affected, but no difference is noticeable near
the mine floor where results are of primary interest.
The effects of gravity in the mine ventilation system are known to produce a separation
layer between the unmixed methane gas and air at sufficiently low air velocities.
The effects of gravity within the porous media of the gob are not well understood. Fur-
thermore, the turbulent equations solved on the appropriate mesh size for a valid porous
media model assumption fail to resolve the separation layer when applying the superficial
velocity formulation and neglecting the inertia resistance term. A solution might be attain-
able with the inclusion of these models, but this is beyond the scope of this dissertation.
The effects of gravity in the gob can be separated into three cases: the formation of a
stable separation layer trapping unmixed methane near the top without the addition of more
methane; the influx of a sufficient amount of methane to overwhelm the development of a
separation layer; and the development of a highly unstable, turbulent separation layer at low
methane concentrations. This third case is more realistic and likely in the mine ventilation
system based on two-dimensional simulations.
The large, three-dimensional models in this application simplify the methane inlet source
as a constant velocity inlet. This fixes the ratio of methane to air in the model to a mole
fraction of 2% methane. A summary of this modeling choice is discussed by Worrall (2012);
the incoming methane velocity has two components as given in Equation 5.24
V = V −V (5.24)
methane pressure density
where V is the actual velocity inlet of methane into the mine, V is the velocity
methane pressure
developed from the pressure driven flow governed by Darcy’s Law and V is the velocity
density
drivenbythedensitygradient. Alinearapproximationofthecontribution, whichthedensity
gradient has on the inlet velocity within the strata layer, can be simplified using a finite
difference of Darcy’s Law as given in Equation 5.25
107
|
Colorado School of Mines
|
k (cid:52)ρg
V = · (5.25)
density
µ (cid:52)h
where k is the permeability of the strata (with a uniform porosity of 10%), µ is the viscosity
of air, (cid:52)ρ is the density difference between air and methane, g is the gravitational constant
and (cid:52)h is the strata height from the top of the gob to the rider coal seam, 28 m (92 ft).
2.9×10−16m2 (1.225kg/m3 −0.668kg/m3)(9.81m/s2)
V = · = 3.3×10−12m/s
density 1.71×10−5Pa·s 28m
(5.26)
The Equation 5.26 equates to six orders of magnitude lower than the 1.31 × 10−6m/s,
defined as inlet velocity at top of the gob. Fixing the mass flow rate would result in an
artificially increased density, and therefore increased pressure at the top of gob, as illustrated
in Equation 5.27
m˙ = ρV A =↑ ρ ↓ V A (5.27)
methane pressure
Following this analysis, the effects of gravity would be localized, compressing the concen-
tration gradients and resulting in the thinning of EGZ zones near the top of the gob. The
effects would further be diminished near the plan view height chosen for result comparisons.
5.5 Explosive Gas Zone (EGZ) Mixture Analysis
An algorithm originally developed by Worrall, 2012 is used to combine the results of
methane and oxygen concentration in each cell into one easily readable data point. The
algorithm based on Coward’s triangle, shown in Figure 5.6, characterizes the mixture and
assigns a value corresponding to the appropriate color zone. The color zone classifications
are below:
(cid:136) Blue – a cell that may become explosive
(cid:136) Yellow – a cell capable of forming an explosive mixture if diluted with air
(cid:136) Green–acellnotcapableofforminganexplosivemixturewithair(darkgreenindicates
a zone inert to the spontaneous combustion process and not studied in this research)
108
|
Colorado School of Mines
|
(cid:136) Red – a cell that is explosive
(cid:136) Orange – an arbitrary buffer zone surrounding the red explosive zone
This algorithm aides in the identification of the presence, location and size of EGZs in
order to compare results. Figure 5.7 shows a diagram of the algorithm’s use, combining the
oxygen concentration and methane concentration into one plot. A thin sliver of EGZ (red)
can be identified as being surrounded by an orange buffer zone. An examination of EGZ
plots reveals the presence and location of EGZs for comparison.
The resulting size of the EGZ is calculated by assigning a value of 1 to each red cell
followed by integrating the total volume of the cell and multiplying by the porosity to
calculate the total EGZ gas volume. A normalized EGZ volume is then used to compare the
effects of a given parameter on the overall trend, where it is assumed that an actual mine
follows the same trends identified in the model. This approach addresses the potentially
large variation in mine characteristics and uncertainty in boundary conditions.
Figure 5.6: Color Coded Coward’s Triangle – Modified after (Coward & Jones, 1952)
109
|
Colorado School of Mines
|
Figure 5.7: Explosive Gas Zone (EGZ) Algorithm Worrall (2012)
5.6 Model Result Post Processing
The CFD results are compared on a plan view plane located 1.5 m (5 ft) from the mine
floor, which is roughly half the coal seam thickness. The plane passes through all mine
entries in order to evaluate the trend in the location of an emerging EGZ. Also, this is
the most common location of methane concentration measurements made by mine workers.
The results of the simulations are plotted on this plane for the following: oxygen, methane,
nitrogen, EGZ, pressure, porosity and resistance (the inverse of permeability). In addition,
the area-weighted average of the methane mole fraction and pressure are reported at each
inlet, outlet and the plane across the regulators (Point 2 in Figure 4.1). Also the mass flow
rate, volumetric flow rate and EGZ volume integral are reported.
5.7 Running on the CSM Mio Supercomputer
The CFD model size is approximately 15.8 million cells and it requires a large amount
of memory to solve. Parallel processing and domain partitioning on a multi-node supercom-
puter architecture are required to solve these models. The NIOSH funded research project
110
|
Colorado School of Mines
|
CHAPTER 6
MODULAR MESHING APPROACH
This chapter discusses the details of a 2015 paper describing the modular meshing ap-
proach developed in this research, published by Gilmore et al., 2015c. The mesh is created
with the ANSYS Meshing software in modular parts in order to maintain high quality and
low skewness.
The cut-cell meshing method is applied to the gob, with the remaining parts free meshed
according to edge and face sizing controls, mapped face controls, and where applicable,
match face control for repeated interfaces. Using this approach, the geometry sections are
individually meshed using custom controls to achieve a minimum quality of 0.20 and a
maximum skewness of 0.85. The mesh used to solve the transport equations affects the
accuracy of the solution and the computational time required to converge a solution. Each
part has an origin, identified in the Section 4.2, that is translated and rotated into place to
complete the fluid domain of the final mesh. Using this method, the mesh files are reused,
reducing the need to store large case files. This also eliminates the need to have computers
capable of re-generating the entire, larger mesh domain files every time a geometry change
is made, which can take days to execute.
The modular meshing approach creates greater control and flexibility over the standard
approach. The standard approach involves representing the fluid domain with a complete
geometry model. The only requirement is the creation of a separate fluid zone for the porous
media; multiple geometries can be interfaced within ANSYS Meshing software not visible to
the Fluent solver. ANSYS Meshing software creates a mesh for each geometry part, one after
the other, using the previous interfacing part as the input to the next. The advantage is the
creation of conformal mesh interface boundaries between parts, however, the non-conformal
algorithm used in Fluent is suitable enough for most interfacing modules.
112
|
Colorado School of Mines
|
6.1 Mesh Creation Cycle
The modular meshing approach produces solvable mesh files that can easily be adjusted
or augmented. The modular meshing of individual, interfacing mesh modules permits quick
geometry changes without re-meshing and rebuilding entire models. This results in greater
control over the resulting mesh quality and skewness; the pre-meshed modules guarantee
a consistent mesh quality. Figure 6.1 shows the model design process, beginning with the
creation of individual mesh modules, which are assembled to create the full mesh. Each new
module is added to the mesh module library from which the modeler can choose the required
components to match the desired mine geometry. The full assembly of modules creates the
CFD mesh for the complete computational domain solved by Fluent. Mesh modules can
be stretched or compressed within certain bounds to approximate a match for the mine
entry geometry. The mesh assembly can be adjusted to match a variety of entry dimensions
and configurations. This meshing approach ensures flexibility when changing a geometry or
ventilation control for the next modeling task. The user simply removes the module to be
changed, and either replaces it with a suitable module from the library or creates a new one.
Figure 6.1: Mesh Creation Cycle
This process takes advantage of the newly released Fluent Version 16.0, released in 2015,
which has Multiple Upstream Mesh systems. Figure 6.2 shows multiple geometries meshed
113
|
Colorado School of Mines
|
individually and connected to a single Fluent solver. This allows completion of a parametric
study when only part of the domain needs to be updated. In addition, the meshing mode in
Fluent can run a journal file of pre-defined commands to create the mesh assembly. The final
mesh assembly consists of the parts shown in Figure 4.2 to create the domain of a bleeder
panel with 46 crosscuts similar to Figure 1.6.
Figure 6.2: Multiple Upstream Mesh Systems (ANSYS, 2014a)
A modular meshing approach allows the domain to be meshed using multiple techniques.
The cut-cell mesh assembly option creates large domains of ideally shaped hexahedron cells
or cubes. The general approach creates tetrahedrons, prism or wedge cells, and may convert
these into hexahedrons frequently resulting in some level of skewness. Figure 6.3 shows the
use of the two techniques applied to the gob (right) and a mine entry (left). The mine entries
contain one layer of inflation that forms wedge cells and forms tetrahedrons in the center.
This resolves the boundary condition applied at the wall and turbulence in the center. The
gob is meshed with the cut-cell method, which produces only hexahedrons in the center, and
slightly skewed cells near the angled edges at the gateroads.
114
|
Colorado School of Mines
|
Figure 6.3: Mesh Cell Types
6.2 Mesh Modules
The mesh modules for each geometry section use unique controls to maintain the quality
and skewness of mesh. Each mesh file is saved in Fluent mesh file format. The following
sections discuss the mesh and metrics for each module used in assembling the bleeder panel.
6.2.1 Gob
Thegobmeshconsistsoftwodifferentsections. Themainsectionusesthecut-cellmethod
with 0.3 m (1 ft) face sizing on the interfaces to the gob-fringe mesh, while the 61 m (200 ft)
gob section behind the face uses a mesh control method for hexahedrons. Figure 6.4 shows
a view looking at the top of the mesh, where the cells at the center have 4.9 m (16 ft) edges
and edge size steps down towards the gateroad edges to 0.3 m (1 ft). Also, the effects of the
0.3 m (1 ft) body of influence shown previously in Figure 4.10(a) can be clearly observed.
The total number of cells is 2.6 million in a single 1,500 m (5,000 ft) module.
The 61 m (200 ft) gob section, shown in Figure 4.11, uses the same 0.3 m (1 ft) face
size control on the interface to the headgate gob-fringe. Figure 6.5 shows the resulting mesh
115
|
Colorado School of Mines
|
6.2.3 Longwall Face
The longwall face uses the cut-cell method, with the leakage gaps interfacing to the gob,
and the same 3.81 cm (0.125 ft) face sizing controls as the headgate, with the maximum
size cell set to 0.3 m (1 ft). Figure 6.9 shows the resulting mesh of the face module with a
total of 460,000 cells. The minimum quality is 0.40 and the maximum skewness is 0.72. The
distribution shown in Figure 6.10 clearly meets the excellent standard for meshes.
Figure 6.9: Face – Mesh
6.2.4 Entries – 60-Feet and 220-Feet
Each entry section has similar mesh controls and the resulting mesh. Two are shown for
comparison, but a third entry section of 33.5 m (110.0 ft) in length is for the tailgate entry
interfacing to the back return gob-fringe. Figure 6.11 shows the 18 m (60.0 ft) entry section
that is used as the initial module in the tailgate and headgate bleeder entries series of mesh
modules. There are 9-divisions in the height and 15-divisions in the width. A single layer of
inflation is shown on the wall with a mapped face mesh control. The resulting mesh has a
minimum quality of 0.22 and maximum skewness of 0.84, with a good distribution, as shown
in Figure 6.12. The mesh uses tetrahedral cells in the center with wedge shaped cells in the
layer of inflation. The total number of cells is 20,849.
119
|
Colorado School of Mines
|
Figure 6.12: 60-feet Entry – Mesh Metrics
Figure 6.13 shows the resulting mesh of a 76 m (220 ft) entry with a total cell count of
76,000 using the same controls as the 18 m (60 ft) entry. The mesh metric distribution is
shown in Figure 6.14 with a minimum quality of 0.21 and a maximum skewness of 0.85. This
demonstrates that the controls set for both length entries are easily scalable to any length
entry. A parameter can be defined for the division number in the length and geometry length
and set equal to each other, with an average count of 1,000 cells per meter (or 345 cells per
foot). The average quality of the wedge cells is 0.4, and 0.8 for the tetrahedral cells, reaching
the good standard for meshes. The wedge cell average skewness is 0.38, and 0.3 for the
tetrahedral cells, which meets the good standard for meshes.
6.2.5 Crosscuts
The crosscut mesh uses the same controls as the entries with 9-divisions in height and 15-
divisions in width, mapped face controls on the walls and one layer of inflation. Figure 6.15
shows the resulting mesh with these controls. The total number of cells is 8,500 with a
minimum quality of 0.20 and a maximum skewness of 0.86. Figure 6.16 shows the mesh
121
|
Colorado School of Mines
|
metrics distributions for the wedge and tetrahedral cells. The average quality for the wedge
cells is 0.25 and 0.61 for the tetrahedral cells, which meets the fair standard for meshes.
The average skewness is 0.62 and 0.43 for the wedge cells and tetrahedral cells, respectively,
which meets the good standard for meshes.
Figure 6.15: 50-feet Crosscut – Mesh
6.2.6 Gob-fringe or Voids
The gob-fringe is meshed as a single 3000 m (10,000.0 ft) module. Figure 6.17 shows the
resulting mesh with 4-divisions at the top and bottom and 42-divisions in height. The length
in feet is equal to the number of divisions in its length. Using these controls, the face cell
size is approximately 1-foot on the interfaces to the crosscuts and gob. Figure 6.18 shows
the mesh statistics for quality and skewness. The minimum quality is 0.28 with an average
of 0.68, and the maximum skewness is 0.52 with a total number of cells of 1.7 million, at 550
cells per meter (170 cells per foot) of gob-fringe. The skewness distribution clearly meets
the good standard.
123
|
Colorado School of Mines
|
Figure 6.18: 10,000-foot Gob-fringe – Mesh Metrics
6.2.7 Back Return Gob-fringe
The back return gob-fringe mesh uses a tetrahedral meshing approach. The module
interfaces with the 3000 m (10,000.0 ft) gob-fringe module and has matching dimensions.
Figure 6.19 shows the resulting mesh employing 28-divisions on the bottom and 11-divisions
at the top. The height uses the same 42-divisions as the gob-fringe, and the mesh results
show the cell size increasing towards the center. The mesh statistics for quality and skewness
are shown in Figure 6.20. The minimum quality is 0.22 with an average of 0.83, and the
maximum skewness is 0.79 with an average of 0.24. The total number of cells is 130,000.
The skewness and quality distribution meet the excellent standard.
6.3 Mesh Assembly
Onceeachgeometrysectionismeshedwithacceptablemeshqualitymetrics, thesectionis
assembled into the final representation of the fluid domain. Each module’s surfaces, interior
zone, and interfaces are given a unique identifier. Interface definitions are assigned to pairs
or sets of faces. The origin point is then translated or rotated in preparation for the input
125
|
Colorado School of Mines
|
CHAPTER 7
MODEL VALIDATION
This chapter discusses validation of the model. The first section presents a comparison
to a tracer gas study that calculates the velocity in the gob. The next section examines
ventilation layout changes made possible by the modular mesh approach; and the final
section presents the results of mesh refinements through adaptation.
7.1 Gob Air Flow Velocity
Exact values of the velocity in the gob are unknown and must fluctuate greatly between
mines due to variance in gas emission rates and gob flow characteristics. However, it is
worth examining a comparison between experimental results taken at an Eastern United
States mine and the predicted values from the CFD simulation.
The analysis of a tracer gas study results yields the average gas velocity between two
points. This is calculated by the differential in release time to the first detected presence
of the tracer gas and the shortest estimated distance the gas may travel. The tracer gas
study by Diamond et al. (1999), conducted in a coal mine located in the Pittsburgh Coalbed
in Greene County, PA injected a tracer gas into an in-taking GVB. Figure 7.1 shows the
longwall panel test area of Test 3-1, where the GVB G3 is used as the point of injection.
The GVBs are located 76 m (250 ft) from the tailgate side of the panel. The analyses of
the results are published by Mucho et al. (2000), and a summary of Test 3-1 is given in
Table 7.1. The calculated velocity from G3 to G2 is 0.008 m/s (1.5 ft/min) and from G3 to
G1 is 0.004 m/s (0.7 ft/min).
Figure7.2showstheCFDsimulationresultsofthevelocitymagnitudeonalogscale. The
velocities in the gob range from 3×10−4m/s and 7×10−3m/s (0.06 ft/min and 1.4 ft/min)
with higher velocities reported near the face. The tracer gas injection point, G3, is located
about 380 m (1,250 ft) from the face, corresponding to about 5.7 crosscuts spaced at 67 m
130
|
Colorado School of Mines
|
(220 ft). The location of G2 is 1,143 m (3,750 ft) from the face, approximately 17 crosscuts
from the face. The average velocity magnitude along this path is 6×10−4m/s (0.12 ft/min).
Diamond et al. (1999) also reported that the tracer gas was detected at the second bleeder
fan (BF2) 5 days after G2 went offline on day 25 of the test. The distance from the GVB
locations, where the tracer gas remained, to the ventilation system is 67 m (250 ft), which
equates to a velocity of 2×10−4 m/s (0.04 ft/min).
Although not a direct comparison to the mine being modeled, the Eastern United States
coal mine measurements in an inactive panel comes within a factor of 2 to 7 of the CFD
simulation results. This comparison of velocities indicates that the CFD results are within
reason given that the unknown permeability may have a range between one to two orders of
magnitude.
7.2 Multiple Geometry Meshing Options and Capability
This section illustrates the use of the modular meshing approach used in assembly of
different mine ventilation network geometries. The studies include multiple panels, single
panels, and partial panels for progressively sealed gobs using U-type ventilation, with and
without a back return, and with changes in nitrogen injection locations.
Figure 7.3 shows the ventilation network layout using two adjacent panels, where a single
bleeder fan outlet draws air through the ventilation network. The panel headgate supplies
47 m3/s (100,000 cfm) from which 33 m3/s (70,000 cfm) is supplied to the face. At the
tailgate, the ventilation system uses an H-type bleeder system, adding additional air at the
tailgate at Point E and C of 4.7 m3/s (10,000 cfm) and 7.2 m3/s (15,000 cfm), respectively.
The startup room and crosscut are wide-open with a moderate restriction placed at Point G,
passing 12 m3/s (25,000 cfm) into the bleeder entries and directs the remaining air flow
across the startup room. The gob flow characteristics are that of Mine W with the startup
room mirroring the recovery room in the completed panel.
The oxygen concentration of the adjacent bleeder panel model is seen in Figure 7.4. The
inactive panel has a significant portion with high oxygen ingress, while the shorter active
132
|
Colorado School of Mines
|
returnmodelwiththefaceairdirectedonecrosscutinbytheface. Thebackreturnventilation
pattern increases oxygen concentration directly behind the tailgate side as compared to
the U-Type ventilation pattern. These results were published by Marts et al. (2014a) and
Gilmore et al., 2015c to develop the larger bleeder panel models.
(a) U-Type Ventilation – Mine C (b) Back Return Ventilation – Mine C
Figure 7.6: Oxygen Concentration of a Progressively Sealed Gob with Nitrogen Injection
and Gob Ventilation Boreholes – Mine C (Gilmore et al., 2015c)
Figure 7.7 shows a progressively sealed gob ventilation pattern with increased panel
length and multiple possible nitrogen injection points on the headgate side (Marts et al.,
2015). The comparison between progressively sealed gobs without nitrogen injection (left)
and with nitrogen injection (right) shows that the development of an EGZ can be eliminated
with a balanced back return flow and the correct selection of headgate nitrogen injection
locations. This forms a “dynamic seal” of low oxygen concentration incapable of forming an
explosive mixture before transitioning to a breathable concentration closer to the face.
Figure 7.8 shows another step towards continually refining the modular pieces composing
the fluid domain of the longwall simulation. Individual shields spanning the longwall face
are created to more effectively model the pressure drop and turbulent nature of the flow.
136
|
Colorado School of Mines
|
Figure 7.7: Progressively Sealed Gob using a Back Return with and without Nitrogen Injec-
tion and Formation of a Dynamic Seal (Marts et al., 2015)
This creates a series of individual shield modules that interface with the next shield, with the
porousmediaofthegobandopenfluidzoneofthegob-fringe. ResultsusingGPGPUprocess-
ing published by Gilmore et al., 2015a revealed that a more complex turbulent model could
be used, which models the transitional and turbulent flows to take advantage of GPGPU
speed up in Fluent.
The results in this section illustrated the variety and scale of modeling made possible by
using a modular meshing approach. Using modules increases the attention to detail of the
mesh controls to match the geometry during mesh creation and to better match the flow
with the expected simulation solution. Also by using the modular meshing approach, the
controls can easily be adjusted to meet the good quality mesh metric, which ensures a more
stable solution and faster solution times. Once a complete modular mesh is assembled to
represent the fluid domain and an initial first-order solution is obtained, then the process
of refinement through mesh adaptation can be used to increase accuracy. The next section
discusses the results of mesh adaptation and refinement.
137
|
Colorado School of Mines
|
(a) 3D-View of Mesh Modules (not to scale) (b) Detail View of Longwall Face Mesh (not to
scale)
Figure 7.8: Individual Shield Mesh Modules (Gilmore et al., 2015a)
7.3 Mesh Independent Study
A mesh independent study was conducted by using mesh adaptation in two stages. The
first stage marked the cells for refinement based on the solution gradients of turbulence,
pressure and species. The second stage marked all the cells with a volume greater than
0.24 m3(8.5 ft3) for refinement. The adaptive mesh refinement is limited by the available
memory on the supercomputer node and since the initial mesh contains polyhedral cells
not capable of adaptation. The polyhedral cells are found in the gob where the cell size
transitions from one size to another within the cut-cell method meshing zones and in the
facemoduleneartheshieldgaps. Inaddition, meshrefinementislimitedbytheporousmedia
model assumption that the mesh size must be greater than the particle size, as discussed in
Section 2.2. The first stage of the study doubled the cell count from 16 million to 32 million
cells and the second stage from 32 million to 39 million cells, at which point computational
limitations occurred. The following results used Mine E gob flow characteristics to evaluate
solution mesh dependence.
Figure 7.9 shows the results of velocity magnitude using the initial mesh plotted with
smooth interpolation between cells. The two insets on the right of the figure magnify a view
of the headgate (bottom) and tailgate (top) near the face. These insets show the face air
138
|
Colorado School of Mines
|
turning the headgate corner, the air as it passes through each approximated shield gap and
the air turning the corner into the back return on its way to the tailgate center entry. The
sharp edges of velocity in the gob on the startup and recovery ends of the panel are clearly
the result of the body of influence shown in Figure 4.10 and the resulting mesh refinement
in Figure 6.4. The inset on the bottom of the figure shows a detailed view of the cell edge
size transitions from 0.3 m to 4.9 m (1 ft to 16 ft). This also shows the high velocity within
the gob-fringe zone next to the gob and air flow in the headgate center entry. The two insets
on the left of the figure show the startup room on the headgate side (bottom) and tailgate
side (top). The majority of the air flows out of the tailgate center entry to the bleeder fan
with the regulated flow at Point 1.
Figure 7.10 shows the mesh after the first stage of refinement where the resulting velocity
plot was shown in Figure 7.2. The mesh is adapted at the corners where the air changes
direction, in the center of the gob where large species gradients occur, and near the shield
gaps. Also in many places the layer of inflation is refined, which requires the scalable wall
function discussed in Section 5.3.3. The resulting velocity plot is smoother in the gob near
the startup room and recovery ends, and the minimum velocity magnitude in the gob has
changed.
Figure7.11showsthemeshafterthesecondadaptationwhenallthecellshavingavolume
greater than 0.24 m3 (8.5 ft3), which can change do. This adaptation is then followed by
another refinement using turbulent solution gradients. Figure 7.12 shows the velocity results
using this mesh, but the resulting solution is unconverged in turbulence. This can be seen by
an examination of the velocity results in Figure 7.12 near the face as compared to Figure 7.2.
The stabilization of the recirculation zone as the face air turns the corner at the headgate
is yet unresolved. This is the result of the adaptation being limited by polyhedral cells near
the shield gaps.
139
|
Colorado School of Mines
|
7.4 Discussion of the Darcy Flow Assumption
This section evaluates the validity of including only the first term in Equation 3.45 to
calculate the flow through the porous medium. Using a custom-field-function to calculate a
Reynolds number (Ward, 1964) (see Figure 2.3) is given in Equation 7.1
ρ×|V|
Re = √ (7.1)
K
Resistance×µ
laminar
This Reynolds number is plotted from the solution of a U-type ventilation scheme in Fig-
ure 7.13 with and without the application of the inertia resistance term. The maximum
value without the inertia resistance term is 10, which is well past the Darcian flow regime
and with the inclusion of this term the maximum value drops to 3, which is just outside the
bounds of this regime. The effects are located in an area behind the shields where the shield
gap assumption has been implemented, and further mesh details would need to be included
before arriving any conclusion.
(a) Reynolds number without Inertia Resistance (b) Reynolds number with Inertia Resistance
Term Term
Figure 7.13: Inertia Resistance Effects for U-Type Ventilation
Figure 7.14 shows the results of the Reynolds number from the Mine E bleeder-ventilated
gob simulation solution. The maximum value is 700 and quickly drops to 2.5 just behind
the shields, after which a value of less than one is maintained throughout the remainder
of the gob. This suggests the need to refine the shield gap details and to use the inertia
144
|
Colorado School of Mines
|
Figure 7.15: Spherical Packing Pore Size
where the pore size, p of the S1 interstitial, is 0.029 m. The porous media assumption
requires that this ratio be much greater than one (see Section 2.2). In this research, the
smallest model cell size in the gob, D2, is 0.3 m as shown in Figure 7.15 in grey, and the
ratio increases to 10.5.
The calculated porosity behind the shields starts at 50% to 40% depending on the model
and rapidly deceases to 30%. The pore size may start out larger than the ideal case, but
due to compaction, the pore size decreases to a smaller size. Therefore, the porous media
assumption is likely less valid near the face and quickly becomes more valid at an inby
location of 30 m to 60 m (100 ft to 200 ft). The effected face area is approximately 1% to
2% of the total model length and would likely have a small effect on the overall development
of EGZs that are forming in the gob.
146
|
Colorado School of Mines
|
7.6 Validation Conclusions
The CFD simulations run in this research represent coal mines located in the Western
United States, which have higher seam heights as compared to coal mines of the Eastern
United States discussed in Section 7.1. A velocity difference within an order of magnitude
can be considered a reasonable comparison.
The modular meshing approach is shown to be beneficial to the development of the first
simulations for many ventilation layouts. Mesh independence using Mine E equation fit is
yet to be achieved due to model size and computational limitations. The trends of EGZ
development using the assumptions made so far about the geometry of the mine are not
likely to change the results with further refinement. Furthermore, it was noted that mesh
independence using Mine C equation is observed in the first adaptation with no change in
the size or location of the EGZ. This is likely due to lower velocities found in the gob because
of greater flow restriction.
Darcian flow in the porous media assumption holds for the majority of the model and
breaks down near the face where other geometry assumptions are made. The details of the
shield and active face geometry assumptions made in this research have reached the limits of
further mesh independence studies and the limits of using the porous media model. Further
research is suggested in these areas as discussed in Chapter 10.
147
|
Colorado School of Mines
|
CHAPTER 8
EXPLOSIVE GAS ZONES IN BLEEDER-VENTILATED GOBS
ThischapterincludesdiscussionofresultspublishedbyGilmoreet al.,2014,2015b. These
results used the porosity and resistance curve fit used by Worrall (2012), for Mine W (see
Section 4.3.5). These results are compared to porosity and resistance curve fits from Mine C
and Mine E (see Sections 4.3.2–4.3.3). Also discussed are the importance of the ventilation
controls at Point 1 and Point 2, and the bleeder-ventilated gob scheme effectiveness in
eliminating the EGZ hazard. The chapter concludes with a discussion on the EGZ hazards
that remain in bleeder-ventilated gobs.
8.1 Bleeder-ventilated Gob – Mine W
The following section includes the EGZ results of the CFD simulation of a bleeder-
ventilated gob using the Mine W porosity and permeability, and also the development of the
ventilation controls necessary to balance the released methane between evaluation points.
Furthermore, a simulation of a ventilation reversal on the tailgate entries shows the EGZ
approaching the face.
8.1.1 Headgate Side Regulator near the Startup Room at Point 1 – Mine W
Figure 8.1 shows the EGZ results for Cases 1 through 3 that adjust the flow through the
regulator at Point 1 with unrestricted flow at Point 2. In Case 1, a stopping is placed at
Point 1 as shown in Figure 8.1(a) that directs the flow through the open crosscuts into the
startup room and first entry inby (Point 2). This sufficiently dilutes the methane build up
in the immediate startup room; however, the fresh air supply is overwhelmed by the time it
reaches the tailgate entry, as seen on the left of Figure 8.1(a). In addition, an EGZ is located
in the first crosscut outby Point 2 as seen on the right of Figure 8.1(a). Also, by placing a
stopping at Point 1, no air is supplied to the bleeder entries outside the model making this
148
|
Colorado School of Mines
|
an invalid operating point. Case 2a in Figure 8.1(b) regulates the flow at Point 1 to 9.4 m3/s
(20,000 cfm), which is the minimum required flow into two bleeder entries. This operating
condition sufficiently sweeps the first entry inby the startup room, but the startup room is
filled with an EGZ. In addition, EGZs are observed in the crosscut accessing the startup
room and the first crosscut outby. Case 3 in Figure 8.1(c) regulates the flow at Point 1 to
16.5 m3/s (35,000 cfm) increasing the size of the EGZ in the startup room.
(a) Case 1: Open Entries at (b) Case 2a: Open Entries (c) Case 3: Open Entries
Point 2 and a Stopping In- at Point 2 and 9.4 m3/s at Point 2 and 16.5 m3/s
stalled at Point 1 (20,000 cfm) at Point 1 (35,000 cfm) at Point 1
Figure 8.1: Case 1, 2a, and 3: Regulator Controls at Point 1 – Mine W (Gilmore et al.,
2015b)
Figure 8.2 shows the resulting normalized EGZ size and methane concentration for con-
trols placed at Point 1 for Cases 1 through 3. Note that the EGZ size is normalized to the
size of the EGZ in Case 3. The EGZ size is minimized with the flow of Case 2a when using
9.4 m3/s (20,000 cfm) flowing through Point 1. Also, the methane concentration ranges from
1.9% to 2.1%, which shows a weak correlation response to the change in the size of the EGZ.
149
|
Colorado School of Mines
|
Figure 8.2: EGZ and Methane Concentration Response to Adjustment of Regulator at
Point 1 – Mine W (Gilmore et al., 2015b)
This ventilation model is incomplete because the common, though not required, practice
istolimitaccess, andthereforeventilation, intothecrosscutsleadingatPoint2inFigure4.1.
Case 2a minimized EGZ size and is selected for further study in the next section.
8.1.2 Headgate Side Flowrates into the Startup Room at Point 2 – Mine W
Figure 8.3 shows Cases 2b and 2c where the flow through Point 1 is fixed at 9.4 m3/s
(20,000cfm)andtheflowthroughPoint2isregulated. Figure8.3(a)uses8m3/s(17,000cfm)
at Point 2 and Figure 8.3(b) has a stopping. The effect of regulating the flow through Point 2
increases the flow through crosscuts outby the startup room on the headgate side, as shown
on the right of the figures. This directs more air into the startup room and eliminates the
methane accumulation. In the crosscuts on the tailgate side are filled with EGZs, and the
EGZ size has significantly increased.
Figure 8.4 shows the EGZ size and methane concentration at Point 1 in response to
the regulated flow through Point 2. As the flow is decreased from 17.5 m3/s (37,000 cfm)
150
|
Colorado School of Mines
|
Figure 8.5: EGZ with Balanced Controls near the Startup Room – Mine W (Gilmore et al.,
2014)
8.1.4 Changing Flow Direction on the Tailgate Entries – Points C and D –
Mine W
A common problem with maintaining an operational bleeder-ventilated gob system is
keeping the tailgate entry open all the way from the face to the startup room. In the United
States, a three-development entry system reduces to a single center entry that must carry all
the face air in addition to any air supplied to the tailgate side of the panel. First reversing
the flow on the tailgate entries and then increasing the flow to the amount of the air supplied
to the face simulates the caving of the tailgate entry.
Figure 8.6 shows a series of steady-state CFD simulations with increased flow out of
the tailgate gate entries. Figure 8.6(a) shows reversing the flow at Point D to 4.7 m3/s
(10,000 cfm) and 7 m3/s (15,000 cfm) at Point C. The EGZ has moved closer to the face and
thetailgateentriesnearthestartuproomarefilledwithafuel-richmixture. InFigure8.6(b),
the flow is increased at the tailgate entries to 9.4 m3/s (20,000 cfm) at Point D and 16.5 m3/s
(35,000cfm)atPointC.TheEGZcontinuestomovetowardstheface; theflowoutoftailgate
entry to the bleeder entries at Point F has reduced and the methane is at a concentration
of greater than 2%. Figure 8.6(c) further increases the flow out at Point C to 28.3 m3/s
(60,000 cfm) and reduces the flow out of Point D to 0.047 m3/s (100 cfm), which has little
effect on the model. In Figure 8.6(d), an increased flow at Point C to 33 m3/s (70,000 cfm)
may seem to have little effect on the overall placement of the EGZ, but the startup room
and the tailgate entry have filled with a fuel-rich mixture. Also a closer examination of the
face as shown in Figure 8.6(e), reveals an EGZ behind the face on the tailgate side.
153
|
Colorado School of Mines
|
8.2 Bleeder-ventilated Gob – Mine E
The gob flow characteristics of Mine E (see Section 4.3.3) are used in the results in the
next sections. The results of adjusting the headgate side regulator are similar to Cases 1–3,
but are studied with an expanded flow range up to 33 m3/s (70,000 cfm) to simulate the
effects of a flow reversal in the startup room. Also, the model sensitivity to the maximum
gob resistance parameter is examined, followed by results of mesh adaptation and the change
in EGZ that occurs. Finally to illustrate the three-dimensional shape of the EGZ in the gob,
vertical cross-sections are shown at every crosscut and a three-dimensional rendered surface
is presented.
8.2.1 Headgate Side Regulator near the Startup Room at Point 1 – Mine E
Figure 8.7 shows the results of changing the flow from 0.047 m3/s (100 cfm), 14.2 m3/s
(30,000 cfm) and 18.9 m3/s (40,000 cfm) with unrestricted flow at Point 2. These results
follow the same progression as seen in Cases 1–3 of the startup room filling with an EGZ.
However, the EGZ fills a greater number crosscuts on the headgate side. Figure 8.8 shows
results of further increasing the flow out at Point 1 causing a reversal of the flow across the
startup room at Point 2. The sequence of reversal begins with the EGZ filling both the
startup room and first entry inby, then the EGZ becomes a fuel-rich mixture, and finally
the fuel-rich mixture is swept with fresh air from the tailgate side. However, the fuel-rich
zone dilutes directly into the headgate side regulator at Point 1 and is a violation of the 2%
methane concentration regulatory limits.
Figure 8.9 shows the EGZ and methane concentration response to the adjustment of flow
through Point 1. The EGZ shows a minimum occurring again at 9.4 m3/s (20,000 cfm)
although the methane concentration does not follow the same trends as shown in Figure 8.2.
The rapid increase in methane concentration is a result of the flow reversal at Point 2.
155
|
Colorado School of Mines
|
Figure 8.9: EGZ Normalized Volume and Methane Concentration Response to Adjustment
of Regulator at Point 1 – Mine E
8.2.2 Permeability Sensitivity Study – Mine E
Thepermeabilityofthegobisthelargestsourceoferrorandvariesgreatlybetweenmines
to the point that every panel can be different and, indeed vary within itself. The following
CFD simulations use 9.4 m3/s (20,000 cfm) of air flowing through Point 1 and the crosscuts
at Point 2 remain unrestricted. Figure 8.10 and Figure 8.11 show a series of EGZ results
when the base viscous resistance is multiplied by a scalar. The maximum viscous resistance
starts at 9.7 × 1041/m2 as shown in the series in Figure 8.10, which is 20% of the initial
value of 4.9×1051/m2, and increases 500 times to 2.4×1081/m2 as shown in the last series
in Figure 8.11. The lowest resistance value in this study is above the maximum milli-Darcy
values used by other researchers (see Table 2.2), and the highest resistance value is of the
same order. A wedge shaped fuel-rich zone exists in many of the results near the startup
room with an accompanying EGZ in the startup room, which may be possible to dilute by
restricting flow at Point 2. These figures show that there is a persistent EGZ in the gob,
and it often fills many crosscuts regardless of the viscous resistance.
157
|
Colorado School of Mines
|
Figure 8.12 shows the effects of the maximum viscous resistance on the EGZ size and
methane concentration as reported at Point 1. The range of viscous resistance values in the
research varies from 105 to 107 resulting in a 20% change in the EGZ size and a variation
in methane concentration at Point 1 from 0.8% to 2.9%. The trend seen in the EGZ results
has a minimum in the viscous resistance range in this research, and therefore the formation
of EGZs persists throughout all ranges of viscous resistance values studied. This further
suggests that the error associated with the choice of viscous resistance is ±10% of the EGZ
and ±1.85% of the methane concentration at Point 1.
Figure 8.12: Resistance effects on EGZ Normalized Volume and Methane Concentration –
Mine E
8.2.3 Mine E Results
Figure 8.13 shows the EGZ results of the full panel using the Mine E equation fit with
9.4 m3/s (20,000 cfm) of air flowing through Point 1 and unrestricted flow at Point 2. The
resulting EGZ near the startup room is similar to the results as shown in Figure 8.1(b),
and the procedure of balancing the flow through the startup room would still be required to
dilute the EGZ. However, the balancing procedure has very little effect on the EGZ in the
160
|
Colorado School of Mines
|
Figure 8.13: Full Panel View of EGZ with 9.4 m3/s (20,000 cfm) through Point 1 and
Unrestricted Flow through Point 2 – Mine E
center of the panel, and the EGZ would still remain continuous as it fills the crosscuts on
the headgate side.
8.2.4 Adapted Mesh – Mine E
Figure 8.14 through Figure 8.17 show the results of the EGZ after the adaptive mesh
process described in Section 7.3. The EGZ has changed in size although there are many
crosscuts on the headgate side that contain an EGZ. The EGZ is in crosscuts starting at the
tenth through the length of the panel. Figure 8.15 and Figure 8.16 show a series of vertical
cross-sections at each crosscut. Figure 8.17 shows the connective nature of the EGZ in a
three-dimensional view of the startup end of the panel. The EGZ first appears as early as
the second crosscut from the face at the top of the gob and becomes continuous across the
panel width by the sixth crosscut. The mesh adaptation process limits further resolution of
the continuous nature of this EGZ.
Figure 8.14: Full Panel View of EGZ using after Mesh Adaptation – Mine E
161
|
Colorado School of Mines
|
33
34
35
36
37
39
40
41
42
43
44
45
46
Figure 8.16: Vertical Cross-sections of EGZ at Crosscuts 33–46 –Mine E
8.3 Bleeder-ventilated Gob – Mine C
The gob flow characteristics of Mine C (see Section 4.3.2) are used in the following EGZ
results. The results of adjustments made to the flow through the headgate side regulator
at Point 1 are similar to Cases 1–3, and the results of the full bleeder panel with the same
operating conditions are presented.
8.3.1 Headgate Side Regulator near the Startup Room at Point 1 – Mine C
Figure 8.18 shows the results of changing the flow at Point 1 from 0.047 m3/s (100 cfm),
11 m3/s (30,000 cfm) and 5.8 m3/s (40,000 cfm) with unrestricted flow at Point 2. The
crosscuts on the headgate side are filled with EGZs or a fuel-rich methane mixture. Also
the corner of the gob on the headgate side remains filled with a fuel-rich mixture instead of
transitioning to an EGZ as shown previously in Figure 8.7. Increasing the flow out at Point 1
and causing flow reversal across Point 2 still has the same effect of clearing the methane in
thestartuproom, whichoverwhelmstheflowatPoint1withhighconcentrationsofmethane.
163
|
Colorado School of Mines
|
Figure 8.20 shows the EGZ and the methane concentration at Point 1 responding to the
flow change at Point 1. The resulting trend in the EGZ is a minima occurring again at
9.4 m3/s (20,000 cfm), but the total range increases to 30% of the maximum. The methane
concentration reported at Point 1 follows the same trend as shown in Figure 8.9, but with
more values above the 2% limit. The observed peak in methane concentration relates to the
flow direction change across Point 2.
Figure 8.20: EGZ Normalized Volume and Methane Concentration Response to Adjustment
of Regulator at Point 1 – Mine C
8.3.2 Mine C Results
Figure8.21showstheEGZresultsofthefullbleeder-ventilatedgobusingMineCequation
fit using 9.4 m3/s (20,000 cfm) of air flowing out at Point 1 and unrestricted flow through
Point 2. The results are similar to those shown in Figure 8.11 of the viscous resistance
sensitivity study with a similar maximum resistance, but the range is larger in Mine C. The
same shape of a fuel-rich zone occurs at the startup end of the panel, and the EGZ in the
first quarter of the panel is nearer the face. The EGZ fills almost all the crosscuts on the
headgate side and extends across the gob in a wide band. This result emphasizes the point
166
|
Colorado School of Mines
|
Figure 8.21: Full Panel View of EGZ with 9.4 m3/s (20,000 cfm) through Point 1 and Open
Entries at Point 2 – Mine C
that although the gob flow characteristics may be the greatest unknown in the system, the
EGZ persists.
8.4 Explosive Hazard Discussion of Bleeder-ventilated Gob Operation
In Figure 8.10 and Figure 8.11, the overall shape of the EGZ is shown to change and
vary greatly in size depending on the gob flow characteristics. However, the accumulation of
methane in the startup room occurs in the models studied. Restricting the flow at Point 2
and forcing more air into crosscuts inby the startup room can mitigate these methane accu-
mulations. It is suggested that there should remain a moderate flow through Point 2 towards
the tailgate side to remove methane in the vicinity of these crosscuts.
The EGZ surrounds the gob and has a wide connecting band from the headgate to the
tailgate side. This connecting EGZ band is independent of ventilation pattern and gob
flow characteristics. The explosive risk posed by this connecting EGZ band is unknown
given the flame propagation characteristics inside the gob, and furthermore, the unknown
continuous nature of the gob-fringe connecting the headgate to tailgate side across the gob.
The ventilation controls near the startup room have very little control on the size of this
band, and further simulation beyond the scope of this research is suggested.
The flow reversal at Point C and D suggest the formation of an EGZ directly behind
the shield on the longwall face. The formation of this ventilation pattern may be the result
of a caving tailgate entry near the face or in the back return crosscut causing an U-Type
ventilation pattern. The EGZ risk is similar to results published by Marts et al., 2014a in
167
|
Colorado School of Mines
|
CHAPTER 9
CONCLUSIONS
The CFD model used simulates a bleeder-ventilated gob in this research. Through a
study of three gob flow characteristics, each examining regulator controls on the headgate
side near the startup room (Section 8.1.1 and 8.1.2, Section 8.2.1 and Section 8.3.1) and
a sensitivity study of scaling the porosity relationship to permeability (Section 8.2.2), has
demonstrated the existence of a persistent EGZ surrounding the gob. Also, a study of a
changing ventilation pattern as the flow reverses at the tailgate demonstrated the efficacy of
the CFD model to predict the development of an EGZ directly behind the shields (Section
8.1.4). The modular meshing approach employed in this research, which was used to create
the models, has progressively advanced the details of the mesh to better represent the actual
mine geometry, and improvements continue with collaboration in the CSM research group
(Marts et al., 2014a; Saki et al., 2015).
Modeling results show the presence of a wide EGZ band crossing the gob that connects
the tailgate to headgate side, which could propagate an ignition event throughout the mine.
Furthermore, an EGZ threat to safe mine operation exists in many crosscuts on both the
headgate and tailgate side. The hazards posed by these EGZs are predicted to occur in the
studied bleeder-ventilated gob models independent of the gob flow characteristics and scaled
permeability.
Ventilation controls near the startup room on the headgate side are essential to eliminate
the EGZ hazard in the startup room and nearby crosscuts. The flow direction from headgate
to tailgate side must be maintained through the crosscuts at the startup room (Point 2),
and if possible in as many crosscuts outby the startup room (Section 8.1.2). However, the
EGZ has been shown to increase in size when forcing sufficient air into the open mine entries
and crosscuts in an attempt to dilute the methane accumulations. The controls placed near
169
|
Colorado School of Mines
|
the startup room on the headgate side (Point 1 and 2) only affect the area surrounding the
startup room and do not have sufficient impact to limit the development of EGZs along the
panel length.
The sensitivity study of the model to the strata geology, overburden depth and strata
in the caving zone forming the gob, which vary significantly between mines, panels or even
within the same panel and result in large variation in the values of porosity and permeability,
has demonstrated that the EGZ remains, but will vary in size and location. These pockets of
explosive gas pose a threat to mine workers and violate the intended purpose of an effective
bleeder-ventilated gob system.
Ventilation reversal caused by a roof fall or other unplanned event on the tailgate en-
tries, forming a U-Type ventilation pattern, draws the EGZ closer to the face, causing the
formation of EGZs directly behind the shields and near the open crosscut of the back return
(Section 8.1.4). The flow out of the tailgate entry to the bleeder fan would no longer be
sufficient to dilute the methane concentration, and EGZs fill the tailgate entry across many
crosscuts.
This research simulates the ventilation system of a bleeder-ventilated gob, using Fluent.
It features a developed meshing approach to model the large scale system of a single bleeder
panel, with the gob flow characteristics of three mines. One conclusion finds that ventilation
controls near the startup room are insufficient to dilute the accumulated methane in the
bleeder-ventilated gob system below the explosive limit required for worker safety. As stated
by Stricklin & Fesak, 2013 on bleeder operations, “Accumulations of methane that are explo-
sive, can become explosive when mixed with air, are approaching the explosive range, or are
irrespirable, may pose a hazard to the active workings whether they occur in accessible areas
or not.” This hazard elimination requirement has clearly not been met in the simulations
in this research, but the studies show instead, that the EGZ and high methane concentra-
tions persist across permeabilities studied and through the ventilation control adjustments
studied.
170
|
Colorado School of Mines
|
CHAPTER 10
SUGGESTIONS FOR FUTURE WORK
The future implications for work and model improvements are detailed in the following
chapter, including the areas of geometry and meshing refinements, gob characterization, and
solution convergence criteria. Furthermore, it is recommended to repeat many of the studies
and sensitivity studies completed previously by Worrall, 2012, which included the study of
face ventilation quantity effects on EGZ size and location.
10.1 Improvements to the Mine Geometry and Meshing
Improvements to the geometric representation and the resulting modular meshing ap-
proach for large and small models would increase the usefulness of CFD as a predictive tool,
decrease solution time, increase accuracy and make results reproducible. Some of the areas
that this applies to are listed below:
(cid:136) The mine entry network mesh representation uses two primary straight entry modules
that poorly resolve the flow in areas of the corners, tee junctions, and cross junctions.
The creation of mesh modules to represent these flow junctions would improve solution
times and convergence. This would isolate the inflation layer to the wall boundary, and
eliminate the computational errors in the center of the flow at the junction between
two modules.
(cid:136) TheuseofANSYSWorkbenchinVersion16.0multipleupstreammeshingwouldcreate
a more usable interface for mesh creation. This would utilize the Fluent meshing
mode, through the use of a custom journal file, to assemble the modules into the
final mesh. Furthermore, user functionality could be increased by creating tools with
ANSYS Software Development Kits that would read a mine map file and build the
representative mesh.
171
|
Colorado School of Mines
|
(cid:136) The longwall face used in this project is an over-simplification of very complex ge-
ometry. Additional modules could be created to represent the shields, armored face
conveyor, tailgate and headgate drives, and the crusher obstructing the headgate entry.
This would result in better resolution of the velocity and turbulent flows in the face
and produce a more realistic pressure drop across the face.
(cid:136) The wall boundary conditions, simplified face and gob-fringe geometry are insufficient
to accurately produce a realistic pressure drop. Therefore, it is suggested to change
the entries’ wall boundary conditions to reflect the appropriate roughness, and increase
the complexity in the shield model. This would result in a substantial increase in the
number of cells in the model and require the use of a complex turbulent flow model
(Gilmore et al., 2015a).
10.2 Improvements to the Gob Characterization
The unknown flow characteristics of the gob are the largest source of error in the system,
which includes the transport modeling equations and the gob-fringe modeling. The following
are some suggested improvements:
(cid:136) Additional refinements to the applied porosity distribution by including a vertical
direction dependence currently vary only on the mining plane. The addition of vertical
direction dependence in the model would require varying the compaction of the gob at
different heights and need to incorporate the variation in the observed block size. The
particle size was assumed constant, but a distribution in the vertical direction would
be more appropriate. The UDF coding used in this research that calculates each step
to determine the porosity and permeability is functionally ready to program a particle
distribution in future work.
(cid:136) The full transport equation used in the porous media (Section 3.7) includes an addi-
tional inertia resistance term, C . The calculation of this value from a relationship to
2
172
|
Colorado School of Mines
|
porosity has been coded into the UDF, but a sensitivity study to examine the effects
it will have near the gateroad and behind the face is still required.
(cid:136) Theresultingflowinthegobisheavilydependentonthegob-fringecharacteristicssuch
assize,geometryandextent. Thegob-fringeconnectsthefluidzonesbetweencrosscuts,
may extend the length of the gob, exist above the shields and near the startup room.
The VSI distribution when including a vertical direction dependence may be used to
better define the shape and extent of the gob-fringe. Assigning a porosity value of one
to the gob-fringe and using the relative velocity formulation would create a smooth
transition from the fluid zone to the porous media. The roughness and large voids on
the surface of the gob material would still require a geometric mesh representation not
practical scale of a full mine.
10.3 Solution Convergence Monitor
It is often challenging to determining when a simulation has converged to a solution and
to correctly identify an important criterion to monitor. This research has identified the EGZ
volume integral, which combines the results of the species concentration of methane and oxy-
gen as the variable of interest. This variable also changes significantly with more iterations
and adaptations of the mesh. To improve solution control, this research recommends the
use of a user defined scalar equation that calculates the EGZ integral on an iteration bases.
This will result in a residual monitor of the scalar equation that can be used by the solver
to automate the identification of a converged solution. This is preferred over calculating the
EGZ algorithm, EGZ integral, and reporting a value at regular intervals.
10.4 Ignition and Explosive Potential of Explosive Gas Zones
The research funding from NIOSH has been extended to examine the scaling of flame
propagation with tube sizes of 0.05 m, 0.13 m, 0.30 m, and 0.76 m (0.16 ft, 0.4 ft, 1 ft,
2.5 ft). The study includes unrestricted laminar flames and flames passing through rock
rubble. Also the CFD simulations of a one-step chemistry model will be validated against
173
|
Colorado School of Mines
|
these experimental results. This will identify the threat posed by the predicted EGZs in this
dissertation.
10.5 Validation of Models with Mine Gas Measurements
The ultimate validation of this research is in the measurement of an existing EGZ in the
mineventilationnetworkatthelocationspredicted. TheidentificationofanEGZinanactive
working mine would result in the immediate shutdown and evacuation of all non-essential
personnel to correct the situation. Furthermore, the measurement would be a violation of
CFR 30 and result in a fine. Mine operators are thus extremely opposed to further research
in this area.
Further validation could be achieved through an examination of explosions that have
occurred in the presence of an effectively operating bleeder-ventilated gob. Brune, 2014
discusses a number of these cases and the need for more research in this area in the following
statement:
In the United States, a targeted, comprehensive research program would deter-
mine whether longwall bleeder ventilation systems can be designed such that
they are truly effective in diluting and rendering harmless accumulations of ex-
plosive methane-air mixtures, or if alternative longwall gob ventilation systems
are needed.
174
|
Colorado School of Mines
|
Grubb, J. W. 2008. Preventative Measures for Spontaneous Combustion in Underground
Coal Mines. Dissertation, Colorado School of Mines.
Grubb, John W, Brune, Ju¨rgen F, Zipf, R Karl, Bogin, Gregory E, Marts, Jonathan A,
Gilmore, Richard C, Saki, Saqib A, & Lolon, Samuel A. 2015. Managing the Risk of
Spontaneous Combustion in Underground Coal Mines. 15th Annual North American Ven-
tilation Conference.
Highton, W. 1979. The Case Against Bleeder Entries and the Reasons for a Safer and More
Efficient Alternative. Pages 437–447 of: 2nd International Mine Ventilation Congress.
Hill, R. W. 1995. Multiseam Mining in South African Collieries. In: 14th International
Conference on Ground Control in Mining.
Hoek, E., & Franklin, J. A. 1968. A Simple Triaxial Cell for Field and Laboratory Testing
of Rock. Transactions of the Institute of Mining and Metallurgy, 77, 22–26.
Itasca Consulting Group. 2013. FLAC3D, Version 5.0.
Jan, B. M., Siagian, U. W., Lee, R. L., & Othman, R. 2002. A New and Robust Simulation
Model for Coalbed Methane Reservoirs. Pages 1–8 of: Society of Petroleum Engineers -
Eastern Regional Meeting.
Jozefowicz, R. R. 1997. The Post-Failure Stress-Permeability Behaviour of Coal Measure
Rocks. Dissertation, University of Nottingham.
¨
Karacan, C. O. 2009a. Degasification System Selection for US Longwall Mines using an
Expert Classification System. Computers & Geosciences, 35(3), 515–526.
¨
Karacan, C. O. 2009b. Forecasting Gob Gas Venthole Production Performances using In-
telligent Computing Methods for Optimum Methane Control in Longwall Coal Mines.
International Journal of Coal Geology, 79(4), 131–144.
¨
Karacan, C. O. 2009c. Reconciling Longwall Gob Gas Reservoirs and Venthole Production
Performances using Multiple Rate Drawdown Well Test Analysis. International Journal
of Coal Geology, 80(3-4), 181–195.
¨
Karacan, C. O. 2009d. Reservoir Engineering Considerations for Coal Seam Degasification
and Methane Control in Underground Coal Mines. Tech. rept. NIOSH.
¨
Karacan, C. O. 2010. Prediction of Porosity and Permeability of Caved Zone in Longwall
Gobs. Transport in Porous Media, 82(2), 413–439.
178
|
Colorado School of Mines
|
Ren, T., & Wang, Z. 2013. Modelling of Respirable Dust and Gas Behaviour on a Longwall
Face. Pages 191–200 of: Mine Ventilation.
Ren, T., Balusu, R., & Claassen, C. 2011. Computational Fluid Dynamics Modelling of
Gas Flow Dynamics in Large Longwall Goaf Areas. Pages 603–613 of: 35th APCOM
Symposium - Application of Computers and Operations Research in the Minerals Industry.
Ren, T. X. 2009. CFD modelling of Longwall Goaf Gas Flow to Improve Gas Capture and
Prevent Goaf Self-heating. Journal of Coal Science and Engineering, 15(3), 225–228.
Ren, T. X., & Edwards, J. S. 1997. Research into the Problem of Spontaneous Combustion
of Coal. In: 6th International Mine Ventilation Congress, vol. 39.
Ren, T. X., Edwards, J. S., & Jozefowicz, R. R. 1997. CFD Modelling of Methane Flow
Around Longwall Coal Faces. Pages 247–252 of: Proceedings of the 6th International Mine
Ventilation Congress.
Ren, T. X., Balusu, R., & Humphries, P. 2005. Development of Innovative Goaf Inertisation
Practices to Improve Coal Mine Safety. Pages 315–322 of: Aziz, N. (ed), Coal Operators’
Conference.
Ren, Ting Xiang, & Balusu, Rao. 2009. Proactive Goaf Inertisation for Controlling Longwall
Goaf Heatings. Procedia Earth and Planetary Science, 1(1), 309–315.
Saghafi, A., Williams, D. J., & Lama, R. D. 1997. Worldwide Methane Emissions from
Underground Coal Mining. Pages 441–445 of: Proceedings of the 6th International Mine
Ventilation Congress.
Saki, S. A., Marts, J. A., Gilmore, R. C., Brune, J. F., Bogin, G. E., Jr., & Grubb, J. W.
2015. CFD Study of Face Ventilation Effect on Tailgate Methane Concentration and
Explosive Mixture of Gob in Underground Longwall Coal Mining. Pages 1–4 of: SME
Annual Meeting and Exhibit. Preprint.
Salamon, M. D. G. 1970. Stability, Instability and Design of Pillar Workings. International
Journal of Rock Mechanics and Mining Sciences, 7, 613–631.
Salamon, M. D. G. 1989. Subsidence Prediction using a Laminated Linear model.
Salamon, M. D. G. 1990. Mechanism of Caving in Longwall Coal Mining. Pages 161–168
of: Rock Mechanics Contributions and Challenges: 31st US Symposium.
182
|
Colorado School of Mines
|
APPENDIX A - FLAC3D DATA CURVE FITTING PROCEDURE
The FLAC3D numerical modeling approach follow the initial developed by Esterhuizen &
Karacan, 2007, initial implementation for this project by Wachel, 2012, and final refinements
by Marts et al. (2014b) that resulted in a calibrated model of gob compaction to surface
subsidence. The surface subsidence profile in the panel length, as shown in Figure 2.6,
matches that of the data of the VSI reported by the numerical model shown in Figure A.1.
Mark are the three different zones that break up the data for curve fitting, as discussed in
Section 4.3, namely, a startup zone, center zone, and recovery zone. A bounded range for the
center zone must be determine in the panel length, and also in the width for super-critical
panels, while a sub-critical panel should not have a center zone across the panel width. The
startup zone ranges from 0 m (0 ft) to 190 m (620 ft), and the recovery zone ranges from
700 m (2,300 ft) to 1,000 m (3,280 ft) that leaves a expandable center zone from 190 m
(620 ft) to 700 m (2,300 ft) for extending the panel length.
Figure A.1: Gob Compaction Profile – VSI
186
|
Colorado School of Mines
|
Figure A.2 shows the behavior of the gob compaction and subsidence across the panel
width. Again, a center zone can be identified from the data ranging from the center of the
panel at 0 m (0 ft) to 500 m (1640 ft), where the data exhibits a less than 1% change from the
maximum value. The remaining distance is defined as the width of the gateroad zone. This
Figure was used to verify the super-critical behavior of the physics used in the numerical
model, and the data used in the two curve fits for Mine C and Mine E used a panel half
width of 200 m (655ft). In the actual data curve fit for this project the gateroad width was
identified to occur at 60 m (200 ft) from the gateroad, as shown in Figure 4.12.
Figure A.2: Gob Compaction and Subsidence Profile across Panel Width
A.1 Data Reading and Parsing
The output of the FLAC3D data must read into MATLAB where numerical spikes and
waviness may be smoothed out. Example output lines for the location data and VSI data
are shown in Listing A.1 and Listing A.2. The code to read the location and VSI values into
MATLAB, are shown inListing A.3. Upon execution, the user is prompted for the names of
the location file and data file.
187
|
Colorado School of Mines
|
APPENDIX B - USER-DEFINED-FUNCTION (UDF) CODE
The User-Defined-Function code is contained in Listing B.14 through Listing B.20. The
instruction for defining the variable in the scheme Fluent programming environment can
be found in the Fluent UDF Manual Section 3.6 Scheme Macros and a brief explanation is
included in the comments at the top of Listing B.14. The VSI values are calculated using
oneoffourDEFINE ON DEMANDfunctionsthatsavethevaluesintoauser-define-memory
location number 4 (udm-4). The viscous resistance values, initial permeability, and porosity
functions are used in the DEFINE PROFILE functions.
Listing B.14: UDF Part 1 – Using the Code and Functions
Underground Longwall Mine Equation Fits for the change in porosity
(volumetric strain increment = VSI) output from FLAC3D Itasca ,
Consulting Group, Inc. Model development and methods published by
Jonathan A. Marts , Juergen F. Brune, Richard C. Gilmore , Gregory
E. Bogin Jr. , and John W. Grubb. February 2014. Dynamic Gob Response
and Reservoir Properties for Active Longwall Coal Mines. Society
of Mining , Metallurgy , and Exploration Symposium Annual Meeting
and Exhibit Pre-print , Salt Lake City , UT.
-- Colorado School of Mines --
Author implementation
Richard C. Gilmore & Dan Worrall Jr. & Jonathan A. Marts
-- Uses with ANSYS FLUENT 15.0 --
Text User Interface Control Commands:
see FLUENT UDF manual Section : 3.6 Scheme Macros
REQUIRED settings :
To set the full panel width and length for the equation to fit
(rp-var-define ’vsi/panel-width 200 ’real #f)
(rp-var-define ’vsi/panel-length 3078.48 ’real #f)
To set the offset from (x=0, y=0) at the center panel (x-directions)
of the recovery room (y-direction) or active face . For example the
center of the panel is locateda t a negative x-direction of 341 meters
(rp-var-define ’vsi/panel-xoffset -341 ’real #f)
The recovery room of the panel is located at a positive y-direction
of 940 meters.
(rp-var-define ’vsi/panel-yoffset 940 ’real #f)
To use multiple panels compile to libudf libraries and set the
displacement variables to the correct value before executing . Also ,
203
|
Colorado School of Mines
|
APPENDIX C - MESH ASSEMBLY JOURNAL FILE CREATION CODE
The follow appendix contains the Matlab code for assembling the meshes of the entry and
crosscutventilationsystem. TheheadgatesidecodeisintheexecutableListingC.22, andthe
tailgate side in Listing C.30. The number entry and crosscuts is 45, however, the headgate
numbering starts at the face with crosscut 1, while on the tailgate side the numbering start
at 0 at the startup room and ending with 44 near the face. The tailgate section must be
rotate 180 degree about the z-axis once assembly is complete. An additional 200-foot section
of mesh modules was later added to represent the open tailgate entry to the center tailgate
entry with a back return ventilation pattern. The final mesh assembly used in this project
was carefully assembled from these sections of headgate and tailgate entry systems with the
gob, face, headgate, tailgate, void, and startup room modules. After assembly it is advised
to check graphically the correct alignment of all interfaces to ensure the desired resulting
mesh.
Listing C.22: Assembly of Headgate Side Entries
% Create Entries and Bleeders for FLUENT journal file assembly of mesh
from the
% parts of selected files .
% Such as Void, Crosscuts , entry 60, 220, 40 and bleeder parts .
% Headgate entries begin at 1 and range to 45, building from the face
towards
% the start-up room.
% Name of journal file to save the TUI commands into :
fileID=fopen(’Create-HG-Entries-and-Void.jou ’ , ’w’) ;
% Absolute path to mesh file locations :
path=’”C:\Mesh-Modules\Bleeder-Mesh-Files \ ’;
num of cuts=0;
% Open an initial mesh file for a starting point rename faces/interface
and solids .
% Change the number times the loop executes to the number desired
Entries
% and Crosscuts along the panel length .
for i=1:45;
235
|
Colorado School of Mines
|
APPENDIX D - RUNNING FLUENT ON MIO WITH GOFLUENT
The following syntax is used to execute GoFluent:
gofluent <number of nodes>x<number of processors> <journal file name.jou> <DAYS-
HOURS:MINS>
The parameter, <number of nodes>, is the request number of nodes to run the job on
Mio and the parameter, <number of processors>, is the minimum number of processors on
each node. The job queries the available number and will run on the total number identified
when the job starts. For example, “4x8” will request the use of four nodes with at least eight
processors, but the job scheduler may run the job on 12-core, 16-core or 20-core machines
using all the available processors. The parameter, <journal file name.jou>, is required as
an input and must contain the suffix .jou, which will become the name of the base working
directory for the Mio jobs. The parameter, <DAYS-HOURS:MINS>, for example, uses the
format, 4-14:02, which means the job will halt un-saved 4 days, 14 hours and 2 minutes from
the time it is started. The files for the case, data and C-file are then moved to the Fluent job
name directory named after the journal file, and then a unique folder is created to contain
the transcript and output files from the job.
To setup a Mio enviroment see Listing D.33, and Listing D.34 for the contents a .bashrc
file with the Fluent module loaded for 15.0. In Listing D.35 a simple script to check the
available Fluent licence useage, and in Listing D.36 the bash script for GoFluent.
Listing D.33: Directory Creation
#!/bin/sh
mkdir /scratch/$USER
mkdir /scratch/$USER/runs
ln -s /scratch/$USER/runs ˜/runs
243
|
Colorado School of Mines
|
measured air inflow of 1033 m3/min and methane inflow (to make the in situ measurements more similar to the
into the goaf of longwall 558 in the amount of 24.5 calculated values), a satisfactory result was obtained,
m3/min, including assumed 3 m3/min leakage from goaf of which is shown in Fig. 4 and Table 3, which includes the
the abandoned longwall. The values for the methane inflow distribution of methane concentration in the goaf of
were determined based on the data obtained from the longwall 558 and the adjacent goaf, as well as the routes
mine's monitoring system during the observation period. carrying the air away from longwall 558 and the
Due to the inability to take direct measurements in the decommissioned longwall.
goaf, the parameters of the flow within the goaf has to be When observing the distribution of methane
determined based on a theoretical model of permeability concentration (fig. 4), there are very clear isolines shown
distribution and the formation of the goaf's height every 2% of the CH, visible in the locations of the
4
(Dziurzyński W. 1998), as well as the information provided by integrated goaf sensors. We have observed a methane
the seam map, properties of rocks, the geological profile of the concentration level identical to the one measured by the
area, the longwall mining plan (geometry, spot heights, seam outlet probe. One can notice that even though both gobs
thickness, type of roof rocks). In order to determine the were separated by a coal pillar, the methane from longwall
parameters of the numerical model of longwall 558, seam 558 leaks through permeable pillar to the gobs of the
510, an initial permeability distribution was adopted, then decommissioned longwall. Based on the obtained
modified in further simulations, in order to make the methane simulation results, the methane balance in the workings
concentration results calculated in the simulation similar to the and goaf in the ventilation area was calculated.
values measured at the inlet and outlet measurement stations. Table 3 shows the workings in the area that illustrate
the distribution of the methane stream transported
3.2 An attempt at reproducing the state of ventilation through the workings. Interestingly, the flow balance
in longwall 558 and in the adjacent goaf indicates, that 13 m3/min. of methane flows from the
goaf of longwall 558 into the goaf of adjacent
As a result of conducting a series of simulations in the
decommissioned longwall.
process of validation and changing the model parameters
(5)
outlet (4)
T
200
150
100
50
0
Heading VIIA east, outlet 50 100 150 200 250 300 350 400 450 500 550 600 Coal pillar
200
(3)
150 integrated goaf sensor
Longwall 558
bed 510Dw 100 Sensor of methane
(2) Anemometer
50
Heading IX east wall inlet 0 50 100 150 200 250 300 350 400 450 500 550 600
(1)
lenght in goaf [m]
Figure7.1:FIigs.o 4l. i nIseolsinoesf omf meetthhaanne econccoenntcraetinont rdaisttiroibnutidonis intr tihbe ugotiaof onf lionngtwhaell g55o8a ifn osefamlo 5n10g, wthea lfllow5 518033in ms3/emainm. 510,
theflow1033m3/min. (Dziurzynski&Wasilewski,2012)
Table 3 Calculation results obtained from the simulation of the propagation of air and methane
Amount of air Methane Amount of methane
No. Working name
m3/min concentration, [%] m3/min
1 Heading IX east longwall inlet 1063 0.32 3.40
2 Longwall 558 outlet 1008 0,904 9.47
3 Heading VIIA east, heading outlet 1357 1.12 15.19
4 Outlet of the goaf of the decommissioned longwall 47.4 33.72 16.01
5 Inclined drift I east area outlet 2523 1.29 32.54
(5 )
T outlet (4)
200 116
150
100
50
0
Heading VIIA east, outlet 50 100 150 200 250 300 350 400 450 500 550 600 Coal pillar
200
(3)
150 integrated goaf sensor
Longwall 558
coal seam 510Dw 100 sensor of methane
(2) anemometer
50
Heading IX, intlet east 0 50 100 150 200 250 300 350 400 450 500 550 600
(1)
lenght in goaf [m]
Fig. 5. Isolines of methane concentration distribution in the goaf of longwall 558 in seam 510,
Figure7.2: Isolinesofmethaneconcenthter falotiwo lnimditeisd ttroi 8b0u0t mio3/nmiinn. thegoafoflongwall558inseam510,
theflo wlimitedto800m3/min. (Dziurzynski&Wasilewski,2012)
Table 4. Calculation results obtained from the simulation of air and methane propagation
Airflow 1033 m3/min. Airflow 800 m3/min.
No. Amount Methane Amount Methane
Working name Flow speed Flow speed,
of air concentration of air concentration
m/s m/s
m3/min [%] m3/min [%]
1 Heading IX east
1063 1.32 0.32 831 1.03 0.41
longwall inlet 138
2 Longwall 558 outlet 1008 1.976 0.904 798 1.976 1.15
3 Heading VIIA east,
1357 1,66 1.12 1137 1.39 1.30
outlet
4 Inlet measuring station 24.5 0.11 0.33 24.5 0.11 0.41
5 Outlet measuring station 18.2 0.072 8.9 18.2 0.072 13.21
3.3 A simulation scenario of decreasing the oxygen 4 Summary
expenditure in longwall 558
The experimental study of monitoring the methane and fire
The positive result of the validation calculations for hazard with continuous measurement of goaf gases using
longwall 558 and its goaf allows us to show consequences an automatic measurement system has confirmed
of a decrease of flow of air in longwall 558a to 800m3/min. (Wasilewski St., Cimr A., Wach M., 2010) that this method
This solution is shown in the form of the methane can be considered complementary to the early fire
concentration isolines in Figure 5. Table 4 shows a detection procedure required by the regulations. This
comparison of calculation results from the simulation of method of monitoring the hazard is independent and allows
the propagation of air and methane in the longwall an objective observation of the composition of goaf air
workings, as well as in the goaf in the locations of inlet and without the participation of sampling personnel and the
outlet probes with limited air inflow into the area from need to transport air samples to the surface.
1033 m3/min, to 800 m3/min. resulting from spontaneous In the analysis of methane ignitions and explosions,
combustion hazard When comparing the distribution of especially in the goaf of caving longwalls, computer
methane concentrations in the goaf for a flow limited to simulation methods have become increasingly useful. The
800 m3/min (Table 4) to the results calculated for the initial effectiveness of simulation and the reliability of
flow of 1033m3/min, one can notice a significant increase calculations is possible, provided that the models are
of the methane concentration in the goaf for longwall 558. validated using in situ data, e.g., obtained in experiments
Moreover, it is very important to note the hazardous or from monitoring systems. Such studies are presented in
approach of the high explosive methane concentrations this paper for the data recorded in the goaf of caving
closer to longwall 558. longwall ventilated using a “U“ system.
117
Inclined
drift
I
east
Inclined
driftIeast
T
T
T
T
|
Colorado School of Mines
|
libudf” no no 0 no 0 no 0 0 0 yes yes ”udf” ”set porosity ::
libudf” no
/ define /boundary−conditions / fluid gob assy−gvb tube 1 fluid no no no
no no 0 no 0 no 0 no 0 no 0 no 1 no no yes yes yes no no 1 no 0 no
0 no 0 no 1 no 0 yes yes yes ”udf” ”set GVB tube perm 1 :: libudf” yes
yes ”udf” ”set GVB tube perm 2 :: libudf” yes yes ”udf” ”
set GVB tube perm 3 :: libudf” no no 0 no 0 no 0 0 0 yes yes ”udf” ”
set porosity :: libudf” no
/ define /boundary−conditions / fluid gob assy−gvb tube 2 fluid no no no
no no 0 no 0 no 0 no 0 no 0 no 1 no no yes yes yes no no 1 no 0 no
0 no 0 no 1 no 0 yes yes yes ”udf” ”set GVB tube perm 1 :: libudf” yes
yes ”udf” ”set GVB tube perm 2 :: libudf” yes yes ”udf” ”
set GVB tube perm 3 :: libudf” no no 0 no 0 no 0 0 0 yes yes ”udf” ”
set porosity :: libudf” no
;
;
; Set perm and porosity for upper coal seam and fractured zone
/ define /boundary−conditions / fluid gob assy−fractured zone fluid no no
no no no 0 no 0 no 0 no 0 no 0 no 1 no no no yes no no 1 no 0 no 0
no 0 no 1 no 0 yes no 1.0133e+12 no 1.0133e+12 no 1.0133e+12 no no 0
no 0 no 0 0 0 no 0.1 no
/ define /boundary−conditions / fluid gob assy−upper coal seam fluid no no
no no no 0 no 0 no 0 no 0 no 0 no 1 no no no yes no no 1 no 0 no 0
no 0 no 1 no 0 yes no 1.0133e+12 no 1.0133e+12 no 1.0133e+12 no no 0
no 0 no 0 0 0 no 0.1 no
;
;
;
;
;
;!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
;DEFINE BOUNDARY CONDITIONS
;!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
;
; Original turbulence BC definition / define /boundary−conditions / velocity
−inlet aa inlet ventilation entry no no yes yes no 1.80360 no 0 no
300 yes no 1 no 1 yes no 0.00000 no 0.200000
; species−0 is CH4, species−1 is O2, remainder is N2
;
;
;Turn Void on headgate side to wall
/ define /boundary−conditions /zone−type aa hg void outlet wall
;/ define /boundary−conditions /zone−type aa inlet gvb 1 pressure−outlet
; P i
|mass target
(kg/s)
/ define /boundary−conditions /zone−type aa inlet gvb 1 velocity−inlet
253
|
Colorado School of Mines
|
/ define /boundary−conditions / velocity−inlet aa inlet gvb 1 no no yes yes
no −0.26182 no 0 no 300 yes no 1 no 1 no no 1 no 0
/ define /boundary−conditions /zone−type aa inlet gvb 2 velocity−inlet
/ define /boundary−conditions / velocity−inlet aa inlet gvb 2 no no yes yes
no −0.26182 no 0 no 300 yes no 1 no 1 no no 1 no 0
;
/ define /boundary−conditions /zone−type aa inlet upper coal seam top
velocity−inlet
/ define /boundary−conditions / velocity−inlet aa inlet upper coal seam top
no no yes yes no 3.25e−06 no 0 no 300 yes no 1 no 1 no no 1 no 0
/ define /boundary−conditions / pressure−outlet aa outlet ventilation entry
no 0.000 no 300 no yes yes no 1 no 1 no no 0.005 no 0.2 no no no
;
; define with intensity and scale
/ define /boundary−conditions / velocity−inlet aa inlet ventilation entry
no no yes yes no 1.80360 no 0 no 300 no yes 3 0.35 yes no 0.00000 no
0.200000
/ define /boundary−conditions / velocity−inlet aa outlet recovery room 1 no
no yes yes no 0.077609 no 0 no 300 no yes 4 0.35 yes no 0.00000 no
0.200000
/ define /boundary−conditions / velocity−inlet aa outlet recovery room 2 no
no yes yes no 0.099137 no 0 no 300 no yes 4 0.35 yes no 0.00000 no
0.200000
/ define /boundary−conditions / velocity−inlet aa inlet n2 1 no no yes yes
no 0.00354 no 0 no 300 no yes 7 0.35 yes no 0.00000 no 0.000000
/ define /boundary−conditions / velocity−inlet aa inlet n2 2 no no yes yes
no 0.00505 no 0 no 300 no yes 7 0.35 yes no 0.00000 no 0.000000
/ define /boundary−conditions / velocity−inlet aa inlet tg n2 1 no no yes
yes no 0.009899 no 0 no 300 no yes 7 0.35 yes no 0.00000 no 0.000000
/ define /boundary−conditions / velocity−inlet aa inlet upper coal seam top
no no yes yes no 3.25e−6 no 0 no 300 no yes 0.1 0.35 yes no 1.00000
no 0.000000
;
; create roughness on entry walls
rough height
in m
/ define /boundary−conditions / wall wall−entries−entries front
0 no 0 no no no 0 no no no no 0.0000 no 0.5 yes yes
/ define /boundary−conditions / wall wall−entries−tg void
0 no 0 no no no 0 no no no no 0.1524 no 0.6
yes yes
/ define /boundary−conditions / wall wall−entries−tg entries
0 no 0 no no no 0 no no no no 0.1524 no 0.6 yes
yes
/ define /boundary−conditions / wall wall−headgate void and entries−hg void
0 no 0 no no no 0 no no no no 0.1524 no 0.6 yes yes
254
|
Colorado School of Mines
|
ABSTRACT
Technologies utilized for the disposal of slurry, paste, thickened, and filtered tailings include
containment within impoundments, open pit backfill, submarine placement, underground backfill, tailings
and waste rock combined disposal, and dry stack placement. Each technology has inherent
environmental, social, and economic drivers to be considered during mine planning. Geotechnical and
geochemical properties of the tailings, and site-specific design constraints will also drive the selection of a
disposal technology. Alternative technologies should be developed to address the increase in volume of
tailings generated as lower grade ore is extracted, and to minimize risk associated with current
technologies. Understanding the key drivers for selection of existing technologies is a necessary
precursor to the future development of alternative technologies.
This thesis presents a qualitative risk assessment of environmental, social, economic,
geotechnical, geochemical, and site-specific elements utilized in selection of a tailings disposal
technology. United States regulations and international guidelines for tailings disposal, and general
industry accepted goals for current disposal technologies are summarized. Copper deposit geology, ore
processing methods and available disposal technologies are described.
Results of the qualitative assessment include the identification of critical, high, moderate, and low
ranked drivers for the selection of each tailings disposal technology. Critical and high ranked elements
are recognized as key drivers for consideration during the selection of each tailings disposal technology.
Key drivers for all technologies include the political and regulatory climate, investor confidence, and the
acid generation and neutralization potential of the tailings. Key drivers for consideration during the
selection of all surface disposal technologies also include land disturbance, construction cost, and the
availability of local materials for construction. Understanding the key drivers for each disposal technology
will assist the selection of a suitable, site-specific technology, and is a necessary precursor to the future
development of alternative technologies.
iii
|
Colorado School of Mines
|
CHAPTER 1
INTRODUCTION
Singer et al. (2005) compiled a database of known ore grades for porphyry copper deposits in
locations worldwide, and reported grades between 0.2 and 0.8 percent copper, with an average grade of
0.5 percent. After copper and other mineral(s) of economic value are extracted from the ore, the
remaining material is disposed of as tailings. Ore processing methods for mineral extraction utilize water
and a variety of chemicals that are likely to be present within the disposed tailings, and must be
accounted for in the selection and implementation of a tailings disposal technology. Tailings differ in
geotechnical and geochemical properties from unprocessed waste rock, and separate disposal practices
for each material have been developed.
Tailings disposal practices have evolved with the advancement of mining and mineral processing
technology, and the development of regulations and international guidance, from downgradient disposal
practices with minimal design to engineered on-site disposal. Mining companies, local communities,
government entities, regulators, scientists, and engineers within the mining industry continuously strive to
improve the design, operation, and closure technologies and practices for tailings disposal. In addition to
minimizing economic cost, minimizing the tailings disposal facility’s environmental and social influence
has become a goal for the selection of a disposal technology. It has been reported that failure of tailings
dams occurs globally at a rate of two to five facilities per year (Davies 2002, Azam and Li 2010). The
recent failure of the Mount Polley copper and gold tailings dam in 2014 highlighted consequences
associated with large disposal facilities, and adversely impacted the general public’s view of the mining
industry’s tailings disposal practices (Schoenberger 2016).
1.1 Problem Statement
Current tailings disposal technologies and practices have inherent environmental, social, and
economic drivers to be considered for selection during mine planning. Geotechnical and geochemical
properties of the tailings, and site-specific design constraints will also drive the selection of a disposal
1
|
Colorado School of Mines
|
technology. Alternative disposal technologies should be developed to address the increase in volume of
tailings generated as lower grade ore is extracted, and to minimize risk associated with current disposal
technologies. A qualitative risk assessment for current copper tailings disposal technologies was
performed to identify the key drivers for selection of each disposal technology, which is a necessary
precursor to the future development of alternative disposal technologies.
1.2 Research Focus
This thesis presents a literature review of current disposal practices for tailings resulting from
copper extraction processes. Tailings are highly variable based on the mineralogy of the ore, and the
physical and chemical processing the ore was subjected to (Osanloo et al. 2008). The scope of research
was restricted to current disposal technologies for copper tailings, in order to gain an understanding of
copper mineral processing, and the physical and chemical characteristics of copper tailings. Information
presented in this thesis has been collected from published articles, conference proceedings, seminars,
and the author’s practical experience with tailings management and disposal facility design. A qualitative
assessment of risks associated with each disposal technology was performed to identify key drivers for
selection, and recommendations for future research are included.
Processing methods for both sulfide and oxide copper ores are described, and the resulting
processed tailings are classified as conventional slurry, paste, thickened, or filtered based on the percent
solids content. The current tailings disposal technologies available include: impoundment, open pit
backfill, underground backfill, dry stack, tailings and waste rock co-disposal and co-mingling, and
submarine placement. Alternative technologies currently in development by others are also briefly
described. Environmental, social, economic, geotechnical, geochemical, and site-specific categories were
identified for the risk assessment. Elements within each category were selected and ranked based on the
potential influence (impact) of the element. Critical and high ranked elements were identified as key
drivers for consideration during selection of a tailings disposal technology.
2
|
Colorado School of Mines
|
1.3 Thesis Outline
This thesis is divided into seven chapters and presents a literature review and qualitative risk
assessment of the mining industry’s copper tailing disposal practices. A brief description of each chapter
is included below.
Chapter 1 – Introduction: This chapter presents a problem statement and research focus.
Chapter 2 – Background: This chapter chronicles the regulations applicable to, and guidance
documents available for tailings disposal; summarizes potential environmental,
social, and economic influences resulting from tailings disposal; and outlines goals
for current tailings disposal technologies.
Chapter 3 – Generation of Tailings from Copper Extraction: This chapter describes the
variety of copper deposits, copper ore processing methods, and characteristics of
the resulting processed tailings.
Chapter 4 – Tailings Characteristics: This chapter describes key geotechnical and
geochemical properties of tailings, laboratory test methods, and design
considerations for the selection of a disposal technology.
Chapter 5 – Tailings Disposal Technologies: This chapter summarizes current tailings disposal
technologies and alternative technologies currently in development by others.
Chapter 6 – Qualitative Risk Assessment of Tailings Disposal Technologies: This chapter
presents a qualitative assessment of risk for current copper tailings disposal
technologies.
Chapter 7 – Conclusions: This chapter concludes the thesis with a summary of work performed
and recommendations for future research.
3
|
Colorado School of Mines
|
CHAPTER 2
BACKGROUND
Mining for copper has been ongoing for the past 10,000 years (ICSG 2014). Copper is present in
several alloys and is found naturally in plants and animals. It is malleable, ductile, corrosion resistant, and
a conductor of electricity (CuDA 2017, Perkins 2001). Copper extraction, processing, and tailings disposal
technologies vary based on site-specific requirements. The advancement of tailings disposal practices
within the United States has been largely driven by an increase in the quantity of tailings material
generated as lower grade ores are extracted, and the establishment of federal regulations developed by
the United States Environmental Protection Agency (USEPA) to protect “human health and the
environment.” Global guidelines have also been developed by collaborative international organizations for
tailings disposal practices and are summarized below.
2.1 Regulations and Guidance for Tailings Disposal
Tailings are mine waste and currently generate no economic value, therefore disposal practices
and technologies have historically been selected based on lowest cost. Early tailings disposal practices
were unregulated and lowest cost options often included the dumping of tailings and other waste
materials wherever convenient, downstream of the mining operation and within natural drainages.
Advances in ore processing and mineral extraction methods over time have made it economical to mine
ores of increasingly lower grade, resulting in an increase in the volume of material excavated to extract
the low-grade ore (Franks et al. 2011). As the global population increases, so does the demand for
extracted resources (Kesler 2000). Mining production rates increase to meet demand, also resulting in an
increase in the volume of material excavated and tailings generated.
Surficial tailings disposal facility footprints, placement heights, and impoundment dams have
increased to accommodate greater tailings volumes. The construction of larger facilities has led to an
increase in the number of reported failures (Azam and Li 2010), resulting in environmental degradation
and loss of life. To minimize environmental consequences from several industries located within the
4
|
Colorado School of Mines
|
United States, including mining, federal regulation began during the 1960s. Tailings disposal technologies
advanced in the 1970s, largely in response to the federal regulations established and collaboration
between tailings practitioners (Caldwell 2011). Current regulations were developed by the USEPA, and
require a mining company to obtain several permits for activities that occur during mine exploration
through mine closure. A mining company must complete permit applications to ensure they understand
the regulatory requirements that apply to the activities listed within the application. A permit is awarded
once the mining company has demonstrated an ability to maintain compliance with regulatory
requirements. Federal regulations a mine operator must comply with include:
Clean Air Act (CAA) of 1970 – this regulation was enacted to address air pollutants with the
potential to be hazardous to human health and the environment, and specifically applies to
equipment exhaust, processing facility emissions, and dust (USEPA 2017c). Mine dust has
been identified as a primary health and environment hazard of tailing disposal facilities and is
regulated under the CAA.
Clean Water Act (CWA) of 1972 – this regulation was enacted to address discharge
pollutants with the potential to adversely impact natural water systems. For application to
mining, the CWA specifically applies to surface water runoff and groundwater infiltration
(USEPA 2017d) within the mine property, including the tailings disposal facility footprint.
Resource Conservation and Recovery Act (RCRA) of 1976 – this regulation was enacted
to provide procedures for waste generation and disposal practices. Most active mines are
exempt from this federal rule. However, active mines may be subject to similar regulations
enacted by local, state governments. This rule often applies to the remediation of abandoned
mine facilities, including tailings disposal facilities (USEPA 2017b).
Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA)
of 1980 – this regulation was enacted to establish a program to fund the remediation of
hazardous waste contaminated facilities (USEPA 2017e). For application to mining, CERCLA
specifically applies to active or inactive mines with contaminant levels exceeding the defined
thresholds for surface water and groundwater, air, and soil. Tailings disposal facilities may be
subject to CERCLA regulation and remediation programs if threshold levels are exceeded.
5
|
Colorado School of Mines
|
Lead and Copper Rule (LCR) of 1991 – this regulation was enacted to address drinking
water pollutants with the potential to be hazardous to human health and the environment. The
LCR specifically establishes acceptable concentration limits for copper in drinking water
(USEPA 2017a). This rule applies to surface water runoff and groundwater infiltration from a
mine facility, including the tailings disposal facility that may adversely impact a drinking water
source.
International guidance and agreement is needed to outline standards of practice for tailings
disposal (Franks et al. 2011). While there are no specific international regulations for the design,
construction, operation, and closure of a tailings disposal facility, several international organizations have
prepared guidance documents including:
Mine Environment Neutral Drainage (MEND) Program, 1991, 1995, 1998 – the MEND
program prepared documents related to the prediction of acid rock drainage and tailings
disposal practices, specifically the “Acid Rock Drainage Prediction Manual” (MEND 1991),
“Review of In-Pit Disposal Practices for the Prevention of Acid Drainage – Case Studies”
(MEND 1995), and “Design Guide for the Subaqueous Disposal of Reactive Tailings in
Constructed Impoundments (MEND 1998).
United States Environmental Protection Agency (USEPA), 1994 – the USEPA prepared
the document “Technical Report – Design and Evaluation of Tailings Dams” to summarize the
design, construction, water management, and potential failure modes for tailings dams
(USEPA 1994c).
United Nations Environmental Program (UNEP) and the International Council on Metals
and the Environment (ICME), 1998 – UNEP and ICME prepared the document “Case
Studies on Tailings Management” to summarize current environmental management, design,
and operation practices for tailings management within the mining industry (ICME 1998).
Department of Mines and Petroleum (DMP), 1999 and 2000 – the DMP, formerly the
Department of Minerals and Energy (DME), of Western Australia prepared the documents
“The Guidelines on the Safe Design and Operating Standards for Tailings Storage” and “The
6
|
Colorado School of Mines
|
International Commission on Large Dams (ICOLD), ongoing – ICOLD is composed of
members from approximately 100 countries, and has prepared numerous publications to
present guidelines and industry standards for the safe and sustainable construction and
operation of dams, based on several performed studies (ICOLD 2017).
International Network for Acid Prevention (INAP), ongoing – INAP is sponsoring the
development of the document “Global Acid Rock Drainage (GARD) Guide” to present an
approach to prevent and mitigate acid rock drainage (INAP 2009). Updates to this document
are ongoing.
A general trend away from conventional tailings slurry disposal practices and toward dewatered
tailings disposal practices is observed, and is likely credited to advanced regulation and the development
of international guidance for tailings disposal. Increased awareness of the environmental consequences
associated with large-scale tailings disposal facility failure has led to the development of current disposal
practices to minimize environmental disturbance. For example, dewatered tailings are generally more
stable, may be engineered to meet specific design requirements, and allow the potential for dewatered
fluids to be recycled back into the mining process (Willis 2006, Davies et al. 2010, Edraki et al. 2014,
Barrera and Caldwell 2015). Highly dewatered tailings may be placed within a dry stack tailings disposal
facility instead of a conventional tailings slurry impoundment (Davies et al. 2010). Processing methods,
tailings characteristics, and disposal technologies are described in Chapters 3, 4 and 5, respectively.
2.2 Economic, Social and Economic Elements of Tailings Disposal
Ore processing, mineral extraction, tailings generation, and tailings disposal practices are multi-
disciplinary and require expertise from the geological, metallurgical, geotechnical, and geochemical
disciplines. Knowledge of these disciplines is required to understand the environmental, social, and
economic elements of tailings disposal (Barrera and Caldwell 2015). Collaboration between practitioners
within these disciplines is often needed throughout the life of a mine project to continually assess
elements, though in practice collaboration regularly occurs in the initial mine development stages and
infrequently occurs once the mine is operating (Edraki et al. 2014). Options to reduce economic, social,
and economic effects resulting from tailings disposal are best identified and implemented during initial
8
|
Colorado School of Mines
|
development stages and are more difficult to implement during mine operation (Edraki et al. 2014). It is
important to effectively communicate all potential influences with stakeholders during the initial mine
development stages and throughout the mine life (Warhurst and Mitchess 2000).
Environmental, social, and economic effects of tailings disposal occur primarily when a facility
fails and there is a release of tailings and process fluids into the environment (Schoenberger 2016).
Failure of a tailings disposal facility can have significant impact including loss of life, costs associated with
remediation, loss of production, loss of reputation, and loss of investor confidence. Environmental
elements of a tailings disposal facility may include land and water disturbance within the disposal facility
footprint, downgradient land and water disturbance, and area pollution resulting from dust, noise, and
carbon dioxide (CO ) emissions. Social elements may include land remediation costs imposed on the
2
local communities, health and safety of community members, political and regulatory climate, aesthetics,
and post-closure land use. Financial elements may include estimation of tailings disposal costs
throughout the mine life, investor confidence and company share value, variances between anticipated
and actual cost and ground conditions, and future resource identification.
Environmental
- Land Disturbance
- Water Disturbance
- Area Pollution
Social
Economic
- Imposed Costs
- Cost Estimate and Variance
- Health and Safety
- Investor Confidence
- Political and Regulatory Climate
- Ground Condition Variance
-Aesthetics and Land Use
- Resource Extents
Figure 2.1 Social, environmental, and economic elements of tailings disposal.
9
|
Colorado School of Mines
|
2.2.1 Environmental Elements
Land and water disturbance will occur within the footprint of a tailings disposal facility
(Schoenberger 2016). For surface disposal, the extent of disturbance area increases with increasing size
of the disposal facility, and may result in alteration of existing surface water flow paths, destruction of
vegetation, and disturbance of wildlife habitats within the facility footprint. Surface water is often diverted
around the tailings disposal facility to avoid alteration of the existing surface water chemistry, and to
prevent tailings erosion and mobilization. Infiltration of stormwater into the tailings disposal facility can
occur, dissolving metals or compounds present within the tailings (Moreno and Neretnieks 2006). This
solution may seep through the underlying foundation and enter the groundwater system, altering the
chemical composition of the groundwater (Franks et al. 2011). When surface water or groundwater comes
into contact with tailings, acid mine drainage (AMD) is likely to form and the precipitation of secondary
minerals can occur, affecting both the surface water and groundwater environments (Grangeia et al.
2011). For underground disposal, the extent of disturbance area is limited to the underground openings,
however surface disposal may be required if the volume of tailings exceeds the available underground
space. Infiltration through tailings disposed underground can still occur and there is a potential for
seepage from the tailings into the groundwater system.
Land and water disturbance may occur downgradient of a tailings disposal facility footprint,
primarily through failure of the facility, surface water runoff, and migration of seepage within the
groundwater system. Tailings and associated processing fluids may be transported for large distances
after failure of a tailings impoundment (Franks et al. 2011). The geochemical makeup of existing water
and sediments immediately downgradient of a disposal facility failure can be significantly altered (Kossoff
et al. 2014). Where failure of a tailings disposal facility occurs near water, dilution may help reduce the
geochemical alteration effects further downstream and away from the failure location. Seepage migration
into the groundwater system can also carry dissolved metals and compounds for a large distance
downgradient and outside of the tailings disposal facility footprint. Failure of historic and abandoned
tailings impoundments can be difficult to remediate where characterization of the tailings and the
deposition method are often unknown (Grangeia et al. 2011).
10
|
Colorado School of Mines
|
Area pollution resulting from dust, noise, and CO emissions can occur at a tailings disposal
2
facility. Preventing dust migration is required to restrict land disturbance to the disposal facility footprint,
and to minimize adverse health effects due to inhalation, water, and soil contamination (Dudka and
Adriano 1995). Maintaining water levels, saturation of the tailings, and spraying water over the perimeter
access roads are common ways to manage dust. Preventing noise pollution and CO emissions resulting
2
from the construction and operation of the disposal facility is required to minimize disturbance to
surrounding wildlife habitats and local communities.
2.2.2 Social Elements
Poor tailings disposal practices can result in unanticipated environmental degradation and
remediation costs may exceed the funds available to the mine owner or operator. Unfortunately, if the
mine is unable to provide funds, these costs to perform remediation may be imposed on the local
community (Franks et al. 2011). A decrease in copper value can permanently suspend mine operations,
providing another avenue for tailings remediation costs to be imposed on the local community. Loss of
income for residents of the local community, and consequent indirect loss of income for local businesses
that provide services to the mine operation, employees, and their family members may be caused by
temporary or permanent suspensions in the mine operation.
Maintaining the health and safety of employees and members of the local community is
important. Tailings disposal facility construction and operation requires the use of heavy mine equipment,
conveyors, and pipelines to transport and place material. Safe operating procedures should be
implemented and training offered to employees to ensure safe practices are employed. The health and
safety of local community members can also be adversely impacted by tailings disposal practices.
Contamination of the local water source and soil can occur from tailings runoff, seepage, and dust (Dudka
and Adriano 1995).
The political and regulatory climate can affect tailings disposal practices, and subsequently affect
the local community. The local government can influence how a mine is constructs, operates, manages,
and closes a tailings disposal facility. If the local government values the health and safety of local
residents and the environment, the mine may select tailings disposal practices that favor preservation of
11
|
Colorado School of Mines
|
those qualities. Alternatively, if regulations are not in place to ensure that safe practices are implemented,
there are no direct consequences for the mine if poor tailings disposal practices are employed. Policies
established by the local government may provide a framework for assessing environmental liabilities and
safe work practices (Warhurst and Mitchess 2000).
Aesthetics of a tailings disposal facility are often important to the surrounding community. Surface
disposal visibly alters the existing, natural topography, and post-closure land use should be considered.
In the early stages of mine planning, it is important to consider what the land will look like once mining
operations have ceased, and what types of activities the post-mine land will be utilized for. Establishing a
post-mine land-use consistent with the interests of all stakeholders can maintain a positive reputation of
the mining company (Warhurst and Mitchess 2000).
2.2.3 Economic Elements
Osanloo et al. (2008) states that the future success of a mining operation depends on the
management of tailings disposal facilities, and that design considerations for these facilities should be
incorporated within the overall mine plan. Processing efficiencies have increased over time, allowing the
mining of decreasing ore grades (Kossoff et al. 2014). As mining for lower grade ores becomes more
common, the material excavated and processed becomes greater, resulting in increased capital and
operating costs (Osanloo et al. 2008). Accurate, reliable estimation of the tailings disposal costs is critical
to the financial success of a mining project, and should consider the overall design, construction,
operation, maintenance, and closure costs for the disposal facility throughout the mine life (Hansen et al.
2008).
Increased environmental treatment, remediation, and monitoring costs due to poor tailings
disposal practices can result in a loss of investor confidence and ultimately a devaluation of a company’s
share value (Franks et al. 2011). The mine operation may experience periods of temporary or permanent
shutdown, additional fines, and litigation costs associated with treatment and remediation (Franks at al.
2011). The potential for future mining opportunities may also decrease. Variances between estimated and
actual costs or ground conditions can also result in periods of temporary or permanent shutdown.
Fluctuating metal prices, ore grades, reserve estimates, processing costs, and water costs have direct
12
|
Colorado School of Mines
|
impact on the cashflow available to a mine. Increased processing and water costs will influence the
selection of a tailings disposal technology.
The location available for tailings disposal can also affect the selection of a disposal technology.
Knowledge of the total resource model, property extents, and potential for future expansion is required
before constructing a tailings disposal facility. Future resource identification or mine expansion may be
limited by the tailings disposal facility, resulting in potential facility removal and relocation, or non-
development of the resource.
2.3 Goals for Current Tailings Disposal Technologies
Industry knowledge of mineral processing techniques and characterization of the tailings has
advanced over time and has led to the development of modern tailings disposal technologies. While a
universal tailings disposal technology is unlikely to be developed due to site-specific variability, common
goals can be applied to each tailings disposal technology. Goals for current tailings disposal technologies
include minimal disturbance to the environment, stability of the disposal facility, compliance with
regulatory requirements and guidelines, and identification of process improvement or the development of
alternative technologies (Caldwell 2011, Franks et al. 2011).
Disturbance to the local environment will occur within the tailings disposal facility footprint and
may occur beyond the facility footprint. The selected tailings disposal technology footprint should favor
minimal disturbance. Environmental influences may include displacement of local wildlife, alteration of the
existing watershed and drainage paths, and contamination of the surface water and groundwater. Site
selection studies should be performed to identify a location for the disposal facility that minimizes the
environmental impacts (Barrera and Caldwell 2015). Post-closure land uses to restore the pre-mine
environment should also be considered. Containment of the tailings and process fluid is required to
minimize disturbance to the surrounding environment (Franks et al. 2011).
Maintaining the short-term and long-term geotechnical stability of a tailings facility within the
design footprint is critical to minimizing disturbance to the local environment. The tailings disposal
technology selected should favor overall geotechnical stability. The disposal facility should be designed to
withstand regional storm and seismic events (Ritchie et al. 2009, Franks et al. 2011). The presence of
13
|
Colorado School of Mines
|
fluids within the tailings can lead to a buildup of pore water pressure, decreasing the stability of the
disposal facility. Consolidation of the tailings due to rapid fluid drawdown will increase the stability of the
disposal facility. Minimizing the erosion potential of the disposal facility will help to maintain stability, and
erosion resistant materials should be selected for construction. Surface water diversions may be
constructed to direct runoff away from the facility. Dust mitigation practices should also be implemented to
prevent migration of the tailings beyond the facility footprint (Edraki et al. 2014).
Maintaining the short-term and long-term geochemical stability of a tailings facility is also critical
to minimizing disturbance to the local environment. The tailings disposal technology selected should favor
overall geochemical stability. The disposal facility should be designed to minimize chemical alteration and
seepage into the environment (Franks et al. 2011). The mineralogy of the tailings must be characterized
to determine the acid generation potential of the material. Oxidation and exposure of the tailings to water
may produce solutions that can enter and degrade the local surface water and groundwater. Tailings dust
migration and seepage will expand the footprint of geochemical influence (Hesketh 2010, Edraki et al.
2014). Implementing tailings disposal practices that render the placed material inert is ideal, though
difficult to achieve (Franks et al. 2011).
Compliance with regulatory requirements is critical for obtaining and maintaining a facility permit
and social license to operate. The tailings disposal technology selected should favor compliance with
regulatory requirements and industry guidance. Non-compliance with regulations or industry guidance
may compromise the environmental integrity of the disposal facility, and can contribute to a negative
public perception.
A qualitative risk assessment was performed to identify key environmental, social, and economic
drivers for selection of a tailings disposal technology. Details and results of the assessment are described
in Chapter 6.
14
|
Colorado School of Mines
|
CHAPTER 3
GENERATION OF TAILINGS FROM COPPER EXTRACTION
The physical and chemical characteristics of tailing material are dependent on the ore deposit
geology and the mineral processing methods, and will ultimately drive selection of a tailings disposal
technology. The mineral processing method is selected based on the ore petrology and mineralogy,
specifically the additional minerals present within the ore that will need to be separated for copper
extraction. Mechanical separation and process solutions alter the ore, resulting in tailings with a variety of
geochemical and geotechnical characteristics.
3.1 Copper Deposits
Copper deposits occur in a number of geologic settings that result in a wide variety of
mineralization. Porphyry, skarn sedimentary, volcanic massive sulfide (VMS), and iron oxide copper gold
(IOCG) are the primary copper deposits that produce two main copper ore styles: sulfide copper ore and
oxide copper ore (Biswas 1994, Willis 2006). Chalcopyrite (CuFeS ) is a copper iron sulfide mineral and is
2
considered one of the most widely distributed copper minerals (NAS 1979). Chalcopyrite is susceptible to
oxidation and will yield several compounds including chalcocite (Cu S), another copper sulfide mineral
2
that is also characteristic of hydrothermal deposits (Perkins 2001). Chalcopyrite and chalcocite are sulfide
minerals present in porphyry, sedimentary, volcanic massive sulfide and iron oxide-copper-gold deposits.
Porphyry deposits tend to have large extents, exhibit low copper grades, are commonly mined as
open-pit, and supply most of the world’s mined copper when compared to other deposits (Osanloo et al.
2008). Porphyry deposits are frequently associated with igneous intrusive rocks and subduction areas,
and are formed when hydrothermal fluids travel through rock fractures and veins (Berger et al. 2008). The
fluid interacts with the surrounding rock, causing mineralization based on the composition of both the rock
and the fluid (Dold and Fontbote 2001). The composition of hydrothermal fluid varies with pressure and
temperature, and can cause variable distributions and concentrations of commodities throughout a
deposit including copper, gold, molybdenum and silver, which may be extracted if there is economic
15
|
Colorado School of Mines
|
value. Skarn deposits are often associated with mineralized zones within porphyry deposits, though
chalcocite minerals are more dominant than chalcopyrite minerals (Meinert 1992). Skarn copper deposits
are formed when hydrothermal fluid is transported along the contact between two lithologies, typically
granite and limestone, dissolving silicate and carbonate minerals.
Sedimentary deposits supply the second greatest quantity of mined copper and are formed after
an initial sediment layer is deposited and prior to lithification (compaction) by an overlying layer (Hayes et
al. 2015). Compaction of the underlying sediment layer forces fluid and air migration upward and out as
the layer consolidates. The fluid interacts with the sediment layer as it migrates upward and out, causing
mineralization based on the composition of the fluid and sediment. Copper minerals such as chalcocite
leach into and are transported upward with the migrating fluid, and are deposited above the original host
sediment layer (Cox et al. 2007).
VMS deposits are formed through the hydrothermal fluid transport of metals and the accumulation
of sulfide mineral precipitates. VMS deposits typically occur in marine-volcanic environments where water
is drawn down beneath the ocean floor, heated, and circulated back up to the ocean floor (Shanks and
Thurston 2012). The circulated and heated water reacts with the surrounding rock and becomes a metal-
rich hydrothermal solution. The rapid cooling of the hydrothermal fluid upon contact with ocean water
results in mineral precipitation.
IOCG deposits are formed through volcanic intrusion, similarly to copper porphyry deposits, but
differ in the presence of iron oxide minerals (Groves et al. 2010). The iron oxide-rich hydrothermal fluids
travel through rock fractures and veins and interact with the surrounding host rock, causing copper
mineralization, typically chalcopyrite and pyrite, based on the rock and fluid composition.
3.2 Copper Ore Processing
Copper deposits are typically developed utilizing open-pit mining methods and produce two main
copper ore types: sulfide copper ore and oxide copper ore (Biswas 1994, Willis 2006). In general, copper
is more easily separated from sulfide deposits, and these typically have higher grades than oxide
deposits. Open-pit copper mining generates large quantities of waste material in the form of waste rock,
spent material from leaching of oxide ores, and tailings from flotation processing of sulfide ores. The
16
|
Colorado School of Mines
|
mineralogy of each type of waste material is highly variable and largely dependent on the ore composition
and mineral processing methods.
3.2.1 Sulfide Ores
Copper sulfide ore deposits are typically formed by hydrothermal fluid flow through rock, and the
most common sulfide ore mineral is chalcopyrite (CuFeS ) (NAS 1979). A typical copper sulfide ore
2
processing system flow diagram is included as Figure 3.1. Waste rock from the sulfide ore deposit is
hauled directly to the waste rock dump, while the ore is sent to the plant to begin processing for copper
extraction. Gravity separation and vat leaching are utilized for high grade sulfide ore deposits, but are less
commonly employed processing methods for sulfide ores (Kordosky 2002), while flotation is the dominant
processing technique utilized to extract copper (Osanloo et al. 2008).
Gravity separation is a processing method that utilizes the specific gravity of individual minerals
for separation. The ore is sent to the plant where it is crushed and ground to meet grain size requirements
for separation within jigs or spiral concentrators. Further processing such as smelting is often required to
increase copper recovery, and the unwanted waste material is disposed as tailings. Vat leaching is a
processing method where the ore is sent to the plant and crushed into a coarse material. The coarse
material it is then placed into a vat with a leach solution and agitated to dissolve the copper from the ore
while keeping the solids suspended. The pregnant leach solution (PLS) containing copper is drained from
the vat and further processed utilizing solvent extraction and electrowinning (SX/EW). During solvent
extraction, the PLS is added to a non-mining liquid and a solvent is added that binds to and removes the
copper (CuDA 2017). The remaining solution of dilute sulfuric acid and residual minerals known as
raffinate may be recycled back into the leach process. However, the presence of metal ions within the
raffinate solution may prevent its reuse for processing. Acid is then added to the solvent and copper
solution. An electrical current is passed through the solution to encourage copper ions to plate directly
onto a cathode. Once the leach solution is drained, the solids are allowed to settle and are disposed as
tailings. The characteristics of tailings resulting from both gravity separation and vat leaching are
influenced by the sulfide ore characteristics and crushing.
17
|
Colorado School of Mines
|
Figure 3.1 Typical copper sulfide ore processing system flow diagram. Adapted from the Copper
Development Association (2017).
Flotation is the dominant processing technique used to extract copper from sulfide ores. The ore
first crushed and ground to the required particle size to liberate the sulfide minerals of interest. The
ground ore is then subjected to flotation, a process wherein the ore is mixed with water (forming a pulp)
and various reagents (chemicals) are added. Air is introduced into the tank, creating bubbles. By surface
chemistry, the bubbles preferentially attach to the liberated hydrophobic copper sulfide minerals (Kossoff
et al. 2014) causing them to float to the surface. The froth is collected as copper concentrate, while the
non-float material (tailings) are extracted from the tank for disposal. The concentrate is dewatered and
typically sent to a smelter to turn it into refined copper for market.
The smelter employs a complex pyrometallurgical process wherein the concentrate is first
oxidized to form a copper matte. Smelting further separates impurities from the matte as slag, a waste
material that can be removed prior to oxidation and formation of a blister copper. Fluxes are added to and
react with the matte during heating, resulting in the slag floating to the surface as the calcine melts, and
the formation of a product with copper purities up to 99 percent (CuDA 2017). The chemical equations for
18
|
Colorado School of Mines
|
the addition of silica (SiO ) as a flux for the production and removal of slag, and the formation of blister
2
copper are included as Equations 3.1 and 3.2, respectively (CuDA 2017).
FeO + SiO 2 → FeOSiO 2 (3.1)
Cu 2S + O 2 → 2Cu + SO 2 (3.2)
The characteristics of tailings resulting from flotation are influenced by the sulfide ore
characteristics and processing. Additional chemicals may be added to during the flotation process that
are often discarded with the tailings including lime, which can be utilized to promote settling, neutralize
acid, and assist with sulfur removal. Alternatively, high clay content within the ore can inhibit flotation and
acid chemicals may be added (Dold and Fontbote 2001). The additional flocculants, water, crushing,
grinding, and variation of the sulfide ore result in a wide range of tailings characteristics.
3.2.2 Oxide Ores
Copper oxide deposits are formed by oxidation and generally have lower grades than sulfide
ores. Copper extraction from an oxide ore is largely dependent on the mineralogy of the deposit. A copper
oxide process flow chart is included as Figure 3.2. Waste rock from the oxide ore deposit is hauled
directly to the waste rock dump, while the ore is sent to the plant for crushing or directly to a stockpile for
leaching. Crushing the ore prior to placement is an additional process cost and is employed to meet
design specifications for solution application and recovery rates. Oxide ores are typically leached using
dilute sulfuric acid that may be generated from the processing of sulfide ore as described in section 3.2.1.
Tailings are not generated through the processing of oxide ores, though the leach stockpile is considered
a type of mine waste once solution recovery is complete.
In general, the oxide ore is placed in a stockpile configuration, overlying a prepared foundation,
and acid is sprayed or dripped across the surface. The acid seeps through the ore stockpile, dissolving
copper and residual minerals into a pregnant leach solution (PLS) that can be collected for further solvent
extraction electrowinning (SX/EW) processing. During solvent extraction, the PLS is added to a non-
mixing liquid and a solvent is added that binds to and removes the copper (CuDA 2017). Acid is then
added to the solvent and copper solution, and placed into a tank with alternating lead anode and cathode
19
|
Colorado School of Mines
|
sheets of either reusable stainless steel or single-use metal-coated carbon fiber. A direct electrical current
is forced through the solution to encourage copper ions to plate directly onto a cathode. The remaining
solution of dilute sulfuric acid and residual minerals known as raffinate may be recycled back into the
leach process. However, the presence of metal ions within the raffinate solution may prevent its reuse for
processing.
Figure 3.2 Typical copper oxide ore processing system flow diagram. Adapted from the Copper
Development Association (2017).
3.3 Processed Tailings
Copper tailings can be divided into three main categories based on the amount of dewatering
achieved, and the resulting percent solids composition of the tailings: conventional tailings slurry,
dewatered paste and thickened tailings, and filtered tailings (Davies 2011). The water and processing
fluid extracted from the tailings during the dewatering process may be returned to the process cycle,
decreasing the water retained within and the overall footprint of the tailings disposal facility (Willis 2006).
However, the presence of metal ions within the fluid may prevent its reuse for processing. Flocculants can
be added to increase the effectiveness of dewatering, and to assist with pipe transport and deposition.
20
|
Colorado School of Mines
|
Dewatered tailings may also be utilized for underground support. Most tailings thickeners and cyclones
utilize gravity to separate particles. Laboratory testing must be performed to characterize the tailings prior
to the dewatering plant design (Franks et al. 2011).
As the tailings are dewatered, the percent solids within the slurry and the density of the tailings
increase, allowing for alternative conveyance system options. Conventional tailings slurry is transported
by a series of pumps and pipes to the disposal facility, while dewatering to produce filtered tailings allows
for transport by truck or conveyor. Pumps and pipes are subject to tailings settling and clogging if the flow
velocity required for transport is not maintained, and the installation of conveyor systems may be limited
by the infrastructure located between the process and disposal facilities. The tailings transport method
and disposal technology are selected based on the characteristics of the processed tailings. In general,
processing costs increase as the degree of dewatering increases to meet tailings transport and disposal
design requirements. The percent solids, dewatering and transport methods for each category of
processed tailings is summarized in Table 3.1. Typical tailings thickeners are depicted in Figure 3.3.
Figure 3.3 Typical tailings thickeners. Adapted from Paterson and Cooke (2017).
3.3.1 Conventional Tailings Slurry
Following flotation, the tailings and processing fluid are removed and placed in a thickener, prior
to pipe conveyance to the tailings disposal facility. The pipe conveyance system is selected based on
transport performance for both the solids and liquid contents of the slurry. The percent solids for a
conventional tailings slurry can range from 30 to 50 percent by weight (Franks et al. 2011). Variability of
the solids and process fluid content can be attributed to the ore characteristics, processing methods, use
21
|
Colorado School of Mines
|
of a thickener, and additional flocculants. A conventional tailing slurry is typically conveyed by a pipe
system to an impoundment, designed to account for both tailings and additional process fluid storage.
The stored process fluids may be extracted from the impoundment and recycled back into the mine
process system to meet varying water demands.
Table 3.1. Tailings dewatering characteristics and transport methods (Barrera and Caldwell 2015, Edraki
et al. 2014, Franks et al. 2011).
Processed Dewatering Percent Solids Dewatering Transport
Tailing Category Class (%) Method Methods
Conventional Conventional High-Rate
30 – 50 Pump and Pipe
Slurry Slurry Thickener
Thickened High-Density
50 – 70 Pump and Pipe
(Non-Segregating) Thickener
Thickened and
Thickened
Paste 65 Cyclone Pump and Pipe
(Sand Underflow)
Deep-Bed
Paste 70 – 80 Pump and Pipe
Thickener
Vacuum >75 Vacuum Filter Conveyor or Truck
Filtered Pressure 85 Pressure Filter Conveyor or Truck
Centrifuge Cake >85 Centrifuge Conveyor or Truck
A high-rate thickener (Figure 3.3) is utilized to produce the conventional tailings slurry. The
thickener tank is cylindrical with a bottom sloped toward the center of the tank. The diameter of the
cylinder is greater than the height, to provide a large enough bottom surface area to encourage settling.
The tailings and processing fluid enter the thickener through a feed pipe. Mechanical scraper blades
(rakes) rotate around the tank at a rate that moves the recently settled solid tailings closer toward an
outlet at the bottom of the tank. The solid material and a portion of the fluid collected at the center of the
tank is removed through an outlet pipe underneath as underflow (Paterson 2004).
3.3.2 Thickened and Paste Tailings
The intent of dewatering is to decrease the amount of water and process fluid transported to and
deposited within the tailings disposal facility. Decreasing the amount of water and process fluids may
have beneficial environmental influences by minimizing the amount of fluid available for seepage
(Schoenberger 2016), and increasing the density of the tailings within a disposal facility. The
differentiating characteristic between paste and thickened tailings is that paste tailings is engineered with
22
|
Colorado School of Mines
|
additives to meet specific design criteria and utilized as structural fill (Tariq and Yanful 2013, Edraki et al.
2014).
The percent solids for thickened tailings can range from 50 to 70 percent by weight (Franks et al.
2011). Thickened tailings are typically conveyed by a pipe system to an impoundment, designed to
account for both tailings and additional process fluid storage, similar to a conventional tailings slurry. The
thickened, higher density tailings typically require a smaller footprint for retention than the conventional
tailings footprint due to a decreased volume of process fluid.
Thickened tailings utilize a high-density thickener or cyclone to separate additional process fluids
and water from the tailings (Figure 3.3). High-density thickeners have greater cylinder heights and bottom
slopes, and produce a non-segregating, higher density tailing underflow than the high-rate thickeners
utilized to produce a conventional tailings slurry. The tailings are allowed to consolidate under their own
weight for a period of time prior to removal as underflow (Paterson and Cooke 2017). Cyclones are
utilized to separate the sand-sized material from the fine silt- and clay-sized material within the tailings
underflow (Barrera and Caldwell 2011). Cyclone thickeners use rotation to force the smaller particles
inside and larger particles to the outside wall. The sand material can then be utilized to construct the
tailings containment structure (Kujawa 2011).
The percent solids for paste tailings can range from 70 to 80 percent by weight (Franks et al.
2011). Paste tailings can be engineered to meet specific design requirements and are often utilized as
structural backfill in underground mine workings or voids to minimize ground subsidence and strengthen
working foundations (Belem and Benzaazoua 2004). Cement is a common structural additive to the paste
tailings (Schoenberger 2016). The placement of paste tailings underground may reduce the
environmental influence by minimizing the surface storage footprint, and the addition of cement may
reduce the particle mobility of the tailings (Tariq and Yanful 2013, Edraki et al. 2014). Chemical
interactions may still occur between the paste backfill and groundwater. Cemented paste backfill may
also be disposed of with mine waste rock.
Paste tailings utilize a deep-bed thickener to separate additional process fluids and water from
the tailings (Figure 3.3). Deep-bed thickeners have greater cylinder heights and bottom slopes, and
23
|
Colorado School of Mines
|
produce a higher density tailing underflow than the high-density thickeners utilized to process thickened
tailings. The tailings are allowed to consolidate under their own weight for a period of time prior to
removal as underflow (Paterson and Cooke 2017).
3.3.3 Filtered Tailings
Removing fluid from the tailings minimizes the potential for chemical reactions within the tailings
disposal facility. Filtered tailings remove additional processing fluid and water than what can be achieved
for thickened and paste tailings, though removal of all fluid from the tailings is not achievable in practice
(Franks et al. 2011, Schoenberger 2016). A tailings slurry is forced through a filter system of coarse
particles, transitioning to fine particles, and finally through a filter cloth to remove fluids. Without the
addition of heat to dry the tailings, some fluid content will remain after filtration. The percent solids for a
typical filtered tailing can range between 75 and greater than 85 percent by weight (Franks et al. 2011).
The resulting filtered tailings can be transported by truck or conveyor to the tailings disposal facility.
Vacuum and pressure filter systems are typically utilized to produce filtered tailings, though
vacuum filters are not ideal for use at high altitude or with tailings that have high concentrations of fine
particles. A centrifuge is costly, but is ideal for dewatering fine particles and can achieve the highest level
of dewatering with greater than 85 percent solids by weight that results in production of a cake (Paterson
and Cooke 2017). Filtered tailings can be stacked on the surface, do not require the construction of large
retention embankments, and are ideal for arid environments where water conservation is critical to the
mining process and in areas of high seismic activity (Schoenberger 2016).
24
|
Colorado School of Mines
|
CHAPTER 4
TAILINGS CHARACTERISTICS
The geotechnical and geochemical characteristics of copper tailings can vary based on the ore
characteristics and processing methods utilized to separate the copper (Dold and Fontbote 2001, Qui and
Sego 2001, Shamsai et al. 2007). For example, copper tailings may contain sand, silt, or clay sized
particles, and additional chemical fluids and water from processing. Tailings transport methods and
disposal technologies are designed to specifically accommodate the unique tailings characteristics, in
addition to other site-specific criteria. Recognizing the variability of copper tailings and the variability that
can be introduced by processing, this chapter aims to present geotechnical and geochemical
characteristics of tailings that, in general, can influence selection of a disposal technology.
4.1 Geotechnical Properties
Standardized testing procedures have been developed by the American Society for Testing and
Materials (ASTM) to determine geotechnical properties of materials. The following geotechnical properties
of the tailings must be defined prior to selection of a transport method and disposal technology (Vick
1990, Shamsai et al. 2007, Kossoff et al. 2014):
Grain Size Distribution
Atterberg Limits
Specific Gravity, Void Ratio, and Density;
Permeability
Shear Strength
4.1.1 Grain Size Distribution
The grain size distribution is utilized to estimate the amount of clay, silt, sand or less common
gravel sized particles (Das 2005) present within the tailings and, with the Atterberg limits, can provide an
indication of how the tailings may behave after placement within a disposal facility. The grain size
25
|
Colorado School of Mines
|
distribution can also be utilized to determine what additives to mix with the tailings to meet engineered
design specifications for disposal. Gravel sized particles are greater than 4.75 millimeters (mm) and are
not commonly present in tailings (Kossoff et al. 2014). Sand sized particles are smaller than gravel sized,
and are between 4.75 mm and 0.075 mm. Silt and clay sized particles are smaller than sand sized, and
are less than 0.075 mm (Das 2005).
The grain size of copper tailings is highly variable and can be difficult to generalize due to the
specific ore processing method utilized. However, if flotation is utilized to process sulfide ores, this
requires that the ore be crushed and ground to sand and silt sized particles to effectively separate copper
(Kossoff et al. 2014, Shamsai et al. 2007). Crushing and grinding the ore to a smaller size increases the
surface area of the ore and the efficiency of copper extraction through flotation, and it is common to
produce tailings with fine particles of less than 0.03 mm (Wei et al. 2009).
The grain size distribution can be utilized to provide initial estimates of other engineering
properties including permeability and strength. Fine tailings typically have a lower permeability than
coarse tailings, can inhibit dewatering, and are susceptible to strength loss when loaded due to pore
pressure increases (Wei et al. 2009). For tailings with a high sand size percentage, it may be economical
to separate the sand from the silt or clay sized particles for use as construction materials. The sand can
be utilized to construct surface containment embankments to retain the smaller sized tailings particles, or
can be utilized to backfill underground voids (Wei et al. 2009).
A common laboratory technique employed to obtain the grain size distribution of the tailings is a
sieve analysis (Das 2005). During a sieve analysis, a tailings sample is passed through a series of
screens and the weight of tailings retained on each screen is recorded to determine the overall
percentage of tailings retained. The sieve analysis testing procedure to determine the grain size
distribution of tailings greater than 0.075 mm is detailed in ASTM Standard D6913-04 Standard Test
Methods for Particle-Size Distribution (Gradation) of Soils Using Sieve Analysis (2009). A laboratory
technique employed to obtain the grain size distribution of fine tailings less than 0.075 mm is a
hydrometer analysis (Das 2005). During a hydrometer test, the tailings are placed in a mixture of water
and a dispersing agent, and the settlement time and distance are recorded to calculate the grain size and
26
|
Colorado School of Mines
|
specific gravity. The hydrometer testing procedure to determine the grain size distribution of tailings less
than 0.075 mm is detailed in ASTM D7928-16 Standard Test Method for Particle-Size Distribution
(Gradation) of Fine-Grained Soils Using the Sedimentation (Hydrometer) Analysis (2016b).
The Unified Soil Classification System (USCS) was established to describe both the texture and
grain size of soil materials, and correlations can be made to predict material behavior, specifically how
tailings will behave after placement within a disposal facility. The USCS utilizes the grain size distribution
and Atterberg limits to classify the tailings as clay, silt, sand, and gravel (ASTM 2011a).
4.1.2 Atterberg Limits
The Atterberg limits include the shrinkage, plastic, and liquid limits (Das 2005) of the tailings and,
with the grain size distribution, can provide an indication of how the tailings may behave after placement
within a disposal facility. The Atterberg limits can also be utilized to determine what additives to mix with
the tailings to meet engineered design specifications for disposal. To determine the Atterberg limits,
moisture is added to the tailings and the water content is recorded as the tailings transition from a solid to
liquid state. The shrinkage limit is the percent water content within the tailings at the point when the
tailings transition from a solid to a semisolid state; the plastic limit is determined as the water content at
the transition from semisolid to plastic; and the liquid limit is determined as the water content at the
transition from plastic to liquid (Das 2005). The laboratory test methods for determining the liquid and
plastic limits of tailings are detailed in ASTM Standard D4318-10 Standard Test Methods for Liquid Limit,
Plastic Limit, and Plasticity Index of Soils (2010b). There is currently no ASTM standard for the
determination of the shrinkage limit. The Atterberg limits and grain size distribution are utilized to classify
the tailings as clay, silt, sand, and less common gravel (ASTM 2010b).
Like the grain size distribution, the Atterberg limits of copper tailings are highly variable and can
be difficult to generalize due to the specific ore processing method utilized. Atterberg limits can indicate
the presence of clay particles within the tailings (Kossoff et al. 2014), and the limit values have been used
in correlation to estimate compressibility and shear strength of tailings, specifically with relation to the clay
and sand percentages of tailings (Boulanger and Idriss 2006). Sand tailings typically behave as non-
27
|
Colorado School of Mines
|
The specific gravity is the ratio of the tailings density (mass of the solids divided by the volume of
solids) to the density of water (Lindeburg 2012). Since tailings are typically less than 4.75 mm in size, the
laboratory test methods for determining the specific gravity should be utilized as detailed in ASTM D854-
14 Standard Test Methods for Specific Gravity of Soil Solids by Water Pycnometer (2014). The void ratio
is the ratio of the volume of voids to the volume of solids, and is calculated by measuring the mass of the
sample before and after drying using the laboratory test methods detailed in ASTM D2216-10 Standard
Test Methods for Laboratory Determinations of Water (Moisture) Content of Soil and Rock by Mass
(2010a).
4.1.4 Permeability
The permeability of tailings is variable, based largely on the grain size distribution (Vick 1990),
and describes the ability for fluid to flow through pore spaces within the tailings. The hydraulic
conductivity is utilized to describe the rate a fluid can pass through a thickness of tailings per a unit of
time (Das 2005). Tailings with a greater percentage of sand sized particles tend to be more permeable
and have higher hydraulic conductivities than tailings with a greater percentage of silt or clay sized
particles, due to the larger void space associated with larger sized particles (Qui and Sego 2001). The
rate of consolidation and seepage are influenced by the permeability and hydraulic conductivity of the
tailings.
The hydraulic conductivity can be determined through a constant or falling head test (Das 2005)
utilizing test methods detailed in ASTM D5084-16a Standard Test Methods for Measurement of Hydraulic
Conductivity of Saturated Porous Material Using a Flexible Wall Permeameter (2016a). A constant head
test is typically utilized for tailings with a high percentage of sand sized particles, while a falling head test
is utilized for tailings with a high percentage of silt and clay sized particles. During a constant head test,
the elevation head between the inlet and outlet is maintained and a constant flow rate of water is passed
through the tailings. The amount of water collected at the outlet over a unit of time is recorded and utilized
to calculate the hydraulic conductivity. During a falling head test, a known amount of water is passed
through the tailings and the change in head elevation is recorded at increments over a determined time
interval.
29
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.