University
stringclasses 19
values | Text
stringlengths 458
20.7k
|
|---|---|
Virginia Tech
|
Table 5.2: Number of pictures taken at each pillar
Pillar Aug. 26 Sep. 16 Sep. 26 Oct. 28
1 0 3 5 0
2 0 11 7 2
3 5 9 8 6
4 0 4 11 7
5 2 13 11 5
6 4 5 0 0
7 12 9 0 3
The photographs were processed through a combination of Agisoft Photoscan [108],
CloudCompare, and Maptek iSite [94]. Photoscan was used to obtain the point clouds and
triangulation surfaces from the photographs. Next, CloudCompare was used to orient and scale
the time-lapse photos. Several 30 cm squares were placed in each scene to provide a reference for
scaling the resultant triangulation surfaces. The squares were not placed in the same location
during subsequent visits, and as a result, would create the appearance of movement on the rib face
at the locations they were placed.
Orienting the photos was performed by locating the same points on the rib or roof between
visits. Exposed rock faces have a significant number of visual features that can be located across
the photographs from different visits and assigned the same three-dimensional coordinate. Four
of these features were chosen in each point cloud to align with a point cloud of the same region at
the next visit. If reference points moved between scenes, such as in the expansion of a rib, this
would cause a systematic error clearly visible when aligning the triangulation surfaces.
Lastly, iSite was used to determine the distances between the triangulation surfaces at
different time periods. The volumes reported are the volumes enclosed by two triangulation
surfaces. The older surface will always be the reference. Negative volumes correspond to the
removal of material from the rib, while positive volumes correspond to an expansion of the rib or
accumulation of material that did not previously exist. The same process was applied previously,
in an underground limestone mine, using laser scanning as the point cloud collection method [112].
These software packages are not uniquely able to perform these functions, nor are the
software packages limited in use to the function presented here. The reason for using each was
the preference of the author. Due to the poor quality of some photographs, a significant cascading
error may exist in some of the point clouds. Imprecisely constructed point clouds can lead to
imprecise scaling and inaccurate referencing, resulting in underestimations or overestimations of
66
|
Virginia Tech
|
volume change. Point clouds, which appear to be significantly noisier than the rest, are noted in
their relevant section.
5.4 Results
The time-lapse analysis of surface change, as calculated by iSite, is shown in Figure 5.2
through Figure 5.9. The colored surface has been overlain on a photograph of the scene at the
earlier time in the analysis. The overlain pictures are approximately aligned by hand; however,
the volumes of change are calculated by iSite from the two referenced surfaces. Warmer colors,
which are positive values, are used to indicate an expansion, or movement of material towards the
camera. Cooler colors, which are negative values, are used to indicated a contraction, or spalling
of material. A green color, which dominates the majorities of overlays, very little or no change
detected between scans.
Figure 5.2: Change at Pillar 1 between September 16th and September 26th
The change observed at Pillar 1, in Figure 5.2, is likely to be artifacts of the
photogrammetry. The September 16th pictures were both blurry and in very low light, resulting in
67
|
Virginia Tech
|
The first time period, from September 16th to September 26th, in Figure 5.8, shows no
significant movement, with the small color variations likely due to the poor quality of the
photographs used. The second time period, from September 26th to October 26th, a large
displacement of rib material was detected. In the area enclosed by Box A, 4.03 m3 was displaced.
The larger displacement, which exceeds the color scale, showed an approximate average
displacement of 34.5 cm. The other sides of this pillar were inaccessible when the next visit
occurred on October 28th.
Figure 5.9: Change at Pillar 6 from Aug. 26th to Sep. 16th
It is difficult to verify the change present in Figure 5.9, due to the blurriness of the
photographs. Change does appear to have occurred in the area enclosed by Box A, however Box
B and Box C are inconclusive. A photograph of this area was taken on September 26th, and it
showed significant damage across the rib face, possibly a result of the movement shown taking
place between August 26th and September 16th. In addition to the pillars shown, observations at
the remaining photographed pillars are summarized in Table 5.3.
74
|
Virginia Tech
|
Table 5.3: Summary of conditions at each pillar
Pillar Condition Notes
Blurriness negatively impacted the
An example of a poorly
1 reconstruction of the Aug 26th to Sep. 16th time
aligned point cloud
period.
A large part of the corner of
the rib was displaced, and
Overlapping between photographs was not high
2 there was significant
enough to capture fully the displacement.
displacement in the weak
shale band
Blurriness negatively influenced the
An isolated 0.52 m3
3 reconstruction of the Aug 26th to Sep. 16th time
displacement occurred
period.
No detectable change occurred during the
4 Good
monitoring period.
Two sides of the pillar were inaccessible
A large 4.03 m3 displacement
5 following the displacement between Sep. 16th
occurred
and Oct. 28th.
Small areas of localized rib Photographs from Oct. 28th show significant
6 spalling and rib expansion are damage in the areas that were expanding, but
detected the photographs were blurred.
Significant rib spalling The Aug. 26th and Sep. 16th photographs were
occurred between Aug. 26th blurry, but less pronounced than at Pillars 1 and
7
and Sep. 16th, but none after 3. Fog may have played a role in reducing the
that time period quality of these photographs.
The pillar behaviors observed can be classified into four categories: spalling, expansion,
weak band failure, and no movement. Nearly all the pillars exhibited spalling, which is shown on
Pillars 2, 3, 5, 6, and 7. This spalling ranges from 0.29 m3 to 0.52 m3 of material being displaced
from the rib. The spalling behavior shown on Pillar 2 and 3 occurred at the corner of the pillar and
was poorly quantified as a result. Due to the light positioning, another set of photos would need
to be taken solely of the pillar corner in order to reconstruct it properly. Photographs were not
taken of the corner, but instead each side of the pillar was reconstructed, leaving the corner
deformation quantitatively unknown. Pillars 3, 5, 6, and 7 all showed some degree of spalling
away from the corners, although the 4.03 m3 displaced on Pillar 5 was the largest by far. This
spalling did not appear preferentially located at certain parts of the rib.
Expansion, or an observed movement of the rib towards the camera, was shown on several
pillars, but this could also be indicative of a poorly aligned or reconstructed point cloud.
Considering the photographic quality at each site, Pillar 6 is the only likely candidate for showing
75
|
Virginia Tech
|
Spalling was observed on four pillars, with three being relatively small amounts of
material, 0.29 m3 - 0.52 m3, displaced between monitoring periods. Another pillar showed
significantly more material, 4.03 m3, displaced between monitoring periods.
One pillar face showed a small amount of spalling while simultaneously showing other
areas of expansion. The expansion was generally less than 15 cm, but the quality of the
photographs was not high enough to locate clear tension cracking as a result. The picture following
the expansion between August 26th and September 16th, on September 26th showed significant
damage in the areas that had been expanding. This damage would be expected if the pillar was
inelastically expanding.
Some pillars experienced either no movement or very little movement across the
monitoring period. Just as it is important to be able to detect material movement, it is also
important to be able to detect the absence of movement, and not suggest that structural instabilities
exist where they do not. Several triangulated surfaces do show false movements between periods,
but these generally show anomalies across the entire surface, and are a result of a poor
photogrammetric reconstruction.
The time-lapse photogrammetric monitoring performed at this site resulted in seven pillars
being successfully reconstructed. The precision of measurements varied with photograph quality,
but expected rib changes were modeled and capable of being quantified. Additionally, the
photographs were taken remotely by a researcher not associated with photogrammetry, who was
at the mine for different purposes. This highlights one of the greatest strengths of photogrammetry
as a rock mass monitoring tool: the ease with which the data collection steps can be communicated
and performed. Photogrammetry has been shown to be uniquely suited for measurement of large
mine areas, and can offer a fast and cost-effective supplementary perspective on the mine behavior
of which traditional point measurement techniques may be incapable.
77
|
Virginia Tech
|
Chapter 6 Time-Lapse Photogrammetric
Monitoring of an Artificially Loaded
Standing Roof Support
B. A. Slaker, Graduate Research Associate
E. C. Westman, Associate Professor
Geomechanics Observation and Imaging, Mining and Minerals Engineering
Virginia Polytechnic Institute and State University, Blacksburg, VA, 24061
6.1 Abstract
Photogrammetry has found many uses is industrial applications, including the mining
industry, but its usefulness as a tool in monitoring standing roof supports in underground coal
mines has not yet been established. Collection of meaningful underground deformation data is
crucial to understanding rock mass behavior, but the ability to do so can be unreliable, expensive,
and time-intensive. This study involves testing using photogrammetry at the Mine Roof Simulator
(MRS) in at the National Institute for Occupational Safety and Health (NIOSH) Research
Laboratory in Pittsburgh, Pennsylvania during the loading of a standing support. At stages during
the standing support’s loading, an ATEX certified explosion-proof digital camera is used to collect
an object panorama and determine the object deformation between convergence stages.
Comparing the photogrammetric results to the MRS stroke measurements, the point cloud support
heights differ from the actual support heights by between 0.2% and 0.6%, depending on the
convergence stage. The strain experienced by the support as calculated through photogrammetry
and measured on the MRS differs by 0.2% and 0.3% after the first and second displacement stages.
While the results may not achieve the same level of accuracy or precision as other industrial
photogrammetric applications, the level of precision does not preclude its use as a monitoring tool.
78
|
Virginia Tech
|
6.2 Introduction
Underground mines commonly experience ground deformations as a result of changing
states of stress in the rock mass surrounding excavation. Monitoring rock deformations is critical
to understanding the mechanical behavior of a rock mass and designing support systems to prevent
unwanted deformations or stress concentrations.
Longwall mines in the United States rely heavily on standing secondary support systems
to prevent the closure and maintain stability of gate roads. A design methodology for secondary
support systems was developed by NIOSH that incorporates ground reaction data into the selection
of the different, available standing support systems, called the Support Technology Optimization
Program (STOP). Without this, a mine is typically required to create a trial section where the new
support system is monitored for its response to abutment loading [35]. To collect the necessary
ground reaction data, in order to properly design a support system for a mine, the displacement of
the individual supports must be monitored.
One way of assessing the support design requirements is by examining the ground response
curve, shown in Figure 6.1. This curve represents how support systems interact with rock mass
expansion and failure when an opening is created. Predicting the shape of the ground response
curve is difficult, due to the heterogeneity and anisotropic nature of the rocks that surround
underground coal mining operations. The only known points along the curve are those that have
been measured along the support reaction line. The design curves for standing supports have been
experimentally determined from laboratory testing, and by measuring the displacement of the
support, the pressure being exerted by it can be estimated.
79
|
Virginia Tech
|
Figure 6.1: Ground response curves and a support reaction (after [16])
Measuring the displacement of a standing support is typically performed using point
measurement devices, such as contact extensometers. Their measurement precision may be high,
but their accuracy is dependent on how they are installed on the support and how the support
deforms. An alternative measurement technique, being proposed in this paper, is photogrammetry.
Photogrammetry is a form of image measurement that derives the geometric properties of an object
or scene from one or more photographs. The primary purpose of this process is to recreate a three-
dimensional object in a digital format [9].
The purpose of this experiment is not to create a perfect scenario in which a standing
support can be modeled with the highest degree of precision. The purpose of this experiment is to
evaluate photogrammetry as a monitoring or measurement tool for deformation in an underground
coal mine. The regulatory, environmental, and operational limitations of underground coal mines
limit camera selection and picture-taking practices. Therefore, it is the goal of this experiment to
closely replicate what is operationally feasible and obtain the most accurate and precise results
possible within those confines. Inducing a load on a standing support, and using photogrammetry
to monitor that change, will constitute a first step in quantifying the reliability of photogrammetry
as a monitoring tool for standing roof supports.
80
|
Virginia Tech
|
6.3 Methods
The experiment will involve artificially inducing a convergence on a standing roof support
and capturing the deformation with photographs at different intervals. The photogrammetric
results can then be compared to the measured convergence to determine their accuracy. There are
three major components to this process: the camera, the support and hydraulic press, and the
software used to process the photographs.
The camera selected for this experiment is a CorDEX ToughPIX II digital camera [106],
shown in Figure 6.2, was designed for use in hazardous environments. It is ATEX and IECEx
certified for petrochemical and mining environments, with an armored LCD screen and flash. The
camera does not have Mine Safety and Health Administration (MSHA) approval, however, it has
camera specifications that could likely be maintained or surpassed while meeting MSHA design
standards, if a manufacturer decided to seek such approval.
Figure 6.2: CorDEX ToughPIX II [106]
Pertinent design specifications for the camera include a 16-megapixel capture resolution,
3x optical zoom, 16-gigabyte image storage, removable and rechargeable battery, and a weight of
0.9 kg. The camera does not offer much flexibility in changing shutter speeds or focal length, but
such concessions must be made when the camera models available are severely limited by the
regulations concerning explosive atmospheres.
The photogrammetry subject is a cylindrical steel support filled with an aerated concrete,
and is a commonly used yielding support in United States longwall mines. These supports are
81
|
Virginia Tech
|
especially common where timber is scarcer, such as in Western longwall mines [41]. Due to the
high deformability of the support, large deformations can be monitored, ensuring that the change
in total length of the support will not fall below the sensitivity of the monitoring capabilities. The
support photographed in this study has a 22.9 cm diameter and stands 274.9 cm tall.
The support will be loaded by the Mine Roof Simulator (MRS), located at the NIOSH
Pittsburgh Research Laboratory. The MRS, shown in Figure 6.3, was manufactured by MTS
Systems Corp, and originally used for testing longwall shields. The MRS is a servo-controlled
hydraulic press that applies loads of 13.3 MN from each of its four actuators. It has a maximum
opening height of 4.9 m, a maximum convergence of 0.610 m, and platens measuring 6.1 m x 6.1
m. The experimental configuration, with the support inside of the MRS is shown in Figure 6.4
[107].
Figure 6.3: Mine Roof Simulator [107]
82
|
Virginia Tech
|
Figure 6.4: Standing support inside the MRS
For this experiment, pictures were taken of the support at three different convergence
levels. The first level is before convergence has started, and the MRS has merely been lowered to
establish a contact with the support. The second level is after 5.1 cm of convergence has occurred.
The third level is after an additional 25.4 cm. of convergence has occurred, or 30.5 cm from the
original height. After the convergence at each stage has occurred, loading was paused and a series
of pictures were taken. A layout of the estimated camera positions is shown in Figure 6.5. Each
picture had a resolution of 4608 x 3456, a focal length of 5 mm, and was captured at a distance of
1 to 2 meters from the support. These distances were chosen because they allowed as much of the
support to be photographed as possible while remaining at a geometrically reasonable distance
from the support considering the dimensions of a coal mine entry. Each camera position shown in
the x-y views consists of either two or five photos, as illustrated in the z-x view.
83
|
Virginia Tech
|
between them, as well as the approximate height of the support as determined by the scaled point
cloud.
6.4 Results
As the MRS converged, the support began to crumple near the top platen. The state of the
support after no convergence (Stage 1), 5.1 cm convergence (Stage 2), and 30.5 cm convergence
(Stage 3) are shown in Figure 6.6. All visually recognizable deformation of the support is
occurring at the top of the support.
Figure 6.6: Stage 1 (left), Stage 2 (middle), and Stage 3 (right)
The calculated convergence is the difference between the top platen of the MRS and the
bottom platen of the MRS. To assist in the visualization of this convergence, the distance between
the top platen and an artificially constructed plane at a z position of 274.9 cm (the measured height
of the support) is shown in Figure 6.7. A triangulation was generated for each stage. To orient
each vertical axis of the triangulation surface with the z-axis, planes of best fit were determined
for the calculated geometry of the top platen. The normal vector for this plane was then rotated to
align itself with the z-axis for all three stages. The three stages were translated by hand to the
same approximate location. The hand translation should not affect height calculations, as they are
independent of x, y, and z position.
85
|
Virginia Tech
|
bottom platen surfaces without vertical deviation. The height of the support could be measured by
finding the perpendicular distance between the two platens at any given point, or by subtracting
the z-coordinates of two areas that appear to best represent the flat surface of the platen. It may
be necessary to arbitrarily choose locations to measure in a field study, but this research benefits
from a convenient plane-to-plane distance that can be used as the height of the support.
The exact height of a standing support in an underground coal mine would be unknown
after any convergence, but would be known to the precision of manufacturer specifications before
installation. When these supports are installed underground, they are commonly prestressed with
timber wedges or inflatable bladders at the support/roof interface [34]. At any subsequent
monitoring stage, the deformation would be unknown prior to collecting measurements. One
proposed method for determining this height through photogrammetry is to determine the roof-to-
floor distance by assigning a z-coordinate position to roof and floor. Isolating the roof and floor
triangulations from the rest of the scene allows them to be analyzed independently to determine
their z-position. An average of the z-coordinates for each vertex in the triangulation could be an
inaccurate representation of position due to skin control issues in the roof or accumulation of
material on the floor. The mode may be a better indicator of position, representing the most
probable occurrence. To lessen the effect of bin sizes when creating a histogram from a set of
continuous distance measurements, the probability density function was smoothed using the kernel
density estimation function in MathWorks’ MATLAB [109].
In order to determine the height of the support at each loading stage, a construction plane
was created at a z-coordinate of 274.9 cm. The z-coordinates corresponding to the points that
comprise the reconstructed top and bottom platens were measured as the distance of each point in
the platen to that construction plane. This creates an array of distance values from which a mode,
or most likely outcome, can be determined through a kernel density estimation. Bandwidths for
the kernel density estimation were adjusted to give each platen location a single, clear peak. This
was most necessary for the top platen at stage 3, which with a rough bandwidth could be considered
trimodal. The maximum values along each of these bimodal curves, shown in Figure 6.8, are
considered to be the z-positions of the top and bottom platens.
87
|
Virginia Tech
|
Figure 6.8: Kernel density estimation for the platen positions
Using the peak-to-peak distance, a height for the MRS at each of the three stages can be
calculated. The height, change, and strain of each MRS displacement stage is shown in Table 6.3.
The measured height is the distance between platens as measured by the MRS. The calculated
height is the distance between platens as estimated through photogrammetry in Figure 6.8. The
remaining measured and calculated columns are derived from the measured and calculated heights.
Table 6.3: Measured vs calculated convergence
Stage Measured Calculated Measured Calculated Measured Calculated Strain
Height Height Change Change Strain Strain (%) Error
(cm) (cm) (cm) (cm) (%) (%)
1 274.88 276.43 N/A N/A N/A N/A N/A
2 269.80 271.60 -5.08 -4.83 -1.85 -1.74 5.94
3 244.40 245.50 -30.48 -30.93 -11.09 -11.19 0.90
Between stage 1 and 2, a 4.83 cm convergence is calculated, and between stage 1 and 3, a
30.93 cm convergence is calculated. These convergences differ from the measured values for
stage 1 to 2 and 2 to 3 by -0.25 cm and 0.45 cm respectively. The magnitude of change between
stage 1 and 2 does not differ proportionally from stage 2 to 3, suggesting that there is an error
independent of the height of the support, and likely instead dependent on the distance between the
88
|
Virginia Tech
|
photographs and the subject as well as camera parameters. All of the calculated heights are larger
than the measured heights, possibly indicative of consistent scaling error.
It is difficult to distinguish between error caused by scaling and inaccuracy or imprecision
in the photogrammetry. The calculated height differs from the measured height by between 0.45%
and 0.56%, which could reasonably fall within an expected margin of error when manually
selecting points for scaling. Despite this error, the measurements, if reproducible in an
underground environment, are of an accuracy that could be useful for measuring roof to floor
convergence.
6.5 Conclusion
The ability to quantify the deformation of underground standing supports is important for
understanding rock mass behavior. In order to see if photogrammetry can be used as a tool for
monitoring underground rock mass movements, a standing support was subjected to different
displacements, photographed, and then modeled. The environment for the experiment is markedly
different from an underground coal mine. There are no issues with inadequate lighting, and the
geometries available for scaling and referencing the scene are favorable. Both lighting and
unfavorable geometries would be significant concerns in an underground environment. However,
the laboratory setting allows for highly controlled displacement of the standing support, which
could not be performed underground, and is an important step in applying photogrammetry to the
monitoring of standing supports underground.
After modeling the standing support at different displacement levels, the results of
photogrammetry showed a calculated convergence of 4.93 cm compared to a measured
convergence of 5.08 cm after the first MRS movement, and a calculated convergence of 30.93 cm
compared to a measured convergence of 30.48 cm after the second MRS movement. The
difference between the calculated and measured convergence is not likely dependent on the size
of the support, but instead of the quality of the photographs and relative size of a pixel to the
support.
The accuracy of measurements required in an underground environment will be mine-
dependent, but the sub-centimeter accuracy calculated from the photogrammetry in this study
should be adequate for many applications of deformation monitoring in underground mine
89
|
Virginia Tech
|
Chapter 7 Underground Photogrammetric
Monitoring of Standing Supports
7.1 Abstract
An underground longwall coal mine in Central Appalachia was studied for ground
movement in response to abutment loading. Photos were taken of a wooden crib and sand-filled
steel support at different visits, with a changing state of stress, for a time-lapse photogrammetric
analysis of displacement. This photogrammetric analysis was performed in conjunction with
extensometer measurements of change during the monitoring period. Both the wooden crib and
the steel support were successfully modeled using photogrammetry in different dynamic lighting
conditions. During the monitoring period, a roof-to-floor convergence of 0.62 cm was recorded
by an extensometer on the modeled crib, with a roof-to-floor convergence of 0.62 cm measured
using the three-dimensional models created through photogrammetry. No extensometer was
placed on the steel support, but a modeled 0.28 cm of roof-to-floor convergence was detected
according to the photogrammetric point cloud. As expected, photogrammetry accuracy was found
to be highly dependent on photograph quality, lighting, and overlap, but modeling of standing
supports was found to be possible within sub-centimeter precision using a point-and-shoot camera.
7.2 Introduction
Underground coal mining was responsible for producing 343 million tons of coal in 2012
in the United States [1], of which 52% was excavated by longwall [2]. Coal remains an important
part of the United States’ energy portfolio, as well as the world’s and it is important to continue
improving coal mine safety. In the United States, falls of ground in underground coal mines have
led to 2300 operator and contractor injuries, as well as 35 fatalities between January 2007 (the year
of the Crandall Canyon mine disaster) and July 2013 [3]. The injury and fatality rate has been
decreasing in recent decades [4], due to better safety practices and increased understanding of
underground mining environments.
91
|
Virginia Tech
|
In longwall mining, large abutment stresses form following the removal of support that
accompanies excavation of panels. These abutment stresses can present ground control problems,
specifically in the gate roads, if not adequately addressed. These abutment stresses are usually
addressed with regularly installed standing supports. The behavior of these supports in laboratory
settings are well documented, and by using observed deformations, the ground response curve for
the rock mass can be determined. The anisotropy and heterogeneity of the typical stratified rock
masses found in underground coal mines leads to conditions that differ widely enough to
necessitate site-specific monitoring and customization of standing support arrangements, typically
with a trial section to determine proper support [29].
Two standing supports were monitored for this research, the strain-softening conventional
timber support and a non-yielding sand-filled steel support. Conventional timber supports, also
known as wooden cribs, were until recent decades the only available support for longwall tailgates.
Wooden cribs provide support over a large deformation, are easy to install, and are generally low-
cost. The steel support used in this study behaves differently than wooden cribs do under load. It
is capable of providing high support pressures, but does not allow more than several centimeters
of convergence before failure.
The means of measuring support deformation was photogrammetry. Photogrammetry is a
method of determining the geometric properties of an object by analyzing photographs. The
purpose of modern digital photogrammetry is to create a three-dimensional object in the digital
space [5]. This study focuses on an application of close-range digital photogrammetry (CRDP),
typically contrasted with aerial photogrammetry. CRDP is a photogrammetric technique where
the object is within 100 meters and the cameras either are contained within, or surround, the object
[53]. Photogrammetry has wide interdisciplinary use, but is also being applied in an underground
mining environment for fracture characterization [57], geotechnical mapping [58][59], and blast
rock volume measurement [60].
Underground mine movements are widely measured using equipment limited to the
response of a single point, such as extensometers. These systems require an extrapolation of rock
mass movement and are subject to local biases and method of installation. These systems usually
come with a high degree of precision, but they lack the ability to capture movements beyond the
one-dimensional axis on which they are affixed. Photogrammetry offers an ability to capture the
three-dimensional movement of an entire scene. The usefulness of photogrammetry as a fast, cost-
92
|
Virginia Tech
|
effective, and precise supplementary tool for monitoring underground mine displacements will be
explored in this research.
7.2.1 Site Description
The underground coal mine visited in this study is located in the Central Appalachian
region of the United States and its primary means of ore extraction is through longwall mining.
The mining depth ranges from 187 m to 193 m, with an average entry height of approximately 2.4
to 2.7 m in the study area. The roof consists of 0.3 to 1.2 m of shale, overlain by 0 to 2.4 m of
sandy shale. The floor strata consists of 0.3 to 2.4 m of shale and fireclay, underlain by 0 to 3 m
of sandstone.
When monitoring first began at the site, little development had occurred, and how the
standing support system would respond to abutment loading was unknown. The standing support
system used to reinforce gate roads consisted of concrete-filled metallic cylinders, wooden cribs,
and sand-filled steel supports. The photogrammetry research was performed coinciding with the
installation of extensometers on standing supports at various locations throughout the mine. These
were installed to monitor the ground response, and improve the standing support plan.
Inconsistencies in the extensometer results led to only two locations providing reliable data, which
are marked on Figure 7.1.
93
|
Virginia Tech
|
Figure 7.1: Approximate mine layout in the monitoring region. The bleeder cribs are located in the bleeder entries and the
headgate cribs are located in what, at the time of monitoring, was the headgate.
It was not common at the mine to install supports far in advance of the longwall face,
however, the mine had a crosscut heavily supported with wooden cribs and steel supports,
approximately 369 m ahead of the longwall face. This area, labeled as “headgate cribs” in Figure
7.1, was supported as a remedial measure in response to unexpected deformation. The headgate
crib area was photographed at different visits to monitor a time-lapse change. The bleeder cribs
were not photographed due to safety regulations, however, extensometer data is available for this
region, which can help to understand the ground response.
7.3 Methods
Two subjects at the headgate site were chosen, one wooden crib with no apparent
deformation and one steel support with significant deformation already present. A photograph of
this crosscut is shown in Figure 7.2, the steel support is shown in Figure 7.3, and the wooden crib
is shown in Figure 7.4.
94
|
Virginia Tech
|
Figure 7.4: Wooden crib photogrammetry subject
7.3.1 Camera and Lighting
Two cameras were used in this experiment. The first is a CorDEX ToughPIX II digital
camera [106], was designed for use in hazardous environments. It is ATEX and IECEx certified
for petrochemical and mining environments, with an armored LCD screen and flash. This camera
is not certified with the Mine Safety and Health Administration, but has representative
specifications that could be expected in a camera designed for use in a hazardous environment.
The camera has a 16 megapixel capture resolution, 3x optical zoom, 16-gigabyte image storage,
removable and rechargeable battery, and a weight of 0.9 kg. The zoom was not used for this study.
The second camera used was a 12 Megapixel Canon PowerShot. The third set of photos was
collected remotely by an employee of the mine who did not have access to the ToughPIX II, and
used the most similar point-and-shoot camera available.
Lighting was provided by either the camera flash or a wearable array of cap lamps, shown
in Figure 7.5. The array was fabricated specifically for this task and consisted of six Polaris
Cordless Cap Lamps [113], set to their brightest setting. The array was worn around the neck to
96
|
Virginia Tech
|
leave the hands free for photography. The lighting was changing throughout the study, although
the changes were minimalized with the cap lamp array. This auxiliary lighting was necessary
because the battery life of the ToughPIX II camera could not sustain flash photography throughout
the experiment.
Figure 7.5: Mobile array of cap lamps for use in underground mine illumination
7.3.2 Image Capture and Software Methodology
Time-lapse photographs were taken of both the steel support and the wooden crib. The
second set of photographs was taken 25 days after the first, with a third set being taken of just the
wooden crib two days after the second set. The light source for the first set of photograph was the
cap lamp array. The camera flash was not used for the first set due to battery concerns. The second
set of photos included both a cap lamp light source and the camera flash, with each having a set
unique to that light source instead of both light sources being used simultaneously. An additional
battery was taken to compensate for the increased power requirements from the camera flash. The
last set of photos was acquired by an employee of the mine using a 12 megapixel point-and-shoot
camera and the onboard flash as a light source. These photograph sets are shown in Table 7.1
97
|
Virginia Tech
|
Figure 7.6: Approximate position of the longwall face relative to the photographed crosscut
The photographs were processed through a combination of software packages. The first,
Agisoft PhotoScan [108], was used to generate a three-dimensional point cloud from the
photographs. The number of photographs used, from Table 7.1, corresponds to the number used
in reconstruction in PhotoScan. A sparse point cloud was generated first, followed by a densely
constructed point cloud. The dense point cloud, depending on the quality of photographs, often
contains many erroneous points, which are removed prior to mesh construction if they are
geometrically unreasonable. Triangulated surfaces were then created for each of the standing
supports in PhotoScan using the dense point cloud as a base for construction. These objects were
then exported into CloudCompare, an open source 3D point cloud and mesh processing software,
for scaling and orienting. The midpoint of each 0.12 m x 0.12 m wooden face was used to translate
and rotate the point clouds onto the same coordinate system. They were then scaled assuming an
average 12.2 cm thickness for the wooden beams. The workflow for processing the photos into
manageable three-dimensional surfaces is illustrated in Figure 7.7.
99
|
Virginia Tech
|
responsible for poor point matching in the photographs, especially towards the center of the
support. Because of this, the November 5th photograph set using the cap lamp lighting was
discarded, and instead the flash-lit photograph set from this time period was compared to the
October 10th cap lamp-lit photograph set. Both the wooden crib and steel support reconstructions
are shown in Figure 7.8 along with the anchor points that were used for roof-to-floor convergence
measurements.
Figure 7.8: Measurement range for the extensometer and photogrammetry
The reconstructed steel support showed a roof-to-floor convergence of 1.10 cm between
October 10th and November 5th. The convergence range on October 10th was calculated to be
227.02 cm, and on November 5th was calculated to be 228.11 cm. The points selected for
comparison on the steel support were not as well defined as the planes created by the individual
beams for the wooden crib, increasing the difficulty with which convergence anchors could be
precisely determined on the October 10th reconstruction.
Extensometer data from October 10th to October 31st on the photographed wooden crib
shows a significant convergence on October 13th, and small movements afterwards that resulted in
101
|
Virginia Tech
|
a slight expansion by the end of the 21 day period. Unfortunately, the extensometer readings did
not continue through to the second set of pictures. The extensometer was vandalized and stopped
producing useful readings until it was repaired on November 5th. The extensometer readings
continued to show no significant trend in convergence until November 8th, when the longwall face
was 25 m from the test site and one day following the last set of pictures. At this point, a clear
convergence is visible throughout the remainder of the data collection period. The extensometer
measurements are shown in Figure 7.9.
Figure 7.9: Extensometer measurements, at 12-hour intervals, for the convergence of the crib throughout the experiment
The measured extensometer convergence from October 10th to October 31st is 0.86 cm,
with a measured expansion of 0.24 cm from November 5th to November 7th. The cumulative
convergence recorded by the extensometer is 0.62 cm, although this does not include any
convergence that may have occurred during the 5 days where the extensometer was not
operational. Each of the photogrammetric reconstructions is shown in Figure 7.10.
102
|
Virginia Tech
|
Figure 7.10: Photogrammetric reconstruction of the wooden crib on October 10th (top left), November 5th (top right), and
November 7th (bottom)
The two locations where the extensometer was anchored were not known in the first set of
photos, and the planar surfaces of the wooden beams made for more convenient photogrammetric
measurement points, shown in Figure 7.8. The different anchor points result in a convergence
measurement range approximately 12 cm larger for the extensometer data. Neither measurements
are true roof-to-floor convergence values, with three wooden beam widths lying outside of the
extensometer measurement range.
Using photogrammetry, and measuring between the planes corresponding to the top of the
2nd and bottom of the 11th beam, 1.16 cm of convergence was detected between October 10th and
November 5th. An expansion of 0.54 cm was detected in the same physical range between
November 5th and November 7th, resulting in a cumulative convergence of 0.62 cm. The absolute
differences between the extensometer and photogrammetry measurements for the first-to-second,
103
|
Virginia Tech
|
second-to-third, and cumulative convergence are 0.30 cm, 0.30 cm, and 0.00 cm respectively. The
first-to-second and cumulative convergence values for the extensometer may be in error; however,
the second-to-third measurement is consistent. The convergence measured on the reconstructed
point clouds are very similar between the steel support and the wooden crib for the October 10th
to November 5th time period, differing by only 0.06 cm, although the measurement range for the
steel support was 19.5 cm larger.
Three additional extensometers were installed in the photographed crosscut. They should
not be used to validate the photogrammetric data, as the local variation in ground response is
unknown. However, they can be used to verify that the extensometer installed on the photographed
crib is measuring movement consistent with deformation in nearby cribs, and not producing
anomalous results, possibly indicative of experimental error. The remaining extensometers are
shown in Figure 7.11 at the point where a converging trend first emerges. The trend develops in
response to the developing front and side abutment load, as the longwall face approaches and
passes. The photographed area shown in Figure 7.6 is the same as the plan view shown in Figure
7.11.
Figure 7.11: Convergence measurements from all four extensometers installed in the monitoring area, starting from the date of a
clear abutment stress-induced movement
104
|
Virginia Tech
|
The extensometers all show a very similar pattern of convergence, with the magnitude
convergence becoming more pronounced as the distance from the longwall panel increases. The
convergence shown in this crosscut is similar to the convergence experienced on instrumented
wooden cribs in the bleeder entries following longwall development.
7.4.1 Other Experiments
In addition to the standing support monitoring, other mine structures were photographed
or recorded through video in different ways, to test the versatility of this close range
photogrammetry application. Two methods of picture capture were explored: Object Panorama
and Straight Line, shown in Figure 7.12. The previous data analysis was performed on
photographs collected using the Object Panorama method, however, this may not always be
possible due to geometric constraints. The Straight Line method also provides the simplest and
quickest method of image capture, often at the cost of an insufficient overlap between photos and
a loss of point information on the obscured side of the object.
Figure 7.12: The Object Panorama and Straight Line methods of image capture
105
|
Virginia Tech
|
A series of 32 photos were taken using the Straight Line method in the supported crosscut
in intervals of 1 m, taking enough pictures to capture the entire roof to floor geometry of the
supports at each interval. The onboard flash was used as the primary source of light, and the
camera settings for these photos are the same as the November 5th photos in Table 7.2. The
reconstructed point cloud as well as a triangulated surface of the photographed supports are shown
in Figure 7.13.
Figure 7.13: A reconstructed standing support triangulation from the camera position (left) and a dense point cloud as viewed
perpendicular to the long axis of the crosscut (right).
Anomalous points have been manually removed from the scene, and the remaining points
show clearly resolved objects. There are several objects in the scene that could provide a scale,
such as the wooden beam widths or steel support labels, however there is no extensometer data for
these supports, nor was there a time-lapse data collection. The resolution with which the camera-
facing side of the supports were reconstructed suggests that they could be used to measure
convergence.
Using video converted into image sequences was explored for its viability in producing
quality photogrammetry images. Using video to reconstruct three-dimensional objects is often
called videogrammetry, but as image sequences are being extracted from these videos and the same
106
|
Virginia Tech
|
methods previously discussed are being applied, it will continue to be referred to as
photogrammetry in this paper. Two experiments were conducted using videography instead of
photography: a Straight Line walkthrough of a crosscut and a steel support Object Panorama.
Two obstacles associated with photogrammetry become significantly more pronounced
when working with video image sequences: motion blur and image resolution. There is a tendency,
when taking photographs, to keep the camera still. When taking a video, the camera is in near
constant motion, the degree to which is controllable by the operator. The motion blur will reduce
the quality of the reconstruction, or render it incapable of being reconstructed altogether. The most
common high-definition video recording resolutions currently available are 720p or 1080p, which
corresponds to an image resolution of 1280 x 720 and 1920 x 1080 respectively. This resolution
is significantly lower than the photograph resolution used for the object panorama.
Both experiments using the image sequences extracted from 720p video proved
unsuccessful. The lower resolution combined with a high degree of motion blur, caused by the
rough walking terrain, resulted in subpar images. Three-dimensional reconstruction was possible
to a limited degree, but always resulted in sparse or incomplete point clouds, unfit for scientific
measurement.
7.5 Conclusions
Monitoring underground mine movements is critical to understanding the rock mass
response. Photogrammetry has been explored as a tool for increasing the availability of mine
deformation information, by providing a faster and more comprehensive data collection technique
than traditional point measurements. An underground coal longwall mine was used as a test site
for monitoring the support response of two standing supports installed ahead of a longwall face.
An instrumented wooden crib support as well as a steel support were photographed in a time-lapse
manner using different lighting conditions.
The photogrammetric reconstruction of the wooden crib showed a cumulative convergence
of 0.62 cm from October 10th to November 7th 2014. An extensometer installed on the crib
recorded a cumulative convergence of 0.62 cm over the same time. Vandalism of the extensometer
resulted in 5 days with no data being recorded between the first and second visit to the support.
Between the second and third visit, the photogrammetric reconstruction showed a 0.54 cm
107
|
Virginia Tech
|
expansion in this time, while the extensometer recorded a 0.24 cm expansion. Larger convergence
values would have been more beneficial for establishing the true precision of the reconstruction.
However, considering the roof-to-floor height is approximately 6.2 m at this location, sub-
centimeter discrepancies in convergence are relatively small.
Additionally, the steel support nearby the monitored wooden crib was also successfully
reconstructed. The difference measured between the October 10th and November 5th visits was
0.28 cm. This change, while not validated by an extensometer on the support, was considered
geologically reasonable, as other nearby extensometers all reported sub-centimeter convergence.
The steel support was more difficult to reconstruct due to a reduced number of features and the
reflectance of the surface. Future photogrammetric studies on similar supports should take care to
diffuse the light source and ensure large overlap between photographs if possible.
Lastly, experiments were performed to test the feasibility of using a Straight Line method
as opposed to the previously demonstrated Object Panorama method of capturing photographs,
and if image sequences extracted from video recordings could substitute photographs. The
Straight Line method was demonstrated to be capable of reconstructing an underground scene,
despite the complex lighting situation. The image sequences extracted from captured 720p video
footage contained too much motion blur and was of too low a resolution to adequately construct
the underground scene. Improving the video resolution and reducing movement while recording
are both within current technological and operational capabilities, and would likely improve
results.
Photogrammetry has been shown to be capable of monitoring support deformation. The
magnitude of deformation measured through photogrammetry is very similar to that measured by
extensometers, which are widely used in research capacities to measure ground movements.
Photogrammetry offers an advantage over extensometers in its ability to capture the entire
geometry of the region being monitored, rather than a single point. Photogrammetry as a tool for
monitoring is not biased by the method of installation nor does it require additional data acquisition
units. Quality photogrammetric results do depend on quality photographs, and monitoring requires
repeated visits to the scene. The limitations of photogrammetry do make its use situational, but
the ability to measure three-dimensional change underground quickly, cost-effectively, and with
accuracy comparable to extensometers makes it a valuable tool for assessing rock mass
movements.
108
|
Virginia Tech
|
Chapter 8 Discussion and Conclusions
8.1 Summary of Work
Photogrammetry and laser scanning were both used to reconstruct the geometry of several
mines and structures within them. Two coal mines, two limestone mines, and the Mine Roof
Simulator (MRS) at the Pittsburgh Office of Mine Safety and Health Research were used as test
locations for these two technologies and their mining application. The goal of each study was to
determine a movement in the scene over time. Three of the experiments involved studying wide-
area rib changes while two experiments focused on monitoring standing support convergence. The
photogrammetry approach to each of these displacement-monitoring studies was performed in a
way considered most practical and cost-effective for application in underground mining.
The rib displacement experiments contained two limestone mine sites and one coal mine
site. The laser scanned limestone mine experiment involved was performed at two scanning
period, which focused on one spalling pillar and another pillar that was scaled between scans.
Calculating the difference between the triangulation mesh reconstructions from the laser scanning
data, volume changes of 2.6 m3 on the spalling pillar, and 2.3 m3on the scaled pillar were detected.
Anomalous change elsewhere was not found, validating the movement that was detected.
The second laser scanning experiment was performed at one of the underground coal mine
test sites, and the displacement being modeled was hand-induced by removing material from the
rib. A scan was performed before and after displacement was created, and the scene
reconstructions were compared for volume change between them. The laser scanner was moved
slightly during the experiment to ensure a marginal need to reference objects in the scene for a
proper orientation. Volume changes as small as 57 cm3 and as large as 57549 cm3 were detected.
Footprints and scaled material were also visible in the time-lapse scans, however the roof and the
undisturbed portions of the ribs showed no movements.
The last rib spalling experiment used photogrammetry instead of laser scanning, and was
performed by another party remotely, with instructions provided for how to best photograph the
scene. Several different behaviors were identified, that corroborated previous observations at the
site: no change during the monitoring period, a weak shale band spalling, small pockets of change,
and one large pillar spall event of 4.0 m3. All measurements were performed using 30 cm square
109
|
Virginia Tech
|
references and without the need for mine coordinates, highlighting the versatility of the
photogrammetry method in the absence of surveying data.
Beginning the standing support convergence experiments, the MRS was used to test the
ability to precisely monitor the movement of a standing support over time. The MRS was
photographed at three different convergence levels, each of which was modeled and compared to
the other two to determine the magnitude of displacement taking place. The support was
photographed in a way that would best simulate the angles possible in an underground
environment. The convergence stages were set at 0 cm, 5 cm, and 30 cm, and the MRS measured
convergence agreed with the photogrammetry measured convergence to within 0.5 cm of change
at both the 5 and 30 cm stage.
The other standing support experiment was performed in an underground coal mine using
photogrammetry, with the goal of detecting convergence caused by an advancing longwall face
and the resulting abutment pressures. This study was performed over 28 days, punctuated by three
visits to capture the deformation on standing supports at each time. This convergence was
simultaneously monitored by extensometers installed on one of the photographed standing
supports as well as others nearby. Convergence measurements as obtained by an extensometer on
a wooden crib matched the convergence measurements of the same wooden crib as obtained from
photogrammetry. The cumulative convergence measured using both methods was 0.62 cm and
0.62 cm respectively. A nearby steel support was also photogrammetrically reconstructed, and
showed a convergence of 1.16 cm, over a slightly shorter time frame than the wooden crib
comparison.
8.2 Discussion of Results
The photogrammetric and laser scanned reconstructions were studied for each test location.
The photogrammetry and laser scanning applied to the limestone mines both revealed small and
large areas of rib change, a result of either scaling or spalling. The reliability of these change
detection studies can be validated by the absence of change in structures that were not expected to
change. The laser scanned limestone mine showed no movement in the mine roof and no
movement across large areas of the ribs, suggesting that the triangulation meshes were properly
oriented. The laser scanned coal mine showed very similar results, with the roof not indicating
110
|
Virginia Tech
|
any movement, and undisturbed areas of the rib and floor also not moving. The roughness of the
coal rib face did not appear to have a large impact on the ability to reconstruct rib features at large
angles of incidence, as demonstrated in the field and laboratory.
The photographed limestone mine also showed areas of rib that experienced clear and
quantifiable displacement. Isolated spalling in a weak shale band was identified as well as
widespread spalling of the limestone, which was verifiable through careful inspection of the
photographs. Not every photograph set, however, was clearly resolved, and some likely showed
anomalous movements. The anomalous movement was characterized by a widespread “inward”
and “outward” movement of the surface, which could be geotechnically interpreted as pillar
spalling and expansion. It was not considered geologically reasonable that the rib was spalling
and expanding in the manner displayed, nor was there visual evidence of the change in the
photographs. The anomalous readings could be traced to poor quality photographs or highly
dynamic lighting conditions.
The standing support modeled in a laboratory setting showed absolute convergence values
that agreed with the instrumentation measurement. Error between the MRS convergence and the
photogrammetric convergence was within 0.5 cm for both measurements. The scene was scaled
using known feature lengths on the MRS rather than an outside scale, which may have contributed
to scaling errors within the model. The standing supports modeled in a field setting also agreed
with the measurement devices, although the exact deformation that occurred could not be
determined, and the precision of the photogrammetry remained vague. Both the steel support and
the wooden crib were difficult to model with the cap lamp light array. An increase in lighting as
well as improved diffusion of the light would likely improve the quality of the photographs.
Speed is a concern when performing any underground measurement because it controls
how much data can be collected and if it can be collected without interfering with operations.
Laser scanning and photogrammetry should be divided into two parts, data collection and data
processing, when discussing the time required to deliver meaningful engineering measurements.
The data collection period, neglecting travel through the mine, for these experiments was
approximately 10-20 minutes, with a visit being defined as the monitoring of one specific area or
object at one instance in time. Data processing takes significantly longer, and varies significantly
based on the quantity and quality of data. Laser scanning data already has a meaningful scale and
accurate reproduction, removing some of the computation time. The laser scanning data from
111
|
Virginia Tech
|
these experiments, if processed again, would likely require 20-60 minutes to obtain rib movement
information. Photogrammetry, including the time to reconstruct the scene, would require
approximately 1-6 hours to process, depending on the number of photographs being used. Much
of the time spent processing in photogrammetry is passive and does not require operator input.
The time estimates made here are assuming familiarity with the photogrammetry and laser
scanning process.
The methods of monitoring mine displacements presented here have largely focused on rib
spalling and roof-to-floor convergence as measured on standing supports. The application of
photogrammetry and laser scanning is not limited to measuring these two conditions, but rather
any visible mine movement is capable of being reconstructed. The spalling and roof-to-floor
measurements were performed because they are most easily controlled and predicted, whereas
floor heave or a full-entry deformation profile would be difficult to validate and predict in a
permissible environment. These results do show quantifiable measurements of rib displacement
as well as extensometer-validated marginal roof convergence, which serves as a foundation for
exporting these techniques to monitoring more varied ground movements or collecting large
quantities of information to establish a thorough ground response.
8.3 Conclusions
Photogrammetry and laser scanning have been shown capable of reconstructing
underground mine scenes with a high level of precision. Several mining and laboratory
environments have been successfully reconstructed using these remote sensing techniques and
monitored for displacement over time. The ability to capture the entire structure as it moves
removes the ambiguity associated with local effects at the anchor points, found with traditional
point measurement instrumentation. In addition, the data collection method allows for large areas
of a mine to be measured quickly and non-intrusively. Monitoring large areas of a mine using
point measurement instruments would require many different installations, and depending on the
instrument installed, may disrupt operations.
When determining the rock mass response, as is typically reflected on the ground response
curve, detailed measurements of the rock displacement behavior in response to changing states of
stress is critical. The discontinuous and anisotropic nature of rock makes site-specific
112
|
Virginia Tech
|
characterization of rock behavior necessary for mine planning, and photogrammetry and laser
scanning are two methods of characterizing that behavior in a precise, fast, and reliable manner.
The underground coal mine photogrammetry and laboratory standing support photogrammetry
showed that measurement accuracy within 0.5 cm could be achieved both quickly and cheaply,
which is important when large amounts of data are required for an accurate quantification of rock
mass response.
The precision of laser scanning is usually provided by the manufacturer, and in these tests,
has not appeared to be significantly affected by the underground mining environment. The
precision of photogrammetric measurements are highly variable, changing with distance to the
subject and the quality of images used. The underground environment presented numerous
obstacles to this method, most notably lighting and a potentially limited selection of cameras
available for use. Despite these obstacles, cameras using typical point-and-shoot specifications
have been used to generate precise point clouds using the flash, and adequate point clouds using a
wearable cap lamp light array.
Rock mass behaviors that were previously difficult to capture, such as rib spalling, are
easily captured and quantified using laser scanning and photogrammetry. Three different tests
using these technologies showed rib displacements ranging from tens of cubic centimeters to
several cubic meters. Just as proper installation of instrumentation is required to model rock mass
behavior, so too should care be taken to obtain the highest quality photographs. It is unlikely that
photogrammetry or laser scanning will match or exceed the theoretical precision of extensometers
in practical use, however, the ability to gather data quickly and cheaply over a large area should
provide an alternative to extensometers where sub-millimeter precision is not required.
8.4 Suggestions for Future Work
This work shows the application of photogrammetry and laser scanning to monitoring
movements in underground mines. The means of obtaining photos are significantly more varied
than what has been presented here, and the movements observed in underground mines extend
beyond roof-to-floor convergence. The methods of obtaining images and the applications of the
technology are two areas that could see significant improvement in future research.
113
|
Virginia Tech
|
Other means of image capture include the exploration of videogrammetry, specifically as
it relates to helmet or vehicle-mounted cameras. The image sequences that were extracted from
the video collected in this paper were marred by a combination of low resolution and motion blur,
which may be mitigated by different cameras and video capture techniques. In addition to this,
improving the lighting conditions during photography will improve results, but generating large
amounts of light in heavily supported areas of underground coal mines may be difficult.
Innovations in hand-held or worn lighting systems, may allow for individuals to reconstruct
underground objects or structures with more precision.
In addition to the rib spalling and convergence measurements taken, other mine behaviors
may be measured using photogrammetry, including numerous modes of pillar failure, roof sag, or
floor heave. Modeling these behaviors could have a large impact on mine planning, and the means
by which these behaviors could be monitored is not fundamentally different from the way roof-to-
floor convergence is monitored.
Lastly, photogrammetry should be compared to traditional, trusted measurement
techniques at every opportunity. As with many remote sensing techniques, demonstrating the
reliability of a new monitoring method becomes more important as the path from data collection
to visual or numerical output becomes more nebulous. Awareness of photogrammetry, and to a
lesser degree laser scanning, is low, and repeated testing of this technology is necessary to improve
operator comfort and trust in the monitoring method.
114
|
Virginia Tech
|
[40] T. M. Barczak, “A retrospective assessment of longwall roof support with a focus on
challenging accepted roof support concepts and design premises,” in Proceedings of the
25th international conference on ground control in mining, Morgantown, WV, 2006, pp.
232–243.
[41] T. Barczak and S. Tadolini, “Pumpable roof supports: an evolution in longwall roof
support technology,” 2005.
[42] Burrell Mining Products, “The CAN.”
[43] E. T. Brown, J. W. Bray, B. Ladanyi, and E. Hoek, “Ground response curves for rock
tunnels,” Journal of Geotechnical Engineering, vol. 109, no. 1, pp. 15–39, 1983.
[44] T. M. Barczak, G. S. Esterhuizen, and D. R. Dolinar, “Evaluation of the impact of
standing support on ground behavior in longwall tailgates,” in Proceedings of the 24th
International Conference on Ground Control in Mining, 2005, pp. 23–32.
[45] T. M. Barczak, G. Esterhuizen, J. Ellenberger, and P. Zhang, “A First Step in Developing
Standing Roof Support Design Criteria Based on Ground Reaction Data for Pittsburgh
Seam Longwall Tailgate Support,” in Proceedings of the 27th International Conference
on Ground Control in Mining, 2008, pp. 349–359.
[46] L. Prosser, T. E. Marshall, S. C. Tadolini, A. T. Iannacchione, and C. Banta,
“Considerations for Using Roof Monitors in Underground Limestone Mines in the USA,”
Retrieved June, vol. 11, p. 2007, 2006.
[47] T. M. Barczak, J. Chen, and J. Bower, “Pumpable roof supports: developing design
criteria by measurement of the ground reaction curve,” in Proceedings, 22nd international
conference on ground control in mining, Morgantown, WV, 2003, pp. 283–93.
[48] B. Shen, A. King, and H. Guo, “Displacement, stress and seismicity in roadway roofs
during mining-induced failure,” International Journal of Rock Mechanics and Mining
Sciences, vol. 45, no. 5, pp. 672–688, 2008.
[49] E. Hosca, M. Karmis, and C. Haycocks, “Influence of support capacity and geometry on
tailgate support,” in International Journal of Rock Mechanics and Mining Sciences and
Geomechanics Abstracts, vol. 32, no. 8, 1995, p. 395A–395A.
[50] J. Dunnicliff, Geotechnical Instrumentation for Monitoring Field Performance. Wiley,
1993.
[51] G. Barrientos and J. Parker, “Use of the pressure arch in mine design at White Pine: 17F,
7R. Trans. Soc. Min. Engrs. AIME, V256, N1, Mar. 1974, P75-82,” in International
Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts, vol. 11, no.
10, 1974, p. 203.
118
|
Virginia Tech
|
[52] E. Esterhuizen, C. Mark, and M. Murphy, “The Ground Response Curve, Pillar Loading
and Pillar Failure in Coal Mines,” in 29th International Conference on Ground Control in
Mining 2010, 2010.
[53] M. Diederichs and P. Kaiser, “Stability of large excavations in laminated hard rock
masses: the voussoir analogue revisited,” International Journal of Rock Mechanics and
Mining Sciences, vol. 36, no. 1, pp. 97–117, 1999.
[54] M. Jeremic, Strata mechanics in coal mining. CRC Press, 1985.
[55] G. Molinda and C. Mark, “Ground failures in coal mines with weak roof,” Electronic
Journal of Geotechnical Engineering, vol. 15, pp. 547–588, 2010.
[56] G. M. Molinda and C. Mark, Coal mine roof rating (CMRR): a practical rock mass
classification for coal mines, vol. 9387. US Dept. of Interior, Bureau of Mines, 1994.
[57] Z. Bieniawski, “Engineering Classification of Jointed Rock Masses,” Civil Engineer in
South Africa, vol. 15, no. 12, 1973.
[58] C. Mark and G. Molinda, “Development and application of the coal mine roof rating
(CMRR),” in Proceedings of the International Workshop on Rock Mass Classification in
Underground Mining, Information Circular, vol. 9498, 2007.
[59] S. Fekete, M. Diederichs, and M. Lato, “Geotechnical and operational applications for 3-
dimensional laser scanning in drill and blast tunnels,” Tunnelling and Underground Space
Technology, vol. 25, no. 5, pp. 614–628, 2010.
[60] D. F. Huber and N. Vandapel, “Automatic three-dimensional underground mine
mapping,” The International Journal of Robotics Research, vol. 25, no. 1, pp. 7–17, 2006.
[61] D. Ferguson et al., “An autonomous robotic system for mapping abandoned mines,”
Advances in Neural Information Processing Systems, vol. 3, 2003.
[62] J. Mah, C. Samson, S. D. McKinnon, and D. Thibodeau, “3D laser imaging for surface
roughness analysis,” International Journal of Rock Mechanics and Mining Sciences, vol.
58, pp. 111–117, 2013.
[63] S. Fekete and M. Diederichs, “Integration of three-dimensional laser scanning with
discontinuum modelling for stability analysis of tunnels in blocky rockmasses,”
International Journal of Rock Mechanics and Mining Sciences, vol. 57, pp. 11–23, 2012.
[64] A. Prokop and H. Panholzer, “Assessing the capability of terrestrial laser scanning for
monitoring slow moving landslides,” Natural Hazards and Earth System Science, vol. 9,
no. 6, pp. 1921–1928, 2009.
119
|
Virginia Tech
|
[78] T. Hagan, “A case for terrestrial photogrammetry in deep-mine rock structure studies,” in
International Journal of Rock Mechanics and Mining Sciences & Geomechanics
Abstracts, vol. 17, no. 4, 1980, pp. 191–198.
[79] F. Tonon and J. Kottenstette, “Laser and photogrammetric methods for rock face
characterization: A Workshop,” in Report on a workshop held in Golden, Colorado, 2006.
[80] S. Rezaei and A. Rahnama, “Application of Close Range Photogrammetry to Monitor
Displacements in Mines,” in SME Annual Meeting, 2013.
[81] Harris Interactive, “Growth in Ownership of Smartphones and Tablets Appears Largely
Android-Driven,” 2013.
[82] J.-A. Beraldin, “Integration of laser scanning and close-range photogrammetry-the last
decade and beyond,” in International Society for Photogrammetry and Remote Sensing,
2004.
[83] “‘Electric Motor-Driven Mine Equipment and Accessories." 30 "CFR" 18. 2014.
[84] S. Foster and D. Halbstein, Integrating 3d Modeling, Photogrammetry and Design.
Springer, 2014.
[85] R. Singh, D. Chapman, and K. Atkinson, “Digital photogrammetry based automatic
measurement of sandstone roof of a mine,” Journal of the Indian Society of Remote
Sensing, vol. 25, no. 1, pp. 47–59, 1997.
[86] T. Luhmann, “Close range photogrammetry for industrial applications,” ISPRS Journal of
Photogrammetry and Remote Sensing, vol. 65, no. 6, pp. 558–569, 2010.
[87] J. Birch, “Using 3DM analyst mine mapping suite for rock face characterization,” in Laser
and Photogrammetric Methods for Rock Face Characterization Workshop, 2006, pp. 13–
32.
[88] R. Christiansson, “The latest development for in-situ rock stress measuring techniques,” in
Proceedings of the International Symposium on In-situ Rock Stress. Trondheim,
Norway:[sn], 2006, pp. 3–10.
[89] J. Kemeny, K. Turner, and B. Norton, “LIDAR for rock mass characterization: hardware,
software, accuracy and best-practices,” Laser and photogrammetric methods for rock face
characterization. ARMA Golden, Colorado, 2006.
[90] M. Lato, M. S. Diederichs, D. J. Hutchinson, and R. Harrap, “Optimization of LiDAR
scanning and processing for automated structural evaluation of discontinuities in
rockmasses,” International Journal of Rock Mechanics and Mining Sciences, vol. 46, no.
1, pp. 194–199, 2009.
121
|
Virginia Tech
|
Operator-mobile lighting is auxiliary lighting that either is held, worn, or moves with the
camera as the camera moves. As the operator moves, so too will the light, causing a different
lighting scenario for each photo. This method of lighting is less favorable, since it alters the visual
features in each photograph, however, it is easier to implement. A method of using cap lamps as
auxiliary lighting was presented in this research, although it leaves much room for improvement.
Cap lamps are easily accessible at most underground mining operations and any means of allowing
a person to use multiple at once may provide enough lighting to use photogrammetry. The cap
lamps will need to be arranged in a manner that does not focus them all on one spot, and a means
of diffusing the light will need to be employed.
The last method of lighting is on-board flash lighting. Most camera flashes provide enough
lighting to illuminate the narrow confines of an underground coal mine, but will not be sufficient
for cavernous environments. An on-board flash is the simplest and cheapest method of
underground illumination, and will likely suffice in most situations if the battery capable of
sustaining repeated flash photography.
Photography
The method of taking photographs should be methodical instead of haphazard, not
necessarily due to computational restrictions, but rather ensuring proper overlap in the scene is
being achieved. Ideally, photographing an object should be performed in an object-panoramic
manner, where the object is circled with photographs being taken at incremental arc lengths. If a
scene is being photographed, such as an entry or room, the scene should still be photographed in a
similar manner, where the camera changes position between photos. There is an inclination to
take photos standing in a central point and rotating about that point without moving camera
positions, which will likely result in a poor photogrammetric reconstruction.
There is no rule for overlap required in the photos, and it will depend on the quality of
photos being used in the reconstruction, but at least 75% overlap in photos has yielded the best
results in these studies. This means that 75% of the subject visible in one photograph is also visible
in the next photograph, accounting for any decreased visibility of an object due to viewing angle
change. Be mindful, that the first and last photos, in an incrementally progressing set, will need
to extend past the area of interest if a full object-panorama was not captured, as they will both
141
|
Virginia Tech
|
contain areas that were not found in any other photo. Care should be exercised when taking the
photos because a single blurry or insufficiently lit photo can cause a gap in overlap that causes the
scene to lose connectivity. Until the camera operator is comfortable with judging the lighting and
level of motion blur in a photograph, as it is taken, taking redundant pictures that increase the
degree of overlap is advisable.
In a low light environment, if possible, the camera should be set to maximize the light
reaching the sensor. The method of doing this was not optimized in this research, as the camera
did not allow for a significant degree of customization. An adjustment of the camera’s ISO was
possible, and a high ISO setting was chosen to accommodate the low-light environment. This
produced a very grainy image, and if possible, adjusting the camera shutter speed or aperture size
may be preferable. These will both affect the depth of field and motion blur of the photographs,
and will likely need to be developed on an application-specific and operator-specific basis.
Photogrammetry often has the ability to use points in the background of the scene for
reconstruction purposes, if they are present in multiple photographs. This results in a stark contrast
between performing photogrammetry in well-lit environments and poorly-lit environments. In the
case of a supported underground coal entry, the background objects may not be visible, resulting
in fewer points available for matching. This is especially pronounced with thin supports when
photographed near their middle, which can be exacerbated if the surface is reflective and the light
source is not diffused. When photographing the top and bottom of thin supports, the roof and floor
appear in the photographs and contain numerous visual features, but these are often nonexistent
when photographing the middle of the support.
The onboard camera flash was sufficient as a light source for the majority of underground
coal mining photogrammetry performed. Photogrammetry in large, cavernous, dark environments
will require an additional light source or the camera will have to be positioned very close to the
area of interest. Due to this different light requirement, it is more likely that in a cavernous mining
environment, a supplemental, stationary light source will be used, instead of a dynamic light source
in an underground coal mine. The stationary light source will improve the photogrammetric
reconstruction, and the overlap or redundant photos can be minimized.
If using lighting to supplement the onboard camera flash, ensure that it is diffuse enough
to not cause the focused light to obscure large portions of the scene. Cap lamps tend to create
highly focused beams of light which obscure the region they are focused on and insufficiently light
142
|
Virginia Tech
|
the surrounding regions. Cap lamps should either be turned off during photography or focused
away from the scene as to not negatively affect the photos. If cap lamps are to be used as a light
source, some means by which to diffuse them must be devised, and it is likely that several will be
required to produce enough for photography.
When using a stationary light source, the effect of shadows on features in the photos
become more pronounced. This is especially pronounced when rounding corners using quasi-
dynamic lighting, such as stationary auxiliary lighting that is moved when the subject becomes too
far from the light source. In these situations, few photographs will be necessary to capture a side
of an object, however, a larger number of photographs will be required when photographing the
corner of the object. A large number of photographs ensures that similar features do exist in the
photos and the photo set is not split into several smaller and distinct photo sets.
Data Processing
Photogrammetric reconstruction can be performed through a variety of different software
packages. This discussion will only concern basic processes that are common to many of the low-
cost options available and point cloud manipulation. All of these practices and issues will not
necessarily apply to specific software options available.
Not every photo taken needs to be included in the photogrammetric reconstruction process.
Often, poor photos will only harm the reconstruction by either causing the camera to be improperly
located, resulting in incorrect points being added to the scene, or causing the process to fail
altogether. Identifying which photographs are harmful can usually be done beforehand, although
removing bad photographs after they were processed, and then reconstructing the scene again is
often possible. There is no universal value that quantifies motion blur, graininess, poor lighting,
overexposure, etc. It will likely require trial and error to determine what a particular software
package requires for image quality.
Photogrammetry can be computationally intensive, depending on the number and
resolution of photographs used in reconstruction. If the overlap in photos was high and the image
quality was high, then redundant photographs may be removed to expedite the reconstruction
process. Additionally, pieces of a scene can be reconstructed individually and later stitched
together, in either the photogrammetry software or point cloud manipulation software. This can
143
|
Virginia Tech
|
help to isolate a problem if the reconstruction is being performed multiple times due to poorly
located camera locations or a failed reconstruction.
After a successful reconstruction, the units associated with the coordinate system as well
as the camera orientations are likely to be incorrect, and will require a coordinate system
transformation. The transformation is usually performed by locating three or more points common
to a reference coordinate system. If the reconstructions are not being used for surveying purposes,
and can function on an arbitrary coordinate system, the most accurate point cloud should be used
as the reference. There are typically numerous features in a scene that can be used for referencing
without the need for external targets, although many software packages will automate the process
if reference targets are used. The software will attempt to translate, rotate, and scale a scene to
match selected points with the reference system, while minimizing tension, or distance error,
between points.
Including as many points as possible in the coordinate system transformation will reduce
the impact of one poorly located point. Other than artificially included reference targets, the best
visual features for referencing tend to be stark coloration differences on flat surfaces. Small,
complicated geometries in a scene are often the most difficult to reconstruct, and choosing a
reference point on them is both difficult to ensure the correct point was selected and it is more
likely to have been poorly located. In an underground coal mine, discolorations on a roof bolt
plate, impurities in wood, or scratches on a metal support often make the best features.
144
|
Virginia Tech
|
DESIGN OF AN EXPERIMENTAL MINE SIMULATOR FOR THE DEVELOPMENT OF A
PROCEDURE FOR UTILIZATION OF MULTIPLE TRACER GASES IN UNDERGROUND MINES
John Robert Reid Bowling
ABSTRACT
An experimental mine simulator was constructed which will be used to conduct
tracer gas experiments in the laboratory. The test apparatus simulates a mine in a tabular
deposit and is modular and simple and can be easily rearranged to represent a variety of
mine geometries. The apparatus is appropriate for the use of tracer gases by being both
airtight and open-circuit (exhausting to the atmosphere) and by maintaining turbulent
flow throughout the model, ensuring the tracer gas is fully dispersed.
The model features ports for injection and sampling of tracer gases, which
represent boreholes present in an actual mine. The model is designed, in part, for the
practice of tracer gas release and sampling methods in the laboratory. Valves on the
apparatus represent ventilation controls, such as stoppings or regulators, or changing
resistances in a mine, such an increase in resistance due to a roof fall or a decrease in
resistance due to stoppings being destroyed. The relative resistances of airways can be
changed by changing the status of the valves to represent different states of the
ventilation controls.
The mine simulator should serve as a tool for identifying and investigating novel
tracer gases, developing a procedure for performing ventilation surveys using multiple
tracer gases, and eventually developing a method for remotely inferring ventilation
changes using tracer gases.
|
Virginia Tech
|
ACKNOWLEDGEMENTS
I would first like to thank my advisor, Dr. Kray Luxbacher, for her patience and
guidance throughout my graduate studies. I would also like to thank Dr. Saad Ragab and
Dr. Erik Westman for their input and support.
Second, I would like to thank Jim Waddell for the hours of help in construction in
the laboratory and Robert Bratton for his technical guidance in selecting instrumentation.
I would also like to thank the lovely and always helpful department administrative
staff, Carol Trutt, Gwen Davis, and Kathryn Dew for signing forms and ordering all my
parts and always greeting me with a smile.
Additionally, I want to thank my fellow graduate students Charles Schlosser,
Edmund Jong, Rosemary Patterson, and Guang Xu for their assistance in tasks related to
the construction of the experimental apparatus.
Finally, I would like to acknowledge the National Institute for Occupational
Safety and Health (NIOSH), especially our program coordinator Dr. Gerrit Goodman, for
providing the primary monetary support for this project.
This publication was developed under Contract No. 200-2009-31933, awarded by
the National Institute for Occupational Safety and Health (NIOSH). The findings and
conclusions in this report are those of the authors and do not reflect the official policies of
the Department of Health and Human Services; nor does mention of trade names,
commercial practices, or organizations imply endorsement by the U.S. Government.
All photographs were taken by the author, 2011.
iii
|
Virginia Tech
|
John R. R. Bowling Ch. 1 Introduction
1. INTRODUCTION
With a continued demand for mined resources, underground mines continue to
become more complex as a result of the increasing difficulty in extracting dwindling and
ever less-accessible reserves. As the complexity of mine geometries increases, coupled
with increasing concern over methane control and dilution in the wake of recent
explosions, continually rising energy costs and the fairly recent introduction of stricter
legislation limiting miners‟ exposure to pollutants such as diesel particulate matter
(DPM), the demand for more complex and reliable mine ventilation is further increased.
As mine ventilation networks become more complex, traditional methods of ventilation
surveys become increasingly cumbersome or, in some cases, ineffective at solving
ventilation problems.
Chemical tracers have been used abundantly to describe fluid flows in both the
natural and man-made environment. Some fields in which chemical tracers have been
used successfully include hydrology, pollution dispersion, urban meteorology, industrial
hygiene, building ventilation and mine ventilation. A tracer gas is a gas which can be
diffused in relatively small proportions in a volume of air and which is detectable in trace
amounts at another point or points downstream. Depending on the particular tracer gas
used, the robustness of the experimental method, and the sensitivity and precision of the
instruments, a tracer gas can be detected and quantified in concentrations ranging from
parts-per-billion (ppb) to parts-per-trillion (ppt).
The primary qualities of a useful tracer gas in a ventilation network are the ability
to diffuse in the airstream (such that the fully diffused gas downstream represents the
flow of the airstream), chemical stability, and the ability to be detected and quantified in
low concentrations. These entail that the chemical is gaseous at the atmospheric
conditions in the ventilation network, the tracer gas does not undergo any chemical
changes or interact with other chemicals in the atmosphere and that the tracer gas is not
naturally occurring (lest slight concentration changes be minor fluctuations of the natural
concentration). For mine ventilation, other qualities determine the usefulness of a tracer
1
|
Virginia Tech
|
John R. R. Bowling Ch. 1 Introduction
gas. A successful tracer gas for mine ventilation should be relatively inexpensive, easily
obtainable, not harmful to miners and not radioactive1.
The concept of using tracer gases has been applied in mine ventilation surveys for
over 60 years. Tracer gases have allowed for the solution of increasingly complex
problems in mine ventilation, previously unsolvable via conventional ventilation survey
methods. The only standard tracer gas used in ventilation surveys, sulfur hexafluoride
(SF ), meets all the above criteria. SF has been used profusely in mine ventilation to
6 6
solve complex problems; however it has been demonstrated both in civil and mine
ventilation that a method for using several tracer gases simultaneously can allow for far
more complex surveys to be conducted in less time. In order for a tracer gas to be
successful for use in a complex survey procedure with SF it must also be detectable
6
using the same method as SF , namely using a gas chromatograph (GC) with an electron
6
capture detector (ECD).
The long-term goals of this research include identifying and investigating novel
tracer gases, developing a procedure for performing ventilation surveys using multiple
tracer gases, and describing a post-disaster mine response based in part on the multiple
tracer gas method. The candidate gases identified will be evaluated based on chemical
properties which are favorable to performance as tracer gases in mine ventilation surveys.
In contribution to that research effort, a laboratory-scale mine simulator was constructed
which is to be used as an experimental apparatus for tests performed with tracer gases.
Simplified tracer gas ventilation surveys will be conducted in the apparatus, which will
allow for the evaluation of the gases both as tracers (the ability to detect them at trace
concentrations) and their ability to be used in conjunction with one another in ventilation
surveys (the ability to separate the gases from one another and from air and other
common mine gases). The location of a test apparatus near the GC in the laboratory
1 Radioactive tracers have been used previously in both building and mine ventilation and can be detectable
at much lower concentrations. The researchers in this study decided to preclude radioactive tracers on the
basis that the increased training, handling and storage requirements make nonradioactive chemical tracers
more desirable for practical application mine ventilation. Additionally, the test methods for “conventional”
chemical tracers are more thoroughly developed and accepted in the mining industry.
2
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
2. LITERATURE REVIEW
2.1. TRACER GASES
Gaseous chemical tracers have been used in many disciplines to better understand
fluid flow. Examples of applications of chemical tracers in fluids, both liquid and
gaseous, are plentiful in science and engineering. As the application of chemical tracers
is a mature field, focus will be placed primarily on the use of gaseous tracer as applied to
ventilation studies.
2.1.1. PROPERTIES OF TRACER GASES
In order for tracer gases to be useful, they must have suitable chemical and
physical properties for use as chemical tracers. A gas which is useful as a chemical tracer
should be
detectable at low concentrations
safe
acceptable to miners (not radioactive)
odorless
chemically and thermally stable
not naturally occurring in the environment
able to diffuse into mine air
easy to handle and store, and
relatively inexpensive [1,2].
Some of these criteria define chemical properties of tracer gases, such as the ability to be
detectable at low concentrations, generally via gas chromatography (GC) [3,4]. Other
criteria define physical properties of the gases like the ability to diffuse into mine air,
which is related to the molecular weight or, for volatile organic vapors, their partial
pressure in air at mine conditions. Some of these properties are more complicated
functions of chemical properties, like the cost or the ease of handling and storage.
Tracer gas surveys often view parts of the ventilation system as a “black box” in
which chemical tracer input and output are measured to determine information about the
4
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
volumetric flow rates, splits, and recirculation within the system. In order for tracer gas
concentrations to be reflective of air flows, the gas must be shown to fully disperse in the
mine air to achieve a uniform concentration throughout the air volume. The gas must
also transport through the ventilation system without being chemically changed or
becoming trapped in the system. The gas must remain chemically stable at mine
atmospheric conditions and not significantly react with nor adsorb onto any materials in
the mine.
The process dominating gas-solid surficial interaction is adsorption. Adsorption
is a process by which gas (or vapor/liquid) molecules adhere to a solid surface without
chemically combining with the solid [5-7]. Adsorption is observed to take place between
every known chemical species of gas and any solid surface at some temperature and
pressure conditions [7]. Adsorption can be divided into two main types, depending on
the process by which the gas molecules are adhering to the surface. The process resulting
in the weakest bonds is physical adsorption or physisorption, in which the adsorption is
also reversible (via a process called desorption) [5]. The process resulting in an
irreversible (via physical processes) process, in which the gas molecules combine
physically with the solid surface, is called chemical adsorption or chemisorption [5,6]. In
chemisorption, there is initially an electron exchange between the gas molecule and solid
surface which precedes the chemical reaction [5]. Chemically stable tracer compounds
interact with mine surfaces through physical adsorption, since they should not react
chemically with any substances in the mine.
2.1.2. SULFUR HEXAFLUORIDE AS A TRACER GAS IN THE MINING INDUSTRY
Prior to its application in the mining industry, SF was previously used as a tracer
6
gas in environmental and meteorological studies using the tracer to study the dispersion
of airborne pollutants [8-10]. Tracer gases were also used in the heating, ventilation and
air conditioning (HVAC) industry to measure the ventilation efficiency of large buildings
[11,12]. It was previously shown by Lester and Greenberg that SF was nontoxic, an
6
important property of a chemical tracer [13]. The case for SF as a tracer gas was also
6
5
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
built when two studies demonstrated that SF could be detected at concentrations as low
6
as 10-5 ppm using GC with an electron capture detector (ECD) [3,4].
Tracer gases were first applied in the mining industry in 1958 when nitrous oxide
was used as a tracer to determine airflow in headings [14]. The first use of SF as a tracer
6
gas by the mining industry occurred in 1972, when SF was introduced into Appalachian
6
oil and gas wells prior to the halting of oil production and subsequent mining through the
wells [15]. SF was injected into the production wells prior to them being plugged and
6
during the period through which the wells were being mined, gas samples were taken
from the mine exhaust air and analyzed for SF as an indicator of leakage from the wells.
6
The first application of SF as a tracer gas mine ventilation aid occurred in 1974
6
when the U.S. Bureau of Mines reported on studies performed using SF [1]. Thimons et
6
al. cite several studies performed in three mines in which the tracer gas method was
applied successfully using SF : in a coal mine, an underground limestone mine and a
6
western vein-type metal mine. Two additional reports followed that year from the
Bureau of Mines, adding to the already broad scope of proven and potential applications
of SF in mine ventilation analysis. In one report, Thimons and Kissell described the use
6
of SF in quantifying recirculation, checking for air leakage, tracing lost air, and
6
measuring the transit time of air through a mine [2]. In another report, Kissell and
Bielicki demonstrated that SF could be used in a model mine to quantify the ventilation
6
air held up in eddies at the working face [16]. Findings from these Bureau of Mines
studies effectively laid the groundwork for the application of tracer gases in mine
ventilation, increasingly referred to in literature as the “tracer gas method.” These studies
also demonstrated the appropriateness of SF as a tracer gas by showing that when fully
6
diffused, it follows the path of the airstream and that SF does not adsorb to coal or
6
sandstone surfaces nor is it removed from ventilation air by mine air cooling plants.
They established standard release and sampling procedures, described the analysis of the
gas concentrations using solid-gas chromatography and documented useful equations.
6
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
In 1975 Kissell and Bielicki used SF as a tracer gas in a full-scale laboratory
6
experiment to demonstrate the recirculation in a coal mine working face due to dust
scrubbers on continuous miners [17]. The study showed that though dust scrubbers did
cause some recirculation of air, it was not a major contributing factor to the buildup of
methane at the working face. The following year, Vinson and Kissell reported on three
ventilation studies conducted by the Bureau of Mines using SF as a tracer gas [18].
6
These three studies used SF to investigate air leakage in a sealed area, the ventilation
6
efficiency of a bleeder system, and leakage across stoppings in parallel intake airways.
Following the analysis of the results of the individual studies, Vinson and Kissell
concluded that the application of SF should be considered to help “solve ventilation
6
problems that do not respond to conventional methods of ventilation analysis.”
Matta, Maksimovic, and Kissell used SF in a tracer decay test as a new method
6
for quantifying leakages through permanent stoppings [19]. Using the tracer gas method,
leakages through stoppings as low as 15 cfm could be measured, whereas using the
Bureau of Mines‟ previously established window brattice method, the maximum
measureable leakage rate was approximately 300 cfm [20]. Another application for the
tracer gas method was established in the evaluation of the ventilation efficiency of
auxiliary ventilation [21].
The role of SF as a tracer gas in the mining industry was solidified by
6
applications in several areas, as documented by a summary of Bureau of Mines studies
[22]. SF was used in coal mining to determine face ventilation efficiency [23], evaluate
6
gob ventilation in three coal mines [24], investigate the effectiveness of a new continuous
miner ventilation system for very deep cuts [25] and to determine the integrity of
escapeways during a fire [26]. SF was applied in a silica mill to evaluate the
6
effectiveness of silica-bagging machine exhaust hood enclosures [27]. SF saw use in
6
uranium mines in evaluating the ventilation of an in-stope uranium flood leaching
operation [28]. Finally, SF was used in a large-opening oil-shale pilot mine to determine
6
and compare the ventilation efficiencies of alternative auxiliary ventilation methods and
to evaluate the effects of blast pressures on large stoppings [29-31].
7
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
2.1.3. MULTIPLE TRACER GASES FOR MORE COMPLEX VENTILATION PROBLEMS
The selection of cited studies prior serves to prove that the tracer gas technique is
a useful tool in solving complex ventilation problems. Airflows in complex networks are
often difficult to trace and/or quantify; both of these problems can be solved by the use of
tracer gases. Further research in both HVAC and mine ventilation has demonstrated that
surveys of significantly complex ventilation networks can benefit from the use of
multiple tracer gases simultaneously.
A considerable amount of work has been done in the field of HVAC using
multiple tracer gases to characterize airflow. In 1973 Foord and Lidwell used multiple
tracer gases to describe the complex airflows in large nonresidential buildings such as
hospitals [12]. Fisk et al. used SF along with five other halocarbons in a system to
6
describe the overall airflow pattern in a large building [32]. The system described by
Fisk et al. featured six independent automated gas release devices and independent
automated gas samplers which capture a volume of air at a regular interval (15 minutes in
their study). Lagus et al. used multiple tracer gases in a study of the ventilation at a
nuclear power facility in Arizona [33]. Lagus et al. applied SF along with
6
bromotrifluoromethane (CBrF ) and perfluorodimethlcyclohexane (PDCH). In a later
3
series of articles, also compiled into a more in-depth publication, Grot and Lagus review
the various tracer gas procedures for use in industrial hygiene, including a lists of
equipment and a list of electronegative halocarbons which make effective tracer gases
[34-38].
After developing their own sampling and release systems for SF , Kennedy et al.
6
at the Cape Breton Coal Research Laboratory (CBCRL) recognized a need for multiple
tracer gases for complex ventilation surveys. Kennedy et al. conducted a series of
experiments with SF , Freon-12 (CCl F ) and Freon-13B1 (CBr F), all of which are
6 2 2 3
detectable using a gas chromatograph with ECD and all of which are of similar molecular
weight [39,40]. Kennedy et al. used a GC with two columns for their chromatography, in
which the SF and Freon-13B1 were split from the Freon-12 and sent into separate
6
columns within the same GC oven [41]. Kennedy et al. described potential problems
8
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
with Freon-12 including sample loss due to adsorption in the syringe and analysis
interference possibly caused by a hydrocarbon propellant in spray paint underground
[39].
Both Fisk et al. and Lagus et al. encountered problems using some halocarbons as
tracer gases since many halocarbons are commonly-used refrigerants or other products
and may be present in the gaseous background [32,33]. Kennedy et al. also described
difficulty with the use of some Freons, as some possibly existed in the mine air
background [39]. Both Kennedy et al. and Lagus et al. examined the use of Freons as
tracer gases and both were able to successfully use Freon-13B1
(bromotrifluoromethane/CBrF ) as a tracer gas alongside SF , although Kennedy chose to
3 6
analyze the Freon on a separate column from SF [33,39].
6
Following the initial 1987 studies using multiple tracer gases conducted by the
Cape Breton Coal Research Laboratory, there have been no major developments in
mining regarding the use of multiple tracer gases. Klinowski and Kennedy of the
CBCRL reiterated the efficiency gained by using multiple tracer gases in a single survey
in a summary of tracer gas techniques [42]. Numerous studies in mine ventilation
applying the tracer gas technique used only SF , including applications in dust control in
6
mineral processing [43,44], auxiliary ventilation [45], controlled recirculation [46,47],
face ventilation in continuous mining [48-51], gob ventilation [52-55], testing leakage
across stoppings [56], turbulence and diffusion modeling [57], narrow-vein shrinkage
stope ventilation [58]. One study used only helium as a tracer gas to investigate
overburden fracturing above longwall gobs [59]. Though their usefulness was well-
demonstrated and they have seen considerable use in other areas such as building
ventilation, it is far from common practice to simultaneously use multiple tracer gases in
mine ventilation studies.
9
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
2.2. FLUID DYNAMICS
A mine ventilation network is a complex system of fluid (ventilation air) in
motion, though it can be compared to – and often significantly simplified as – a pipe or
duct network. Individual airways (branches) can be described using the Bernoulli energy
equation and the Atkinson equation, while the whole network can be represented as a
network of resistances and solved via application of Kirchhoff‟s laws [60,61]. As with
any physical system, the physical laws governing the system offer a means of qualitative
solution of the system variables. Table 2-1 lists common variables used in mine
ventilation along with their units [60].
TABLE 2-1: SELECTED VARIABLES USED IN MINE VENTILATION
Physical Quantity Variable English Units SI Units
Pressure psi (lb/in2) (k)Pa
Pressure head* (Head loss) in. w.g. mm w.g.
Velocity fpm (ft/min) m/s
Quantity (volumetric flow rate) cfm (ft3/min) m3/s
Length, Diameter, Perimeter ft m
(Cross-sectional) Area ft2 m2
Density lb /ft3 kg/m3
m
Specific weight** lb/ft3 N/m3
Acceleration of gravity ft/sec2 m/s2
Atkinson friction factor lb∙min2/ft4 kg/m4
*Pressure head is often used in place of pressure, or interchangeably.
**Specific weight is more common than density in the English unit system.
2.2.1. IMPORTANT RELATIONSHIPS IN MINE VENTILATION
Through the application of fluid dynamics equations, a number of important
physical and thermodynamic relationships have been adopted for use in mine ventilation.
Each of these equations is derived from the governing equations of fluid mechanics: the
continuity equation, the conservation of momentum, and the conservation of energy.
10
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
Generally, flows for which can be described as moving in sheets
(laminae), so the flow regime is considered laminar [60,61,64]. Flows for which
behave irregularly and exhibit a great deal of mixing; this flow regime is
described as turbulent. Flows between these two regimes, for which
, exhibit elements of both laminar and turbulent flow (partially developed
turbulence, limited eddies); this third flow regime is appropriately referred to as
transitional, as the flow appears to be in transition from laminar to turbulent.
When a fluid flows past a solid surface, some of the fluid is slowed by friction
with the surface in an effect called friction drag [64]. Since some of the fluid has slowed,
the adjacent fluid experiences a shear stress, the magnitude of which increases with the
viscosity of the fluid (a property of the fluid determining the resistance to shear stress)
[64]. This shear stress has the effect of slowing an even thicker layer of fluid, and thus
the boundary layer develops. The boundary layer is the layer of fluid in which viscous
forces have an effect on fluid flow [64]. When a fluid flows within a conduit, a velocity
profile taken across the diameter of the conduit reveals a parabolic velocity distribution,
with the highest translational fluid velocity (velocity in flow direction) in the center of the
conduit and lower velocities near the walls of the conduit. This velocity gradient near the
walls is a result of the viscous boundary layer. Conduit flow (also channel flow, pipe
flow) is the condition in which the boundary layer encompasses the conduit radius. The
parabolic velocity profile generally remains the same throughout a conduit, except near
the inlet and where the flow is affected by other changes to the conduit such as
contractions or expansions, bends, valves, splits, etc. [64].
When a fluid enters a conduit, it does not immediately demonstrate the velocity
profile defining conduit flow. The region through which fluid travels immediately
following the entrance is called the entrance region, during which the viscous boundary
layer progressively thickens until it encompasses the conduit radius [64]. The distance in
a conduit before which the conduit flow velocity profile is fully-developed is referred to
as the entrance length, [64]. Reynolds noted that the entrance length is a function of
conduit diameter and estimated that turbulent flow, as estimated by the development of
13
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
eddies in a dye stream injected into the conduit, developed approximately 30 diameters
into a conduit [61-63]. Lien et al, in a recent study of estimates for entrance length,
found literature estimating entrance lengths ranging widely between less than 30 to
greater than 200 and recommends conservatively assuming an entrance length of 150
diameters [65]. The entrance length for a conduit in which flow is turbulent, as defined
by the Reynolds number, is much shorter than for a conduit in which flow is laminar.
Young et al. present the following equation for the entrance length for turbulent flow as a
function of Re and conduit diameter [64]
( ) ⁄ Eq. (2.11)
where is entrance length, in ft (m)
is conduit (hydraulic) diameter, in ft (m)
is Reynolds number, dimensionless.
2.2.3. VELOCITY/QUANTITY MEASUREMENT USING PITOT-STATIC TUBES
Several options are available for anemometry, including plate orifices, vane
anemometers, Pitot-static tubes, hot-wire anemometers, laser-Doppler systems, and
ultrasonic anemometers. A Pitot tube is a metal tube whose open end (usually rounded to
be more aerodynamic) is parallel to the flow direction and whose other end is connected
to a pressure-measuring device such as a manometer or electronic pressure transducer. A
Pitot tube is usually bent at 90° to protrude into an airstream or duct from within a
housing or outside the duct. A Pitot-static tube, also called a Prandtl tube, consists of two
concentric tubes, one of which measures the stagnation (total) pressure from the sensing
tip, and the other of which measures the static pressure of the moving fluid through static
pressure taps (holes exposed perpendicular to the flow) behind the sensing end. A
diagram of a Pitot-static tube is displayed as Figure 2.1 [66]. A Pitot-static tube is
generally more useful than a Pitot tube since the difference between total and static
pressure, velocity head, can be measured with one instrument. Although technically
incorrect, the term Pitot tube is often used to describe the more common Pitot-static tube.
14
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
2.3. SCALE MODELS
Scale modeling has proven useful in numerous fields of science, engineering, and
design to make predictions about a structure or system. Architects use scale models to
design buildings, including the reflection and diffusion of light from various surfaces.
Naval architects have long used water basins to study the performance characteristics of
watercraft. Similarly, wind tunnels are used extensively to study the performance of
automobiles and all manner of air- and spacecraft. Hydrology, geophysics and
meteorology are also fields which have benefitted greatly from the application of scale
models. Nearly every engineering discipline concerned with physical systems has made
extensive use of scale models for studying very large or very small structures or systems
[64,67]. Scale models are “experimental models structured to mirror the true physical
behavior of an original phenomenon, or a prototype” [67]. The system represented by the
model is called the prototype. The model is usually much smaller than the prototype so
as to be easier to handle, be more easily accessible or controllable, cost less to build and
operate, use fewer materials, and/or be generally simpler to understand [64,67].
With the rapid increase of the computational abilities and availability of
computers, numerical models are increasingly reinforcing and replacing physical scale
models [68]. Better understanding of physical phenomena and the development of novel
numerical methods for the solution of systems of partial differential equations (PDEs)
have contributed to the development of increasingly robust numerical models of physical
problems. A very successful numerical model for the analysis of dynamic fluid systems
is Computational Fluid Dynamics (CFD). Over the past few decades, CFD simulation
has become much cheaper while wind tunnel testing has become more expensive, making
CFD increasingly instrumental in the field of fluid dynamics [68].
Although numerical modeling tools such as CFD have made great advances over
the past few decades, they still have not entirely replaced physical experimentation.
CFD, as with all numerical solution methods, has its inherent drawbacks. A CFD
simulation is only as accurate as the information used to create the model. Complex
16
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
behavior such as eddies and chemically reactive flows rely heavily on submodels based
on generalized assumptions [68]. Physical scale models are still necessary to provide the
proper input, especially boundary conditions, into numerical models. A CFD model can
achieve a high degree of precision, but with inaccurate input data may be
unrepresentative of the actual physical system.
2.3.1. MODELING THEORY: SIMILITUDE
When creating scale models of systems containing fluid flow, it is important to be certain
that the measurements made on the model system represent behavior of the prototype
system; to achieve this goal, the concept of similitude is applied [64]. Most relevant
forces in a system of dynamic fluids often do not scale linearly with length, making the
responsible use of any scale model difficult without attention not only to geometric
similarity (length ratios), but also to kinematic similarity (velocity and acceleration
ratios) and kinetic similarity (force/moment ratios) [64,67,69]. In model theory, the
Buckingham pi theorem is applied in order to maintain similitude between the model and
prototype. Pi theorem involves applying dimensional analysis to derive unique
dimensionless groups (ratios) of variables which are used to maintain similarity between
the model and prototype, referred to as pi terms. Pi terms are constructed from variables
which determine the system behavior (e.g., pressure, velocity, length, viscosity and
density) and from the physical laws relating them (e.g. velocity = length/time, pressure =
force/area, kinetic energy = ½ × mass × velocity2) [67,69]. Many useful pi terms in fluid
dynamics are force ratios such as those shown in Table 2-2 [64].
17
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
TABLE 2-2: SOME COMMON DIMENSIONLESS GROUPS IN FLUID MECHANICS
Dimensionless Interpretation (Index of
Group Name Force Ratio Indicated) Types of Applications
Reynolds Generally of importance in all types
number, Re of fluid dynamics problems
Froude number,
Flow with a free surface
√ Fr
Euler number, Problems in which pressure or
Eu pressure differences are of interest
Mach number, Flows in which the compressibility
Ma of the fluid is important
Variables: Acceleration of gravity, ; Characteristic length, ; Density, ; Pressure, ; Speed of
sound, ; Velocity, ; Viscosity,
Models which maintain similitude (geometric, kinematic and kinetic similarity)
by having all pi terms equal between model and prototype (i.e., , ,
etc.) are referred to as true models [64]. Models which do not meet all the similarity
requirements (e.g., , ) are referred to as distorted models. Often in
order to maintain similitude between the model and prototype, the working fluid
(specifically, the density and viscosity of the working fluid) must be changed in the
model, e.g. by representing air in the prototype system with water in the model.
Distorted models, while if applied improperly can yield misleading information about the
behavior of the prototype, are often necessary. Such cases could arise when maintaining
perfect similitude is impossible or impractical, such as when a fluid with the required
viscosity or density is unreasonable to use or nonexistent. In such cases where a distorted
model is used, care must be taken in extrapolating model behavior to make assumptions
about the prototype.
2.3.2. LOW-SPEED WIND TUNNEL DESIGN
Experiments performed with scale models require a testing environment in which
the fluid flow can be controlled and the performance of the model can be tested. In fields
requiring the use of aerodynamics such as aeronautics and automobile design, the
working fluid is usually standard air and the apparatus to perform experiments with
18
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
moving air is called a wind tunnel. All modern wind tunnels share four common
features: an area for experimentation in which the flow parameters can be sufficiently
controlled called the test section or working section, an effuser prior to the test section, a
diffuser after the test section, and a source for air motion (often a fan) [70,71]. The
effuser is usually a contraction which serves to increase the air velocity and usually
contains screens to condition the flow (for more uniform flow). The diffuser serves to
efficiently convert velocity energy back into pressure energy to minimize the power
required to move the air [71].
Wind tunnels can be of drastically varying sizes, simulate various conditions and
have a variety of geometries. Wind tunnels can be classified based on their recirculation
of air, test section configuration, and operating speed ranges. Wind tunnels have one of
two configurations; they can be either open-circuit of closed-circuit [70]. Open-circuit
wind tunnels generally intake air from and exhaust air into the atmosphere; the air is not
recirculated. In closed-circuit wind tunnels, the air path makes a closed loop in which air
leaving the test section passes again through the fan and enters the test section again; air
within a closed-circuit wind tunnel is not exchanged with the atmosphere [70]. The test
section configuration is defined by how many solid boundaries enclose the test section;
generally they can be described as closed, open, or partially enclosed (such as by one or
more slotted walls) [70]. Wind tunnels are also generally defined by the operating speed
range: low-speed (generally less than 300 mph) or subsonic (less than the speed of sound,
Mach 1), transonic (operating range crosses Mach 1), supersonic (above Mach 1), and
hypersonic (higher Mach numbers and often high altitude conditions) [70,71].
When designing wind tunnels, scaling parameters which apply to models apply
also to the flow maintained within the tunnel. Airflow velocity (Mach number) and
Reynolds number are the two most critical parameters for wind tunnel design [70].
Another key design metric for wind tunnels is flow quality, a term used to describe
uniformity within the flow over the test section. Flow quality can be maintained through
the use of a number of engineered features of the wind tunnel. Vanes are used in corners
to reduce flow separation following the curve [64,70,71]. Screens, often honeycomb-
19
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
shaped or similar, are used to reduce eddies and overall vorticity (swirling) in the flow
prior to the test section [70,71].
2.3.3. SCALE MODELS IN MINE VENTILATION
Mines are very complex and irregular systems, so scale models must be applied
with relative caution to mines as opposed to aircraft or automobile design. Unlike the
geometry of an aircraft or production automobile, the geometry of a mine airway is less
regular and much less repeatable; no two shafts are bound to have the same dimensions
and features and no set of mine entries, excavated via blasting or via continuous mining
from natural ground, can ever be accurately considered identical. In this sense for the
purpose of modeling airflow, a scale model of an entire mine ventilation network could
never be considered as accurate as, for example, a scale model of the Space Shuttle.
With these irregularities in mind, scale models are applied in mining in a much more
general sense than they would be in aeronautical or naval engineering. Scale models
have been employed by mine ventilation engineers to analyze relatively specific systems
common to mine ventilation networks, such as shafts and working headings [72-83].
Scale models have been used by mine engineers to study shaft resistances, face
ventilation, auxiliary ventilation, and gob ventilation. One early study applying scale
models in mine ventilation focused on shaft resistances. Gregory noted that shafts are
often made to perform the roles of both ingress/egress and production and simultaneously
used as part of the ventilation system [72]. Gregory held that poor aerodynamic design
represented significant annual cost in the form of ventilation power requirements. He
used a smoke tunnel to study the aerodynamic drag on various shapes of individual
buntons (horizontal shaft dividers). Gregory also created a 20-ft long 1:12 scale model of
a representative section of shaft to measure the resistance resulting from various bunton
configurations. Gregory also stated that given the range of Reynolds numbers over which
the wind tunnel operated (from 0.3×106 to 1.0×106) and given that the maximum flow
speed was 80 ft/sec, the requirement of dynamic similarity did not apply to the model
20
|
Virginia Tech
|
John R. R. Bowling Ch. 2 Literature Review
since the pressure loss coefficient for which was tested is independent of Reynolds
number.
Stein et al. with the Bureau of Mines used a 1:10 scale model of an auger-type
continuous miner in a low coal seam to qualitatively investigate the effects of auxiliary
ventilation on dust dispersion near the mining face [74]. Breslin and Strazisar with the
Bureau of Mines performed a series of studies using scaled models of mining machines in
coal mines [73]. These studies used 1:5 scale models of a drum-type continuous miner
and a double-drum longwall shearer and a 1:6 scale model of a twin-borer-type mining
machine and made use of methane as tracer gas to model the dispersion of dust around
the mining machines.
Gillies also used a scale model to investigate coal face ventilation in the
laboratory [75,76], using colored smoke and bits of string to visualize airflow patterns in
a 1:10 scale model of a working heading with a continuous miner. Gillies repeated the
need for geometric and dynamic similarity but accepted a difference in the Reynolds
number between the model and the prototype mine heading, stating that there was little
difference in flow patterns between the model and the prototype. Tien similarly used a
1:10 scale model to study airflow in a continuous miner heading, this time while
developing a crosscut [77]. One ½-hp fan was used to provide primary ventilation while
another ½-hp fan was used to simulate the scrubber unit on the continuous miner. The
model was fitted with a regular grid of holes through the rib, at several heights and
distances from the working face, into which Pitot-static tubes could be inserted at various
depths to yield a regular three-dimensional array of sample points throughout the model.
Uchino and Inoue used both actual and reduced-scale models to study the airflow
patterns resulting from forcing auxiliary ventilation in a coal mine entry [78]. The
roughly 1:16 scale model used water as a working fluid and used laser light illumination
to qualitatively visualize flow patterns in the heading. Konduri et al. also used both
actual and reduced-scale models to study airflow resulting from jet fans for face
ventilation [79-81]. Konduri et al. also used the Reynolds criterion for scaling and used
21
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
3. DESIGN OF EXPERIMENTAL APPARATUS
3.1. DESIGN CRITERIA AND CONSTRAINTS
Once candidate tracer gases have been identified, an experimental mine simulator
will be useful for performing experiments to verify the function of the gases as tracers in
a ventilation network. The experimental apparatus will be useful to simulate tracer gas
surveys in a mine ventilation network. Such a model would allow for the injection and
sampling of tracer gases as would be performed in an actual mine ventilation survey. The
model would also allow for the measurement of air velocities at the points of injection
and sampling, as such data would also be measured in an actual mine ventilation survey.
Recalling Eqs. (2.12-14), atmospheric conditions (temperature, relative humidity and
barometric pressure) provide a means for improving the accuracy of the velocity
measurements using Pitot-static tubes. Additionally, these atmospheric conditions and air
velocities will be valuable data for building a numerical model for application to actual
mines, which is part of the overall project goal.
Designing, preparing and performing experiments in underground mines requires
a great deal of planning and, from Virginia Tech, a fair amount of travel, such that
regular visits to an underground mine for simple tracer gas experiments would be
exceedingly difficult, if not time-prohibitive. The ability to perform experiments in a
local laboratory setting would be far more convenient for researchers investigating these
new tracer gases. An experimental apparatus for laboratory-scale testing of multiple
tracer gases was deemed necessary for both convenient access to a test area and
controllable conditions for experiments. In addition to the convenience of a local
laboratory in which experiments would be conducted, the conditions within a laboratory
are more conducive to initial investigation of novel tracer gases. Controlled conditions
would allow for more accurate quantification of concentrations emitted and sampled,
more stable and directly controllable atmospheric conditions, and control or at least far
better knowledge of potential contaminants within the atmosphere. Controlled conditions
would allow experiments in the laboratory to be conducted with better accuracy and
simpler procedures limiting the introduction of error from unknown sources.
23
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
The procedures involved in tracer gas analysis are complex and the accuracy and
precision of any gas analysis results are highly dependent on the technique of the
analysts. Prior to field trials, which, as previously stated, require a great deal of
preparation, travel and cost, researchers should have developed the proper technique to
confidently expect repeatable results from their analyses. Through the performance of
many tests on the experimental apparatus, located in close proximity to the GCs,
researchers can practice routine sampling and analysis and develop a consistent
sampling/analysis methodology prior to conducting field trials.
Several criteria for this tracer gas study preclude sole reliance on a CFD model.
The first and most important reason for a physical model is for developing the technique
of gas sampling and analysis. Recalling from the literature review, CFD has some
shortcomings as a modeling tool. Boundary conditions should be well-understood prior
to the creation of a model. CFD also has difficulties modeling turbulence, which is
extremely important in the description of the dispersion of tracer gases. Furthermore, the
diffusion and dispersion of tracer gases are controlled by the molecular weights and other
properties of the gases. These properties may be ill-defined for gases which have not
been used extensively as tracers and may be not be accurately reflected in the
assumptions included in the CFD program.
3.1.1. EXPERIMENTAL GOALS
Given the expectations of the laboratory experimental apparatus, a simple list of
necessary criteria can be laid out for the apparatus. The experimental apparatus for tracer
gas investigation should
1) simulate a mine in a tabular deposit,
2) allow for the injection and sampling of appropriate amounts of tracer gases,
3) simulate changes in ventilation (as after a mine disaster) by incorporating
simple variable ventilation controls, and
24
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
4) allow for the measurement and monitoring of air velocities (quantities) and
atmospheric conditions (temperature, relative humidity and barometric
pressure) within the apparatus.
This list of criteria should be considered the goals of the apparatus; they represent the
criteria which must be met in order for the apparatus to be considered a sufficient
model mine for the purposes of tracer gas testing.
3.1.2. DESIGN CONSTRAINTS
A number of constraints were placed on the design of this experimental apparatus.
The apparatus should simulate a mine in a tabular deposit topologically and by
maintaining fully-developed turbulent flow throughout the apparatus. The apparatus
should have ports for the injection and sampling of tracer gases in the system. The
ventilation system should feature an appropriately-sized fan and functioning ventilation
controls. The ventilation system must also be airtight and open-circuit so that tracer
gases can be exhausted from the laboratory atmosphere rather than building up within the
laboratory gaseous background and obscuring concentration results. The design should
also be able to simulate various mine topologies in order to perform experiments on a
wide variety of mine layouts. In order to simulate many mine layouts, the experimental
apparatus would need to be altered. To limit the amount of effort associated with altering
the experimental apparatus, the apparatus should be modular in construction so that it can
be changed quickly and with little reconstruction of permanent fixtures.
3.2. EVALUATION OF DESIGN ALTERNATIVES
3.2.1. ALTERNATIVE DESIGNS
Given the four goals and additional constraints defined for the experimental
apparatus, a number of alternative designs were evaluated based on their ability to meet
those requirements. With focus on the geometry of a mine in a tabular deposit, various
design geometries were considered. The geometries under consideration ranged from
25
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
completely dimensionally representative of the prototype mine (most geometrically
similar to the prototype mine) to more loosely representative (less geometrically similar).
The first design considered was an entirely flat grid which would be most
representative of a mine in a tabular deposit. For this design, a flat plate of material such
as Plexiglas would be used for the top and bottom of the deposit (the upper and lower
bounds of the mine entries) and rectangular inserts would make up the walls. Entire
sections of a mine could be modeled this way while maintaining geometric similarity to
the prototype mine. The plates bounding the top and bottom of the model could be
notched or grooved in a grid such that the walls could be easily inserted and removed.
Holes could be drilled through the top bounding plate through which velocity
measurements could be made and tracer gases could be injected or sampled.
The flat grid design has a number of disadvantages, however. The flat design
would necessarily require a large footprint, especially if constructed horizontally. The
entire model could be used vertically near a wall or horizontally in an area near the
ceiling. Even if the mining height in the model was 0.5 in., the expanse of a scale model
representing an operating longwall mine would be at least several to tens of feet in each
direction. These dimensions would preclude vertical orientation (as the ceiling height is
less than ten feet) and horizontal orientation near the ceiling (where the largest footprint
is available) would leave the model very inaccessible. Additionally, sealing the apparatus
in a way which renders it airtight could be difficult, especially in separating the airways
from one another. Though some leakage between parallel airways would be
representative of an actual mine ventilation network, controlling internal leakages in the
apparatus could prove very difficult. Such a model, though modular, would not be very
simple.
A simpler version of this full-scale model could be constructed in which the
airways are based on a simpler network topology. A simplified model between two flat
plates would have the same advantages as the larger scale model except for its good
geometric similarity. A simpler model would use less material and could remain
26
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
modular, with its parts able to be moved fairly simply. Such a model would still be fairly
difficult to seal, however, with many loose parts needing to line up precisely or sealed to
be made airtight. Having to seal many dividers serving as walls would make the
disassembly and reassembly of the apparatus take more difficult and require more time,
but may be necessary to keep the model airtight.
Furthermore, in a flat model, regardless of scale, ventilation controls would all be
uniquely constructed and more difficult to seal and maintain. Placing variable ventilation
controls such as a regulators or stoppings which can simulate damage will likely require
making openings in one of the flat plates bounding the apparatus, rendering it harder still
to seal. Ventilation controls for a flat model would need to be uniquely constructed and
as such would be difficult to maintain.
Relaxing the constraint of the apparatus being entirely planar allows for more
options for the apparatus. If the vertical displacements are kept to a minimum (as they
would have to be in a laboratory with vertical extent of less than ten feet), some extension
into the vertical plane would allow sections of the airways to overlap one another,
allowing for more airway length with a smaller footprint. Though this design could carry
on a similar planar form with multiple levels, the planar concept could be discarded
altogether while maintaining topological fidelity to a prototype mine. In this sense the
geometric constraints could be relaxed until it could be represented by a network of
discrete airways. Recall that mine ventilation networks are often, for the purpose of
modeling and analysis, represented by a network of discrete airways. Such a network of
discrete airways is frequently constructed in HVAC as a duct network.
A duct network could represent a simplified mine geometry by being
topologically identical while allowing for some geometric variation, particularly by
relaxing the constraint that the model be entirely coplanar. Duct networks are easily
rendered airtight and, depending upon the materials used, ventilation controls such as
dampers or valves are ubiquitous for ducts of most sizes and shapes. This would mean
that individual ventilation controls need not be uniquely fabricated and then made
27
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
airtight. Ducts are available in many materials and cross-sectional shapes and sizes.
Ducts are easily integrated with fans and can be easily assembled and disassembled.
The disadvantage of using a standard home or commercial duct system is that the
ducts have a relatively large cross-section (101~102 in.2). Creating a network from
typical home ventilation ducting with multiple branches that is long enough to represent a
mine ventilation system could require a significantly large volume, even if tightly wound
upon itself. Also most available duct material does not have the same cross-sectional
shape as a coal mine entry: some ducts are rectangular, some are square and some are
circular. Pipe could also be used instead of circular duct; pipe would have approximately
the same advantages and disadvantages as ventilation duct but is available in a generally
smaller cross-section.
A summary of the design alternatives is provided in Table 3-1. The Planar, to
scale option represents the most geometrically similar model to a working mine in a
tabular deposit but in turn would be the most difficult to work with. The Simplified pipe
network represents the design with the geometric similarities most relaxed but which is
the simplest and easiest to work with and make airtight. Between those two options are
design options representing two inner points on the spectrum, the Planar, simplified and
Simplified duct network. Keeping in mind the overall goal of the apparatus serving in
tracer gas experiments, the topological fidelity of the model to the prototype should take
precedence over the geometric similarity. It is most important that the tracer profiles
created in the apparatus be representative of those resulting from actual mine tracer gas
surveys. Furthermore, the ability to be made airtight with relative ease is very important
in the laboratory. Having tracer gases leak slowly from the apparatus will negatively
affect all the tracer gas experiments conducted in the lab. Finally, the ability to
potentially make use of off-the-shelf parts makes the Simplified duct network or
Simplified pipe network options the most preferable. In choosing duct or pipe, the
many options for materials should be evaluated.
28
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
TABLE 3-1: COMPARISON OF DESIGN ALTERNATIVES
Model Type Advantages Disadvantages
Full mine to scale too large to fit in
Geometrically similar
lab
Planar, Parallel airway leakages easily
Difficult to make airtight
to scale included
Ventilation controls difficult to make
Modular
and seal
Airway shape similar Difficult to make airtight
Planar, Parallel airway leakages easily Ventilation controls difficult to make
simplified included and seal
Modular Difficult to access
Modular parts/ventilation controls
Simplified Airway geometry not identical to
available off-the-shelf
duct mine geometry
Easy to make airtight
network Most duct too large
Simple to work with
Modular parts/ventilation controls
Simplified Airway geometry not identical to
available off-the-shelf
pipe mine geometry
Easy to make airtight
network Metal pipe could be most expensive
Simple to work with
3.2.2. MATERIALS SELECTION
As mentioned along with the design alternatives, various materials were evaluated
for their appropriateness for constructing the apparatus. Since a network of discrete
airways was decided to be preferable to a modular flat grid, sections of duct or pipe
would be appropriate materials. Duct and pipe are available in many different sizes
(diameters) and cross-sectional shapes. The materials evaluated were galvanized steel
duct, aluminum duct, Plexiglas, copper tubing, steel tubing, and PVC pipe. Given that
the apparatus was designed to be disassembled and reassembled, the effort in achieving
an airtight seal and the availability of connections or flow controls off-the-shelf were
important factors. Cost of materials and components was also considered.
Galvanized duct is available with either a rectangular or round cross-section,
which could approximate the rectangular cross-section of a mine airway. The duct could
also be reshaped to better represent mine airway dimensions, but would then be more
difficult to maintain as airtight. The galvanized duct is simple to connect with other
sections and tees and wyes are available off-the-shelf, so long as the duct is not reshaped.
29
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
Dampers manufactured for galvanized duct are available off-the-shelf but would need to
be uniquely constructed if the duct was reshaped. Flow controls for home HVAC also
tend to allow some leakage, so stoppings inside the apparatus may not meet the same
resistance requirements as controls in a mine. Galvanized duct for home HVAC also
tends, as previously mentioned, to have a relatively large cross-sectional area. This
suggests that building a model with a length-to-diameter ratio approximately matching
that of a mine would require a great deal of large-diameter duct, which would require a
large volume of space to build.
Aluminum duct would have approximately the same characteristics as galvanized
steel duct. Though aluminum duct is available in most of the same dimensions as
galvanized duct, it may be available in a slightly different array of possible shapes and
sizes. Aluminum duct would be somewhat lighter and easier to work with, but likely a
bit more expensive.
Plexiglas or other acrylic materials are available in round and square tubes as well
as sheets, from which tubes of any cross-section could be fabricated. Plexiglas and
acrylic tubing could be easily cut to length and easily attached to other pieces and made
airtight. Most connections and ventilation controls for a Plexiglas system would need to
be uniquely fabricated, which detracts from the modularity of a design using Plexiglas
components. Plexiglas is lightweight and easy to work with and offers the benefit of
being transparent, which would make it easier to understand and explain the way the
model works.
PVC pipe is available in round and, to a limited extent, square cross-sections.
PVC pipe is the most inexpensive material investigated and is easily available off-the-
shelf. An extensive range of fittings for round pipe from reducers to wyes and tees and
several various valves are also readily available off-the-shelf. PVC is also lightweight
and very easy to assemble and to render airtight. PVC is very simple to cut to length
though it is not very flexible.
30
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
Copper pipe is available in round cross-section with relatively smaller diameter.
Copper pipe has fittings available off-the-shelf, including a wide variety of valves.
Copper pipe is more difficult to work with than the other materials and requires either
soldering or special fittings to create airtight connections. With the difficulty in making
connections, copper pipe is not conducive to a modular design. Copper is also the most
expensive of the materials and the simple no-solder connectors are very expensive.
Steel tubing is available in round, square or rectangular cross-section in various
diameters. Steel tubing is heavy and difficult to work with and also fairly expensive.
Steel tubing is similar to the HVAC duct but is thicker and not made to be connected
without welding or bolting together. Although some connectors may be available for
steel tubing, their availability and cost depend highly on the dimensions used.
Ventilation controls from ducts could be useful with steel tubing, but are less likely to fit
exactly and more difficult to make airtight.
A comparison of the attributes of various materials is included as Table 3-2.
Copper pipe and steel tubing were eliminated based on cost and difficulty to work with.
The ducts were eliminated based on the volume they were likely to occupy. Plexiglas
would have proved more difficult fabricate, especially to fit with airtight ventilation
controls. PVC pipe (2-in. diameter) was selected as the material from which the model
would be constructed. There are many benefits to PVC pipe over other materials: PVC is
considerably cheaper than aluminum or steel tubing, simple to handle due to its light
weight, simple to cut to desired length and simple to connect via couplings, elbows and
tees. A wide variety of valves, reducers and other fittings are available in local hardware
and plumbing stores and are inexpensive enough to maintain a supply of fittings in the
lab. PVC can withstand the relatively small pressures applied by the simple laboratory
ventilation system and can easily be made airtight by applying sealant, caulk or duct tape.
Sealed in such a way, PVC pipe is also simple to separate, clean, and reassemble for
rearranging the geometry of the model. For these reasons, PVC makes an excellent
material for the tracer gas test apparatus.
31
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
TABLE 3-2: COMPARISON OF MATERIAL ALTERNATIVES
Cross- Difficulty in
Material Section Cost Assembly Ventilation Controls
Galvanized rectangular, some available,
medium low
steel duct round hard to make airtight
Aluminum rectangular, some available,
medium low
duct round hard to make airtight
square,
Plexiglas/ must be uniquely fabricated,
round, medium medium
acrylic hard to make airtight
(sheet)
round, wide variety available off-the-
PVC pipe low medium
(square) shelf
Copper wide variety available off-the-
round high high
pipe shelf
round,
Steel could use duct controls,
square, high high
tubing difficult to make fit/airtight
rectangular
Additional concerns arise for the materials selection to include appropriateness for
use in the apparatus with tracer gases. Two material properties which affect how tracer
gases behave in the apparatus are the surface roughness (ability to create and sustain
highly turbulent flow) and the adsorption characteristics of the gas and material. The first
property, surface roughness, represents only a minor hurdle in the design. The flow
regime in the apparatus is affected by the roughness of the walls but also controlled by
the Reynolds number. As was demonstrated in the discussion of similitude, turbulent
flow is created and maintained throughout the apparatus.
The second materials property, adsorption characteristics, is an interfacial
property that is controlled by the physical and chemical properties of both the gas and the
solid surface. Recall that initial results from early tracer gas experiments did not show
any significant adsorption of SF onto mine surfaces [1]. Recall also, that adsorption is
6
always observed to occur between any gas-surface pair [7]. No significant measurement
of adsorption of SF has been demonstrated in laboratory results, though there is some
6
anecdotal evidence that SF adsorbs onto PVC, glass and plastics including chemical
6
containers and syringes [10].
32
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
3.3. LOCATION AND ORIENTATION OF APPARATUS
The space available for the construction of the apparatus, in the ventilation
laboratory at the Mining and Minerals Engineering Laboratory on Plantation Road,
provided some spatial constraints. While the size of the room is certainly large (19 ft ×
48 ft), there is a roughly 4-ft-high, 4-ft-deep shelf along three sides of the room, leaving
significantly less floor space in the center of the room for equipment and experiments.
Figure 3.1 shows the floorplan of the ventilation lab from architectural drawings. In
addition to the space at one end set aside as a classroom, the lab also contains two
workbenches atop which the gas chromatographs (GC) and a PC rest, an area adjacent to
the GCs for gas cylinder storage, two workbenches for experiments, a desk for
manometers and tools, and four permanent wind tunnel experiments set up in the space,
each with a fan (either centrifugal or axial-vane) with an electric motor and a length of
duct.
FIGURE 3.1. FLOORPLAN OF VENTILATION LABORATORY
With limited remaining floor space in the laboratory, a vertical orientation was
chosen for the apparatus layout in order to minimize the footprint of the apparatus. The
resulting configuration appears as dual, vertical stacks of tubes, connected continuously
and bolted to the frame of the large wind tunnel. The layout of the new tracer gas test
apparatus with respect to the other existing wind tunnels is shown in Figure 3.2. The
proximity of all the equipment in the laboratory equipment is apparent, as one wind
33
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
mine ventilation networks are most frequently modeled as networks of resistances rather
than individual airways; rarely is very accurate modeling of individual airways of
significant importance to the numerical solution of the entire network. Given the nature
of mine ventilation surveys, in which air quantities are accounted for via a network
balance, fidelity to the topology of a mine should be the primary priority of the criteria
for accurate representation of a mine. The similitude requirements represent a secondary
priority for assessing the validity of the model. The significance of the geometric and
dynamic similarity requirements is primarily in demonstrating that similar flow
conditions are present in the model as occur in an actual mine; those flow conditions are
fully-developed highly turbulent conduit flow. For this reason and for the sake of
thoroughness are similitude and entrance length considered conditions for valid
representation of the mine by the model.
3.4.1. TOPOLOGICAL FIDELITY
A typical longwall coal mine ventilation system was chosen as the basis for the
first simple mine model. A simplified ventilation network with one active panel and one
gob panel was selected for the model and the model was constructed from 2-in.-diameter
PVC pipe. Figure 3.3 shows a layout of the mine used as a basis for the experimental
model. Figure 3.4 is a schematic of the model as constructed in the laboratory. The
figures are color-coded to highlight the topological similarity between the simplified
mine geometry and the physical model. For the purpose of simplicity, only flow around
the gob panel was modeled, rather than flow through the gob panel. Modeling flow
through the gob panel would add a number of complications, including considering a
porous medium which is representative of a longwall gob on the model scale and
accounting for the significant pressure drop across such a medium and the potential
adsorption of the tracer gases onto the porous medium.
35
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
Intake
Stopping
Regulator Regulator
(Valve 1)
(Valve 4) (Valve 2)
Sample Point 4
Active
Gob Panel
Roof Fall
(Valve 3)
Rescue Borehole
(Sample Point 2)
Existing Explosion
Borehole Damages
(Sample (Valve 5)
Point 3)
FIGURE 3.3. SIMPLIFIED MINE GEOMETRY USED AS BASIS FOR MINE MODEL
Front
Sample/Injection Point 1
Inlet
A
B Valve 2: Regulator
Sample/Injection Point 2
Valve 3: Roof Fall
Back
Sample/Injection Point 4 Valve 1: Stoppings in MainsA
Valve 4: Regulator
B Sample/Injection Point 3
Valve 5: Explosion Damages
Fan
Key
-1 0 1 2 5
Airflow Direction Mains Active Panel Gob Area
Exhaust SCALE (IN FEET)
FIGURE 3.4. MINE MODEL SCHEMATIC WITH COLORS CORRESPONDING TO SIMPLIFIED MINE GEOMETRY
It is important to recognize how the model and simplified mine geometry are
topologically identical. Both feature parallel intake and exhaust entries (mains), which
are connected via a normally close valve representing stoppings in the main entries. An
actual coal mine which is mined via room and pillar mining has several parallel entries
making up the mains, of which some are used for the transportation of intake (fresh) air
and some are used for the transportation of return (exhaust) air. Those entries which
conduct air of the same type in the same direction (either all of the intakes or all of the
returns) are connected such that for the purpose of simplicity, they can be modeled
36
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
collectively as one branch in the network. There is a very small but usually realistically
non-negligible leakage from the higher-pressure intakes to the lower-pressure returns,
which underground takes place across hundreds of parallel stoppings separating the
intake and return airways. Topologically, and again for simplicity, these are collectively
represented by one branch of variable resistance (a valve). In the simplest case, this
valve could be completely closed, though it could be slightly open to simulate the small
leakage quantity across the stoppings. In the case simulating the destruction of some
stoppings due to high air overpressure resulting from an explosion, the valve could be
opened to simulate the loss of some stoppings. The resulting network topology, with a
much-lower resistance connection effectively short-circuiting airflow between the main
intakes and returns, is shown in Figure 3.5.
Airflow
Intake
Stopping
(opened)
Exhaust
Active
Gob Panel
FIGURE 3.5. MINE MODEL GEOMETRY AND AIRFLOW RESULTING FROM LOSS OF STOPPINGS ALONG MAINS
Both the working panel and the gob panel are represented by independent loops
which run parallel to the main intake/exhaust series. For each panel, the branch
representing the panel is fitted with a regulator, as is the branch on the mains parallel to
that panel. These resistances should be adjusted to achieve a representative quantity split
such as would be present in an actual mine ventilation network. In a mine, the desired
quantity split would be controlled similarly by adjusting the resistance of the working or
gob panel split via regulators. The valves in these panels can also be changed to reflect
the conditions resulting from a disaster, such as an increase in resistance in an airway
37
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
resulting from a roof fall. To represent such an increase in resistance, the valve on a
particular airway need only be closed more significantly.
Locations for injection and/or sampling of tracer gases were selected in the intake
airway, in the exhaust airway, in the active mining panel and in the (entries around the)
gob panel. The injection of tracer gas could take place within the intake entry or at the
top of the intake shaft at an actual mine. Similarly, the sampling of air for tracer gases
could take place within a return entry or at the exhaust shaft at an actual mine. Air
sampling at the intake might take place for purposes of establishing a background
concentration, but otherwise sampling at the intake or injecting into the return would
serve no purpose for a tracer gas ventilation survey. Within the active mining and gob
panels, gases would most likely be injected (but could, for some purposes, be sampled).
The injection or sampling in the active panel would take place along or near the face
under normal conditions or possibly through a rescue borehole following a disaster after
which the mine personnel have been evacuated. Sampling or injection around the gob
panel could take place within the entries exhausting the gob or via an existing gob gas
monitoring borehole, if such boreholes exist at the particular mine.
3.4.2. SIMILITUDE IN THE APPARATUS
Recognizing the importance of similitude in engineering models, similitude
between the model and a typical longwall mine was studied. Recalling the pi terms for
similarity, the geometric scale should be considered along with the Reynolds number.
Examining a typical coal mine entry, assuming airway dimensions of 5 ft high by 20 ft
wide, the area is given as
( ) ( )
Eq. (3.1)
and the hydraulic diameter, from Eq. (2.10), is given as
( )
( )
Eq. (3.2)
.
38
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
An estimate of the length-to-diameter ( ⁄ ) ratio of an underground mine, with a total
airway length of several miles, can be found as
( ) ( )
( ) ( )
Eq. (3.3)
⁄ .
The result of Eq. (3.3) shows that in order to successfully model an underground mine,
the diameter ( ⁄ ) ratio of the models‟ airways should be very high, perhaps on the
order of 103. Literature estimates Reynolds numbers in mine airways as generally
exceeding 104 and stresses that fully turbulent airflow nearly always prevails in mine
airways [60,61].
PVC pipe of 2-in. diameter was used in an exploratory model to investigate the
viability of a centrifugal fan in the lab. Using 2-in. diameter pipe, in order to maintain an
acceptable degree of similitude, the total length of pipe needed for the model can be
calculated by
( )( ) ( )( ) Eq. (3.4)
.
This result shows that for a model using 2-in. diameter pipe, ~102 ft (hundreds of feet) of
pipe can represent 105 ft (miles) of mine airway. Given that the model currently contains
approximately 160 ft of pipe, the model is sufficiently geometrically similar to a mine
airway a few miles in length.
Examining the air velocities in the model, the Reynolds number of the flow can
be calculated. Given that average velocities in the model were measured to be
approximately 8 m/s (26 ft/sec), the Reynolds number can be found by applying Eq. (2.9)
( )( ⁄ )
Eq. (3.5)
( )
Using the dynamic similarity criterion
39
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
Eq. (3.6)
it can be concluded that the model Reynolds number of is
representative of the flow regime in a mine airway. Furthermore, given that
, Eq. (3.7)
the flow in the model airway can be considered within the fully-developed turbulent
regime. Maintaining turbulent flow throughout the model is important to insure that
tracer gases disperse fully in the airstream so that measured gas concentrations are
representative of the whole airstream.
It should be noted that the geometric similarity of airway cross-sectional
dimensions ( ⁄ ) is not conserved in the model; the model airway is circular while the
mine entry is rectangular in cross-section. In this sense, the model cannot be considered
an entirely true model by definition. However, given that the ratio ( ⁄ ) is somewhat
variable among mines with different mining heights and entry widths, this constraint
should be considered far less important than the vital constraints of the ( ⁄ ) ratio and
the Reynolds number. Based on similarity between these two values, the experimental
model should be considered a nearly true model, distorted only in the cross-sectional
geometric ratio.
It is not the goal of this model to perfectly simulate mine airflow in any given
section of the model or mine. It should be considered sufficient that the model
topologically represents a realistic mine and that turbulent flow is maintained throughout
the model because the primary interests are in the tracer profiles and ensuring mixing that
is similar to mine conditions. A detailed CFD model of the apparatus was made to help
predict tracer gas profiles under various ventilation control conditions. Initial
experiments using tracer gases, along with the CFD model showing turbulent flow in the
apparatus, demonstrated tracer concentrations which reflect complete diffusion of the
tracer in the airway [84].
40
|
Virginia Tech
|
John R. R. Bowling Ch. 3 Design of Experimental Apparatus
3.4.3. ENTRANCE LENGTH
Given that turbulent flow is necessary for the complete dispersion of tracer gases,
it is worth estimating the length of duct required before fully-developed duct flow
dominates. Recalling Eq. (2.11), the entrance length required for conduit flow to fully
develop can be estimated by
( ) ⁄ ( ) ⁄ . Eq. (3.8)
Although Eq. (3.8) estimates an entrance length of 126 diameters, recall that Lien, et al.
recommend a conservative estimate of 130-150 diameters for the development of fully-
turbulent flow [65]. Conservatively, turbulent flow can be assumed to be fully-developed
at a distance from the inlet of
( ⁄ ) . Eq. (3.9)
( )
Given that the inlet section is just over 20 ft long and followed by a section of turns, it
can therefore be assumed that so long as the velocity is such that turbulent flow is
maintained ( ), fully-turbulent flow dominates throughout the
rest of the apparatus.
41
|
Virginia Tech
|
John R. R. Bowling Ch. 4 Construction and Instrumentation
4. CONSTRUCTION AND INSTRUMENTATION
4.1. CONSTRUCTION
Recalling the design constraints and the four criteria comprising the experimental
goals, the apparatus was constructed such that it sufficiently satisfied the experimental
goals while maintaining the ability to be rearranged with relative ease. The ideal
constraints of maintaining simplicity and modularity were respected throughout the
construction. For the sake of simplicity, the fewest materials should be used to create the
simplest reasonable configuration. By keeping the apparatus modular, the fewest
materials should be wasted (via cutting, etc.) in order to disassemble the apparatus and
reassemble it in another configuration representing another mine geometry. The
connections should also be as simple as reasonably possible and require the least amount
of preparation to be constructed and later disassembled and reassembled.
4.1.1. STRUCTURE OF PIPE NETWORK
To meet the requirements for modular design and simple rearrangement, the
sections of pipe representing various airways should be mounted to a frame which allows
for both the easy removal and the secure mounting of the pipes sections. A frame of
slotted angle iron was selected, unto which the pipe sections could be secured using U-
bolts. Figure 4.1 shows the inlet end of the apparatus, in which the support frame and
mounting U-bolts are clearly visible. The support frame has more space for expansion to
build more complicated models in the future. In order to maintain modularity, an attempt
was made to maintain the 10-ft sections of pipe intact to minimize cutting and waste and
maximize their reusability regardless of what fittings were placed in line with the pipe
sections. In order to then compensate in the differences in lengths of pipe sections which
needed to be combined, sections of flexible shop vacuum hose, also visible in Figure 4.1,
were used to connect the pipe sections which did not line up precisely. Vacuum hose was
selected with an O.D. matching the O.D. of the PVC pipe so that the hose could be used
with the existing pipe fittings. Care was taken to fully seal the vacuum hose connections
using silicone since the ribbed surface seemed less likely to seal inside the pipe fittings.
42
|
Virginia Tech
|
John R. R. Bowling Ch. 4 Construction and Instrumentation
the sampling point. A sampling/injection port with adjacent Pitot-static tube is shown in
Figure 4.2.
Pitot-static tube, with sensing
Injection/sampling port,
tip inserted into duct
sealed with septum
FIGURE 4.2. SECTION OF TEST APPARATUS SHOWING INJECTION/SAMPLING PORT AND PITOT-STATIC
TUBE (SIDE VIEW)
When mounting a Pitot-static tube, the orientation and position of the sensing tip
in the duct are important factors which have a result on the accuracy of the velocity
measurement. Since the insertion depth of the Pitot-static tube (6 in.) is far longer than
the radius of the pipe (1 in.), most of the Pitot-static tube (~5 in.) protrudes outside the
pipe and must therefore be supported. The Pitot-static tube is arranged such that the
sensing tip is in the center of the airway, so the tube must be secured with respect to the
insertion depth into the pipe. The velocity measurement is also sensitive to the yaw of
the Pitot-static tube (the angle the sensing tip makes with airflow direction), so the tube
must be secured against rotation. The brace for the Pitot-static tube is a section of ½-in.-
diameter PVC pipe split longitudinally and cemented to a section cut from 2-in.-diamter
PVC pipe. Between the Pitot-static tube brace and the pipe making up the model airway,
a section of rubber gasket is used to form an airtight seal around the hole through which
the Pitot-static tube protrudes. Hose clamps are used to hold the Pitot-static tube brace to
44
|
Virginia Tech
|
John R. R. Bowling Ch. 4 Construction and Instrumentation
the model pipe and serve to compress the gasket, which forms a tighter seal around the
Pitot-static tube. The long stem of the Pitot-static tube is attached to the brace with cable
ties. A disassembled cross-sectional diagram of the Pitot-static tube, gasket and brace is
provided as Figure 4.3.
Pitot-static tube
Brace
1/2" PVC pipe
cut longitudinally
Rubber gasket
Airway (pipe)
FIGURE 4.3. DIAGRAM OF PITOT-STATIC TUBE BRACE
(DISASSEMBLED, FRONT VIEW)
4.1.3. OPEN-CIRCUIT VENTILATION OF THE APPARATUS
As stated previously, one of the goals of the experimental apparatus is that it be
open-circuit and exhaust to an area outside the ventilation lab. This is necessary so that
tracer gases do not build up in the gaseous background within the lab and complicate or
add error to tracer gas sampling over time. The ventilation system selected for the
apparatus is an exhausting system for two reasons. First, the exhausting system is most
common in coal mines, of which the test apparatus should be representative. Second, by
being set up as an exhausting system, the ventilation system of the apparatus is
maintained as a negative-pressure system. By maintaining a negative pressure system,
any leaks in the system would result in lab air leaking into the system, rather than air
containing tracer gases leaking from the apparatus into the lab air.
As part of the initial construction of the apparatus, a small centrifugal fan was
used was used to provide the ventilation. This fan was fairly small and light and, as such,
45
|
Virginia Tech
|
John R. R. Bowling Ch. 4 Construction and Instrumentation
could be placed directly in the fume hood. Figure 4.4 shows the original centrifugal fan
operating in the fume hood. Although raised for clarity in the photograph, the sash would
be lowered while using tracer gases in the experimental apparatus to reduce the
possibility of tracer gas diffusing into the laboratory atmosphere.
FIGURE 4.4. SMALL CENTRIFUGAL FAN PLACED DIRECTLY IN FUME HOOD
4.2. INSTRUMENTATION
Since reliable measurements for air velocity and air density were desired,
instrumentation would be used for the apparatus. The instrumentation would be
electronic and automated such that air velocities throughout the mine model and
atmospheric conditions could be monitored and recorded while researchers were
performing other tasks, such as coordinating injection and sampling of the tracer gases.
Given that a small cross-sectional area would be used for the airways to simulate the very
large ⁄ ratio in a mine, the instrumentation measuring velocities would need to be
fairly small in order to measure the velocity without seriously disturbing airflow. For this
46
|
Virginia Tech
|
John R. R. Bowling Ch. 4 Construction and Instrumentation
4.2.2. RELATIVE HUMIDITY, TEMPERATURE AND BAROMETRIC PRESSURE
In order to achieve more accurate velocity measurements using Pitot-static tubes,
the velocity calculations should be corrected for current air density. Three parameters are
needed to calculate air density in real time; these are temperature, barometric pressure
and relative humidity (or the partial pressure of water vapor). Most important in
calculating the air density are the air temperature and barometric pressure. The accuracy
of this can be increased incrementally by factoring in the contribution of water vapor to
air density. As far as sensing technology applies, temperature and relative humidity are
both commonly collected data for HVAC systems, while quality barometric pressure
transducers and generally comparably expensive. For the experimental apparatus, a
simple combination relative humidity (RH) and temperature sensor (RH/T sensor) was
purchased. This sensor package from Dwyer Instruments was available in a simple
plastic enclosure designed for wall-mounting, as shown in Figure 4.10. The ranges for
both temperature and relative humidity far exceed those likely to occur in the climate-
controlled lab, so a costly sensor package was deemed inappropriate. For barometric
pressure measurement, a PX-409 barometric pressure transducer configured for 24V DC
excitation, 16-32 in. Hg measurement range and 4-20mA analog output was purchased
from Omega Engineering. The barometric pressure transducer is shown in Figure 4.11.
FIGURE 4.10. RELATIVE HUMIDITY/
TEMPERATURE SENSOR (WALL-MOUNT HOUSING)
52
|
Virginia Tech
|
John R. R. Bowling Ch. 4 Construction and Instrumentation
FIGURE 4.11. BAROMETRIC PRESSURE TRANSDUCER
4.2.3. PROGRAMMABLE LOGIC CONTROLLER
As an interface between the PC and the sensors and fan control, a programmable
logic controller (PLC) was selected. A PLC was selected because it is modular, versatile,
and simple to program. A PLC can perform the multiple roles of data acquisition, digital
switching, and analog and digital control. The PLC selected is an Allen-Bradley
MicroLogix 1100, a compact logic controller with embedded analog and digital I/O
(relays), an embedded LCD screen error-reading and an embedded web server with the
capability for webhosting and online editing. The MicroLogix 1100 can be expanded
with up to four 1762-I/O Modules, of which four appropriate modules were also selected.
Many of the relevant technical specification are included in the section Appendix:
Selected Technical Specifications. Three analog input modules were selected to monitor
the sensor inputs. One analog output module was also selected for analog control of
future devices, such as variable speed fans. The PLC and expansion I/O modules are
shown as installed in Figure 4.12.
53
|
Virginia Tech
|
John R. R. Bowling Ch. 4 Construction and Instrumentation
in.-diameter PVC pipe was run from its outlet directly outside the building, maintaining
the airtightness required by the experimental goals.
FIGURE 4.16. A 1/2-HP BLOWER FAN WAS SELECTED TO
REPLACE THE SMALLER CENTRIFUGAL FAN
4.3.2. VARIABLE FREQUENCY DRIVE
The ½-hp-rated Baldor VS1MX10P5-2 VFD selected as the motor drive is shown
in Figure 4.17. The VFD transforms 115V 60 Hz single-phase AC power into 230V
three-phase AC power at a controllable frequency; the speed of an AC electric motor is
directly proportional to the drive frequency. This VFD allows the user to vary the air
velocity in the apparatus, which may be useful in experiments to set up specific
conditions. The VFD selected has both a relay control, which is included as a safety
interrupt and must be closed for the drive to run, and an analog input for speed. These
features allow the drive to be hardwired to an on/off switch and a speed control knob
(trim potentiometer) or, as with this apparatus, controlled by a logic controller. Using the
analog input, connected with the analog output module of the PLC, the user can set a
desired air velocity at a specific location. The fan speed could then be adjusted by the
PLC until the target velocity is reached.
59
|
Virginia Tech
|
John R. R. Bowling Ch. 4 Construction and Instrumentation
Re ( N∙s/m )
( kg/m )( m)
Eq. (4.2)
.
At Point 1, the measurement point in the inlet airway, the air velocities were
measured as a function of fan speed. These velocities, measured over what should be
considered the normal operating range of the fan, are displayed in Figure 4.18. Although
the fan could theoretically be operated at speeds approaching zero RPM, practical
limitations (primarily the current draw in the VFD) establish a lower limit. For this
reason, the practical operating range of the fan should always produce air velocities
above the critical air velocity of 1.28 m/s. This is the velocity in the inlet, so should the
air split into several airways, it can still be maintained above the critical velocity. A plot
of the fan speed as a function of the motor frequency is provided as Figure 4.19.
Operating the fan for extended periods outside its designed operating range can shorten
its life, so caution is taken keep the motor frequency below 120 Hz (twice the design
speed). Though the upper limit of the VFD output is 5000 Hz, operating the fan at much
higher speeds could tend to damage the fan more quickly and should be avoided. Using
the quantities, computed via Eq. (2.1) by applying the correction factor in Eq. (4.1), the
characteristic curve of the apparatus was also computed. The characteristic curve, which
shows the pressure drop over the apparatus as a function of the quantity forced through it,
is provided as Figure 4.20. Also plotted is the best-fit polynomial of power 2; recalling
Eq. (2.4), the head loss of an airway or network sum of airways is understood as a
function of the square of quantity. Using the characteristic curve, the equivalent
resistance of the apparatus can be calculated
( )
( )
Eq. (4.3)
.
61
|
Virginia Tech
|
John R. R. Bowling Ch. 5 Summary and Conclusions
5. SUMMARY AND CONCLUSIONS
5.1. SUMMARY OF PROJECT
An experimental apparatus was constructed which will be used to conduct tracer
gas experiments in the laboratory. The test apparatus is modular and simple and can be
easily rearranged to represent a variety of mine geometries. The apparatus is appropriate
for the use of tracer gases by being both airtight and open-circuit (exhausting to the
atmosphere) and by maintaining a turbulent flow regime throughout the model. The test
apparatus simulates a mine in a tabular deposit by being topologically identical to a
simple longwall mine and by demonstrating similitude with a typical longwall mine with
adherence to modeling theory (following the Buckingham pi theorem).
The model features ports for injection and sampling of tracer gases, which
represent boreholes in the mine prototype. The sampling/injection ports can be pierced
with a syringe or solid-phase microextraction (SPME) fiber so that many of the methods
for the release and sampling of tracer gases can be practiced in the laboratory, albeit on a
smaller scale and in a more controlled environment. Valves on the apparatus represent
ventilation controls, such as stoppings or regulators, or changing resistances in a mine,
such an increase in resistance due to a roof fall or a decrease in resistance due to
stoppings being destroyed. The relative resistances of airways can be changed by
changing the status of the valves to represent different states of the ventilation controls.
The model, which is simple and versatile, can be used to represent a working
mine ventilation system for the purpose of tracer gas sampling. A wide variety of mine
plans can be simulated by reconstructing the modular model into different geometries.
Once a particular model geometry has been established, various ventilation states should
be identified. By performing tracer gas tests in the model under “normal” ventilation
operating conditions, baseline tracer gas profiles can be constructed. Researchers should
then apply changes to the state of ventilation controls in the model by adjusting the
airway resistances using the valves. Performing tracer gas tests after changes have been
made to the ventilation network, the tracer gas profiles can be compared to infer the
64
|
Virginia Tech
|
John R. R. Bowling Ch. 5 Summary and Conclusions
nature of the changes from the normal ventilation conditions. The mine simulator
described in this paper is meant to help researchers develop a method for inferring
ventilation changes remotely using tracer gases.
5.2. CONCLUSIONS
Four criteria were determined which the apparatus should meet to be considered
successful for the purposes of tracer gas experimentation. The experimental apparatus
for tracer gas investigation should
1) simulate a mine in a tabular deposit,
2) allow for the injection and sampling of appropriate amounts of tracer gases,
3) simulate changes in ventilation (as after a mine disaster) by incorporating
simple variable ventilation controls, and
4) allow for the measurement and monitoring of air velocities (quantities) and
atmospheric conditions (temperature, relative humidity and barometric
pressure) within the apparatus.
The apparatus meets the goals, as described above, via the following.
1) The apparatus is topologically identical to and demonstrates similitude with a
simplified longwall coal mine. Turbulent flow is maintained in the apparatus.
2) The apparatus has ports through which chemical tracer gases can be injected
and sampled while maintaining an airtight seal to prevent leaks into the
laboratory atmosphere. The apparatus allows for precise measurement of
tracer gases, as shown in initial experiments [84].
3) The apparatus contains several valves at key locations which can be used as
ventilation controls. They can be opened or closed to represent the type of
damage which might occur in a mine disaster, such as a roof fall or explosion.
4) The apparatus is instrumented with velocity and climate measurement devices.
Instruments measure temperature, relative humidity and barometric pressure
to compute air density and the air density is used to correct air velocity
measurements, made at each sampling/injection port.
65
|
Virginia Tech
|
John R. R. Bowling Ch. 5 Summary and Conclusions
5.3. FUTURE WORK
For the future use of the apparatus, some suggestions are presented for the
improvement of the apparatus as a research tool. One such improvement which would be
beneficial is the replacement of the opaque PVC pipe at the sampling/injection ports with
a transparent pipe, such as a pipe constructed from Lexan or a similar transparent plastic.
This pipe could also be fitted with a depth gauge for measuring the protrusion depth of
the Pitot-static tube and sampling instrument in the airway. The Pitot-static tube could
then be inserted at various depths into the airway and the velocity at each depth could be
measured in order to define the velocity profile across a section of the duct. The velocity
profile could be compared among sampling/injection ports to determine whether the flow
at each sampling/injection port section is fully-developed. Additionally, accurately
defining the velocity profile across the duct could also assist in determining the average
velocity in the airway as a function of the insertion depth of the Pitot-static tube, rather
than by applying a literature-value coefficient for velocity measured along the centerline.
The depth gauge, readable through the transparent pipe, would also assist in
reliably sampling from precisely the same point or points in the airflow. In an actual
mine airway, a mine ventilation engineer would establish a regular sampling pattern
likely consisting of sampling either the center of the airway or sampling over a specific
traverse of the airway. Similarly, the researcher sampling the air in the model could draw
samples from the same depth in the airway each time, with the aim of establishing a
similar protocol for repeatability.
In order to define a tracer gas profile from a slug test, several samples must be
drawn over the time in which the slug passes the sampling point. Plotting the
concentrations of those samples, a tracer gas concentration profile can be created. In a
slug test, a tracer gas profile under given ventilation conditions is observed and can be
compared with the profile under different conditions to draw conclusions about the
ventilation control changes. Currently, the residence time of a slug (fixed volume) of
tracer gas in the apparatus is too short for measurement using typical slug test techniques.
66
|
Virginia Tech
|
John R. R. Bowling Ch. 5 Summary and Conclusions
In its current form and with the operating velocities described, a tracer gas slug takes less
than five seconds to pass from the inlet to the exhaust of the apparatus. With a minimum
practical sampling frequency of approximately two seconds, the concentration cannot be
plotted with enough resolution to accurately define the tracer profile. In order to provide
a longer residence time for tracer gas slug tests, the apparatus should be extended in
length and the air velocity should be slowed (while still maintaining highly turbulent flow
conditions for good mixing).
Another upgrade which should be added to the apparatus is an electronic mass-
flow controller. More accurate measurements of the injection rate of SF and other
6
tracers could improve the accuracy of experimental results. The PLC should be used to
initiate the flow of gas into the apparatus so that its release rate and time can be precisely
controlled. The experimenter could also set a delay and a visual or audio cue to begin
sampling from the apparatus when the mass-flow controller begins injecting gas into the
apparatus.
Using this experimental mine simulator, researchers will seek to develop
standardized tracer gas release and sampling procedures with which multiple tracer gases
can be used simultaneously. If the method of tracer gas inference is proven successful in
laboratory studies, it is hoped that the method can be scaled up and applied to operating
mine ventilation networks. Following the development of standardized tracer gas release
and sampling procedures, tracer gas surveys should be conducted in actual mines before
and after changes are made to the ventilation network and researchers will try to correlate
changes in the tracer profiles to changes in the mine ventilation network. This technique
could ultimately allow mine engineers to conduct a tracer gas survey following a disaster
in which unknown changes occurred in the mine ventilation network and, by comparing
the tracer gas profiles to the baseline normal operating profiles, infer the nature of
changes in the ventilation network. This would allow for more rapid assessment of the
ventilation conditions following a disaster and, in the case of a rescue operation, help to
better inform mine rescue teams earlier about the status of the mine ventilation.
67
|
Virginia Tech
|
John R. R. Bowling References
[79] I.M. Konduri, M.J. McPherson, and E. Topuz, “Simulation of Induced Jet
Ventilation in a Mine Face Using a Laboratory Model,” Proceedings of the 7th US
Mine Ventilation Symposium, 1995, p. 6.
[80] I.M. Konduri, M.J. McPherson, and E. Topuz, “Model and field investigations of jet
fans for face ventilation,” SME Transactions Volume 302, 1997, p. 5.
[81] I.M. Konduri, M.J. McPherson, and E. Topuz, “Experimental and Numerical
Modeling of Jet Fans for Auxiliary Ventilation in Mines,” Proceedings of the 6th
International Mine Ventilation Symposium, 1997, pp. 505-510.
[82] A.D. Jones, S. Lowrie, and J. Edwards, “Aerodynamic Scale Model Simulations to
Investigate the Consequences of Changes in Mining Conditions for Gas Control,”
Proceedings of the 7th US Mine Ventilation Symposium, 1995, pp. 287-292.
[83] A.D. Jones, Z. Pokryszka, S. Lowrie, C. Tauziede, and P.-M. Dupond, “A Physical
Scale Model of Flows in the Waste of a Retreat Longwall Coalface,” Proceedings of
the 6th International Mine Ventilation Congress, 1997, pp. 231-236.
[84] G. Xu, J. Bowling, K. Luxbacher, and S. Ragab, “Computational Fluid Dynamics
Simulations and Experimental Validation of Tracer Gas Distribution in an
Experimental Underground Mine,” SME Annual Meeting 2011, 2011, pp. 1-5.
74
|
Virginia Tech
|
John R. R. Bowling Appendix: Selected Technical Specifications
TABLE A-5: FEATURES OF THE ALLEN-BRADLEY 1762-IF4 ANALOG INPUT MODULE
Parameter Value/Note
Dimensions 90mm × 87mm × 40mm (H×D×W)
Number of analog inputs 4 differential (bipolar)
Analog input type Voltage: -10 to +10V DC
Current: 4 to 20mA
Full scale analog ranges Voltage: -10.5 to +10.5V DC
Current: -21 to +21mA
Analog input resolution 15 bit (bipolar)
Repeatability ±0.1%
Typical overall accuracy ±0.3% FS (0 to +55°C)
±0.24% FS (at 25°C)
A/D converter type Successive approximation
Common mode voltage ±27 V
Common mode rejection >55 dB at 50 and 60 Hz
TABLE A-6: FEATURES OF THE ALLEN-BRADLEY 1762-OF4 ANALOG OUTPUT MODULE
Parameter Value/Note
Dimensions 90mm × 87mm × 40mm (H×D×W)
Number of analog outputs 4 single-ended (bipolar)
Analog output type Voltage: 0 to 10V DC
Current: 4 to 20mA
Full scale analog ranges Voltage: 0 to 10.5V DC
Current: 0 to 21mA
Analog output resolution 12 bit (unipolar)
Repeatability ±0.1%
Typical overall accuracy ±1% FS (0 to +55°C)
±0.5% FS (at 25°C)
D/A converter type R-2R ladder voltage switching
Module update time 2.5 ms
77
|
Virginia Tech
|
The Assessment of Sonic Waves and Tracer Gases as Non-Destructive Testing (NDT) Methods
for In-Situ Underground Mine Seals
Kyle T. Brashear
ABSTRACT
In 2006, two tragic mining incidents occurred in the United States, resulting in the loss of
life for 17 coal miners from explosions in underground coal mines. As a result, legislators passed
the MINER Act of 2006. In addition to the numerous new regulatory requirements, the strength
requirement of both monitored and unmonitored in-situ seals were increased to 50 and 120 psig,
respectively. The new strength requirements of these seals serve an important safety purpose, but
there is currently no mandatory monitoring or testing program for the structural condition of the
seal themselves. Civil and structural engineers have been using non-destructive testing (NDT)
methods for nearly a century to evaluate the condition of both concrete and non-concrete
structures. The NDT work with concrete has allowed engineers to measure the thickness of
structures, detect flaws, delaminations (or voids in the subsurface), measure the corrosion of
metal reinforcements that may be part of the structure, and even characterize the physical
properties of the structure, all without having to disturb or damage the specimen. One of these
NDT methods, the impact-echo method, has been widely used in concrete evaluation and has the
potential to assess the structural condition of in-situ mine seals. While the impact-echo method
has been successfully used for nearly 30 years in evaluating civil structures, the concept of
tracking the movement and concentrations of tracer gases is a previously untested NDT concept
for both seals and concrete structures. Tracer gases, specifically sulfur hexafluoride and
perfluorinated tracer compounds, have been used to map the ventilation characteristic of
underground mines. A novel NDT method can potentially combine the two methods, where the
injection of a tracer, and the flow of the tracer through the seal material may provide information
on the structural condition of the seal. This paper details the development and assessment of these
two potential NDT methods for the evaluation of in-situ underground mine seals. The assessment
was carried out through a series of small, laboratory experiments and transitioned to both large
and full scale experiments located in working underground mines, accompanied with
supplemental computer modeling to assist in confirmation of perfluorinated tracers moving
through the seal material.
|
Virginia Tech
|
Attribution
The following thesis would not be possible without the attribution and help from numerous
people. Below is of those who assisted me with co-authorship for Chapter 3: Assessment of Sonic
Waves and Tracer Gases as Non-Destructive Testing Methods to Evaluate the Condition and
Integrity of In-Situ Underground Mine Seals, as well as Chapter 6: Technical Note: Modeling the
Movement of Perfluoromethylcyclohexane (PMCH) through Underground Mine Seal material
with PCF3D and Avizo®.
Chapter 3 Co-Authors:
Sponsored me for project, oversaw
Virginia Polytechnic Institute
Kray Luxbacher experiment design and analysis, and
and State University
edited chapter
Sponsored me for project, oversaw
Virginia Polytechnic Institute
Erik Westman experiment design and analysis, and
and State University
edited chapter
Supervisor who oversaw project, provided
Cardno Marshall Miller & small scale sonic wave experiment
Cary Harwood
Associates samples, and material for tracer gas
experiments
Provided underground lab space in
Braden Lusk University of Kentucky
Georgetown, KY
William Weitzel University of Kentucky Poured small scale sonic wave samples
Chapter 6 Contributors:
Virginia Polytechnic Institute Assisted in writing and developing
Drew Hobert
and State University PFC3D code
Virginia Polytechnic Institute Assisted in SkyScan data collection and
Joseph Amante
and State University Avizo® simulation
iv
|
Virginia Tech
|
Chapter 1: Introduction
In 2011, coal mines in the United States produced a total of 1,096 million short tons of coal in
both surface and underground mines. Of the over a billion tons of coal produced, 31.5% was mined in
underground coal mines. Of all coal producing mines, 38.3% are classified as underground operations
(U.S. Energy Information Administration, 2012). While underground mines may not represent a majority
of the coal mining industry, it is likely that the number of underground coal operations will increase as the
surface reserves are mined out and environmental and as social impacts of surface mines continue to face
legislative and public struggles. Often, when comparing between surface and underground mine
development, apart from the economic concerns, underground mines have less environmental impacts and
reclamation costs associated with them than surface mines. Although surface mining is generally cheaper,
the reduced cost of surface mining may not be enough to overcome the social costs of operating on the
surface (Hartman & Mutmansky, Intoductory Mining Engineering, 2002). Looking forward, the coal
industry will eventually have to invest more into the underground sector, as 57.3% of the estimated
recoverable coal reserves in the United States are specified as underground coal reserves (U.S. Energy
Information Administration, 2012). During the same 2011 time period, coal use was responsible for
20.1% of the United States’ total energy consumption, and 28.4% of the total energy production of 97.301
and 78.096 quadrillion Btu’s, respectively (U.S. Energy Information Administration, 2012). The 2011
Annual Energy Outlook projects the production of coal and domestic energy consumption increasing
steadily through 2035 (U.S. Energy Information Administration, 2012), leaving a need for the
development of more underground mines in the United States.
Despite the recent increased popularity and funding in sustainable energy solutions, coal mining
in the United States is projected to continue to be one of the major factors in the U.S. energy distribution
and consumption. As previously mentioned, an increasing number coal mines in the U.S. will need to
become underground operations in order to access the underground coal reserves, totaling 148,084
million short tons (U.S. Energy Information Administration, 2012). According to United States Bureau of
Labor Statistics (BLS), between 2003 and 2012, an average of about 17 deaths occurred annually in U.S.
underground bituminous coal mines. Only two times during that span did the overall number of fatalities
rise above 20 (2006 and 2010). During those two years three major mine explosions/fires occurred in
underground coal mines in West Virginia and Kentucky. Below, in Table 1-1, the BLS data can be seen.
Table 1-1. Fatality data for U.S. underground bituminous coal mines between 2003-2012
Fatalities Caused by Fire
Year Total Fatalities
and/or Explosions
2003 19 0
2004 14 0
2005 7 0
2006 33 17
2007 20 0
2008 9 0
2009 5 0
2010 38 29
2011 11 0
2012 12 0
1
|
Virginia Tech
|
While the obvious goal is to eliminate all fatalities in U.S. coal mines, and while the high
numbers in 2006 and 2010 are from single events, a sound conclusion from Table 1.1 is the need to
prevention of explosions and fire propagation in underground coal mines. Coal mine explosions are a
constant concern for operators, as methane and other combustible gases naturally occur and emit from the
coal. With the addition of machinery and electric components, there are numerous scenarios and
possibilities for ignition to occur. While the actual explosion and fire can be devastating to personnel,
equipment, and support structures in the mine, it is the loss of oxygen and inhalation of toxic gases, such
as carbon monoxide, that are typically the cause of fatalities in underground coal explosions or fire
disasters. Ignitions in coal mines can be caused by many different factors, such as misuse or poor
maintenance of mechanical or electric equipment, frictional ignition caused by mining machinery,
welding, underground blasting, and even lightning strikes on the surface. Even when fires occur in areas
away from personnel and equipment, open fires in underground mines and the expansion of air due to
heating from the fires causes a “choke,” or extension of air in the opposing direction of the engineered
ventilation, as well as the reduction of the density of air, resulting in potentially hazardous effects on the
overall ventilation plan of the mine. These effects can be countered by increasing the overall airflow in
the mine, but doing so will also increase the propagation rate of the fire (McPherson M. J., 1993).
It is widely agreed that the best mitigation against fire and explosion is a well-designed
ventilation program and maintenance of the friction-inducing equipment. One of the most commonly used
ventilation engineering designs to prevent the propagation of fires and explosions in underground coal
mines and also provide adequate airflow to the working sections of the mine without extraneous demand
on the main fan(s) is the construction of underground mine seals. These structures fit across the
dimensions of mine entries and isolate the working section of the mine from the non-working section
(Weiss, Slivensky, Schultz, Stephan, & Jackson, 1996). While the placement of these seals are important
for the proper ventilation and safety of the mine, the proper construction of these seals is equally as
important. Improper construction of these structures can lead to air leakage, exposing the working section
of the mine to the hazardous and potentially combustible atmosphere typically found behind the seals.
Improper construction can also result in the structural failure of these seals if an explosion, and resulting
force, was to occur within the sealed area (Mine Safety and Health Administration, 2008). Because of the
need to maintain the structural condition of the seals, it became necessary to develop and assess methods
to evaluate the condition of the structures. The assessment of these methods required a series of small-
scale, large-scale, and, eventually, full-scale experiments in functioning underground mine environments.
2
|
Virginia Tech
|
Chapter 2: Literature Review
2.1 Mine Seals
2.1.1 Ventilation and Seal Purpose
Unquestionably, the most important auxiliary operation in any underground mining operation is
the ventilation of the mine. Ventilation has been an invaluable mining operation, for over two millennia,
from early B.C. mining in England and Greece to the writing of De Re Metallica in 1556 by Georgius
Agricola (McPherson M. J., 1993). Today, ventilation is required for three main components in order to
efficiently and safely work in underground mining conditions — air quantity, air quality, and temperature
control. When looking at the air quality control concern, it is important to note that in every mine, both
coal and metal/no-metal mines, dust and gases naturally exist in either the local geology or result from
industrial equipment used in underground mines (Hartman & Mutmansky, 2002). In previous years, the
quantity and quality of air entering and leaving the mine was the primary concern of ventilation
engineers, but now the comfort and tolerance of the human workers has become a more significant
priority (McPherson M. J., 1993). While prevention of these hazards is a primary goal, dilution of
contaminants to safe levels can be accomplished by supplying the mine with an appropriate quantity of
air. Air quantity controls exist to supply the mine and mine workers with a continuous flow of fresh air to
facilitate normal respiratory functions and disperse chemical and physical contaminates (heat, dust,
humidity, etc.). Temperature controls also play an important part in ventilating deep underground mines,
where the geothermal gradient of the local geology increases with depth. Chilled water is prepared at the
surface; , then in heat exchangers, this water can be used to cool and dehumidify the air going into the
mine (Hartman & Mutmansky, Intoductory Mining Engineering, 2002). As mining progresses, more air is
needed to provide adequate ventilation to the workings, as well as to continue to dilute and remove hazard
contaminates that increase as more surface area is exposed in the mine.
In order to avoid moving more air though the mine to provide appropriate quantity and quality
controls, mines often seal abandoned areas or portions of the mine that are no longer working sections.
The seals allow proper ventilation to affect the working sections of the mine, while not being wasted in
the abandoned sections and eliminate exposure of personnel. Continuing to ventilate abandoned mine
sections can become a costly enterprise that involves continuously increasing the total air quantity
entering the mine (Zipf, Sapko, & Brune, 2007). Abandoned areas are sectioned off by constructing seals
at the entrances of the connecting airways (McPherson M. J., 1993). Historically, seals were two walls 16
to 48 inches thick (reinforced concrete seals) made with a variety of materials across entry dimensions of
up to 288 square feet. The area between the walls were filled with run-of-mine and other fill material to
make a barrier with a total thickness of about 12 to 20 feet. Modern seals are made with variety of solid
incombustible materials such as poured concrete, concrete blocks, cementitous foams, and other novel
materials with thickness of about 12 to 20 feet. (Kallu J. R., 2009). According to MSHA, there are over
14,000 seals installed in active U.S. coal mines, with multiple applications associated with them. The two
most common types of seals used in underground coal applications are panel and district seals. As the
name indicates, panel seals are typically constructed parallel with panels in both longwall and room and
pillar mines. Once a panel or group of panels has been mined-out, panel seals are constructed to restrict
the ventilation away from the mined-out area. District seals are used once a mining district (made up of
multiple panels) is mined-out and are usually designed for higher strength parameters because of the large
volume behind them. An example of these two types of seals can be seen in Figures 2-1 and 2-2 below,
for both room and pillar mining and longwall mining applications (Zipf, Sapko, & Brune, 2007).
3
|
Virginia Tech
|
2.1.2 Explosions and Seals
While seal construction is a necessary operation in properly and effectively ventilating
underground mines, they also serve as a protective barrier between explosive areas of the mine and mine
personnel and equipment. Spontaneous combustion is a phenomenon that is can occur when the
percolation of air through organic material, such as coal, result in a measurable increase in temperature.
Thermal equilibrium is reached when the airflow is sufficient enough to reach a balance between the rate
at which heat is produced and the rate at which heat is removed from the material by the airflow and can
be difficult to maintain. Materials that are known to spontaneously combust have known minimum self-
heating temperature (SHT) — the lowest temperature that will result in a sustained exothermic reaction.
Behind seals in underground coal mines, if the temperature of the coal reaches the SHT before it can
reach a thermal equilibrium due to the lack of air flow, the oxidation process will accelerate. At a certain
rate of oxidation, the coal will become incandescent, begin to smoke, and produce gaseous products of
combustion (McPherson M. J., 1993). The explosive risk in underground mines is present when
spontaneous combustion and heating occur in an area with high levels of methane accumulation. The
initial atmosphere behind mine seals typically consists of 21% oxygen, 79% nitrogen, and less than 1%
methane. Once the ventilation to the mined-out area has ceased, the methane levels can increase as
methane accumulates behind the seal. Methane is typically explosive over a range of 5-16%, depending
on the oxygen levels, and sealed areas can reach the upper explosive limit in a matter of days or weeks,
depending on the methane liberation rate (Zipf, Sapko, & Brune, 2007). When methane is in an explosive
range, an explosion can take place when sufficient oxygen and an ignition spark — from a roof fall,
lightning strike, mechanical electronics, welding equipment, etc. — occur within the explosive
atmosphere. Based on the explosive range of methane, decreasing the oxygen content below 12% would
not provide enough fuel for the methane to combust behind the mine seal (Cowards & Jones, 1952). The
application of seals allows for the region to eventually develop a low-oxygen atmosphere incapable of
spontaneous combustion. However, even after the methane concentration has exceeded its upper
explosive limits or oxygen depletion has created an inert atmosphere behind the seal, leakage around the
boundary of the seals can create explosive atmospheres along the edges of the seams. This hazard can be
reduced by providing sufficient flow of air to the active side of the seals to prevent methane accumulation
(Zipf, Sapko, & Brune, 2007).
The spontaneous heating nature of coal is a naturally-occurring phenomenon that must be
considered when looking at potential explosion hazards in both abandoned and working mine sections of
underground coal mines. Another well documented natural occurrence responsible for methane-based
explosions underground is lightning strikes. Methane based underground coal mine explosions can occur
when lightning strikes cause electric sparks with sufficient energy in an atmosphere with an explosive
concentration of methane. There are two documented modes of transportation that allow lightning to
penetrate underground mines — through the over lying strata and through metallic structures connecting
the surface to the mine (Geldenhuys, Erickson, Jackson, & Raath, 1985) via. (Novak & Fisher, 2001). The
depth of lightning propagation through the overlying strata was shown to be proportional to the resistivity
of the soil, where lightning will penetrate greater depths through soils with a higher resistivity. Large
conductive structures that are grounded and geological faults/discontinuities in the overburden can distort
the current distribution (Berger, 1977) via (Novak & Fisher, 2001). The second mechanism of lightning
propagation is through a direct strike to a metallic structure on the surface that extends into the mine.
Examples of these types of structures include, but are not limited to: cables, conveyor belt structures,
water pipes, and borehole casings. The attenuation of the strike depends on the surge impedance of the
structures and how well they are effectively grounded (Novak & Fisher, 2001).
2.1.3 History of Explosions in Sealed Areas (U.S.)
Since 1986, there have been at least 12 documented explosions in U.S. coal mines that occurred
within the sealed areas and resulted in numerous seals being destroyed or damaged. Table 2-1 on the
5
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.