An integrated methodology for monitoring spontaneous combustion of coal waste dumps based on surface temperature detection

Applied Thermal Engineering 122 (2017) 27–38

Contents lists available at ScienceDirect

Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

An integrated methodology for monitoring spontaneous combustion of coal waste dumps based on surface temperature detection Zhenqi Hu ⇑,1, Qing Xia 1 China University of Mining and Technology (Beijing), Beijing 100083, China

h i g h l i g h t s
A new methodology for constructing surface temperature distribution is proposed.
The procedure of data collection and data processing is studied.
The model obtained presents spatial coordinates and temperature information.
This research is capable of giving an early warning for preventing fire disasters.

a r t i c l e

i n f o

Article history: Received 29 August 2016 Revised 19 April 2017 Accepted 5 May 2017 Available online 6 May 2017 Keywords: Spontaneous combustion Coal waste dumps Close-range photogrammetry Infrared thermography Fire disaster

a b s t r a c t Coal mining wastes are prone to spontaneous combustion, causing burning environmental issues and a threat to human safety. To warn about such spontaneous combustion, mapping surface temperature distribution and locating surface anomalous zones is needed. The new methodology developed in this study was achieved following four primary steps, including field investigation procedures, data pre-processing procedures, data coupling, and 3D visualization. Eventually, a 3D temperature distribution model was presented, and the observed zones were classified into three categories basing on different temperature levels. This new methodology may be useful in monitoring and locating the potential risk zones in advance and making an early warning and allowing to prevent spontaneous combustion. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Since China has currently become the world’s largest coal production and consumption country [1–3], coal wastes generated from coal mining have turned into one of the primary pollution factors in China. So far, approximately150 km2 of land has been abandoned due to accumulation of 4.5 billion tons of coal wastes. These coal wastes are commonly deposited in the vicinity of mining areas, forming 1700 coal waste dumps [4–7]. Coal wastes consist of claystones (40–98%), coal shales (2–40%), mudstones (2– 40%), and sandstones (<33%). Carbonaceous rocks and conlomerates are rarely present within their mass. In terms of mineralogical composition, typical coal wastes are composed of clay minerals (50–70%), quartz (20–30%), other minerals (10–20%), and carbonaceous matter [8–10]. In ambient conditions, organic matter, which is represented by all three groups of macerals, i.e., huminite, lip-

⇑ Corresponding author. 1

E-mail address: [email protected] (Z. Hu). Zhenqi Hu and Qing Xia contribute to this work equally.

http://dx.doi.org/10.1016/j.applthermaleng.2017.05.019 1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.

tinite, and intertinite, of coal waste may react with atmospheric oxygen and spontaneous combustion is more likely to occur when heat generated by the exothermic oxidation of the organic matter is larger than the dissipated one [6,11–13]. This spontaneous combustion poses a serious threat to the environment by releasing toxic gases and chemicals, e.g., carbon monoxide, hydrogen sulfide, sulfur dioxide, sulphate, liquid hydrocarbons, and heavy metal materials, to surrounding soils and ground waters [14]. In addition, it can also lead to coal-mining-related geological disasters, such as landslides, large collapses, rock bursts, and gas explosions, which endangers miners’ and nearby people’s safety and property [15]. Therefore, finding and locating surface anomalous zones is important, as it enables taking preventive measures before spontaneous combustion incidents, which then leads to avoidance of the occurrence of geological disasters and loss of lives and property. In the early days, thermometers have been used to measure surface temperature of coal waste dumps with the temperature ranging from 40 °C to 1200 °C. However, this technique is timeconsuming due to a point-to-point method, and it is dangerous for operators in burning zones with disadvantages of contact

28

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

[16]. Many studies have been carried out to locate underground coal fires by analysing of remote sensing images, e.g., Priyom Roy used remote sensing images from ASTER and Landsat-8 to detect coal fires in Jharia coal field, India [17–22]. However, it is difficult to apply large-scale remote sensing images into monitoring the surface temperature in small-scale coal waste dumps due to relatively low spatial resolutions. Infrared thermography has gradually become the most common technique on a global scale for temperature detection with the advantages of non-contact, high accuracy and low cost. However, thermographic images can only display the surface temperature distribution without topographic information. Some attempts have been made to detect surface temperature in coal waste dumps using topographic surveying techniques and infrared thermography [15,16,23,24]. However, these studies mainly emphasized on presenting surface temperature distribution, resulting in difficulty concerning locating anomalous zones because of the lack of accurate spatial coordinates. Although topographic surveying techniques have been attempted to locate the regional anomalous zones by integrating with infrared thermography, the results are still unable to reconstruct 3D visualization models of entire coal waste dumps. In this study, a new methodology for constructing a surface temperature distribution model of a coal-mining waste dump by integrating infrared thermography and close-range photogrammetry, was proposed. The close-range photogrammetry is a surveying technique using image capture to obtain topographic information and this operation is normally performed without any contact with the object under scope [25–27]. For the field data collection, some key issues on how to (1) set control points, temperature-marked points (TMPs), location of an infrared thermographic camera, location of a close-range photogrammetry camera, and (2) capture thermographic images and close-range photogrammetry images, were addressed in detail. Data pre-processing procedures for two different kinds of images, data coupling, and 3D visualization were also discussed. This study emphasized on providing a feasible procedure for detecting and locating thermally anomalous zones in coal waste dumps. This methodology was applied in a small test coal waste dump in northern China to verify its feasibility and applicability. The integrated method with two techniques involved is capable of easily monitoring and locating potential risk zones, especially for small-scale coal waste dumps, compared with the previous studies.

2. Field investigation procedure 2.1. An overview of the technology used Infrared thermography allows to detect surface temperature of an object by measuring its emitted electromagnetic radiation. This goal is achieved without the need of a physical contact. The radiation detected is converted into electrical signals. Then a thermographic image or a visible image is displayed on a screen of an infrared thermographic camera [28–31]. A thermographic image represents a map of relative temperature variation where the highest and the lowest temperature is identified by hot colors (reds) and cold colors (blues), respectively. The TH9100MV/WV thermographic camera used in this study has a resolution of 320 (H)  240 (V) pixels and operates in the 8 to 14 lm wavelength band range. The device has adjustable temperature measuring ranges from 0 °C to +500 °C with an accuracy of ±0.06 °C. It is equipped with a commercial MikroSpec4 software for thermographic images processing [32]. Close-range photogrammetry is a technique used to determine position, size and shape of an object by using images; by applying it one can metrically reconstruct objects in 3D by using accurate

imaging techniques [25,33]. The close range refers to an object distance of up to about 300 m [34]. For the purpose of this study’s close-range photogrammetry, we have applied a digital camera and the lensphoto 2.0 software [25,35–38]. A Canon EOS 5D Mark digital camera with a 28 mm focal length lens was used to capture images, and then these images are processed with the lensphoto 2.0 software to provide topography information for an area. This software can provide lens calibration before image processing. 2.2. Investigation methods The experiment site was situated at a discarded coal waste dump in the Changping District of Beijing, China. The test coal waste dump was formed as a cone with a base diameter of 2.5 m and a height of 1.8 m, weighing approximately 2767 kg. The coal wastes consist of subbituminous coal, sandstone, clay, quartzite, sulfide mineralization, and solid wastes. Field investigations reveal some potential high temperature points at the coal waste dump surface and lack of vegetation. The location of the study area is shown in Fig. 1. The first step of the experiment involved setting control points for the close-range photogrammetry. It is necessary to carry out an investigation of the test site (such as area, topography, elevation, presence of vegetation, and so on). In order to make the identification of control points easier, the points were marked with white crosses. First, the test site was divided into four different flanks (Fig. 2). For each flank, at least four control points were established and were evenly distributed at each corner (points a, b, c, and d in the Fig. 2). Additional control points were widely and uniformly distributed over each flank (points e, f, and g in the Fig. 2) to improve the accuracy during data processing. In the Fig. 2, numbers 1–4 represent close-range photogrammetry camera station. Each camera station captures an image, which is equal to each dashed frame. 1 & 2 is the overlapping region between image 1 and image 2, and 3 & 4 is the overlapping region between image 3 and image 4. For two adjacent flanks, both flanks must share at least three control points in the overlapping region because the theory of matching two adjacent flanks is based on the collinearity equations used as observation equations in a least-squares adjustment according to the Gauss-Markov model [39]. The second step involves setting the temperature-marked points (TMPs) for infrared thermography. Since the captured images (from the infrared thermographic camera) were too small to cover a single flank, a single chosen flank was vertically divided into multiple subareas, defined as 1st, 2nd. . . Nth, as shown in Fig. 3. In general, the quality of image stitching depends on the number of subareas. The more subareas distinguished, the lower accuracy. It is necessary to share at least three TMPs between two vertically adjacent subareas (black stars in the Fig. 3). For a single chosen subarea, it is necessary to share more than two points and the extent of the overlap should be higher than 10% between two adjacent thermographic images (black rectangles in the Fig. 3). The reason why we used two points was that the theory of image stitching between two adjacent thermographic images is based on a rigid transformation [40,41]. The transformation equation is described below.



x2 y2



 ¼

cos h

sin h

sin h

cos h



x1 y1



 þ

tx ty

 ð1Þ

where (x1,y1) is pixel coordinates before image stitching, (x2,y2) is pixel coordinates after image stitching. (h; t x ; t y ) is three unknown parameters. In order to calculate these three parameters, two points are needed. 10% overlapping is an empirical value taken from the MikroSpec4 software instruction after many field tests. It is difficult to distinguish TMPs in the thermographic image, so hot metal coils

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

29

Fig. 1. Location of the study area in China.

Fig. 2. Distribution of the control points for an example flank and location of the close-range photogrammetry camera. Numbers 1–4 represent close-range photogrammetry camera station. The dashed frames in different colors (black, red, blue and orange) represent the cover region of each image. Point a–f, represent the control points.

should be made and fixed at where TMPs should be as highertemperature marks. That means TMPs represent hot metal coils (black stars and black rectangles in the Fig. 3 represent hot metal coils). A total of 32 TMPs are set around the test site. The third step involves elaboration of an appropriate method of image capture by the close-range photogrammetry camera. The NTS-352 type total station was used in the study to measure spatial coordinates of control points in a relative system oriented with

respect to the north. The spatial coordinates of the control points are the referencing parameters for further image process. The images must be taken in accordance with the principles of closerange photogrammetry and they are subject to the requirements of the lensphoto 2.0 software. For each flank, the camera was moving from left to right and parallel to the flank on a distance of 5 m. Technically speaking, it is required at least two images to locate spatial coordinates in stereo photogrammetry for each flank. In

30

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

Fig. 3. Distribution of TMPs for an example flank and the position of thermographic camera: black stars represent TMPs located between two adjacent subareas, black rectangles represent TMPs between adjacent thermographic images horizontally, triangles represent control points.

our preliminary results, three images worked better than two images. Therefore, a minimum of three images were suggested to be taken for each flank. It is necessary to share at least one control point between two adjacent images to accomplish a stereo image pair identification in stereo photogrammetry. It is highly recommended to take extra images to improve image process accuracy. The images should be taken considering the following points: one image should be taken at one place; the elevation angle should not be higher than 45°; the extent of the overlap between two adjacent images must be at least 80%; the object depicted must occupy more than 80% of each image (Fig. 2). The 45° elevation angle and 80% overlapping are empirical values taken from the lensphoto 2.0 software instruction. The same procedure is used for other flanks of the test site. The fourth step involves capturing images by the infrared thermographic camera. First, the test site was divided into four different flanks, and then a single chosen flank was divided into several subareas (Fig. 4). For a chosen subarea (e.g., 1st subarea), the thermographic images were taken from left to right on a distance of 5 m. The other subareas followed the same procedure above. The distance depends on the quality of thermographic images and the size of images. All the thermographic images in each subarea were obtained sequentially by parallel shooting. After taking all the images for a chosen flank, the other flanks of the test site followed the same procedure above to take thermographic images. 3. Combining topographic data with infrared thermographic data Since images obtained from close-range photogrammetry and infrared thermography shared the same TMPs, the spatial coordinates of TMPs extracted from the former could be added into the ones from the latter technique. After temperature calibration and image stitching with the MikroSpec4 software, a stitched thermographic image was created as an array or a matrix (the matrix is defined as the temperature grid) (Fig. 5). The matrix dimensions relate to the number of pixels in the stitched thermographic image, columns and rows. The cells of the matrix relate to the pixels of the stitched thermographic image. The coordinate system in the matrix is the same as in the stitched thermographic image and it is being represented by the row, column number a, b coordinate respectively. If (a, b) coordinate is known, it means that a specific cell within the temperature grid is known and the numerical value of the temperature can be read.

Each cell in the temperature grid recorded three properties: pixel coordinates (a), (b) and temperature value (T). Moreover, the spatial coordinates (Xb, Yb, Zb) of TMPs in a thermographic image obtained from close-range photogrammetry were placed in the temperature grid. Delaunay triangulation was then constructed basing on the location of the TMPs in the temperature grid [42– 44]. According to the obtained spatial coordinates of TMPs, the spatial coordinates of all points in the Delaunay triangulation were computed by a bilinear interpolation algorithm [45–47]. The spatial coordinates of all points were stored in the temperature grid. Finally, a 3D model, including final spatial coordinates (X, Y, Z) of all points for surface of the test site obtained from a bilinear interpolation and temperature (T), was constructed to visualize the surface thermal anomalies of the test dump site (Fig. 6). 3.1. Pre-processing of the close-range photogrammetry data A data processing procedure is important when integrating close-range photogrammetry with infrared thermography, because the accuracy of the model depends on the quality of the image processing. Lensphoto2.0 software allows to work with images obtained from a digital camera in an automated way and can provide possibility of calibrating cameras and lens, interior orientation, exterior orientation, and 3D model construction. In this software, there are many modules, but the ones used in this study included engineering management module, aerial triangulation module, point cloud generation module, and point cloud product module (Fig. 7). Before processing the images, the interior orientation parameters were computed. These parameters were used to create zero distortion images as an intermediate step. To begin, a new project was set up in the engineering management module. The following step was to carried out relative orienting [25,35]. It means that the corresponding homologous points on different close-range photogrammetric images are referenced to each other. This operation is called image-to-image matching and was conducted in the aerial triangulation module. After that, bundle adjustment was carried out to analyse the matching accuracy. The next step was absolute orientating. Once the accuracy met the requirements, it was necessary to adjust the model to the ground’s coordinate system. Control point coordinates were thus introduced. Subsequently, an overall adjustment was performed to check the overall accuracy. All the procedures above had to be done in cycles until the overall adjustment error was below half a pixel. The last step was generation of point clouds. A point cloud is a set of data points, which represents

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

31

Fig. 4. Procedure of taking infrared thermographic images: black rectangles represent TMPs between adjacent thermographic images, triangles represent control points.

the shape of an object by X, Y, Z coordinates in a three-dimensional coordinate system. The information contained in the images could be updated automatically marking all of the elements of interest on the images. The point clouds were generated for the test site and irrelevant point clouds obtained from interfering subjects

rather than our target were deleted in the point cloud generation module. Finally, the spatial coordinates of all the points, including TMPs, were ready to be exported into a MS-ExcelTM chart in order to be visualised and processed. The point clouds model of the study area is shown in Fig. 8.

32

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

Fig. 5. An illustration of the temperature grid.

Fig. 6. A brief outline of a data processing procedure.

33

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

temperature values of all the points in the temperature grid were exported into a MS-ExcelTM chart after pre-processing. The interface of the software is shown in Fig. 9 and the image pre-processing procedure is shown in Fig. 10. The mosaic images for the test site are shown in Fig. 11. 3.3. Data-coupling algorithm In this study, a Delaunay triangulation for a set of TMPs in the temperature grid was generated [42–44]. Spatial coordinates of any points within the Delaunay triangulation can be interpolated from the spatial coordinates of the three sample TMPs at the corners of the triangle into which each point falls. As shown in the Fig. 12, a cdef square represents the temperature grid. Taking the triangle A1A2A3 within the Delaunay triangulation as an example, the spatial coordinates of the points A1, A2, A3 constituting the triangle A1A2A3 are known. We want to interpolate the spatial coordinates of the point B. The triangle A1A2A3 is split into two triangles A1A3A4 and A2A3A4 by a line A3A4 through the point B. The spatial coordinate value X of the point A4 can be calculated by bilinear interpolation algorithm described by Eq. (1) [45–47].

f 4 ða4 ; b4 Þ ¼ q=p  f 2 ða2 ; b2 Þ þ ð1  q=pÞf 1 ða1 ; b1 Þ Fig. 7. A flowchart of the close-range photogrammetry data pre-processing with the lensphoto 2.0 software.

3.2. Pre-processing of the infrared thermographic images The thermographic images were imported into the MicroSpec4 software. Before image analysis, some parameters, such as the emissivity of the measured object (0.9), the ambient temperature (29.8 °C) and the distance (5 m between the thermographic camera and an example flank), were input into this software for adjusting temperature to improve temperature accuracy after thermographic images have been taken. Then, the image stitching, which created a large size thermographic image mosaic by combining multiple thermographic images, was achieved basing on TMPs and image feature matching. Finally, the pixel coordinates and

ð2Þ

where (ai,bi) (i = 1,2,. . .) are pixel coordinates in the temperature grid, f1(a1,b1) and f2(a2,b2) are the X coordinate values of the points A1 (a1,b1) and A2 (a2,b2), respectively. q is the distance between the points A1 and A4 and p is the distance between the points A1 and A2. The distance p and q can be calculated from Eq. (3). 2 1=2

q ¼ ðða4  a1 Þ2 þ ðb4  b1 Þ Þ

2 1=2

p ¼ ðða2  a1 Þ2 þ ðb2  b1 Þ Þ

ð3Þ

For the point B, its X coordinate value can be computed using Eq. (4).

f ðx; yÞ ¼ n=m  f 3 ða3 ; b3 Þ þ ð1  n=mÞf 4 ða4 ; b4 Þ

ð4Þ

where f3(a3, b3) and f4(a4, b4) are the X spatial coordinate values of the points A3(a3, b3) and A4(a4, b4), respectively, m is the distance

Fig. 8. Point clouds of the test site.

34

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

Fig. 9. Interface of the MicroSpec4 software (an initial thermographic image inside this software).

Fig. 10. Procedure of thermographic images pre-processing.

Fig. 11. Mosaic thermographic image for the whole test area: the crossing lines represent TMPs.

between the points A3 and A4, n is the distance between the points B and A4. The distance calculation can be described by Eq. (5).

The Y and Z spatial coordinates can similarly be calculated using the Eq. (6).

2 1=2

m ¼ ðða4  a3 Þ2 þ ðb4  b3 Þ Þ 2 1=2

n ¼ ððx  a4 Þ2 þ ðy  b4 Þ Þ

ð5Þ

By substituting Eq. (5) into the Eq. (4), one may obtain the X spatial coordinate value of any point in the triangle A1A2A3 from the following Eq. (6): 2

2

2

2

1=2

1=2

f 4 ða4 ; b4 Þ

f ðx; yÞ ¼ ½ððx  a4 Þ þ ðy  b4 Þ Þ=ðða4  a3 Þ þ ðb4  b3 Þ Þ 2

2

2

2

þ½1  ½ððx  a4 Þ þ ðy  b4 Þ Þ=ðða4  a3 Þ þ ðb4  b3 Þ Þ

f 3 ða3 ; b3 Þ ð6Þ

3.4. Plotting a 3D visualization model The temperature values and spatial coordinates (X, Y, Z, T) of all the points were implemented in the temperature grid and a 3D visualization model was constructed with the use of the MATLAB software. A colour dataset in which each colour corresponds to one temperature value was created to visualize temperature values. Meanwhile, a cursor was set up to make a connection with spatial coordinates and temperature values. Therefore, 4D information (X, Y, Z, T) for each point in the temperature grid was dis-

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

35

Fig. 12. Interpolation algorithm for the spatial coordinates.

played. Different views of the 3D visualization model are shown in Fig. 13.

4. Discussion Infrared thermography applied in this study allowed to present the surface temperature distribution by thermographic images with the method’s advantages of non-contact character, high accuracy and low cost. However, the major limitation of this technique is incapability of locating potential surface anomalous zones due to the lack of spatial coordinates. Close-range photogrammetry was thus used with the intention of associating accurate topographic survey with thermographic images. Compared with the previous studies [15,23,24], the main advantages of integrated methodology are discussed below. Each technology attained a specific result, and then the data coupling method was developed to interpolate spatial coordinates from thermographic images. Importantly, some key issues about field operations and data process were addressed in detail. This methodology was applied at a test site and a 3D visualization model was constructed. The integrated method proved a valid procedure to follow for surface temperature distribution of a coal waste dump. Once the coal waste dump was modelled by a 3D visualization, the infrared thermography technique contributed to the identification of the thermally anomalous zones, while permitting the access to the accurate coordinates of these detected zones within the model. For better verification of the spatial coordinates, the results were extensively compared with reference data. Some TMPs (points 1, 2, 12, 18, 21, 23, 29, 30) were selected as reference data and their spatial coordinates (x, y, z) were measured by the total station in field investigations. Corresponding interpolated spatial coordinates were extracted from the final 3D visualization model. The comparison results were listed in Table 1. It can be seen that

the maximum error was 0.012 m and the comparison showed satisfactory agreement with spatial coordinates accuracy. After modelling the coal waste dump, with the purpose of easy fire-fighting countermeasures for different level of coal waste susceptibility to spontaneous combustion, an appropriate scale with three or four classes is usually used, basing on different risk degrees. In this study, the temperature is classified into three categories: low, medium and high self-combustion risk zones, as follows (Fig. 14). (1) Low self-combustion risk zones: these zones are characterized by the temperature ranging from 16 °C to 35 °C. It is not necessary to take any fire-fighting measures for these low temperature zones because these zones have a low risk to occur coal geological disaster. (2) Medium self-combustion risk zones: these zones are characterized by temperature ranging from 35 °C to 80 °C. Based on a law called Disaster Prevention and Control Standard on Coal Waste Dumps in China the temperature of 80 °C is considered to be critical [48]. 80 °C was used as a threshold for separating medium and high self-combustion risk zones. There are two reasons for this. One reason is that administrators in coal mines were asked to keep surface temperature below 80 °C based on a law mentioned above. The other reason is that once temperature exceeds 80 °C, it accelerates exothermic oxidation reactions, which leads to temperature increase. The temperature rapidly goes up, described as thermal runaway, until it reaches the selfignition stage [10,16]. If this process proceeds unchecked, a runaway ignition event can ensue and subsequently initiate a fire. For the temperature ranging from 35 °C to 80 °C, temperature monitoring and measurements should be conducted timely and periodically due to the potential geologically hazardous phenomena such as large collapses, rock bursts, and gas explosions.

36

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

Fig. 13. Different views of the obtained 3D visualization model. A - full view with detailed temperature and spatial coordinate information; B, C and D are different views.

Table 1 Comparison between the real values and interpolated values of the spatial coordinates. No.

1 2 12 18 21 23 29 30

Total station measurements

Interpolation

Error

x

y

z

X

Y

Z

m

988.502 989.017 987.623 986.241 987.013 987.241 988.876 989.002

1017.578 1018.426 1016.867 1016.824 1018.024 1019.274 1018.462 1018.179

99.106 97.899 99.021 98.947 98.435 98.976 98.861 98.629

988.511 989.025 987.631 986.253 987.009 987.252 988.887 989.014

1017.565 1018.412 1016.858 1016.835 1018.018 1019.283 1018.453 1018.188

99.109 97.889 99.014 98.934 98.442 98.968 98.872 98.624

0.009 0.010 0.008 0.012 0.005 0.009 0.010 0.009

(3) High self-combustion risk zones: these zones are characterized by the temperature exceeding 80 °C. For the zones where the temperature exceeds 80 °C, subsurface spontaneous combustion is more likely to occur [10,49]. The accelerated oxidation reactions releases heat, which raised temperature rapidly. When the accumulative temperature reaches self-ignition of different coals, spontaneous combustion occurs. The temperature of self-ignition of coal is 150 °C for subbituminous coal, 200 °C for bituminous coal, 250 °C for coke and 300 °C for anthracite [10]. In the high selfcombustion risk zones, some of them are burning, and the others are the zones where self-heating has been detected, but fire has not yet started. For those zones, some precau-

tions should be taken to avoid the occurrence of spontaneous combustion. These highly problematic zones are identified, various fire-fighting techniques should be employed to extinguish fires. If fires are subsurface, injections such as grout injection, gel injection, and insert gas injection are the common methods to extinguish fires after borehole drilling [50]. The main purpose of these injections is to lower the temperature and to seal the air pathway. In case of surface fires, sand-bentonite slurry flushing, surface sealing with soil, loess coverage and coverage with some other non-combustible material have been used to prevent air leakage into extinguished fire zones, and to provide a soil basis for reclamation in the future [51].

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

37

Fig. 14. Classification maps corresponding to the Fig. 13. Red represents high self-combustion risk zones, yellow represents medium self-combustion risk zones, blue represents low self-combustion risk zones.

5. Conclusions This study aimed to show how two different techniques could be integrated to pinpoint and visualize the potential surface thermally anomalous zones in a coal waste dump. Although each technique can be considered to be well recognized and developed, many integrated issues in field investigations, such as control points distribution, TMPs distribution, image shooting methods of two techniques, were explored. After field investigations, the pre-process procedure of different images and data-coupling algorithm were also discussed in detail. Finally, a 3D model containing spatial coordinates and temperature values of the surface temperature distribution was constructed. This study is beneficial to the safety of miners due to the non-contact character and applicability in inaccessible regions. It is shown that this methodology is effective and feasible and allows for constructing a 3D visualization model, which may contain both spatial coordinates and temperature values for the entire coal waste dump. Spatial coordinates within the 3D model were shown to be in a satisfactory agreement with those collected via total station measurements, with the maximum error of 0.012 m. In the final 3D model constructed, the temperature was classified into three classes based on different risk degrees. The classification map allows for taking various measures to control and extinguish fires in different risk zones basing on the knowledge of the thermally variable zones. When applying the classification map to other coal waste dumps, the temperature range for low self-combustion risk zones needs to be redefined basing on daily minimum and maximum temperatures measured at surface of other coal waste dumps. It is not necessary to take any fire-fighting measures for these low temperature zones because of lacking possibility of the occurrence of the spontaneous combustion. For the medium self-combustion risk zones, the classification map can provide an early warning to mining authorities to take precautions against spontaneous combustion. For the high self-combustion risk zones, where self-heating has been detected,

but a fire has not started yet, spraying coating material over coal surface, should be deployed to eliminate the risk [52]. For those burning zones, various fire-fighting measures, such as grout injection, gel injection, surface sealing with soil, and coverage with some other non-combustible material, should be taken to control and extinguish fires in order to avoid the occurrence of coalrelated geological disasters. Acknowledgements This work was funded by the National Natural Science Fund in China (No. 41371502). The authors would like to express appreciation to members of the research group in China University of Mining and Technology for providing great help in terms of the field operations. The authors want to give their special thanks to Jennifer Lawrence at the Virginia Polytechnic Institute and State University (USA) for her language consulting. The authors would like to thank the anonymous reviewers for their remarks that have improved the paper. References [1] S. Voigt, A. Tetzlaff, Integrating satellite remote sensing techniques for detection and analysis of uncontrolled coal seam fires in North China, Int. J. Coal Geol. 59 (2004) 121–136. [2] B. Shen, Y. Lei, Y. Guo, Progress of coal science and technology in China, J. China Coal Soc. 36 (2011) 1779–1783. [3] T. Xu, S. Simon, C. Benjamin, H. Mikael, Clean coal use in China: challenges and policy implications, Energy Policy 87 (2015) 517–523. [4] Y. Zhao, J. Zhang, C. Chou, Y. Li, Z. Wang, Y. Ge, Trace element emissions from spontaneous combustion of gob piles in coal mines, Shanxi China, Int. J. Coal Geol. 73 (2008) 52–62. [5] Y. Wang, Y. Sheng, Q. Gu, Infrared thermography monitoring and early warning of the spontaneous combustion of coal gangue pile, The international archives of the photogrammetry, remote sensing and spatial information sciences, Beijing, China, 2008. [6] Z. Bian, J. Dong, S. Lei, H. Leng, S. Mu, H. Wang, The impact of disposal and treatment of coal mining wastes on environment and farmland, Environ. Geol. 58 (2009) 625–634.

38

Z. Hu, Q. Xia / Applied Thermal Engineering 122 (2017) 27–38

[7] G. Fan, D. Zhang, X. Wang, Reduction and utilization of coal mine waste rock in China: a case study in Tiefa coalfield, Conserv. Recy. 83 (2014) 24–33. [8] K. Skarzynska, Reuse of coal mining wastes in civil engineering-part 1: properties of minestone, Waste Manage. 13 (1995) 3–42. [9] X. Querol, M. Izquierdo, E. Monfort, E. Alvarez, Environmental characterization of burnt coal gangue banks at Yangquan, Shanxi Province, China, Int. J. Coal Geol. 75 (2008) 93–104. [10] M. Misz-Kennan, M. Fabianska, Application of organic petrology and geochemistry to coal waste studies, Int. J. Coal Geol. 88 (2011) 1–23. [11] P. Li, Z. Hu, J. Wu, Research and discussion on the hazards and greening technology of coal gangue dump, Min. Res. Dev. 26 (2006) 93–96. [12] I. Sy´korová, W. Pickel, K. Christanis, M. Wolf, G.H. Taylor, D. Flores, Classification of huminite-ICCP System 1994, Int. J. Coal Geol. 62 (2005) 85– 106. [13] G.H. Taylor, M. Teichmüller, A. Davis, C.F.K. Diessel, R. Littke, R.Robert, Organic petrology, 1st ed. Berlin, 1998. [14] Z. Bian, H.I. Inyang, J.L. Danels, F. Otto, S. Struthers, Environmental issues from coal mining and their solutions, Mining Sci. Technol. (China) 20 (2010) 215– 223. [15] J. Pandey, D. Kumar, R. Mishra, Application of thermography technique for assessment and monitoring of coal mine fire: a special reference to Jharia coal field, Jharkhand, India IJARSG 2 (2013) 138–147. [16] M. Misz-Kennan, A. Tabor, The thermal history of selected coal waste dumps in the Upper Silesian Coal Basin (Poland), Coal and Peat Fires: A Glob. Perspect. 3 (2011). [17] R. Slavecki, Detection and location of subsurface coal fires, in: Proceedings of the third symposium on remote sensing of environment, institute of science and technology, University of Michigan, Michigan (Ann Arbor, MI: ERIM), (1964) 537–547. [18] A. Cracknell, S. Mansor, Detection of sub-surface coal fires using landsat thematic mapper data. Int. Arch. Photogramm. Remote Sens. 29 (1992) 750– 753. [19] A. Prakash, R. Gens, Remote sensing of coal fires, Coal-combust. Geol. 1 (2010) 231–253. [20] T. Martha, A. Bhattacharya, K. Vinod, Coal-fire detection and monitoring in Raniganj Coal Field, India: A remote sensing approach, Curr. Sci. 88 (2005) 21– 24. [21] R. Chatterjee, Coal fire mapping from satellite thermal IR data: a case example in Jharia coal field, Jharkhand, India, ISPRS J. Photogram. Remote Sens. 60 (2006) 113–128. [22] R. Priyom, G. Arindam, K. Vinod, An approach of surface coal fire detection from ASTER and Landsat-8 thermal data: Jharia coal field, India Int. J. Appl. Earth Obs. 39 (2015) 120–127. [23] O. Carpentier, D. Defer, E. Antczak, B. Duthoit, The use of infrared thermographic and GPS topographic surveys to monitor spontaneous combustion of coal tips, Appl. Therm. Eng. 25 (2005) 2677–2686. [24] Y. Sheng, Y. Wang, Mining spatial information in surface temperature measurements of coal gangue with infrared thermography, Infrared. Technol. 29 (2007) 59–61. [25] P. Arias, E. Herra, Control of structural problems in cultural heritage monuments using close-range photogrammetry and computer methods, Comput. Struct. 83 (2005) 1754–1766. [26] A. Mostafavi, C. Cold, M. Dakowicz, Delete and insert operations in Voronoi/ Delaunay methods and applications, Comput. Geosci. 29 (2003) 523–530. [27] J. Zhang, 3D detection and visualization of underground coal fires. Int. Conf, China, (2005) 410–426. [28] C. Balaras, A. Argiriou, Infrared thermography for building diagnostics, Energy Build. 34 (2002) 171–183.

[29] N. Avdelidis, A. Moropoulou, Applications of infrared thermography for the investigation of historic structures, J. Cult. Herit. 5 (2004) 119–127. [30] N. Eddie, Y. Kawb, W. Chang, Analysis of IR thermal imager for mass blind fever screening, Microvasc. Res. 68 (2004) 104–109. [31] S. Bagavathiappan, B. Lahiri, T. Saravanan, J. Philip, T. Jayakumar, Infrared thermography for condition monitoring – a review, Infrared Phys. Technol. 60 (2013) 35–55. [32] R. Vardasca, J. Gabriel, P. Plassmann, Towards a medical imaging standard capture and analysis software, in: 12th Int. Conf. on Quantitative Infrared Thermography, Bordeaux, France, 2014 162–168. [33] J. Lerma, S. Navarro, Terrestrial laser scanning and close range photogrammetry for 3D archaeological documentation: the Upper Palaeolithic Cave of Parpallo as a case study, J. Archaeol. Sci. 37 (2010) 499– 507. [34] P. Wolf, B. Dewitt, Elements of Photogrammetry, With Applications in GIS, 3nd ed., McGraw-Hill, Boston, 2000. [35] Z. Zhang, J. Zhang, Generalized point photogrammetry and its application. Archives of ISPRS 2004 Congress B. 5 (2004) 77–81. [36] http://www.lensoft.com.cn/?COLLCC=2366020578&#. [37] W.D. Li, N. Lin, X. Chen, Research on 3D tunnel modeling based on close-range photogrammetry, Adv. Mater. Res. 2014. [38] X. Wu, W. Feng, K. Wang, Application of multi-baseline digital close-range photogrametry technique in 3D reconstruction of underground tomb, Adv. Mater. Res. 346 (2012) 847–851. [39] T. Luhmann, S. Robson, S.A. Kyle, Close range photogrammetry: principles, techniques and applications, 1st ed. Whittles, 2006. [40] L.G. Brown, A survey of image registration techniques, ACM Comput. Surv. (CSUR) 24 (1992) 325–376. [41] J.B.A. Maintz, M.A. Viergever, A survey of medical image registration, Med. Image Anal. 2 (1998) 1–36. [42] F. Aurenhammer, Power diagrams: properties algorithms and applications, SIAM J. Comput. 16 (1987) 78–96. [43] S. Lo, Delaunay triangulation of non-uniform point distributions by means of multi-grid insertion, Finite. Elem. Analy. Des. 63 (2013) 8–22. [44] B. Wang, B. Khoo, Z. Xie, Z. Tan, Fast centroidal Voronoi Delaunay triangulation for unstructured mesh generation, J. Comput. Appli. Math. 280 (2015) 158– 173. [45] P. Smith, Bilinear interpolation of digital images, Ultramicroscopy 6 (1981) 201–204. [46] K. Gribbon, D. Bailey, A novel approach to real-time bilinear interpolation. Field-Programmable Technology, Proceedings, IEEE Int. Conf., 2004. [47] P. Fredericp, F. Agnes, Bilinear Interpolation, 1st ed. Germany, 2010. [48] H.J. Wu, H.J. Wu, H.J. Wu, H.J. Wu, H.J. Wu, F.Y. Zeng, H.F. Yao, Danger evaluation and control technology of coal mine gangue spontaneous combustion. Coal, Sci. Technol. 41 (2013) 119–123 (in Chinese). [49] P. Lu, G.X. Liao, J.H. Sun, P.D. Li, Experimental research on index gas of the coal spontaneous at low-temperature stage, J. Loss Prev. Process Ind. 17 (2004) 243–247. [50] Z. Song, C. Kuenzer, H. Zhu, Z. Zhang, Y. Jia, Y. Sun, J. Zhang, Analysis of coal fire dynamics in the Wuda syncline impacted by fire-fighting activities based on in-situ observations and Landsat-8 remote sensing data, Int. J. Coal Geol. 141– 142 (2015) 91–102. [51] Z. Song, C. Kuenzer, Coal fires in China over the last decade: a comprehensive review, Int. J. Coal Geol. 133 (2014) 72–99. [52] R.V.K. Singh, V.K. Singh, Mechanised spraying device – a novel technology for spraying fire protective coating material in the benches of opencast coal mines for preventing spontaneous combustion, Fire Technol. 40 (2004) 355–365.