Applied Thermal Engineering 25 (2005) 2677–2686

www.elsevier.com/locate/apthermeng

The use of infrared thermographic and GPS topographic

surveys to monitor spontaneous combustion of coal tips
O. Carpentier, D. Defer *, E. Antczak, B. Duthoit
Laboratoire d’Artois de Mécanique Thermique Instrumentation (LAMTI), FSA—Université d’Artois,
Technoparc Futura, 62400 Bethune, France

Received 9 February 2004; accepted 30 November 2004

Abstract

Aerial infrared thermography is one of the methods most widely used for monitoring coal tips. Although
it offers the advantage of providing wide coverage of the site and clearly displaying areas at risk on the ther-
mal image, it is costly, sensitive to atmospheric conditions and cannot be used to monitor combustion reac-
tions. It is for these reasons that the LAMTI, in liaison with the Charbonnages de France Group, has
developed a ground-level method based on a combination of GPS topographic and infrared thermographic
surveys that is intended to be more precise and less expensive and may eventually enable combustion reac-
tions to be monitored.

2004 Elsevier Ltd. All rights reserved.

Keywords: Infrared thermography; Coal tips; Digital terrain model; Monitoring

1. Introduction

Coal tips (Fig. 1) are conical or truncated-cone hills formed by piling up the waste from mining
operations. Some of them still possess 5–15% of residual coal, the percentage varying according to
the sorting techniques used. The oldest tips thus often contain more residual coal. The porous

*
Corresponding author. Tel.: +33 3 21 63 71 55.
E-mail address: [email protected] (D. Defer).

1359-4311/$ - see front matter
2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2004.11.017
2678 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686

Fig. 1. An example of conical coal tips.

nature of the tip enables air and water to circulate, causing the coal to oxidise. Any iron sulphides
such as marcassite and pyrites present in the tip will tend to catalyse the exothermal oxidation
reaction, causing spontaneous combustion of the residual coal [1]. In north of France, 10% of coal
tips (above 30) are undergoing a process of spontaneous combustion, the effects of which are as
follows [2]:

• Slipping of the tip, accompanied by clouds of inflammable dust.

• Formation of cavities.
• Emission of noxious gases.
• Creation of water–gas, a compound that explodes in air.

Most failures occur during heavy rains, with penetrating water having the following additional
direct effects:

• Increase in the density of the materials.

• Drop in shear strength.

Fig. 2. Slips effects of spontaneous combustion in a conical coal tips.

 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686 2679

• Increase in pore pressures.

• Downward movements, creating destabilising forces.

The effects can be particularly catastrophic. Slips (Fig. 2) are not classic rotational ones but, on
account of the high water content in the schists, are more similar to mud slides; these can be
deadly, as experienced in Virginia in the 1940s. There is also the added problem of major atmo-
spheric pollution due to the release of sulphurous fumes [3]. As it is very expensive to fight coal tip
fires, attempts have been made since the second half of the 20th century to prevent such fires dur-
ing the formation of coal tips by taking precautions, such as for example not creating them above
carbonate materials or spreading them out in order to reduce internal pressures. In the case of tips
formed before this period, spreading them can be dangerous and costly if high-risk areas are not
known, and monitoring then appears to be a necessary step.

2. Monitoring of coal tips

Today there are many methods for monitoring soils, such as radar methods or electrical scanning.
These have been applied to coal tips in order to draw up surface maps or profiles that link resistivity
to soil stability. The main drawback of these methods is that they involve working in high-risk areas,
as cavities formed at the surface can collapse under the weight of a man. Optical fibre methods [4]
have also been used, but they appear to be more suited to detecting spontaneous combustion in

Fig. 3. The monitoring of coal tips by aerial infrared thermography (Charbonnages de France Group files).
2680 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686

mining galleries, which have a more slender profile than massive coal tips, where a surface approach
is more appropriate. It is a known fact that the state of a soil subjected to stress (thermal, mechan-
ical, etc.) can be observed by infrared methods [5]. This is especially true of rocks and coal deposits.
Spontaneous combustion in coal deposits has already been modelled [6] as well. Here, on the basis of
external information, the internal development of the combustion phenomenon is determined.
Infrared thermographic monitoring can therefore display the internal activity in a coal tip on its sur-
face. The Charbonnages de France Group uses aerial infrared thermography (Fig. 3) to monitor
coal tips. Moreover, these are not the only igniting masses to be monitored. Fermenting public
dumps, in which the formation of internal heating can be compared to that of coal tips, are also
monitored by aerial infrared thermography [7], while the usefulness of ground-level monitoring
for checking anomalies is also recognised. When conducting aerial monitoring surveys, it is very dif-
ficult or even impossible to take into account certain data such as the topography, the relative
humidity or CO2 level, the influence of the atmosphere, or the directional aspect of emissivity.

3. Combined topography and thermography

Combining thermography and topography (Fig. 4) means that it is possible to work over a large
surface area and quickly. Once anomalies have been determined precisely, they can be monitored
with regard to changes in temperature and displacement of the source. Information gathered in
the field indicates the environmental conditions. Subsequently, the site can be monitored and it
will be possible to tell whether temperature measurements indicate internal activity within the
tip or are rather the direct result of environmental conditions.
Three stages are involved:

• A topographic and thermographic survey in the field.

• Processing of the topographic data, pixels and infrared images to produce a 3D thermogram.
• Post-processing of the temperatures, incorporating the environmental components.

3.1. On the field

An AGEMA 570 FPA monitoring camera (Fig. 5) was used with a resolution of 320 · 240 pix-
els working in the 8–14 lm wavelength band, which is suited to low temperatures. The point set is

Fig. 4. Combination of IR and GPS methods.

 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686 2681

Fig. 5. IR camera on the field.

gathered with a Trimble type GPS accurate to within a centimetre. The environmental data are
collected with a portable probe.
The topographic survey must be precise enough to offer a satisfactory reconstruction of the site
so that the projection error between the images and the DTM (Digital Terrain Model) is as small
as possible. The thermal camera must be treated as both a monitoring tool and a theodolite. It is
first necessary to carry out an investigation of the site as areas subject to slips (Fig. 6) can some-
times be identified with the naked eye (previous slips, fumes escaping, absence of vegetation).
In this case, the thermographic survey will help especially to detect any movement of the source
and locate future risk areas. Fig. 6 clearly shows the slip area, due to spontaneous combustion,
over a wide strip running from the summit eastwards, but it is also possible to detect areas of heat-
ing to the south and north-east that are not visible to the naked eye and that perhaps correspond
to movement of the source or the opening of a vent from the source to the surface. Surface tem-
peratures collected by the heat probe also confirm that these apparently inert areas have surface
temperatures of 50 C and temperatures reaching 90 C at a depth of 20 cm.

Fig. 6. Landslide due to spontaneous combustion of residual coal.

2682 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686

During the field investigation stage, the camera must be levelled as a topographic instrument
since the quality of the reconstructed image will depend on the orthogonality of the images.
All points of the survey are determined by global positioning system (GPS), including the camera
stations and image references. It is the link between the surveyed reference determined by GPS
and simultaneously displayed by the camera that will enable thermographic information to be
added. With regard to the actual monitoring, coal tips undergoing combustion generally provide
good contrasts. However, work should not be carried out when it is raining as this tends to make
surface temperatures uniform, or in strong sunlight, since the addition of atmospheric radiation is
a source of error in surface temperatures. Lastly, low angles should be avoided, as the directional
nature of emissivity means that its value decreases with distance from the normal angle of obser-
vation. The optimum conditions for measurements of this type are generally found at night as this
is when the best contrast is obtained and the effect of atmospheric radiation is much lower. Sur-
face temperature measurements made with a probe can nevertheless be used to correct the results
obtained with a thermal camera. The raw data obtained in this way must therefore be processed,
and this can only be done by collecting information on the site.

3.2. Plotting a 3D thermogram

In the area of research and application, the DTM has become a common surveying technique
for identifying forms of terrain. Infrared thermographic scanning creates specific geometric prob-
lems. Backing up imagery with topographic data (DTM) is a way of achieving a better under-
standing and higher degree of accuracy [8]. To prepare a 3D thermogram, a topographic
method (GPS survey) is used with a view to forming a base for future modelling. The location
of coal tips and their simple nature overcomes the main problems arising with the GPS, namely
multipath and mask phenomena. A GPS survey is collected in the form of a set of points. As the
thermographic survey is combined with the topographic survey, it is possible to obtain georefer-
enced images and pixel coordinates can therefore be defined as 3D polar coordinates in relation to
the camera, within a system chosen by the user (either national or local).
Two successive grids will be used to obtain the DTM, namely a triangular grid obtained by a so-
called Delaunay triangulation, which will be used for the actual topographic model, and a quad-
rangle grid that will transpose the quality of the thermogram reconstruction into three dimensions.
In the case of the Delaunay triangulation, the associated convex envelope is covered by a set of ele-
ments, in this case triangles. This method is based on the Voronoi diagram [9]. The point set con-
sists of a set of vertices P containing n points Pi with (x, y) coordinates. A Voronoi polygon is then
associated with each point, each polygon being associated with a site Pi of region Vor(Pi), and
each region being all the points (x, y) closest to point P such that the nearest site to each point
of P is Pi. Unifying all the Voronoi regions produces the Voronoi diagram on which the reconstruc-
tion is based. All defined vertices are grouped into triplets. Each triangle is defined by its vertices
[Ki = (Pi, Pi+1, Pi+2)]. After triangulation in the plane, the sides are elevated along Z [Zi = f(Pi,
Pi+1, Pi+2)]. The quadrangle grid is superimposed on the triangle grid (Fig. 7), providing a working
base for adding the thermographic information. (x and y axis in Figs. 7, 9, 10 are in Lambert na-
tional topographic system, 1 · 10
3 variation of coordinates on graph represents 100 m on field.)
As the images are georeferenced, they can be superimposed on the DTM (Fig. 8) by considering
each pixel of each image taken during the monitoring survey. However, optical deformation by
 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686 2683

Fig. 7. Coal tips DTM.

Fig. 8. IR data link to DTM.

the camera lens introduces constant angular errors that distort the image in proportion to the dis-
tance from the central pixel. Part of the thermographic image must therefore be eliminated. The
part to be eliminated will be determined by the tolerance applied to the angular distance in rela-
tion to an image that has not been distorted in any way. As an example, an error of one pixel at
100 m produces a difference of 13 cm. A tolerance of 20 cm applied to the plotted points elimi-
nates about 15% of the pixels from the image.
This is therefore not thermal imagery but a thermographic data base that can be processed as
required, in particular for views (Figs. 9 and 10), but also and especially for temperatures, which
are collected by infrared camera and determined on the basis of the directional luminance of black
bodies in a spectral band [10].
Z kb
c1  k
5
LCN
ðk;T ;hÞ ¼  dk ðW m
2 sr
1 Þ
ka expðc 2 =kT Þ
1
2684 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686

Fig. 9. East view of 3D thermogram with localisation of combustion areas.

Fig. 10. Aerial view simulation.

k: wavelength (m),
c1 and c2, Stephan constants,
c1: 1.191062 · 10
16 W m
2 sr
1,
 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686 2685

c2: 1.4388 · 10
2 m K,

T: temperature (K).

This apparent temperature is not the temperature of the object. Real objects are not Lamber-
tian bodies, and therefore the directional emissivity of the object is taken into account. This may
be defined as follows:
R kb c1 k
5
0 ka expðc2 =kT m Þ
1
 dk
eðk;h;T Þ ¼ R kb c1 k
5
ka expðc2 =kT r Þ
1
 dk

Tm: measured temperature (K),

Tr: real temperature (K).

As raw data were measured on site, the parameters can be included in the radiometric equation
for the camera [11].
Lmes ¼ s  e  Lobj þ ð1
eÞs  Lenv þ ð1
sÞLatm ðW m
2 Þ

Lmes: measured luminance,

Lobj: object luminance,
Lenv: environment luminance,
Latm: atmosphere luminance,
e: object emissivity,
s: atmospheric transmission factor.

The three-dimensional aspect of the measurement means that the luminance values associated
with the topography can be weighted on the basis of the directional emissivity curve of a material:
Pn i 2 i
Pn i i 2
i¼1 s  ðeh Þ Lmes i¼1 Lenv  ð1
eh Þ
Lðx;y;zÞ ¼ P n i
þ P n i
þ ð1
sÞ  Latm ðW m
2 Þ
e
i¼1 h i¼1 ð1
eh Þ

3.3. Post-processing of temperatures

Post-processing can take place after reconstruction, as the coordinates and temperatures of
all the points are known. This means, for example, that several emissivity values can be selected
from the same thermogram (coal tip, vegetation, etc.). The environment data must then be
included. As the environment plays a predominant role in the in situ measurements [12], it was
necessary to incorporate parameters such as the atmospheric transmission factor and the notion
of thickness of the atmospheric layer into the measurement processes. These parameters are cur-
rently being looked at in the laboratory in the context of a study of ground-level thermographic
monitoring specifically in relation to the 8–12 lm spectrum band. General nomographs [13] can
also be used. The atmospheric temperature and relative humidity can also be determined by
the probe. Measurements are taken from the camera without any internal compensation (e = 1
2686 O. Carpentier et al. / Applied Thermal Engineering 25 (2005) 2677–2686

and RH = 50%). Compensation is carried out during post-processing of the data by including
environmental data.

4. Conclusion

Ground-level infrared thermographic monitoring has the advantage of being more rigorous,
more flexible and less costly than aerial monitoring surveys. Characterising the emissive character
and modelling the micro-climate of the study environment are the keys for proposing a method
for monitoring coal tip combustion reactions. It should be noted that an automatic system is cur-
rently being developed; this should enable the two methods to be linked without there being any
need for the processes to be simultaneous, and should also minimise the need for operators in the
field or even eliminate it if reconstruction is done on the basis of existing topographic data.

Acknowledgement

This work takes part of an Urban Engineering Concerted Research Action and authors are
grateful to the Charbonnages de France Group and to the Nord-Pas de Calais Region for sup-
porting this work and allowing the publication of this results.

References

[1] D. Harrington, J.H. East, Incendies des terris in Annales des mines de Belgique 48 (1948) 427–434.
[2] A. Monjoie, C. Schrodoer, Instabilité de versant de terrils en relation avec lÕautocombustion du charbon résiduel,
Revue française de géotechnique 95–96 (2001) 91–102.
[3] F.H. Hebley, J. Lecomte, La lutte contre les feux de terrils, Annales des mines de Belgique 48 (1948) 709–712.
[4] M.J. Wilett, J.S. Wykes, G.A. Powell, M.J. Gibson, A. Cooke, The use of a distributed optical fibre temperature
sensing system to monitor underground mine roadways for spontaneous combustion, Mining Technology 1 (1995)
301–305.
[5] L. Wu, J. Wang, Infrared radiation features of coal and rocks under loading, International Journal of Rock
Mechanics and Mining Sciences 35 (7) (1998) 969–976.
[6] N.P. Cheremisinoff, Spontaneous combustion in bed of coal, Handbook of Heat and Mass Transfer, 4, GulfPub,
1986, pp. 1071–1148.
[7] F. Guigoures, Characterisation of the thermal behaviour of a solid–fluid interface—Application to the monitoring
of waste by aerial infrared thermography, Ph.D. Thesis, 2001.
[8] D.-L.-B. Jupp, Directional radiance and emissivity measurement models for remote sensing of the surface energy
balance, Environmental Energy & Software, 1998.
[9] P.-L. Georges, Triangulation de Delaunay, 1997.
[10] W. Rohsenow, J. Hartnett, Y. Cho, Handbook of Heat Transfer, 1998.
[11] L. Ming Te, W. Jyi, Predicting the integrity of existing prestressed concrete bridge using sonic method, Journal of
Marine Science and Technology 9 (1) (2001) 45–52.
[12] T. Hamrelius, Accurate temperature measurement in thermography, AGEMA Infrared system AB Department 13
(3–4) (1997) 341–351.
[13] X. Maldague, Infrared thermography for Nondestructive Testing, 2001.