Comparison of methods to estimate the rate of CO2 emissions and

International Journal of Coal Geology 86 (2011) 95–107
Contents lists available at ScienceDirect
International Journal of Coal Geology
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j c o a l g e o
Comparison of methods to estimate the rate of CO2 emissions and coal consumption
from a coal fire near Durango, CO
S. Taku Ide a,⁎, F.M. Orr Jr. b
a
Energy Resources and Engineering, Stanford University, Green Earth Sciences Building, Room. 65, 367 Panama St, Stanford, CA, 94305-2220, United States
Energy Resources and Engineering, Stanford University, Precourt Institute for Energy, The Jerry Yang & Akiko Yamazaki Environment and Energy Building, Mail Code 4230,
Room 324, 473 Via Ortega Stanford, CA 94305-4230, United States
b
a r t i c l e
i n f o
Article history:
Received 14 June 2010
Received in revised form 13 December 2010
Accepted 13 December 2010
Available online 20 December 2010
Keywords:
Coal fires
CO2 emissions
Magnetometer
Rate estimation
Natural convection chimneys
Surface subsidence
a b s t r a c t
Subsurface fires in coal beds consume coal resources and contribute to the global emissions of CO2 and air
pollutants. Many of these fires are found in China, India, Indonesia, and the United States. Combustion product
gases at these coal fires exit through surface fissures that form over fires. These fissures are created when
subsurface subsidence causes preexisting fractures in the area to widen. Fissures act as both inlets for air and
exhaust for combustion gases. While remote sensing approaches have been used to quantify the rate of coal
consumption and CO2 emissions at large scale fires that extend over large distances, methods for estimating
the coal consumption and CO2 emissions values based on surface observations are less well established.
In this paper, a coal fire near Durango, CO, is described. A combination of fissure mapping, thermocouple
temperatures, and a cesium-vapor magnetometer survey was used to delineate the aerial extent of the current
combustion zone and previously burned zones.
Three methods were then used to estimate combustion rates at an active region at the site. In the first method,
time-lapse, high-resolution topographic surveys were used to relate surface volumetric losses over the active
region to coal consumption and rates of CO2 emission. In the second method, measured temperatures, gas
compositions, and dimensions of an exhaust fissure were used in a simple natural convection chimney model
to estimate rate values. The third method estimated coal consumption and CO2 emission rates by measuring
the velocity of exhaust gases, gas compositions and exhaust fissure dimensions. For the second and third
methods, 13C isotope signatures were used to determine the fractions of CO2 that were emitted from coal and
CH4 combustion or from CO2 in the native gas in the coal seam. A flux accumulation chamber was also used to
quantify CO2 leakage rates from non-fissured regions over an active fire region. The three methods produced
roughly consistent estimates of coal combustion and CO2 emission rates.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
Uncontrolled subsurface fires in coal beds account for significant
releases of CO2 to the atmosphere. A single coal fire documented in
Wuda, China is estimated to consume approximately 200,000 tons of
coal per year (Kuenzer et al., 2005), equivalent to around 0.60 Mt of CO2.
In addition to the problem of CO2 emissions, gases released into the
atmosphere from these fires are often toxic. Furthermore, the loss of coal
volume in the subsurface can lead to significant surface subsidence and
fissures, resulting in damages to near-surface or surface infrastructures.
Coal bed fires are burning in many locations in China, Indonesia,
India, and the United States (Stracher and Taylor, 2004). They can be
started naturally by forest fires that burn near a coal outcrop, by
lightning strikes that ignite trees that subsequently ignite an outcrop,
⁎ Corresponding author. Tel.: + 1 650 868 6575 (mobile), + 1 650 725 0801 (office).
E-mail addresses: [email protected] (S.T. Ide), [email protected] (F.M. Orr).
0166-5162/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.coal.2010.12.005
by human activities, or by spontaneous exothermic reactions of pyrites
(DeKok, 1986). Spontaneous combustion of coal can occur when
coal oxidizes slowly in air, which generates heat. When this heat
accumulates–usually in confined settings–it can cause volatiles to
evolve from the coal. These volatiles can then react with available O2 to
provide more heat to sustain and propagate combustion. This type of
ignition is more likely in coal refuse piles in which coal particles are
surrounded by air.
Forest fires in Indonesia in 1997 and 1998 ignited hundreds of
coal fires at outcrops (Brown, 2003). In the U.S., a subsurface fire
near Centralia, Pennsylvania, was started in 1962, when the local
government decided to burn an unregulated trash dump in an
abandoned strip mine to reduce trash volume and control rodents.
The fire ignited an anthracite outcrop, eventually connected to and
spread through underground tunnels, and has been burning since.
A combination of subsidence and emissions from fissures has caused
the town of Centralia to be abandoned (DeKok, 1986; GAI Consultants,
1983). Fires that occur in abandoned coalmines such as the one in
Centralia obtain their supply of oxygen and convect exhaust gas away
96
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
from the combustion region through mine tunnels and shafts (GAI
Consultants, 1983; Kim and Chaiken, 1993).
Subsurface fires can also occur in unmined coal beds when an
outcrop ignites, and the resulting combustion front subsequently
burns into the formation away from the outcrop. In these natural fires,
O2 inlets and exhaust gas outlets are not as well defined as those in
mine fires. Exchanges of gases between the surface and the coal seam
occur through fissures that connect the surface and the coal seam.
Fissures are formed when ash and void areas that result from coal
combustion and gasification collapse under the overburden pressure
(Ide et al., 2010).
Most abandoned underground mine fires stay close to old mine
tunnels, but they can sometimes burn away from the old mining
network and burn into the formation (GAI Consultants, 1983). Once
the fire burns into the formation, surface features similar to those of
natural coal bed fires are observed. In the United States, nearly 100
abandoned underground mine fires, excluding natural coal bed fires,
across 10 states were documented by the Department of Interior
(DOI) in 1988. The costs to control or extinguish these fires were
estimated to cost around $741 million at that time (Kim and Chaiken,
1993). Since natural coal bed fires do not fall under the category of
Abandoned Mine Land (AML), these fires are often not monitored by
the DOI.
Several authors have suggested a range of CO2 emission rates from
coal fires in China (Kuenzer et al., 2005; Rosema et al., 1993). For large
coal fires that use remote sensing methods such as Landsat-7 ETM+, an
infrared satellite, to quantify CO2 emissions is appropriate (Prakash and
Gupta; 1998, Kuenzer et al., 2007a). When the fires are small, however,
satellite images typically do not offer sufficient spatial resolution to
quantify coal combustion accurately. In this paper we consider a small
coal fire located along the Hogback Monocline of the San Juan Basin
that spans an area of roughly 600 m × 200 m. The best deformation
information that could be obtained for this area using InSAR satellite
data is at a resolution of 50 m × 50 m. That resolution is too coarse to
allow accurate estimates CO2 emission or coal consumption rates.
To overcome spatial resolution issues, we first test methods to
determine more precisely the current extent of the combustion
zone. We used three approaches to identify the current location of the
active combustion zone and the previously burned zone. In the first
approach, the spatial distribution of surface fissures was mapped using
a portable GPS unit, and the temperature of gas present at each fissure
was measured. The second approach used subsurface temperatures
measured using thermocouples installed in boreholes drilled into the
coal seam. The third approach delineated the combustion region using
a cesium vapor magnetometer. As we show below, the magnetometer
measurements provided the most useful determination of burned,
currently burning, and unburned areas. The use of magnetometer data
to detect coal fire boundaries has been reported by several investigators
(Bandelow and Gielisch, 2004; Gielisch, 2007; Hooper, 1987; Schaumann and Yu Change, 2005; Sternberg, 2004; Sternberg et al., 2008),
though the O(1) m resolution over a coal fire of this size is more detailed
than those previously described. In addition, the magnetometer results
over the North Coal Fire have been corroborated using well-logs, driller's
logs, core samples, and temperature measurements. Finally, data have
been filtered by removal of diurnal fluctuations, by signal amplitude and
by pole reduction.
Based on the estimated locations of current combustion, previously burned, and unburned zones, three methods were used to
estimate quantitative rates of CO2 emissions and coal consumption.
The first method estimates coal consumption rates from time-lapse,
high-resolution surface deformation measurements at different
points. Data acquisition points were obtained near the hottest fissures
found over the North Coal Fire. The second method uses measured
velocities of exhaust gases from a fissure over an active region and
combined that estimate with estimates of amounts of CO2 seeping
from the less fractured areas over the active fire. The third makes use
of an analogy between coal fires and natural convection chimneys
to estimate a range of CO2 emission rates. The dimensions and the
measured subsurface temperatures of the most active fissure over the
site were used to set the chimney geometry and the thermal gradient.
In the first method (the subsidence method), the coal consumption
rate was calculated by taking the volume of subsidence and
converting it to mass by using the density of coal. In the second and
third methods, CO2 emission rates were calculated first. For those
methods, 13C isotope signatures were used to determine the fractions
of CO2 that were emitted from coal combustion/gasification, from
CH4 combustion, and from CO2 in the native gas in the coal seam. A
simple stoichiometric relationship was then used to convert to and
from coal consumption rates and CO2 emission rates. The results
presented provide first-order estimates of the rates of CO2 production
and coal consumption from this fire.
2. Fire location
A detailed description of the San Juan Basin, local geologic setting,
and fire location is given by Ide et al. (2010). The fire considered here
is located in the San Juan Basin, along the Hogback Monocline, about
50 km southwest of Durango, CO. There are four known fires along the
Hogback Monocline. This particular fire is termed the North Coal Fire
to distinguish it from a South Coal Fire that lies two miles to the south.
The coal layer that is burning is in the Fruitland Formation (Fassett,
2000). It is one of three coal layers in the Fruitland separated by
shales, sandstone, and clay layers, though at this location, the top two
coal layers have been eroded away. The coal is overlain by fractured
sandstone and shale, and is underlain by the Pictured Cliffs sandstone.
The coal layer dips to the southeast at about 11°. The depth to the
top of the coal in the zone that is burning is about 12–15 m, and the
coal layer thickness is approximately 5–7 m. The Fruitland Formation
outcrops along the Hogback Monocline, and it is believed that trees
struck by lightning or a forest fire at the outcrop ignited the coal seam.
As the combustion front moved subsequently into the formation,
the loss in structural integrity in the burned coal seam resulted in
subsidence. Many surface fissures that formed due to subsidence are
observed over the North Coal Fire today. Exhaust gases produced from
the coal combustion, some as hot as 1000 °C, flow out of some of these
surface fissures, while others appear to be air intakes or are inactive.
CH4 and CO2 are also present in the coal in areas that have not been
affected by combustion. These gases desorbed from the coal, and they
support coal bed methane production in the San Juan Basin. They also
contribute to the observed outflow of gases from the fissures at the
North Coal Fire, as the isotope signatures reported below document.
3. Extent of the North Coal Fire
3.1. Surface fissure distribution
Surface fissures observed at the North Coal Fire are similar to those
observed at coal fires around the world (Cao et al., 2007; Huang et al.,
2001; Kuenzer et al., 2007b; Wessling et al., 2008), and are surface
manifestations of the coal fire burning below. Fissures observed over
the North Coal fire have apertures ranging from 0.02 m to N ~1.5 m,
and they connect the surface and the coal seam. These fissures were
mapped in 2007. Fissures are observed only in areas affected by the
fire, and they are not present in neighboring areas where the coal
seam is intact. At the North Coal Fire, fissures with narrower apertures
typically vent hot exhaust gases, while the wider fissures are at
ambient temperatures at the surface. The geomechanical mechanisms
of formation of these fissures are discussed in detail by Ide et al.
(2010). Fissures at the North Coal Fire were mapped using a packmounted Trimble ProXH GPS unit, with accuracy better than 0.5 m
after correction. Fig. 1 reports the spatial distribution of fissures over
the North Coal Fire. The fissures shown in the figure are color coded by
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
Fig. 1. Fissure distribution over the North Coal Fire. Fissures are not observed in
neighboring regions where the coal seam has not burned. The blue fissures in the figure
are at ambient temperatures, while the red fissures indicate those that are venting hot
gases. Two active fire regions can be observed in this figure, one in the north (Crestal
Extension Fire, circumscribed by a red box), and one in the south. In this and
subsequent figures, the location of the outcrop of the Fruitland Formation is between
the green diagonal lines. Boreholes drilled in 2007 are shown in black, those drilled in
2010 are shown in blue and red triangles.
thermal anomalies. The blue fissures are at ambient temperatures,
while the red fissures are those that are venting gases with elevated
temperatures above approximately 65 °C.
The distribution of fissures indicates that the surface over the
North Coal Fire is highly fractured, allowing for air and combustion
gases to exchange freely between the surface and the coal seam.
Thermal signatures presented in Fig. 1 show that there are currently
two active combustion regions, one to the north, which will be
referred to as the Crestal Extension Fire, and one to the south. The
Crestal Extension Fire is circumscribed by a red box in Fig. 1. In this
paper, only the Crestal Extension Fire will be considered for rate
estimations, so the estimates obtained here reflect a portion of the
overall emissions and coal consumption at the North Coal Fire.
3.2. Thermocouple temperature results
Approximately 40 boreholes were drilled previously over the North
Coal Fire at locations with an average spacing of about 30 m in 2000,
and all of these boreholes were equipped with subsurface thermocouples. In 2007, an additional 15 boreholes were drilled at the site,
with thermocouples installed in each. As of 2010, 18 thermocouples
were functional over the Crestal Extension Fire region of the North
Coal Fire. The thermocouples are Type-K and Type-E thermocouples
produced by Fluke Inc. Interpolation of the temperatures measured at
the 18 thermocouples does not give a particularly precise location of
the combustion front, since any temperature information between the
boreholes cannot be determined. Nevertheless, a contour map created
from thermocouple temperatures gives an approximate location
of the heated zone that can be compared with other measurements.
A temperature contour map constructed from the temperatures
measured in 2010 is shown in Fig. 2. Temperature changes have
97
Fig. 2. A temperature contour map constructed using 18 thermocouple temperatures
measured in 2010 over the Crestal Extension. Black squares denote thermocouples
deployed in 2000. Black triangles indicate thermocouples installed in 2007. Colored
triangles indicate thermocouples emplaced in 2010. The red (hot) and white (cold)
lines are fissures as in Fig. 1. The magenta dashed line encloses an area where snow
had melted on the day of a snowstorm in 2010; snow remained on the surface outside
that area.
been recorded at this fire since 2001, which show a steady migration of
the fire in the NW direction towards the current hot spot in 2010. Since
2007, the location of the hottest zone has not changed substantially,
but the temperatures observed in that zone have increased.
The magenta dashed line in Fig. 2 is a snowmelt boundary
measured using a pack-mounted GPS on the day of a snowstorm in
2010. There was no snow inside of the snowmelt outline due to the
heat conducted to the surface from the subsurface fire. Areas outside
of the snowmelt outline were covered by snow. Note that most of the
thermally elevated fissures (red fissures) lie within the snowmelt
boundary, while the fissures at ambient temperatures (white fissures)
lie outside of it. High temperature regions delineated by the snow
boundary are consistent with the high temperature regions identified
by thermocouple temperatures.
3.3. Cesium vapor magnetometer
A higher-resolution contour map of the current location of the
combustion zone was obtained using a cesium vapor magnetometer.
Interpretation of the survey was based on the following assumptions.
First, since the Fruitland Formation was deposited in a shoreline
environment, it is reasonable to assume that the depositional
environment resulted in random or only slightly aligned magnetic
orientations of any magnetic material now present in the sandstone.
Second, we assume that once deposited there were no external
mechanisms–heat, mechanical, or chemical–with sufficient energy to
alter the magnetic orientations. Third, because the temperatures of the
gases emitted from the fissures can be as high as 1000 °C, we assume
that rocks in the subsurface that are in the proximity of the coal bed
98
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
fire would reach temperatures higher than the Curie temperature
of many common iron-containing ferrimagnetic minerals present in
typical sandstones.
Magnetite, a common magnetic mineral found in sandstones, has a
Curie temperature of 585 °C (Walden et al., 1999). Magnetites can also
be created in the presence of heat and low O2 concentrations (Hooper,
1987). These newly formed magnetites would have no magnetization
if they are formed at temperatures above their Curie temperature.
Furthermore, any magnetization that may have existed in the
overburden prior to the heating would also be lost. When the
combustion zone and surrounding rocks cool below the Curie
temperature, the magnetic orientations of pre-existing and newly
formed magnetites align with the Earth's magnetic field. Such an
alignment gives rise to a net magnetic moment that did not exist prior
to the coal bed fire if there are sufficient amounts of ferrimagnetic
materials (1 ~ 2% by mass) present in the sandstone. This resulting
magnetic alignment can be measured using a portable magnetometer,
which measures the magnetic potential of the ground below the
device. When the ambient magnetic field is subtracted from the field
measurements, zones that are currently hot, burned and cooled, and
unburned can be differentiated. A detailed map of magnetic anomalies
near and over the North Coal Fire was constructed by walking a
grid pattern with O(1) m spacing and using filter algorithms to process
the data. The fissures at the surface do not affect the magnetometer
readings.
A magnetic anomaly map over the entire North Coal Fire region is
shown in Fig. 3. In this figure, the light green to yellow areas represent
unaffected coal seam, the blue zones show regions that have
previously burned, and the red areas denote locations where the
rocks above the coal fire are hot today. We assume that the heated
zone is a reasonable indication of the current combustion zone, though
sufficient time is obviously required for cooling of the overburden
Fig. 3. A map of magnetic anomalies, in nano-Teslas (nT) over the North Coal Fire. The
green to yellow regions indicate unburned coal seams, the blue regions show
previously burned areas, while the red regions represent currently burning regions.
The superimposed white lines are snowmelt boundaries on the day of a snowstorm
in 2010, while the magenta lines are snowmelt boundaries a week after a snowstorm
in 2009.
rocks once the combustion zone has migrated away. The magnetic
anomaly data indicates that the Crestal Extension Fire area, which is
circumscribed by a red box, shows decreased magnetic correlation,
indicating the existence of heating due to an active combustion zone.
Fig. 4A shows an enlargement of the magnetometer results along
with the snowmelt boundary over the Crestal Extension area. Fig. 4B
shows the fissure distribution and the snowmelt outline. The white
(in Fig. 3) or green (in Fig. 4B) lines are snowmelt data taken on the day
of a snowstorm in 2010, while the magenta line in both figures is a
snowmelt data recorded a week after snowfall in 2009. The fire did not
progress significantly over the time period when the two snowmelt
boundaries were mapped. The results in Figs. 3 and 4 are consistent
with the temperature observations reported in Fig. 2 and the
snowmelt data (indicated again in Fig. 3 by the white dotted line).
3.4. Conceptual picture
Fig. 5 is a conceptual picture assembled from field observations,
measurements, and simulations that relate the subsurface collapse
and the fissure formation at the surface (Ide et al., 2010). Fires along
the Hogback Monocline most likely start near the outcrop, and in
the early stages, air is drawn in from the outcrop. As the fire continues
to burn, however, the weight of the rock above the coal seam
causes the roof to sag and eventually collapse. That collapse causes
cracks (fissures) to form where the rock layers are already fractured.
Although fissures can form due to rock failure, the systematic
orientations of fissures over the North Coal Fire suggest that these
fissures are due to preexisting fractures widening under tensile
stresses caused by the subsurface collapse. The fissure orientations are
consistent with preexisting fracture orientations that are observed
along the Hogback Monocline where fires are not present (Ide et al.,
2010). The formation of these fissures can be a result of sudden
rock mechanical failure or a more continuous process. Some fissures
have opened abruptly, while others have widened and lengthened
gradually since 2007 at the North Coal Fire. The rates at which failures
occur can vary depending on both the distribution of pre-existing
fractures and on subsurface combustion patterns. Areas where the
overburden is less fractured can behave more elastically, allowing the
overburden to deform slowly over time.
Fig. 5 is sketched at a time when the fire has moved sufficiently
far from the outcrop so that it obtains most of its air from cooler,
chimney-like inlet fissures. Fissures that form near the advancing
combustion front provide a pathway for hot combustion product
gases to escape. As the combustion front advances past these fissures,
air flows in through one or more of these fissures in the previously
burned zone. Rock cuttings, well logs, drillers' logs, down-hole video
observations, and surface observations suggest that the Crestal
Extension Fire is now drawing air from cool fissures in the area.
Three potential air supply channels exist. The first is the small
cracks in the overburden, the second is from the outcrop, and the third
is from wide, ambient temperature fissures in previously burned areas
near the combustion zones. Of the three possible channels, the path
of least resistance is likely through the fissures. Ambient fissures close
to the combustion zone can have apertures as large as O(1)m, while
cracks in the overburden are typically three orders of magnitude
smaller. Boreholes drilled near the outcrop showed that the ash layers
there were compacted, and the region lacks large void regions
and fractures to allow significant airflow. While airflow from the
overburden or the outcrop is likely not entirely absent, it is more
probable that air is being delivered to the subsurface through surface
fissures, since it provides the path through which air can flow most
easily. Due to the abundance of cool fissures and their large apertures,
the velocity of air inflow at any one of these fissure can be quite low.
When boreholes were drilled in 2010 in between the cool fissures
and hot fissures, a zone of fractured rocks was encountered immediately
above the coal seam. The fractures were observed using a down-hole
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
99
Fig. 4. A) A magnetic anomaly map in nano-Teslas (nT) of the Crestal Extension Fire, and B) surface features over the Crestal Extension Fire. The green line shows the snow boundary
on the day of a snowfall in 2010, the magenta line shows the snow boundary a week after snowfall in 2009. The blue lines are ambient fissures, and the red lines are fissures that are
venting warm to hot gases. The black circle at approximately x = 340 m and y = 500 m is the outline for a steel water tank at the surface, and the lines extending from the circle are
steel water pipes. A portion of the magnetic anomaly in that area was removed due to the presence of the tank and pipes.
camera. Hot air flowed through these fractures, suggesting that air being
drawn into the subsurface is being heated as it flows towards the
combustion zone. It is reasonable to assume that ambient fissures and
the combustion fissures are connected by a region of multiple fractures
in the subsurface just above the burned coal seam. The apertures of the
fractures above the coal seam are O(10− 2~5 × 10− 2) m, which allow
high volumes of air to feed the combustion zone. These fractures likely
resulted when the overburden collapsed under its own weight.
Fig. 5. Conceptual picture of a coal fire with subsidence. Fresh air and hot exhaust gases enter and leave the subsurface through high permeability fissures that connect the coal seam
and the surface.
100
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
4. CO2 emission and coal consumption rate estimates
The fire boundary identified in Section 3 helps define where
measurements and modeling efforts should be concentrated. In
this section, three methods to estimate the rate of CO2 emissions
and the coal consumption rates from the Crestal Extension Fire region
are described. Each method requires a relationship between coal
consumption rates and CO2 emissions rates to convert from one to the
other. While the kinetic complexity of the coal combustion process
cannot be captured fully using a single-step, global reaction, the
overall chemical balances are useful for deriving an estimate of coal
consumption rates from CO2 emission rates, and vice versa. Eq. (1)
describes a stoichiometric combustion reaction between coal and O2.
The molecular formula for the coal was approximated as CH0.9.
Laboratory analysis showed that coal from the North Coal Fire region
has an average formula of CH0.867N0.018O0.096S0.003.
2CH0:9ðsolidÞ þ 2:45ðO2ðgasÞ þ 3:76N2ðgasÞ Þ→2CO2ðgasÞ þ 0:9H2 OðgasÞ þ 9:212N2
ð1Þ
This one-step, global reaction relates coal consumption to CO2
emission, but it cannot be used to represent accurately the kinetics
of the complex sequence of reactions that take place during coal
combustion. Those steps include pyrolysis in the presence of heat (but
in the absence of O2), devolatilization in the presence of heat and O2,
char gasification when hot char reacts with CO2 and/or H2O, and char
combustion when hot char reacts with O2 (Anthony and Howard,
1976; Bredenberg et al., 1987; Gavalas, 1982; Larendeau, 1978;
Thorsness et al., 1978). There are also numerous secondary gas phase
reactions that take place as volatiles and are chemically driven off of
the coal. Temperatures and O2 concentrations in the subsurface at the
coal fire are not uniform, which means that the rate of reaction will be
spatially variable.
Instead of modeling the rates of reaction and numerous complex
chemical reactions involved in reality–which would require initial and
boundary conditions that cannot be well constrained by the measured
data–the rate of reaction was ignored and the complexity of the coal
combustion process was simplified to the stoichiometric relationship.
In effect, we assume that the reaction is fast compared to the flow
process that feeds the O2 to the combustion zone, and the results
presented state that for a given amount of CO2 produced a fixed
amount of coal was consumed under stoichiometric conditions.
In the subsidence approximation approach, where the mass of coal
burned is estimated first, the stoichiometric relationship in Eq. (1) is
applied to convert the mass of coal consumed to the CO2 emissions
rate. In the last two methods, where the mass rate of CO2 is estimated
first, the CO2 stream must first be separated into three groups before
the relationships in Eq. (1) can be applied to convert the CO2
emissions rate to equivalent coal consumption rates. This requirement
stems from the fact that there are three sources of CO2 at the Crestal
Extension Fire: native CO2 flow that is flowing up-dip from the Central
San Juan Basin, CO2 produced from oxidizing the native CH4 flowing
from the Central Basin, and CO2 produced from the coal combustion
reaction. The CO2 contribution from the coal is determined using δ13C
isotope ratios of the various gases.
The isotopic signature of a mixture, δ13C (a measure of the ratio of
stable isotopes 13C:12C, reported in parts per thousand (per mil, ‰)),
is a function of mole fractions of the components in the mixture. Thus
the measured δ13C of the carbon in CO2 and CH4 present in the exhaust
gases can be expressed using Eqs. (2a) and (2b), respectively.
Eqs. (2a) and (2b) state that the δ13C values for the CO2 or CH4 in a
gas mixture are volume fraction or molar fraction weighted sums of
the values for the individual components, assuming an ideal gas
(Keeling, 1960). The unknowns in the first two equations are the five
mole fractions, which are represented by the variables, xcomponent
,
source
where the sources of the components–CO2 or CH4–are coal, native
CH4, or native CO2. Three additional constraints, Eqs. (2c) through (2e),
CH4
2
are needed. The two variables, mCO
exhaust and mexhaust, are measured
mole fractions of CO2 and CH4 in the exhaust gases, respectively.
Eqs. (2c) and (2d) state that the mole fractions relevant to the
average δ13C values in Eqs. (2a) and (2b) must sum to one. Eq. (2e)
reflects an assumption that the ratio of native CO2 mole fraction to
that of native CH4 mole fraction (native ratio) stays approximately
constant. The numerator is the mole fraction of native CO2 in the
exhaust gas at the hot fissures. The denominator is the sum of the
mole fraction of unburned native CH4 in the exhaust gas and the mole
fraction of the native CH4 that has been oxidized to CO2 in the exhaust
gas. The 1:1 mole ratio of CH4 and CO2 in the second term of the
denominator accounts for the fact that one mole of CO2 was created
per mole of native CH4 consumed. An average value of the native ratio
was calculated from the gas compositions of native gases that were
measured in the vicinity of the coal fire but where the coal seam was
unburned (see Table 1). The average value was 0.939, with a range
from 0.873 to 1.043. That average value is consistent with the ratios
observed for gas production wells in the Central San Juan Basin.
13 CO
nativeCH4 oxidized 13 CH4
δ Cnative
coal 13 CO
2
δ Cexhaust
= xCO2 δ Ccoal2 + xCO2
+
13 CH
ð2aÞ
nativeCO2 13 CO2
xCO2
δ Cnative
coal 13 CH
nativeCH4 13 CH4
δ Cnative
4
= xCH4 δ Ccoal4 + xCH4
δ Cexhaust
coal
nativeCH4 oxidized
coal
nativeCH4
xCO2 + xCO2
xCH4 + xCH4
nativeCO2
+ xCO2
ð2dÞ
nativeCO2
2
nativeCH4
4
=
CO
2
d mexhaust
nativeCH4 oxidized
CH
4
d mexhaust
+ xCO
2
native−CO
xexhaust 2
native−CH
xexhaust 4
ð2cÞ
=1
=1
xCO
xCH
ð2bÞ
=
CO2
xnative
CH4
xnative
CO
2
4
d mexhaust
d 1molCH
1molCO
2
ð2eÞ
= 0:939
In the calculations, average values from each of the sources listed
in Table 1 were used to solve the Eqs. (2a)–(2e). To obtain the δ13C of
the CO2 derived from coal, two coal samples were burned to generate
CO2 for subsequent isotope analysis. One sample was unburned coal
obtained from a cored borehole. The second was thermally altered
Table 1
Measured gas compositions and δ13C values measured from boreholes drilled in and
around the North Coal Fire. Gas compositions are in mole fractions, and isotope
CO2
signature values are in parts per thousand (‰). The values of δ13Ccoal
were measured,
CH4
CO2
and the values of δ13Ccoal
were assumed to be equal to δ13Ccoal
.
Source
2
mCO
source
4
mCH
source
2
mN
source
mother
source
Mole fraction
2
δ13CCO
source
4
δ13CCH
source
‰
‰
Exhaust
Well #1
Well #9A
Average burned
0.19
0.17
0.18
0.02
0.01
0.01
0.72
0.77
0.75
0.07
0.05
0.06
− 17.13
− 17.81
− 17.47
− 29.25
− 33.38
− 31.32
Unburned area
Well #3
Well #7
Well #11
Average unburned
0.45
0.49
0.47
0.47
0.52
0.47
0.52
0.50
0.02
0.02
0.01
0.02
0.01
0.01
0.00
0.01
14.00
13.71
14.09
13.93
− 45.36
− 45.14
− 45.23
− 45.24
Collected coal samples
Unburned coal
Thermally altered coal
Average coal
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
n/a
− 27.12
− 26.47
− 26.80
− 27.12
− 26.47
− 26.80
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
coal recovered from a borehole in a region where the top portion of
the coal had burned, but coal below that zone had been heated but
not burned. The measured isotope signatures of the unburned and
the burned coal samples were consistent (− 27.12 and −26.47‰),
which indicates that no fractionation occurred. The isotope values,
CO2
CH4
δ13Ccoal
and δ13Ccoal
in Eqs. (2a) and (2b) are assumed to be equal, since
both the CO2 and the CH4 originate from the coal. The locations of the
boreholes listed in Table 1 are found in Figs. 1 and 2. The resulting
solution indicates that the average CO2 in the exhaust gas stream was
composed of approximately 42.8% (39.2 ~ 51.1%) from native CO2,
44.0% (36.0 ~ 46.0%) from CH4 oxidation, and 13.2% (2.9 ~ 24.9%) from
coal, by mole fractions. Similarly, 75.5% of the small amounts of CH4
in the exhaust gases came from the coal, while the remaining 24.5%
came from the native gas.
4.1. Rate estimation using surface subsidence results
A Nikon DTM 521 total station was used to measure deformation
changes at 171 points over the Crestal Extension Fire with a 7-month
interval between measurements. The first survey was conducted in
April, 2009, and the second in November, 2009. The same 171 points
(±0.0125 m) were used in both surveys. The vertical resolution
measuring the surface deformation was 3.18 × 10− 3 m. The survey
was conducted by Jim Flint of Performance Engineering and Surveying.
Fig. 6 is a contour map of the differences in surface elevation, in meters.
Values between the measurement points were interpolated linearly.
In Fig. 6, some large (O(10− 1) m) subsidence is observed in the current
combustion zone, at x = 350 m, y = 500 m, where fissures are emitting
hot gases. The largest subsidence occurs approximately at x = 380 m,
y = 460 m, which is a zone that was previously burned and now
cooled, according to magnetometer results in Fig. 4. The timing of
magnetometer and surface deformation measurements does not allow
Fig. 6. A contour map of the surface deformation that occurred over a 7-month period
(April–November, 2009). The deformation is in meters. Surface deformation is most
apparent near the hot (red) fissures. The blue dots are the 151 locations where the
measurements were taken.
101
us to determine whether the subsidence occurred during combustion
in the area or thereafter. There are likely new fissures that resulted due
to the surface subsidence that was measured in 2009. The fissures
shown in Fig. 6 were mapped in 2007, and the figure does not include
any fissures that resulted due to the 2009 subsidence.
A lower bound estimate of the rate of coal consumption can be
constructed based on the assumption that the surface deforms exactly
by the volume of coal consumed in the subsurface. To calculate the
change in surface volume, the interpolated subsidence contour map of
the Crestal Extension Fire was gridded using 0.5 m squares. A single,
interpolated deformation value in the z-direction was assigned to
each grid square. A deformation volume was calculated for each grid
square, and the deformations were summed to calculate the surface
deformation volume over the entire area. The calculated volume loss
at the surface using this method was 104 m3 per 7 months, or 178 m3
per year. This rate calculation assumes that the subsidence is more
or less continuous, which would be a reasonable assumption if the
overburden is viewed as an elastic medium that deforms slowly and
continuously under the stress induced by its own weight. At the other
end of the spectrum, the overburden may be so fractured such that
the subsidence takes place only periodically and abruptly. Under such
circumstances, the rate of volume loss, as calculated, would not be
applicable. If the surface volume change were equal to the subsurface
coal consumption volume, and that the density of coal in this region
was assumed to be around 1400 kg/m3, the rate of coal consumption
is about 249 metric tons of coal consumed per year. The ash content of
the coal is accounted for in the density used. Using the stoichiometric
relationship in Eq. (1), this coal consumption rate corresponds to a
CO2 emissions rate of 849 metric tons per year.
4.2. Estimation of rates using exhaust gas flow velocity and flux chamber
accumulator measurements
A second estimate of the rates of CO2 emission and coal consumption was obtained from measurements of the exhaust gas velocity
from a fissure using a video camera sensitive to volatile organic
compounds (VOC) in combination with measurements of gas fluxes
from non-fractured regions over the Crestal Extension Fire using
a flux chamber accumulator. The plume footage documented by
Premier NDT's VOC camera is shown in Fig. 7. Each frame in the figure
represents approximately 0.05 s, and the red tick marks in the center
of the frame are spaced 1 ft (0.305 m) apart. The green arrow in each
frame shows the location of a point on the plume being tracked to give
an indication of velocity. The distance traveled by the plume was
divided by the elapsed time to obtain an approximate velocity of gas
exhaust from this fissure. The measured exhaust gas velocity was
multiplied by the exhaust gas density and the cross-sectional area
of the fissure in order to convert it to a CO2 mass flow. As Fig. 1 shows,
there are many hot fissures over the Crestal Extension area. However,
only two of them are venting exhaust gases at a rate large enough, O(1)
m/s, to be captured by the VOC camera. The fissure where the VOC
video footage was collected was the more active of the two fissures by
an appreciable amount. Because the plume data were obtained at only
one fissure out of the many that exist over the Crestal Extension Fire,
the estimate of the CO2 emissions from the fissures must be smaller
than actual emissions. Other fissures were emitting combustion gases
at flow velocities that were too low to be captured using the VOC video.
Based on the VOC video footage collected by Premier NDT in 2009,
the approximate average velocity of exhaust gases was 1.66 m/s. The
fissure from which the gases were venting measured approximately
11.6 m long, with an average aperture of 0.03 m. The temperature
of gases emitted varied from 315 to 1080 °C (600 °F to 1970 °F),
depending on the location along the fissure. The exhaust gas density
for the temperature range is about 0.281 to 0.610 kg/m3. Taking the
average density of 0.446 kg/m3 over this range, the mass flow rate of
the exhaust was calculated to be 8125 metric tons per year. The CO2
102
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
Fig. 7. Still frames from footage of the exhaust gas plume recorded at a fissure over the Crestal Extension Fire using a Volatile Organic Compound camera (VOC camera). Each frame is
roughly 0.05 s, and the red dashes in the center of each frame are spaced by 1 ft (0.305 m) increments (Premier NDT, personal communication, 2009).
mass flow rate is then approximately 2112 tons per year since CO2
was around 26% of the exhaust gas by mass in the burning zones based
on the mole fractions reported in Table 1. Flow of CO2 from other hot
fissures would add to this total, so here again, this estimate is a lower
bound.
Additional CO2 mass flow rates were measured using a flux accumulation chamber over the non-fractured areas over the Crestal
Extension Fire by LT Environmental in 2008. Measurements were
taken only in areas where the temperature of the gases inside the flux
accumulation chamber remained close to ambient temperatures. The
fluxes of CO2 measured, in mols CO2/(m2-day), are shown in Fig. 8.
Fluxes were measured at the nodes of a 50 ft × 50 ft grid. These values
were then interpolated over the Crestal Extension Area using Kriging
by the surveying company, who reported that the total CO2 flux
from the area was approximately 136 MCFD (1000 ft3 per day) or
3851 m3 per day. The area circumscribed by the red circle in Fig. 8
corresponds approximately to where the subsurface temperatures are
high according to the thermocouple measurements shown in Fig. 2.
While the subsurface temperatures were high, the gases flowing from
the non-fissured overburden were cool enough to be measured by
the flux accumulation chamber. The green diagonal line in Fig. 8
corresponds to the top of the Fruitland Formation outcrop.
Since most of the samples were collected around 300 K and at
around 7000 ft, the volumetric flow rate was converted to a mass
flow rate of 5353 kg per day by using 1.39 kg/m3 as the CO2 density.
This amounts to an additional 1954 tons of CO2 emitted per year from
the less fractured areas over the Crestal Fire Extension. Adding this
CO2 mass flow rate to the mass flow rate of CO2 from an active fissure
results in a total CO2 mass flow rate of 4066 tons per year from the
Crestal Extension Fire. This value suggests that ignoring the CO2 flux
from the less intensely fractured regions could underestimate the CO2
emissions rate by more approximately a factor of two.
Using the stoichiometric ratios obtained from Eq. (1) and the
isotope results which showed that approximately 13.2% of the total
CO2—or 538 tons per year–of the exhaust stream–was produced from
the coal, the rate of coal consumption was calculated. The ultimate
analysis of the coal that is being consumed at the Crestal Extension Fire
shows that it is composed of 72.33% carbon by mass, and the remainder
composed of ash, moisture, nitrogen, hydrogen, sulfur, and oxygen.
The results from the ultimate analysis are shown in Table 2. To get the
mass rate of consumption of coal, the mass of CH0.9 corresponding to
538 tons of CO2 was first calculated from the stoichiometric ratios of
Eq. (1). This value was then divided by 0.7233 to account for the other
species contained in coal. The rate of coal consumption equaled
approximately 218 tons per year based on this approach.
4.3. Estimation of rates using a natural convection chimney model
Next we consider a simple model of the flow of air and exhaust gas
based on a natural convection chimney. Fig. 9 is a simplification of the
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
103
Fig. 9. The Crestal Extension Fire represented as a simplified natural convection
chimney. Gas flow induced by the thermal buoyancy forces occurs mainly through the
zone of ash and overburden rock just above the burned or burning coal seam that has
been fractured by subsidence and collapse.
Fig. 8. CO2 volume flow rate measured from the non-fissured regions over the Crestal
Extension Fire. The samples are approximately spaced 50 ft (15.24 m) apart. The values
are reported in mols CO2/(m2day) (LT Environmental, personal communication, 2008).
conceptual picture shown in Fig. 5. Field observations and a borehole
camera lowered to the coal seam at the North Coal Fire suggest that air
is flowing to the combustion zone through the fractured overburden
that lies immediately above the burned-out coal seam. Multiple
fractures with apertures of O(10− 2 ~ 5 × 10− 2) m that lie above coal
seams likely resulted when collapse occurred in the subsurface.
Strong flow of air and ash was observed through these cracks. Based
on the geometric characteristics of these fractures–some which have
apertures on the order of a few centimeters–the flow in the subsurface
is likely governed by the resistance to flow in these fractures. That
resistance is quite small when the fracture apertures are large O(10− 2
to 10− 1) m, and thus even a small pressure gradient caused by a
thermal buoyancy of the hot combustion product gases is sufficient
to induce flow through the subsurface.
In Fig. 9, the coal fire is modeled using only one inlet and one exhaust
vent. The field situation is more complex, in that there are several
fissures that emit hot gases, but only a few of those vent combustion
gases at a significant flow rate. When air with density, ρair, enters the
chimney in Fig. 9 at temperature, Tambient, and pressure, Patm, the air is
heated and pulled through the fractured zones of the chimney by flow
induced by the density difference between hot combustion gases and
ambient air. The air reacts with the coal to produce exhaust gas at
the combustion zone, which is a located length, L, meters away from
Table 2
Results of the ultimate analysis of a coal sample collected from an unburned region near
the North Coal Fire.
Ultimate analysis
(wt.%)
Moisture
Carbon
Hydrogen
Nitrogen
Sulfur
Ash
Oxygen
3.25
72.33
4.91
1.49
0.57
11.06
6.39
the chimney (fissure). Once the combustion gas reaches the fissure, it
flows through an elbow-joint-like geometry, and travels up the fissure,
which has a height, H, meters. As the combustion gas travels upward,
the gas will likely travel along a tortuous path along the fissure. This
tortuous path can be modeled using more elbow-joints. Heat loss from
the system is neglected in this simple model, so all of the heat produced
in the combustion zone rises through the fissure. The temperature of
the exhaust stream, Tg, in the exhaust fissure is assumed to be constant
throughout the length of the chimney, and hence the gas density of
the exhaust gas can be calculated from Tg. The combustion gas leaves
the system at a mass flow rate, m, through a fissure with an aperture,
d, meters and a length, Lf, meters (not shown in Fig. 9).
Gas rising through the fissure is assumed to be isothermal, and the
gases are assumed to have a consistent density within the combustion
zone. The change in the combustion gas density between the
combustion zone and the surface as it exits an exhaust fissure is so
small such that the constant-density assumption is appropriate
(Rosner, 2000). The friction losses are only considered after the air
reacts with the hot coal. All friction losses on the cooler air side are
ignored based on the assumption that air is reaching the coal seam
through channels that are bigger than the channels through which
exhaust gases are escaping. Apertures of fissures over previously
burned areas (cf. blue regions in Fig. 4) can be orders of magnitude
larger than the apertures of fissures that are currently emitting gases.
The overall flow rate of air and combustion product gases through
the subsurface is determined by the buoyancy of the hot gases, which
is balanced by the resistance to flow in the chimney and in the zones
where air is flowing toward the combustion zone. To assess whether
the frictional resistance to flow upstream of the combustion zone is
important, we compared the pressure difference due to buoyancy to
the pressure drop in a channel with dimensions 12 ~ 18 m in length
at the surface, an average aperture of 0.01 m, and 30 m in distance
that the warm air must travel to reach the combustion zone. If there
were only one such channel through which all of the air was supplied
to the combustion zone, then the upstream pressure drop is quite
significant, about five times the buoyancy pressure difference. Direct
observations at boreholes drilled in the previously burned zone but
relatively near the current combustion zone indicate that the limiting
case of a single fracture path cannot be correct. All of those boreholes
flowed warm air, which indicates that they all intercepted flow paths
conducting air to the combustion zone. If there were only a few flow
paths, then the probability that a borehole would intercept them
104
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
would be small. The fact that all the boreholes did intercept flow
paths suggests that the air flow takes place in some combination of
connected fractures, voids, and ash zones.
To examine the effect of more flow paths we calculated the
upstream pressure drop as a function of the number of fracture paths
like the single fracture described above. If there are 20 such paths, a
number that is likely to be much smaller than the actual number of
paths, then the upstream pressure difference is on the order of 1.25%
of the buoyancy pressure difference. We conclude, therefore, that it is
reasonable to ignore the resistance to air flow through the previously
burned zone and focus instead on the resistance to flow in the fissures
that allow hot gases to escape from the combustion zone.
An expression for the exit gas velocity and the combustion gas
mass flow rate using a macroscopic mechanical energy balance is
available (Bird et al., 2001; Rosner, 2000), and this is presented in
Eq. (3). In Eq. (3), the exit gas velocity, vg, is expressed as a function of
fissure geometry, roughness coefficient of the fissure walls, pressure
gradient induced by the thermal buoyancy force, and the density of
the combustion gas. Temperature appears implicitly in the relationship through both the pressure gradient and the density terms. Eq. (4)
can be used to obtain the mass flow rate of the exhaust gases. The
mass flow rate from the chimney is the product of gas velocity, the
cross section area of the fissure, and the density of the combustion gas.
0
11
2
B2g ρair −ρg H
C
1
B
C
vg = @
d
A
λðL + HÞ
ρg
+ 1:5nelbows
dh
0
ð3Þ
11
2
B2ΔP
C
1
C
=B
@ ρ d λðL + H Þ
A
g
+
1:5n
elbows
dh
ð4Þ
ṁg = vg Lfissure dρg
In Eqs. (3) and (4), g is the acceleration gravity, λ is the D'ArcyWeisbach friction coefficient, dh is the hydraulic diameter of the
fissure that is approximated by a rectangular duct with a cross
sectional area Lfissure meters by d meters, and nelbows is the number of
elbow joints or turns that the gas goes through as it migrates upward
to the surface from the coal seam. All other variables are as previously
defined. Methods to obtain λ from the pipe roughness coefficient
(typically referred to as k, in units of meters) and Reynolds number
can be found in Bird et al. (2001) and Rosner (2000).
The values of the variables in Eqs. (3) and (4) were estimated from
field measurements, but they are obviously not determined precisely.
To assess the sensitivity of the exit gas velocity and the exhaust gas
mass flow rate with respect to variation in other variables, a range
Fig. 10. A distribution of exhaust gas velocity from a fissure based on 10,000 simulations
with various input value combinations. Input values for each simulation run are
chosen at random from the ranges listed in Table 3. Mean = 1.64 m/s, standard
deviation = 0.28 m/s.
of input values for each of the variables was assumed. A distribution
of exit gas velocities and exhaust mass flow rates was obtained by
running multiple simulations with varying combinations of input
values. The ranges all had uniform distributions, chosen based on
observation and measurements made in the field. Table 3 lists all of
the variables, their definitions, and the ranges of values assumed. For
each simulation run, a random number generator was used to pick
a set of input values from the defined ranges. Fig. 10 reports the
resulting distribution of combustion gas velocities calculated from
10,000 solutions of Eqs. (3) and (4) based on the ranges of values in
Table 3. The average of this distribution is 1.64 m/s with a standard
deviation of 0.28 m/s. Fig. 11 is the corresponding CO2 mass flow rate.
The exhaust gas mass flow rate was obtained using Eq. (4), and that
value was multiplied by 26% to account for the fact that the exhaust
gas contained roughly 26% CO2 by mass at the North Coal Fire. The gas
compositions, in mole fractions, of the gases vented from the fissures
over the Crestal Extension Fire are listed in Table 1. The mole fractions
of the exhaust gas can be converted to mass fractions to show that
Table 3
Range of values that each input variables can take for the chimney model. The ranges
are determined based on field observations and measurements.
Variable Units Definition
H
L
(m)
(m)
Tair
Tgas
nelbow
(°C)
(°C)
(–)
k
(m)
Lfissure
(m)
d
(m)
Height of fissure
Horizontal length between combustion zone
and fissure
Temperature of ambient air
Temperature of combustion gas
Number of elbow-joints required to model
tortuosity
Roughness of duct, used to calculate D'Arcy–
Weisbach friction coefficient
Length of the fissure where exhaust gases are
venting
Average aperture of the fissure where exhaust
gases are venting
Range
12.2 ~ 18.3
0 ~ 10
0 ~ 20
315 ~ 1000
1~5
0.005
11.6 ~ 18.3
0.01 ~ 0.03
Fig. 11. CO2 emissions rate from a fissure based on the exhaust gas distribution shown
in Fig. 10. At the Crestal Extension Fire, CO2 makes up about 26% of the exhaust gas
stream by mass. The mean rate of CO2 emissions is equal to 1616 metric tons per year
with a standard deviation of 350 metric tons per year.
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
105
how much lower this modeled rate is in comparison to actual rates
would require additional data concerning fissures and surfaces fluxes
over the full area of the fire.
5. Discussion
Fig. 12. Rate of coal consumption based on the mass flow rate of CO2. Mean coal
consumption rate is 80.4 metric tons per year and a standard deviation of 17.4 metric
tons per year.
the exhaust stream is approximately 26% CO2 and 69% N2, with small
percentages of CH4 and H2 as well as ppm levels of CO and H2S by
mass. The mass rate of CO2 calculated in Fig. 11 includes CO2 from coal,
native CO2, and CO2 resulting from CH4 oxidation. The mean of the
distribution of the total CO2 mass flow rate is 1616 metric tons per
year with a standard deviation of 350 metric tons per year.
The rate of CO2 emissions was converted to the rate of coal
consumption using the stoichiometric relationship in Eq. (1), the
results from the isotope analysis, and the results from the ultimate
analysis, as outlined in Section 4.2. Fig. 12 presents the distribution of
the rate of coal consumption, with a mean of 80.4 metric tons per year
and a standard deviation of 17.4 metric tons per year. The distributions in Figs. 10 and 12 are slightly different, because to calculate a CO2
emission rate or coal consumption rate, each calculated velocity was
multiplied by a fissure cross-sectional area. The cross-sectional area
varied from one simulation run to the next.
4.4. Comparison of results and air requirements
Table 4 lists the annual rate of total CO2 emissions (including
native San Juan Basin gases), rate of CO2 emissions only from coal, coal
consumption rate, and the rate of air required for the stoichiometric
combustion reaction. Values obtained from the VOC method and
the chimney methods are similar. The flux chamber measurements
account for the CO2 that is escaping into the atmosphere from the
non-fissured areas over the active coal fire region.
Here we have chosen to report rates based on some direct observation in a limited area over the North Coal Fire. Therefore, these
amounts are almost certainly lower than actual rates. Determining
Table 4
Estimates of CO2 emission rates, coal consumption rates, and air requirement rates for
each of the three models.
Method/rate
CO2 total
(metric tons/year)
CO2 from coal
(metric tons/year)
Coal consumption
(metric tons/year)
Surface subsidence
VOC
Chimney
Flux Chamber
VOC + Flux
chamber
Chimney + Flux
chamber
–
2112
1616
1954
4066
849
279
213
258
537
249
113
87
105
218
3570
471
191
Given the uncertainties in many of the variables and assumptions
made, it is worth considering the limitations of the rate estimates.
For example, in the surface subsidence estimate of the rates of CO2
emissions and coal consumption, we assumed that the loss of surface
volume was exactly offset by the volume loss in the subsurface. This
assumption is likely to be an underestimate of the amount of coal
consumed in the subsurface when subsidence is not uniform.
Numerous fractures were observed when a down-hole camera was
lowered into one of the boreholes drilled in a previously burned
region. The existence of O(10− 2 ~ 5 × 10− 2) m fractures above the coal
seam suggests that the overburden does not entirely compact the void
spaces left by the burned coal seam, at least not immediately. In fact, if
subsidence were perfect, such that all of the void regions created
above the coal seam due to combustion were compacted by the
overburden, then there would be no channels through which air could
travel to get to the combustion zone under a very small pressure
gradient. It is also possible that subsidence occurs long after
combustion takes place in the subsurface. Comparing the results of
the surface subsidence in Fig. 6 to the fire zones presented in Fig. 4,
large zones of surface occurred where the fire is not currently the
most active. This result suggests that the subsidence observed today
may be over regions that burned previously, and not where it is
burning today. Thus the surface subsidence may not reflect the
present rate of coal consumption. In addition, subsidence does not
take place until a critical radius of coal (for a given overburden
thickness) is consumed by the fire. Until this critical length is attained,
the surface may not show any signs of subsidence. Once the critical
length is exceeded, the surface may subside suddenly including areas
that may have been consumed previously.
In the method of estimating rates using the gas plume footage and
the flux accumulation chamber, the VOC camera was only stationed at
one fissure, so CO2 emissions estimated from the video footage does
not account for CO2 from other fissures that are venting hot gases.
However, flows of exhaust gases from other fissures were smaller, at
least qualitatively, than the flow from the fissure at which the video
footage was recorded. The flux chamber accumulation measurements
made over the non-fissured areas over the Crestal Fire Extension
suggests that flow of CO2 coming from areas other than fissures cannot
be neglected. The flux accumulation chamber measurements do not
suggest that gas is diffusing or flowing at an appreciable rate through
the matrix of the sandstone overburden. The pressure gradient
induced by the thermal buoyancy and gravity is not large enough to
induce appreciable gas flow through low-permeability (O(10)md)
sandstones. Instead, the flux chamber measurements suggest that the
transport through the overburden takes place through the network
of fractures, even in locations where fissures are not readily visible at
the surface. It is also likely that the 30 cm of soil that covers the site
hides many smaller fissures in the area through which CO2 from the
subsurface is escaping.
The final method of calculating rates using the natural convection
chimney analog is based on the assumption that there is only one
active fissure, and the model ignores flow from other fissures. The
model also ignores any flow from non-fractured areas, which leads
again to an underestimate of both the rates of CO2 emissions and coal
consumption calculated. The distributions of exhaust gas velocity, CO2
mass rate, and coal consumption rate obtained from the suite of
chimney simulations suggest that despite the range of possible input
values, the resulting rate estimations are bounded within an order
of magnitude. Finally, the exhaust gas velocity calculated for the
dimensions and conditions measured at the coal fire agrees reasonably
106
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
well with the exhaust gas velocity measured by the VOC camera,
which suggests that modeling fissures as chimneys is a reasonable
approach.
Finally, the isotope material balance results suggest that only
13.2% of the CO2 in the exhaust comes from the coal, while the
remainder of the CO2 in the exhaust is roughly equally split between
the native San Juan Basin CO2 and the oxidized native San Juan Basin
CH4. This result suggests that methane combustion plays a significant
role in this coal fire. This result is roughly consistent with the idea
that native gases are drawn in towards the combustion zone from the
edges of the burned zone. As the native gases migrate toward the
most active fire zones, they mix with air that is drawn in through
fissures. Methane in the native gases oxidizes in the air to produce
CO2 as the gas mixture approaches the combustion zone and heats
up to temperatures approaching 1000 °C. The resulting combustion
gases escape out of the exhaust fissure. Other sources that contribute
to the CO2 in the exhaust stream are the combustion products of
CO and CH4 that devolatilize from the coal and the combustion
gases from char combustion. However, the contribution of CO2 from
burning the native CH4 is significant because the native gases are
likely exposed to the highest concentrations of O2 at the periphery
of the fire.
This picture is also consistent with the observation that a much
larger fraction of the CH4 in the hot exhaust gases comes from the coal
than from the native gas (75.5% vs 24.5%). The CH4 from the coal is
liberated by thermal alteration of the coal in zones closer to the vent
fissures, where little oxygen remains, while the native CH4 oxidizes at
lower temperatures upstream of that zone.
The rate of native gas (CH4–CO2 mixture) required to produce the
gas composition and the isotope signatures at the exhaust fissures
is also consistent with the rate of native gases produced from
production wells near-by and down dip of the North Coal Fire. At
these production wells, gases flow between 30 ~ 300 MCFD. Native
gases that are not intercepted by these production wells migrate
updip towards the Hogback monocline where the North Coal Fire is
currently located. The flow of CO2 and CH4 into the combustion zone
may or may not be representative of other coal fires around the world,
but isotope measurements similar to those described here can be used
to determine whether that is the case.
6. Summary and conclusions
In this paper, the current boundary of the combustion zone at the
North Coal Fire, which is located along the Hogback Monocline in
the San Juan Basin, was established using three methods. Of those,
the cesium vapor magnetometer survey produced the highest
resolution map of the combustion zone boundary in the subsurface.
The magnetometer results were distinguished between previously
burned, currently burning, and unconsumed coal. The results of the
magnetometer surveys agree well with other data obtained at the
North Coal Fire site: subsurface temperatures, snow melt boundary
measurements, fissure mapping, well logs, core samples, drillers' logs,
and down-hole borehole videos.
Three methods were used to quantify CO2 emissions and coal
consumption rates for one active combustion area at the North Coal
Fire, the Crestal Extension Fire. Those methods gave roughly
consistent lower bound estimates of coal consumption and CO2
emissions for this small fire. The estimates indicate that the Crestal
portion of the fire accounts for emissions of at least 4066 metric tons
of CO2 per year, and that a significant fraction of the CO2 emissions
come from combustion of CH4 flowing into the combustion zone from
the unburned coal. Estimates of the rate of consumption of coal range
from about 249 tons per year based on subsidence measurements
to 218 and 191 tons per year from based on surface observations
of flow velocities from a single fissure and on a simple chimney model
of flow of combustion product gases through a fissure, respectively.
In the last two estimation methods, CO2 leakage rates that were
measured using a flux accumulation chamber were included before
converting the CO2 emission rates to coal consumption rates.
These observations and estimates lead to the following conclusions:
1) Cesium vapor magnetometer data provided the most precise
location of the underground coal fire when data was collected at
high resolution and appropriate filters were applied to remove
noise and asymmetry;
2) Other data, such as well-logs, well-bore imaging, subsurface
temperature distributions obtained from thermocouples, surface
fissure mapping, and snow-melt data are completely consistent
with the magnetometer results and are also useful in establishing
subsurface flow geometry and direction;
3) When there are multiple sources of CO2, carbon isotope signatures
can be used to quantify the relative contributions of the various
sources of CO2 to the composition of the gases being emitted over a
coal fire;
4) The chimney model calculations suggest that the North Coal Fire
can be modeled as a natural convection chimney, and that the flow
resistance in the chimney determines the rate of air flow through
the system;
5) Three independent methods of rate estimations provide roughly
consistent rates of both CO2 and coal consumption. While these
estimates are certainly lower bound estimates, they are useful
first-order estimates associated with the North Coal Fire.
Acknowledgments
This project was made possible by the Southern Ute Indian Tribe,
which provided access to the site as well as financial and considerable
technical support for the field portions of this project. The authors
especially thank Bill Flint and Kyle Siesser of the Southern Ute Indian
Tribe Department of Energy for their help and insights and many
helpful discussions. We also thank Sam Krevor for performing isotope
measurements at the field site and helping us interpret them, Erik
Mischker of Premier NDT for measuring the gas plume velocities, LT
Environmental for their CO2 flux accumulation chamber measurements and data, and Jim Flint of Performance Engineering and
Surveying for the surface subsidence measurements and data. The
Global Climate and Energy Project at Stanford provided support for
the authors.
References
Anthony, D.B., Howard, J.B., 1976. Coal devolatilization and hydrogasification. AIChE
Journal 22 (4), 625–656.
Bandelow, F.K., Gielisch, Hartwig, F., 2004. Modern exploration methods as key to
fighting of uncontrolled coal fires in China. 2004 Denver Annual Meeting.
Bird, R.B., Stewart, W.E., Lightfoot, E.N., 2001. Transport Phenomena, second ed. John
Wiley & Sons, New York.
Bredenberg, J.B-son, Huuzka, M., Vuori, A., 1987. Latest advances in thermal and
catalytic reactions of the ether bond in coal and related model compounds. In:
Volborth, A. (Ed.), Coal Science and Chemistry. Elsevier Science Publishing
Company, New York, pp. 1–30.
Brown, K., 2003. Subterranean coal fires spark disaster. Science 299, 1177.
Cao, Daiyong, Xinjie, Fan, Haiyan, Guan, Chacha, Wu., Xiaolei, Shi, Yuerong, Jia, 2007.
Geological models of spontaneous combustion in the Wuda coalfield, Inner
Mongolia, China. In: Stracher, Glenn B. (Ed.), Geology of Coal Fires: Case Studies
from Around the World. The Geological Society of America, Boulder, pp. 23–30.
DeKok, D., 1986. Unseen Danger: a Tragedy of People, Government and the Centralia
Mine fire. University of Pennsylvania Press, Philadelphia.
Fassett, J., 2000. Geology and coal resources of the upper cretaceous Fruitland
formation, San Juan Basin, New Mexico and Colorado, U.S. Geological Survey,
Professional Paper 1625B, Chapter Q.
GAI Consultants, Inc., 1983. Engineering Analysis and Evaluation of the Centralia Mine
Fire, 2. U.S. Department of Commerce, National Technical Information Service,
Springfield.
Gavalas, G.R., 1982. Coal Pyrolysis. Elsevier Scientific Publishing Company, New York.
Gielisch, H., 2007. Detecting concealed coal fires. Reviews in Engineering Geology 18,
199–210.
Hooper, R.L., 1987. Factors affecting the magnetic susceptibility of baked rocks above a
burned coal seam. International Journal of Coal Geology 9 (2), 157–169.
S.T. Ide, F.M. Orr Jr. / International Journal of Coal Geology 86 (2011) 95–107
Huang, J., Bruining, J., Wolf, K., 2001. Modeling of gas flow and temperature fields in
underground coal fires. Fire Safety Journal 36, 477–489.
Ide, T.S., Pollard, D., Orr Jr., F.M., 2010. Fissure formation and subsurface subsidence in a
coal bed fire. International Journal of Rock Mechanics and Mining Sciences 47,
81–93.
Keeling, Charles, 1960. The concentration and isotopic abundances of carbon dioxide in
rural and marine air. Geochimica et Cosmochimica Acta 24, 277–298.
Kim, A., and Chaiken, R., 1993. Fires in abandoned coal mines and waste banks. United
States Department of the Interior, Bureau of Mines, Washington, Information
Circular 9352.
Kuenzer, C., Voigt, S., Morth, D., 2005. Investigating land cover changes in two Chinese
coal mining environment using partial unmixing. Gottinger Geographische
Abhandlungen 113, 31–37.
Kuenzer, C., Zhang, J., Li, J., Voigt, S., Mehl, H., Wagner, W., 2007a. Detecting unknown coal
fires: synergy of automated coal fire risk area delineation and improved thermal
anomaly extraction. International Journal of Remote Sensing 28, 4561–4583.
Kuenzer, C., Zhang, J., Tetzlaff, A., van Dijk, P., Voigt, S., Mehl, H., Wagner, W., 2007b.
Uncontrolled coal fires and their environmental impacts: investigating two arid
mining regions in north-central China. Applied Geography 27, 42–62.
Larendeau, N.M., 1978. Heterogeneous kinetics of coal char gasification and
combustion. Progress in Energy and Combustion Science 4, 221–270.
Prakash, A., Gupta, R.P., 1998. Land-use mapping and change detection in a coal mining
area: a case study in the Jharia coalfield, India. International Journal of Remote
Sensing 19, 391–410.
107
Rosema, A., Genderen, J.L. van, Schalke, H.J.W.G., 1993. Environmental monitoring of
coal fires in north China. Project Identification Mission Report, BCRS 93-29, ISBN
90-5411-1054.
Rosner, D., 2000. Transport Processes in Chemically Reacting Flow Systems. Dover
Publications, New York.
Schaumann, G., Yu Change, Chun, 2005. Airborne and ground geophysical investigations over the coal fire area of Wuda, inner Mongolia. Land and Resources, 100083.
Sternberg, R., 2004. Magnetic surveys over clinkers and coal seam fires in Western
North Dakota. 2004 Denver Annual Meeting.
Sternberg, R., Sparks, A., Knutson, O., 2008. Magnetic surveys over burning and
remediated coal seam fires in Western North Dakota. Proceedings of the Symposium
on the Application of Geophysics to Engineering and Environmental Problems.
SAGEEP, p. 21.
Stracher, G., Taylor, T., 2004. Coal fires burning around the world: thermodynamic
recipe for environmental catastrophe. International Journal of Coal Geology 59,
7–17.
Thorsness, C.B., Grens, E.A., Sherwood, A., 1978. A one-dimensional model for in situ
coal gasification. Lawrence Livermore Laboratory, Livermore. UCRL-52523.
Walden, J., J. P. Smith, and F. Oldfield. 1999. (eds) Environmental Magnetism, A Practical
Guide. Quaternary Research Association. Technical Guide No.6. 243p.
Wessling, S., Kuenzer, C., Kessels, W., Wuttke, M., 2008. Numerical modeling
for analyzing thermal surface anomalies induced by underground coal fires.
International Journal of Coal Geology 74, 175–184.