Ignition Studies on High-Vitrinite and High-Inertinite Coals Using TGA/DSC, DTIF, EFR, and 20 L Dust Explosive Chamber

The aim of this work was to study the ignition behaviour of eight coals of different coal ranks, petrographic compositions, and places of origin. The research allows us to gain deeper insight into the ignition mechanism and the relationship between certain properties of coals and their behaviour during ignition. The methodology utilised standard fuel ASTM data, petrographic analysis, pyrolysis and oxidation reactivity, and ignition characteristics generated through lab-scale tests using various ignition measurement methods. The results show that, in the dust explosion, a homogeneous ignition of coal dust took place. The ignition potential was the highest for coals with a high content of liptinites and a low content of inertinites. The ranking of coals in terms of ignition potential under these conditions can be determined on the basis of the measurements of the devolatilization rate. During the combustion of coal dust in TGA/DSC, a dust cloud, and a pulverised fuel stream, the ignition of particles was performed according to a heterogeneous mechanism. The study showed that the reflectance index may be the most reliable method of predicting and comparing ignition temperatures of both vitrinite-rich and inertinite-rich coals. Due to the lack of regularity in the ignition temperatures of some coals, depending on the proportion of inertinites, the petrographic composition of coal cannot be used to predict ignition temperatures during the combustion of coal dust. The ranking of the coals according to their ignition potential can be determined using TGA/DSC.


Introduction
Solid fuel ignition is defined as a transition from a slow fuel oxidation rate to the rapid oxidation of either the volatiles or the particle surface. The ignition of coal particles is an important preliminary step in determining fire and explosion hazards in the fuel combustion process. Hence, more research, and the development of research procedures, would allow us to predict the behaviour of a given fuel with regard to its ignition efficiency in pulverised fuel combustion (p.f.) and to determine fire and explosion hazards in industrial dust installations. Many fundamental research studies have been conducted regarding solid fuel ignition [1][2][3][4][5]. These processes also require a deep knowledge of the basic relationships between the physicochemical properties of fuel, its chemical composition, and the ignition parameters. Most of the experimental research undertaken in the last 20 years has concerned the determination of the relationship between the ignition characteristics of various coals, their coal rank, the particle size, the oxygen concentration, or the mineral content. The developed relationships seem to be useful to some extent but usually only if a limited range of coals of one geological origin is used. However, these relationships are insufficient when used to characterise a full range of coals from different parts of the world [6,7]. Coals from the Northern and Southern Hemispheres differ in their chemical properties and petrographic composition. Although coals from the Southern Hemisphere are more likely to contain large amounts of inertinite, this inertinite can be different from that of the Northern Hemisphere [8,9]. There is not enough information on how the behaviour of coals from the Northern and Southern Hemispheres differs during ignition, and there is limited literature on the effect of coal petrographic composition on ignition [6,[10][11][12]. Results of research on the influence of petrographic composition on coal ignition behaviour are rarely published and unclear [6,[13][14][15]. The insufficient number of experiments covering the influence of petrographic composition on coal ignition is probably the result of the assumptions that the coals behave in the same way both during ignition and during combustion.
A serious obstacle to the research on fuel ignition is the lack of universal and commonly accepted testing procedures and techniques. Thermogravimetry and differential scanning calorimetry (TGA/DSC), flat flames burners, the standard 20 L sphere test, single particle tests, and tube ignition furnaces, in which a dust cloud or particles dispersed in a stream are tested, are commonly used research methods of ignition phenomena. The closest to real conditions occurring in the pulverised coal flame are those occurring during ignition in the dust cloud or in the dust-air stream. Although TGA techniques have been widely used to determine the behaviour of coal during combustion [16][17][18][19], they operate under different conditions than those found in the p.f. combustion chambers. Ignition in TGA occurs at a very low heating rate (10-30 • C/min), whereas, in the pulverised coal flame, the heating rate of the particles is in the order of magnitude 10 5 -10 6 • C/s, depending on their sizes. Another serious difficulty related to TGA/DSC tests is the lack of a precise definition of ignition. For instance, Cloke et al. [20] defined the ignition temperature as the temperature at which the burning rate rises to 1% weight/min, whereas Chen et al. [21] proposed that the ignition temperature is the point on the TG plot at which the TG combustion line detaches from the TG pyrolysis line. The key feature of the TGA method is its simplicity, determination speed, and accurate control of sample weight loss by controlling temperature, oxidant concentration, heating rate, particle size, and pressure. A more sophisticated ignition testing device is the wire-mesh reactor, which is characterised by relatively high heating rates (of the order of 1000 K/s) and the accurate control of time history and particle temperature [15,22,23]. Other commonly used ignition test devices encompass drop-tube furnaces (DTIF), entrained flow reactors (EFR), fluidized-bed reactors, and pulverised flames, which more accurately simulate the combustion conditions of industrial combustion chambers (i.e., the continuous feeding of coal, the residence time of particles, the heating rate, and particle interactions) [7,20,[24][25][26][27][28][29][30]. It is not clear whether the ignition temperatures obtained from TGA and the above-mentioned tests are connected. In other words, the question is whether the measured ignition temperature of coal particles using thermogravimetry and differential scanning calorimetry (TGA/DSC) can be used to predict the ignition potential of any coal under the conditions of a dust cloud or a p.f. stream. Furthermore, it is also unclear whether the results of these tests can be transferred to the behaviour of a given coal in a pulverised flame.
Research on the ignition process is also restricted by the incompletely known complex ignition mechanism and the fact that it is affected by many factors related to the type of coal and its origin, the chemical and physical properties of the fuel, its petrographic composition, the content of inorganic components, the conditions of the gas environment (composition chemical gas, heating rate), the intensity of turbulence, and the interaction between particles. Despite the progress and implementation of more and more sophisticated measurement techniques in ignition and flame-stability studies, there are still several vital issues concerning the ignition mechanism of pulverised coal particles, which need to be addressed. It is now generally accepted that coal-particle ignition is a multistage process [1,31]. The ignition theory points out that there are three ignition mechanisms, i.e., the homogeneous, heterogeneous, and hetero-homogeneous ignition mechanisms during the ignition process of coal in air. In the case of homogeneous ignition, the primary stage is the ignition of volatile matter, followed by the secondary ignition of char after the end of pyrolysis. Heterogeneous ignition is the result of a direct oxidation reaction of the surface of a coal particle.
The aim of this paper is to study the ignition behaviour of various coals of different coal ranks (represented in this work by vitrinite reflectance R 0 in the range from 0.47% to 0.85%), of different petrographic compositions (inertinite content in the range 28-70%), and of different places of origin (Polish coals, from the Carboniferous age, and South Africa, from the Permian age). The research aims are as follows: • To gain a deeper knowledge of the ignition mechanism; • To establish the relationship between some properties of individual coals (coal rank, petrographic composition, and chemical and physical properties) and their behaviour during ignition; • To explain to what extent the different ignition testing methods allow one to directly assess the ignition potential of different coals and assess whether the ignition parameters determined by them are mutually dependent; • To explain whether there are similarities in the behaviour of coals during ignition when different ignition criteria and ignition measurements are used. The purpose of this study was to use data from four types of lab-scale test reactors, each with different heating characteristics under conditions where the particle interaction takes place.
The methodology utilises standard fuel ASTM data, petrographic analysis, and ignition characteristics generated through lab-scale tests using various ignition measurement methods (non-isothermal TGA/DSC, EFR, DTIF, and a 20 L chamber for testing dust explosibility).

Coals
Four inertinite-rich South African coals, SA_1, SA_2, SA_3, and SA_4, and four vitrinite-rich Polish coals, P_1, PL_2, PL_3, and PL_4, of varying ranks were selected to be representative of extremes in fuel characteristics experienced by South African and Polish utilities. These fuels were selected taking into consideration their availability and suitability for combustion, as well as their potential future use. Following the study requirements to characterise the ignition of dusts, the samples were pulverised in stages using a Retsch Cutting Mill SM100 to <500 µm and a Retsch Ultra Centrifugal Grinding Mill ZM200 using a 200 µm sieve. A size fraction of 63-90 µm, which is the size range usually used in pulverised coal-fired boilers (with 70%w. < 75 µm), was produced and then used in all measurements. All fuels were analysed for their composition through elemental (using a LECO Truspec CHNS analyser; oxygen content was calculated by difference) and proximate analysis. The High Heating Value (HHV) of all samples was determined in an IKA bomb calorimeter to the specifications of PN-ISO 1928. The ultimate and proximate analyses of these coals are given in Table 1.  [8,9], vitrinite was formed under conditions of deep flowing water in which aerobic oxidation was minimised. On the other hand, inertinite in South African coals was generally formed in shallow stagnant water, which allowed aerobic oxidation. Inertinite deposition conditions favour concurrent mineral matter formation, and inertinite-rich coals are found to have a higher ash content than vitrinite-rich ones. Due to the high content of inertinite macerals in South African coals, they are also characterised by a high level of fixed carbon, FC (64.5-73.0% daf). The high volatile yield (27.0-35.5% daf) in all tested coals is associated with high aliphatic-hydrogen content with a high percentage of oxygen in carboxylic and ether groups. The almost twice as high content of sulphur (2% weight, daf) for South African SA_1 coal is noteworthy. Polish coals, due to the lower content of thermal ballast in the form of ash, have the highest HHV (23.4-28.1 MJ/kg). The calorific value of South African coals ranges from 18.0 up to 26.5 MJ/kg.

Petrographic Analysis and Vitrinite Reflectance Measurement
The petrographic composition was determined in line with the standard. The maceral analysis was undertaken following ICCP Accreditation Certificate ICCP/SCAP/-127/AB; ICCP/CBAP-127 using a qualified certified petrographer and the following standards ISO 7404-3:2009 [33]. The methodology of conduct allows one to determine the content of basic macerals, the technical groups of macerals, and the mineral matter in coal in the reflected light. Table 2 shows the results of these determinations for individual technical groups of macerals, vitrinites, liptinites, and inertinites, as well as their total composition. Apart from testing their petrographic composition, the range of vitrinite reflectance (which is said to determine the degree of coal coalification-rank) was also analysed. Reflectance (R 0 , %) is the percentage-weighted ratio of the intensity of unpolarised light (with its specific wavelength) reflected against a given polished surface in relation to the intensity of light falling on it perpendicularly. The determination of vitrinite reflectance was conducted in accordance with the ISO 7404-5:2002 standard [33]. The results are summarised in Table 2. The maceral content was determined on a mineral-free basis. The expanded measurement uncertainty was determined for a confidence level of 0.95.

Ignition Measurements
A wide variety of experimental conditions with different definitions of the ignition are used in different techniques [16][17][18][19][20][24][25][26][27][28]. The results are hence experiment-or devicedependent. The ignition parameters are not an inherent property of the fuels, as they are dependent on the operating conditions (temperature, heating rate, surrounding atmosphere, etc.). In order to understand the mechanisms of inertinite-rich coal ignition and to find optimum solutions for the ignition of these coals, the ignition measurements were carried out using: • A drop-tube ignition furnace (DTIF); • A 20 L spherical explosion chamber; • An entrained flow reactor (EFR).
These techniques were selected because they allowed the ignition behaviour to be studied under conditions similar to those encountered in p.f. firing or in industrial powder installations, with realistic particle/sample sizes, a particle residence time, and heating rates. For the purposes of comparison, the coal ignition parameters were also determined in a thermogravimetric analyser (TGA) and differential scanning calorimetry (DSC).

A Drop-Tube Cloud Ignition Furnace
The ignition tests were carried out at the test stand described in [34]. It is a drop-tube furnace with a working chamber length of 1 m and a diameter of 0.2 m. The working chamber allowed for testing in the temperature range from 400 to 1000 • C with a stable measuring zone, which was ensured by four independently heated and controlled furnace sections. The detection system consisted of two photodiodes. A fuel sample of weight approximately 2 × 10 −4 kg (which corresponds to an average dust concentration of 4.8 × 10 2 kg/m 3 during ignition) was injected through a water-cooled probe. The ignition moment was identified as the time at which the rate of changes in the detector's voltage signal reached its maximum. The ignition tests were conducted in air at atmospheric pressure. At the time of testing, the following parameters (among others) were recorded:

•
The minimum temperature at which ignition of fuel occurs (T cloud ign , • C); • The time at which ignition is recorded (t i , s).
The minimum ignition temperature of a dust cloud, T cloud ign , is defined as the lowest temperature of the hot inner wall of the furnace at which the ignition of the dust cloud in the air contained within the furnace occurs.

20-L Explosion Chamber
The ignition and explosion hazards of organic-dust air-cloud particles are determined by means of maximum explosion pressure P max , the maximum rate of pressure rise (dP/dt) max , and the Minimum Explosible Dust Concentration (MEC) in lab-scale test vessels (20 L sphere). These are functions of various parameters representing properties that are specific to the chemical nature of dust-air mixtures or reflect the sensitivity of ignition and explosion to flow properties and to the heat transfer of dust-air clouds. The current measurements of ignition and explosion hazards use 20 L spherical stainless steel vessels with the central ignition of a turbulent dust-air mixture. The methods for the determination of explosion characteristics are outlined in the PN-EN 14034-1 + A1:2011 and PN-EN 14034-2 + A1:2011 standards [35,36]. A water jacket served to dissipate the heat of explosions and to maintain thermostatically controlled test temperatures. After dust dispersion, the ignition source with a total energy of 10 kJ was activated in the center of the sphere. Two piezoelectric pressure sensors were used to record pressure-time histories. The parameter characterising ignition is the Minimum Explosible Dust Concentration (MEC), which is defined as the specified minimum amount of dust per unit volume of air below which a flame cannot propagate in a cloud of dust [37]. Starting from very small amounts of dust and repeating the test with constantly increasing amounts, a critical value at which the dust cloud finally ignites was determined. The critical dust mass divided by the volume of the explosion chamber was adopted as the Minimum Explosible Dust Concentration (MEC). The ignition criterion serves as the value of the maximum explosion pressure in the chamber, which should be at least 1.5 bar.

An Entrained Flow Reactor
The coal ignitions were performed in the entrained-flow reactor described in [38] according to a procedure, which bears some similarity to that of Faúndez et al. [28]. The ignition temperature of the flame propagation initiation was determined for the studied coals by means of tests done in an electrically heated 3 m-long entrained-flow reactor (EFR). The coal samples with a particle diameter of 63-90 µm were fed with a feeding rate of 5-10 g/min. The gases were preheated to the oven temperature before being introduced into the reactor. During the ignition tests, the reactor was heated at the rate of 15 • C/min from 400 to 800 • C. The experiments were conducted under stoichiometric conditions in air. The criterion for determining the ignition temperature is based on the derivative curves of the profiles of the gases produced (CO, CO 2 , NO, and O 2 ).

Reactivity and Ignition Measurements in the TGA and DSC
Ignition measurements were made using a Setar Setsys Evolution 18 thermo-balance TGA equipped with a differential scanning calorimetry (DSC) sensor cooperating with the Shimadzu IRAffinity spectroscope, respectively. Samples of 20 mg were heated in nitrogen (pyrolysis) or in air (oxidation) at a rate of 10 and 30 • C/min to a final temperature of 900 • C and kept at this temperature until a constant sample weight was obtained. The temperature, sample size, and air flow rate were chosen to provide the kinetic control of pyrolysis and the combustion process. The application of TGA and DSC techniques enabled the determination of combustion and pyrolysis profiles, from which the ignition temperature was obtained. Table 1 shows that the tested coals differ significantly in their coal ranks (represented by the R 0 ,% parameter) and their petrographic compositions. For Gondwana coals, which have a high inertinite content, the volatile matter and carbon content are not accurate measures of their coal rank. Based on the data presented in Table 1, it appears that, according to the coal rank, the coals belong to a similar group of bituminous coals, which occupy a central position in the coal rank (HVB, HVC, and HVD). The reflectance of the vitrinite group (R 0 ) for the tested coals ranges from 0.47% (PL_3) up to 0.85% for SA_4 coal. The SA_1 and PL_1 coals have comparable coal ranks (R 0 = 0.72%) and inertinite content (30-39.4%). Similarly, the PL_2 and PL_3 coals have comparable inertinite content but different coal ranks. On the basis of the data presented in Table 2, the tested South African coals are characterised by a diverse and high content of inertinites. The high content of inertinites indicates a high level of aromaticity. With the increase in the coal rank, the role of petrographic components becomes less significant due to comparable vitrinite and inertinite coal ranks.

Characterisation of Coals
According to Table 2, there are seven basic constituents distinguished in the composition of inertinites. Among them, fusinite, secretinite, funginite, and micrynite are characterised by their relatively low and similar content in individual coals. Other macerals occur at a diverse and high level in the tested South African coals. The content of semifusinite ranges from 15.9% up to 23.1%, macrinite from 1.7% to 12.2%, and inertodetrinite from 4.8% to 33.7%. Commonly, semifusinite accounts for most of the inertinite macerals of Carboniferous coals, whereas a higher proportion of inertodetrinite compared to semifusinite was found in the PL inertinites. In the group of the shown submacerals, some of them can be classified as reactive and some neutral. The term "reactive" in this context refers to those macerals that ignite and combust rapidly in the presence of oxygen. Inert macerals are those macerals that are slow to change. They require more heat and oxygen in order to ignite and take a longer time to burn out. The petrographic composition of Polish coals is characterised with a high content of vitrinites (from 55.4% to 60.5%). The macerals in the tested coals generally possessed a low content of liptinite (<8%) except for the PL_2-PL_4 coals, for which liptinite content was in a range from 9.5% to 11.7%. The high liptinite content in these coals was accompanied by a high content of hydrogen, which occurred in the form of an aliphatic functional group within the organic structure (from 4.7 to 5.5% daf) and volatile matter (from 37.0 to 38.8% daf). The high liptinite content in Polish coals indicates the low level of their aromaticity and fewer cross-linked structures of these samples [9].

TGA/DSC Ignition Measurements
The non-isothermal TGA procedure (with a heating rate of 30 • C/min) was used in the research on the devolatilization and oxidation of the original coals and the char that was separated from them. The devolatilization was performed in an inert atmosphere, whereas the oxidation was performed in air. In order to examine the effect of the heating rate on the ignition temperature values, more tests were performed at a heating rate of 10 • C/min. A single peak corresponding to the maximum mass loss rate due to volatile matter separation and solid phase oxidation could be distinguished on all the DTA profiles. For all tested coals, it was possible to observe, during their oxidation at temperatures below the beginning of oxidation temperatures, an increase in sample mass caused by oxygen chemisorption.
The ignition temperature was determined geometrically based on TGA measurements similarly as in the work of Wang et al. [39], where the temperature was defined by the intersection point of two straight lines: the one tangent to the point of maximum sample mass loss rate and the other horizontal line tangent to the sample mass loss line below the temperature of oxidation onset. Figures 1 and 2 show how the ignition temperatures T TGA ign depend on the coal rank, expressed by vitrinite reflectance R 0 , and on inertinite content for original coals and chars obtained from them.   Figure 1 shows what is known from the literature [17,28], namely that the ignition temperature of coals as well as chars obtained from them rises along with the increase in their coal ranks. More aliphatic low-rank coals have lower ignition temperatures than highrank coals, for which the aromatization level of the organic structure is higher. A similar effect can be observed, regardless of the geological origin of the coal, for inertinite content, whose rise also leads to the growth ignition temperature (Figure 2). The only exception here is the SA_2 coal, with the highest inertinite content (70.05%) among the tested coals. Its ignition temperature is 60 • C lower than predicted by the trend line, and the difference between the expected and measured values of ignition temperature of the char obtained from it is about 30 • C.
Another subject of the investigation was to study the influence of the heating rate on the ignition temperature. The results are presented in Table 3 and Figures 1 and 2. According to them, during slow heating in 10 • C/min, the determined ignition temperatures were lower on average by about 20 • C than the ignition temperature determined at the heating rate of 30 • C/min. Assuming that the oxidation of fuel samples is governed by the first order reaction, the oxidation rate is given by [40].
with an initial condition expressed by where fractional weight conversion α is where the reaction rate coefficient is given by the Arrhenius equation k = k 0 exp(−E/RT), m is the current sample mass, and m 0 is the initial sample mass. If furnace temperature rises linearly with constant heating rate T = T 0 + βt, then the above equations have an approximate solution [40]: where β is a constant heating rate ( • C/min). According to the above equation for a constant mass loss value (1−α), increasing the heating rate β will result in a shift of the mass loss profiles, and simultaneously of all associated characteristic temperatures, to the right, towards higher temperatures. For ignition conditions, one may assume that α → 0, and the ignition temperature will then result from Equation (4): The above equation can be rewritten in a simpler form applying a simplification proposed by Cassel and Liebman [41]: where γ = E/RT is a parameter resulting from the expansion of the term e −E/RT ign into a power series. According to the above formula for a given conversion degree α, the ignition temperature should rise together with the rise in the sample heating rate. A comparison of the devolatilization onset temperatures T in.dev obtained in TGA measurements in an inert atmosphere with obtained ignition temperatures T TGA ign (Figure 1), where the former temperature is higher, would suggest a homogeneous ignition mechanism of the tested coals under the TGA conditions. As in [42], the temperature at which the weight loss reaches one-fifth of the maximum devolatilization rate was assumed as the temperature of the beginning of devolatilization T in.dev . Considering that, under slow heating conditions in the presence of an oxidiser, the devolatilization process slows down due to the deterioration of the thermoplastic properties of the tested coals [12,23,42,43]; the adoption of the above criterion as decisive for the ignition mechanism seems unreliable.
The DSC analysis may be a useful tool to study the mechanism of the ignition process. The results of the research on the example of SA_4 coal are shown in Figure 3. The analysis of the other tested coals showed that their profiles have similar trends to SA_4 (not shown here). In this technique, the ordinate value of an output curve at any given temperature is directly proportional to the differential heat flow between a sample and a reference material, in which the area under the measured curve is directly proportional to the total differential calorific input.
The presence of O 2 results in two exothermic reaction stages, which are demonstrated by the DSC curves. The first stage of reaction, observed at temperatures ca. 200-380 • C, represented by the shoulder left to the big peak, is due to low heat evolution caused by oxygen chemisorption [18,44]. The sample mass rises slightly during oxygen chemisorption. Figure 3 contains also a dDSC plot on which characteristic peaks indicate the fastest changes (increase or decrease) in the rate of heat release. The temperature at which the onset temperature of the second big exothermic peak in oxygen atmosphere T DSC ign1 (equivalent to the first inflection point between two exothermic peaks in the DSC profile) is possibly a measure of the susceptibility of coals to ignition. According to Figure 1 and Table 3, temperatures T DSC ign1 , depending on the type of coal, are lower by the value 8-60 • C from the ignition temperature of oxidation in the TGA experiment T TGA ign . According to the DSC measurements, the order of ignitability increased regularly with the increasing rank of the original coal. Similarly, ignition temperature T DSC ign1 also increases with increasing inertinite content. The exception here is the SA_2 coal for which the temperature was about 50 • C lower than the expected value. Figure 3 and Table 3 also show temperatures T DSC ign2 at which the rise in heat release during oxidation is the highest, which corresponds to the second inflection point. Figures 1 and 2 show that the nature of temperature changes depending on the coal rank and inertinite content and is similar to the one presented earlier. These temperatures are the closest to the ignition temperatures obtained from TGA measurements T TGA ign and on average higher by about 20-30 • C. This would suggest that the best temperatures for assessing the potential for the ignition of a given fuel by DSC measurements are the T TGA ign or T DSC ign2 temperatures, which correspond to the maximum rate of heat release due to the oxidation reaction. For assessing the potential for dust explosibility, the lowest oxidation onset temperature T DSC ign1 should be used. An ignition temperature corresponding to a second inflection point results from the heat balance equation for two crucibles, one containing the studied sample and the other being a reference crucible, which is empty. The presented model is similar to the models known from literature [45][46][47]. The energy balance for the crucible with the tested sample is as follows: The energy balance for the reference crucible is as follows: The initial condition is The furnace temperature rises linearly with time T w = T 0 + βt. The assumption that ignition temperature corresponds to the inflection point leads to the following equation: The sample temperature can be determined from the following equation: When mass loss is negligible during oxidation, it can be assumed that α → 0. Applying the simplification proposed by Cassel and Liebman [41], the ignition temperature can be determined from Equation (11): According to Equation (11), the ignition temperature will increase when the heat flux is released (expressed by the heat transfer coefficient on the sample surface area) from the fuel sample into the environment increases.

Dust-Cloud Ignition
The dust ignition temperature, T cloud ign , and the time of ignition, t i , were determined by means of the ignition of dispersed coal-particle cloud tests performed in the drop-tube furnace, which allows for interaction between particles. For furnace temperatures below a certain critical value (defined as the minimum ignition temperature, T cloud ign ), ignition delay times t i increase to infinitely large values. Table 3 presents the values of the determined minimum ignition temperatures.
The determined values of the ignition temperatures, T cloud ign , and the ignition time depend on the type of coal being tested. The lower the temperatures and ignition times, the more sensitive the coal is to ignition. The results summarised in Figure 4 verify the relationship between the critical ignition temperature and rank parameter R 0 , and those in Figure 5 verify the relationship between critical ignition temperature and inertinite content.  The ignition temperature of the studied coals increased with the increasing rank of original coal. This trend can be attributed to the lower observed oxidation rate with the increasing coal rank. This decrease in ignitibility with the decrease observed in reactivity R obs is consistent with the heterogeneous model of cloud-particle ignition [34]. For 60 types of coal with various ranks, from brown coal, to bituminous coal, and to anthracite under conditions of particle interaction (cloud ignition), it was concluded that, under experimental conditions, the ignition of the dust clouds of fuels with a volatile matter content below 65-70% daf (anthracites, bituminous coals, and some brown coal) occurred according to the heterogeneous mechanism. Based on the analysis of the heterogeneous ignition of the coal-dust cloud, the ignition temperature and ignition delay time is proportional to 1/R obs , so high-rank coal will tend to ignite at higher temperatures and for longer ignition times because of its lower reactivity. A similar character of changes of the critical ignition temperature indicates a similar ignition mechanism of particles in the dust cloud under critical temperature conditions and in TGA.
A similar relationship between ignition temperature and inertinite content is illustrated in Figure 5. The only exception here is the SA_2 coal, which has the highest inertinite content. The SA_2 coal, as was observed during the TGA/DSC measurement, behaves differently and has an ignition temperature T cloud ign of about 80 • C lower than it would appear to have from the depicted relationship. It follows from this that a comparison of vitrinite-rich Polish coal with other inertinite-rich South African coals of various ranks and inertinite concentrations will indicate that the coal rank rather than reactive maceral (or inertinites) is a dominant factor in the ignitability of coal in TGA and in a drop tube, which will be shown in further sections describing dispersed coal-particle stream measurements. Under these conditions, vitrinite-rich coals are easier to ignite than inertinite-rich coals. Although rank can indicate the general ignition behaviour, some of the scattered results observed in the drawings indicate that the assessment or prediction of ignition behaviour based on rank alone may not be sufficient.

Ignition of Dispersed Coal-Particle Streams
There are a few works [28,30,38,48] related to the ignition of dispersed coal particles in a stream, which is more comparable to the actual process in a coal boiler, regarding flame stability, flame shape, flame length, and pollutant formation. In the coal-particle stream, coal particles interact with each other and the surrounding gas flow. These factors affect the particle motion, heat and mass transfer, and chemical reaction.
The research shows that, during ignition, a rapid decrease in CO production, a significant O 2 consumption, and an increase in the production of CO 2 are observed. Similarly to [28,38], here, the criterion for determining the ignition temperature, T stream ign , was based on the first derivative curves of the gas composition. Table 3 and Figures 6 and 7 show the values of the ignition temperatures for the tested coals obtained from their corresponding CO 2 , O 2 , and CO curves. As can be observed, the ignition of the lower-rank coals took place at lower temperatures than the ignition of the higher-rank coal. A positive trend of decreasing ignition temperature T stream ign with a decreasing coal rank was found. Similarly, the ignition temperature matched well with the values of the inertinite content of the tested coals ( Figure 7). Likewise, it was observed that during the ignition carried out in TGA/DSC and DTIF (cloud) tests, in the case of low-rank coal SA_2 with high inertinite content, temperature T cloud ign was substantially lower (175 • C) than what would result from its inertinite content. It can be inferred from measurements of the gas composition in the stream test that the mechanism of ignition for the tested coals is heterogeneous, as reflected by the single peak in the CO profiles, by the constant increase in CO 2 , and by the constant decrease in O 2 , produced after the ignition. As shown earlier, on the basis of dust-cloud ignition research, the results obtained for the ignition mechanism and for the variation of the ignition temperature with coal rank were similar to those obtained in the coal-particle streams.

Dust Ignition and Explosibility
The ignition and explosibility parameters for the studied coal are given in Table 3. The dusts have been ranked in terms of their ignition, represented by MEC and explosibility dust indices P max and (dP/dt) max . According to the data presented in Table 3, in terms of these ignition and explosion parameters, the Polish PL_3 and PL_4 coal dusts are the most hazardous of the dusts listed. The theoretical maximum explosion pressure is a thermodynamic parameter related to the heat of the mixture combustion. Higher peak pressures P max indicate higher flame temperatures. The flame temperature in sphericalvessel explosions with a central ignition is close to the adiabatic temperature, as the only heat losses from the flame are radiation to the wall or the dust ahead of the flame. Hence, the highest values of pressure are observed for PL coals with a calorific value of 24-28 MJ/kg, and the lowest values of P max for SA_2 coal are observed with the lowest caloric value, equal to 18.0 MJ/kg.
The variability of (dP/dt) max in Table 3 is partially due to size variability resulting from composition differences and partially due to turbulence in the dust cloud. The turbulence increases the concentration of volatiles in the preheat zone and consequently the amount of fuel that is actually being combusted by the flame. The fraction of the total volatiles evolving from the dust particles during the pre-explosion time determines the effective yield of fuel that participates in the flame-propagation process. The (dP/dt) max results showed that Polish coals were more reactive than South African coals. This was attributed to the higher volatile content, the higher rate of the evolution of the volatiles, and the fact that the tested PL coals released volatiles at lower temperatures than other coals. This was supported by TGA analyses.
The data in Table 3 show that, below a certain dust concentration level (the Minimum Explosible Dust Concentration (MEC)), ignition is not observed. This means that, at concentrations below the minimum limit of explosibility, the heat released from the combustion of the particles near the ignition source is not sufficient to ignite neighboring particles, so flame propagation does not occur. In addition to the previously determined values of minimum ignition temperature, the minimum dust-explosion concentration is assumed to be another parameter that characterises the ignition of dust. The minimum explosibility limit (ignition) concentration in air varied strongly for each of the tested coals. According to Table 3, for the tested coals, the measured MECs in the 20 L chamber are in the range of 150-450 g/m 3 . The lower the concentrations needed to ignite the coal dust, the more sensitive the fuel is to ignition. The MEC is lower for Polish coal compared with South African coals, and this indicates higher reactivity and ignitability. On the basis of the MEC values for the tested coals presented in Table 3, it appears that no exact relationship can be established between these values and the coal rank. This means that, during the ignition of the coal particles, a different mechanism applies to that in the previously described TGA/DSC, DTIR, and EFR ignition conditions. Table 3 shows that the lower the explosive limits of MEC are for coal dust with a higher volatile content, the higher the rate of the evolution of the volatiles, the lower the volatiles' evolution temperatures, and the higher the content of the reactive macerals (a lower content of inertinite) in the parent coals. Figure 8 shows the relationship between the parameter characterising the volatile matter release rate, R max.vol , determined by TGA measurements, and Figure 9 shows the inertinite content and MEC data.  The figure shows that, despite the large scattering of the results data, an increase in devolatilization results in a decrease in the value of the Minimum Explosible Dust Concentration (MEC). This devolatilization in ignitability with the rising volatile-matter content is consistent with the findings of Hertzberg et al. [49]. According to them, it is presumed that only the volatile matter is consumed in an explosion (homogeneous ignition) and that the role of the less reactive material, the char, is that of a heat sink. During homogeneous ignition, depending on the particle size, dust concentration, volatile matter content, devolatilization rate, fuel type, relative velocities of the particles with respect to a gaseous environment, and other factors, the released volatile matter burns according to one of two scenarios: (1) in the vicinity of each individual particle and (2) in the gas cloud away from the particles. According to the first scenario, the ignition of a dust cloud with a high content of volatile matter, high volatilisation rates, and the volatile matter Stefan flow out of the particles can cause a dilution of the dust cloud and a lower interaction between particles in the cloud. Thus, according to this scenario [34], the minimum limit of the coal-dust-cloud homogeneous ignition increases with the devolatilization rate R max.dev . In the second case, according to the model presented by Zhang and Wall [50], volatile matter is released from dust particles into the gas environment and does not ignite in the surroundings of the particle but in the gas cloud (homogeneously), far away from the particles. Under these conditions, the minimum limit of the coal-dust ignition is proportional to 1/R max.dev . The dependencies of MEC on the devolatilization rate R max.dev presented in Figure 8 seem to indicate such an ignition mechanism. This means that the relationship is the basis for a theoretical interpretation of the experimentally obtained influence of the rate of devolatilization on the minimum ignition concentration, as shown in Figure 8.

Discussion
The work aimed to clarify whether there are relationships between the coal rank, the petrographic composition and place of origin of the coals, and their ignition parameters determined by different experimental methods. Figure 10 shows a comparison of ignition temperatures depending on the coal rank determined with TGA/DSC and with cloud and stream measurements. Although the operating conditions in TGA, DTIF, and EFR are very different, a similar course of changes in ignition temperature depending on the coal rank was obtained. The regular relationship between ignition temperatures and the coal rank suggests that this parameter turned out to be the main factor enabling the prediction of ignition of both vitrinite coals from the Northern Hemisphere and geologically atypical inertinite coals from the Southern Hemisphere. Lowering the reactivity and increasing the ignition temperature of coals along with their rank results from the ordering of their molecular structure and the reduction in the concentration of active centers due to the coalescence of polyaromatic rings. The increase in the coal rank results in the removal of aliphatic and alicyclic groups, which results in the reduction of the volatile content and progressive aromatization. As the presented results have shown, there is a slight deviation from this rule. The ignition temperatures of coals with the lowest rank (PL_3 and PL_4) and their chars were on average 20-30 • C higher than those resulting from the trend line. This suggests that, regardless of the coal rank, the ignition temperatures of these two tested coals will also be affected by other factors, including the properties associated with the petrographic composition. Vitrinite coals contain more alkyl side chains and cationic groups, while inertinite coals are more aromatised. However, as has been shown in the example of coal SA_2, there are also deviations from this rule.
An interesting fact is that the ignition temperatures determined for the cloud and the pulverised coal stream had, regardless of different conditions and methods of temperature determination, very similar values and a similar nature of change in relation to the coal rank. The determined ignition temperatures of the original coals were on average 200-280 • C higher than the temperatures determined under TGA/DSC conditions and were similar to those determined by [28]. In this paper, the differences, for similar coal ranks, were reported to be in the range 300-400 • C in the case of TGA measurement and 600-700 • C in the case of EFR measurements, and the average temperature difference was around 300 • C. Figure 10 shows that the ignition temperatures determined in the cloud and stream were higher by about 70 • C for the low rank coals, up to 150 • C for the high rank coal SA_4, than the ignition temperatures determined in TGA/DSC measurements for the char obtained from these coals. These values were similar to those determined by Zhang et al. [47], where the difference between the ignition temperature of char determined in the DSC (585 • C) and the ignition temperature of the original coal determined in the EFR (662 • C) was 77 • C. The main reason for the temperature differences seems to be due to different heating rates under TGA/DSC conditions (10-30 • C/min) and stream conditions (10 5 • C/s).
The knowledge of the ignition mechanism of dust in a flame or during an explosion is a basic prerequisite for developing a mathematical model that allows one to link ignition or explosion parameters with any process conditions. The similar nature of the relationship between ignition temperatures and the coal rank of original coals determined in different conditions (TGA/DSC, DTIF, and EFR) suggests that there is a heterogeneous ignition mechanism common for all the tested coals. This is also indicated by the similar nature of the same relationship observed for the original coals and the chars obtained from them, which are practically devoid of volatile matter. It seems that, during the heating of the studied coals in the rank range R 0 = 0.47-0.85%, the formation of a thermoplastic phase [51] effectively worsens the conditions of devolatilization; as a result, ignition takes place on the surface of the solid phase. Under these conditions, the volatile matter does not significantly affect the ignition process. The ignition temperature does not depend on the amount or the rate of volatile matter release but on the rate of oxidation (reactivity) of the fuel, i.e., kinetic parameters describing the intrinsic oxidation reactivity of the solid phase and the rate of oxidant diffusion. No effect of diffusion processes on reactivity was confirmed by carrying out additional measurements under the non-isothermal conditions with the application of TGA/DSC for the most reactive coals with a sample mass of 10 mg and particle fractions < 63 µm. The slightly higher ignition temperatures of the two lowest-rank coals, PL_3 and PL_4, and the char separated from them, were observed as a result of the lower reactivity of these fuels. The reason for the lower intrinsic reactivity than would result from their coal rank could be attributed to the content of macerals (a high content of liptinites). According to [12], in the presence of large amounts of liptinites, interactions between macerals resulting from the increased release of volatile matter from liptinites are possible. They affect the reactivity of the remaining macerals through adsorption on their surface, reactions, and their agglomeration. Another reason for the lower intrinsic reactivity of PL_3 and PL_4 coals could have been the weathering process caused by a storage period of several years [52].
A strong dependence of the ignition temperature on the coal rank is observed (Figure 10) in the case of tests carried out during measurements with high heating rates that take place in the stream (EFR) and in the cloud (DTIF) and is much stronger than that observed in the case of TGA/DSC measurements. This means that, under these conditions, the ignition temperature, assuming that the intrinsic chemical reactivity under TGA and the EFR and DTIF tests remains the same, will also depend on other factors, especially oxidant diffusion. As shown in [53], ignition in a dust cloud and in a stream takes place in the II reaction zone, i.e., when the reaction rate is controlled by chemical reactions and diffusion in the pores, which is different from what occurs during ignition under TGA conditions. Therefore, according to [53], the ignition temperature of a particle in a cloud or stream will be inversely proportional to the intrinsic reactivity (expressed by pre-exponential coefficient k 0 ) and the oxidant diffusion process in the pores (expressed by fuel porosity ): According to Equations (6) and (11) under TGA/DSC ignition conditions, when the process depends only on the intrinsic chemical reactivity, the increase in reactivity (expressed by the pre-exponential coefficient k 0 ) will result in a decrease in the TGA/DSC ignition temperature, T TGA ign /T DSC ign . Another ignition mechanism can be observed for the dust explosion of the tested coals in a 20 L sphere. Here, two alternative ignition mechanisms are possible, homogeneous or heterogeneous. So far, it is not fully settled under what conditions a given mechanism occurs and what it depends on [37,49,54]. In this paper, the results from the TGA analysis were used to explain the ignition mechanism of coal dust during the explosion in the 20 L chamber. During the explosion, very high heating rates take place. As the particles are very fine, the volatile matter can be released even before the ignition of the char. In these circumstances, ignition should strongly depend on volatile matter content and the rate at which it is released. Based on the measurements taken (Table 3), it was found that, unlike the content of volatile matter, the devolatilization rate does not depend on the coal rank. On the other hand, it was observed from the performed TGA measurements that the devolatilization rate regularly decreased with increasing inertinite content in the tested coals. In other words, the amount of volatile substances released during the pyrolysis from coal of a similar rank decreases as the content of components changes from reactive macerals (liptinites and vitrinites) to inert inertinites. The decisive factor in the high ignition capability is the high amount of liptinites due to their high content of hydrogen and volatile matter. The conducted research [13] shows that liptinites contain twice as many volatile parts as vitrinites. Taking into account the coal rank, the composition of macerals should therefore be the key property determining the rate, the amount of released volatiles, and, consequently, the ignition. As shown in the results presented in Figures 8 and 9, the ignitability (defined by the lower explosive limit) increases as the devolatilization rate increases and decreases as the content of inertinites increases. The lowest values of the lower explosion limit were achieved by the PL_3 and PL_4 coals with the highest content of liptinite.
Due to the fact that some coals exhibit a lack of regularity between the ignition temperature and the content of inertinites, the latter cannot be used to predict the ignition temperature of coal. It seems that the main reason for the deviation of some coals from the trend line (observed in Figures 2, 5, and 7) is the incorrect determination of the share of the reactive fraction in the total inertinite content. The TGA/DSC, DTIF, and EFR analyses show that low-rank SA_2 coal (R 0 = 0.56%), which possess the highest inertinite content (70.5%), has higher oxidation reactivity, based on the TGA studies and the lower ignition temperatures of both the original coal and the char obtained from it, than would result from its content of inertinites. Figures 2, 5, and 7 show that, for SA_2 coal, in order for the ignition temperatures to be on the trend line, the actual content of inertinites should be within the range of approximately 30-40%. On the other hand, the remaining share of inertinites, i.e., approximately 30-40%, should be classified as the reactive part of inertinites. It seems that this is due to the fact that a certain part of inertinites (reactive inertinites), especially those found in low rank coals, may be, apart from liptinites and vitrinites, classified as reactive macerals characterised by good reactivity towards combustion. Such an assumption seems to justify some of the results presented in the literature. It describes attempts to quantify the actual content of inertinites and the portion of them that can be considered reactive. According to some authors [13,[55][56][57][58], these inertinites, which has a vitrinite reflectance below 1.3-1.8%, can be classified as reactive. According to Shibaoka et al. [59], Gondwana's coals contained more low-reflection inertinites than European ones, which, on the other hand, were rich in high-reflection inertinite. Furimsky et al. [60] define reactive macerals as the sum of liptinite and vitrinite as well as one-third of the semifusinite content. According to Thomas et al. [61], inertinites occurring in coals from the Southern Hemisphere, which contains more than 40-50% of the inertinites, may be 90% more reactive, so the assumption made by Furminski et al. [60] concerning the 30% reactivity of semifusinite may lead to erroneous results. Moreover, according to Falcon [8], reactive semifusinite occurs mainly in Gondwana coals and less frequently in the Carboniferous coals. Some authors have defined the threshold for the presence of reactive inertinites by analysing the image at any reflection boundary at the grey level to be 190 [20]. The grey level of 190 corresponds to the vitrinite reflectance of about 1.20% [57]. The share of the reactive part of inertinites determined in this way was significant. In the three low-rank coals tested by them, it was 83%, 47%, and 61%, respectively. The authors [61] investigated the proportion of (fusible) reactive inertinites using the laser microreactor technique. Using this criterion, inertinites were divided into thermoplastic fusible and infusible macerals, with a specific reflectance value for each coal. The calculations with the use of a full coal reflectogram allowed them to determine the shares of fusible inertinites. The average share was 75% for the four Gondwana coals and 51% for the two Carboniferous coals.

Conclusions
This work is concerned with determining the parameters and mechanism of the ignition of coal particles with different ranks and different petrographic compositions under conditions where particle interactions take place. To this end, temperature-programmed pyrolysis/combustion TGA/DSC, a drop-tube furnace, a laminar entrained-flow reactor, and 20 L explosion chamber operating at high coal-feed rates were used in order to promote interactive effects between the particles. The subject of the research was to correlate petrographic data and rank parameters with ignition parameters of four South African coals (Gondwana) and four Polish coal (Carboniferous). Based on the ignition studies conducted under slow and fast heating conditions of coal particles, some general trends related to coal rank and the content of macerals were observed. They can be summarised as follows:

1.
Coal rank (determined by the R 0 ,% parameter) and the content of macerals, including the share of inertinites or reactive macerals, are important parameters describing the ignition conditions and allow for the ignition potential of various coals to be directly assessed.

2.
Under the conditions of the dust explosion, the homogeneous ignition of coal dust took place. The ignition potential was the highest for coals with a high content of liptinites. The lowest potential was characterised by coals with a high content of inertinites. The share of inertinites can be used to predict ignition. On the other hand, the ranking of coals in terms of ignition potential under these conditions can be determined on the basis of the measurements of the devolatilization rate using the TGA technique.

3.
During the combustion of coal dust under the conditions of measurements in TGA/DSC, a dust cloud, and a pulverised fuel stream, the ignition of particles was performed according to a heterogeneous mechanism. Under these conditions, the ignition potential of vitrinite and inertinite coals depended on the degree of coal rank and the content of inertinites. The study showed that the vitrinite reflectance may be the most reliable method of predicting and comparing ignition temperatures of both vitrinite-rich and inertinite-rich coals. The ranking of the coals according to their ignition potential can be determined by simply measuring the ignition temperature using the TGA/DSC technique. In other words, the measured ignition temperature of coal particles using thermogravimetry and differential scanning calorimetry can be used to predict the ignition potential of any coal under dust cloud or p.f. stream conditions in a flow reactor, i.e., under conditions closest to those prevailing in p.f. flames.

4.
TGA analyses together with DSC, despite their limitations, can provide a useful initial insight into how a given coal will ignite in a dust explosion and in a p.f. flame.

5.
Increasing the heating rate will increase the ignition temperature. Inertinites worsened the ignition potential of the coals. However, due to the lack of regularity of the ignition temperatures of some coals, depending on the proportion of inertinites, the petrographic composition of coal cannot be used to predict the ignition temperatures during the combustion of coal dust. It seems that the main cause of deviations of some coals lies in the incorrect determination of the reactive fraction share of inertinites in the total content of inertinites. The inertinite-rich low-rank Gondwana coal has good ignition properties due to the presence of low-rank reactive inertinite. This calls for additional tests aimed to determine which inertinite sub-components, and in what parts, can be classified as intrinsic reactive macerals. Informed Consent Statement: Not applicable.

Data Availability Statement:
The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Acknowledgments:
The authors would like to thank ICHPW, Poland, for petrographic analysis and vitrinite reflectance measurements.

Conflicts of Interest:
The authors declare no conflict of interest.

Abbreviations
The following abbreviations are used in this manuscript: