Underground coal gasification (UCG) is a coal utilization method that permits coal resources to be exploited in situ using high-temperature conversion reactions (Perkins 2018a). The process makes it possible to utilize more coal resources as environmentally sustainable gas and liquid fuels instead of solid fuel (Lavis, Courtney, and Mostade 2013; Mostade 2014). In this technology, an injection well and a production well are drilled into the coal seam, and an airflow connection is then created between the two wells before gasification commences (Figure 1).
A UCG gasification chamber can be separated into three regions: the oxidation zone, the reduction zone, and the dry distillation zone (Figure 1). The oxidation zone is dominated by heat-generating chemical reactions that involve coal and oxygen; the reduction zone is subjected to an environment deficient in oxygen, which leads to the gasification chamber having a zonal coal progression. The zonal coal progression areas are affected differently by the coal conversion process (Figure 1). The separation of the zones depends on key chemical reactions, temperature (related to the coal composition), and the gas composition occurring in each zone (Bhutto, Bazmi, and Zahedi 2013; Lavis, Courtney, and Mostade 2013; Perkins 2018a; Pershad et al. 2018).
The main reactions involved in the coal gasification zones are drying, pyrolysis, gasification, and combustion (Perkins 2018b; Pershad et al. 2018) (Figure 1, Table 1). The UCG process begins with heating of in situ coal, and the moisture is evaporated at around 100°C (R1), with pyrolysis occurring at higher temperatures of between 350 and 600°C (Perkins 2018b). At the pyrolysis stage (R2), volatile gases are driven out of the coal and enter the gasification channel, leaving behind char and ash. The chemical reactions (R3–R7) occur at the surface of the coal, and the secondary reactions (R8–R10) occur in the gaseous phase. The gasification chamber is therefore exposed to different chemical and physical processes, leading to a zonal coal profile due to progressive heating that results in different gaseous products (Figure 1).
In UCG, the reactivity of the coal is a function of the diffusion of the reactants and products, as well as the coal properties. Coal petrography enables the assessment of coal, gasification, combustion, coking, and residue products to enhance the understanding of the gasification processes (Wagner et al. 2008; Malumbazo, Wagner, and Bunt 2012; Wagner, Malumbazo, and Falcon 2018). Vitrinite reflectance (RoV) is one of the primary parameters utilized to assess the degree of the coalification of organic substance in coal and can also be used to estimate the susceptibility of coal to coking process (Róg 2018) and other coal conversion processes (Wagner et al. 2008; Wagner, Malumbazo, and Falcon 2018).
An understanding of the chemical, mineralogical, and petrographic properties of the coal, degasified coal, char, and ash (minerals) from the UCG gasification zone is fundamental to a number of investigations needed to assess the effectiveness of the process, and to quantify future risk, such as groundwater contamination (Mokhahlane, Gomo, and Vermeulen 2018a). For example, post gasification, the ash (possibly containing sulfur and other potentially toxic minerals and elements) and char (possibly containing heavy-molecular-weight products such as uncracked or partially cracked hydrocarbons) in the UCG cavity may be leached by the rebounding groundwater into surrounding aquifers, potentially resulting in groundwater contamination (Mokhahlane et al. 2018). Therefore it is necessary to conduct a comprehensive analysis of coal relics from a spent UCG chamber.
Thus, the current study aims to provide a mineralogical and petrographic characterization of UCG residue obtained from the Eskom Majuba UCG pilot plant site in South Africa. Samples were selected from a verification borehole (VH3) drilled at the site. The series of reactions Table 1 are unlikely to follow a linear path through the coal seam, resulting in a mixture of residue products ranging from ash, char, devolatilized coals, and even unburnt coal. The information gathered from the analysis of the residue products can be used to determine the temperature profile of the gasification process following Róg (2018), and the geochemistry and mineralogy, hence enabling the understanding of the efficiency of the conversion process. The effect of coal rank on UCG success is well documented (Perkins 2018a), but there is limited information pertaining to the behavior of the coal macerals during UCG, specifically inertinite-rich coals typical of South African Permian Karoo Basin coal seams. Conversely, there is a very good understanding of coal particle pyrolysis and gasification reactions regarding South African coals (Bunt, Wagner, and Waanders 2009; Malumbazo, Wagner, and Bunt 2013), and many models exist regarding external surface reactions that govern coal behavior (Perkins 2018b). In this paper, the results pertaining to coal, degasified coal, and ash samples selected from the verification borehole core are discussed, verifying that coal conversion did occur, and providing an estimate of the temperature reached at this location in the gasification chamber. The mineralogy and potential for acid leaching is discussed.
2. Study Area
The UCG pilot plant near Majuba Power Station in Mpumalanga Province, South Africa (Figure 2), completed Phase 1 of the gasification trial in 2010 (Pershad, Pistorius, and van der Riet 2018). This is the first power plant in Africa to utilize UCG-generated syngas. The syngas was successfully co-fired with coal in a pulverized coal boiler (Pershad, Pistorius, and van der Riet 2018). As this was a pilot-scale trial, gasification ceased in September 2011 by assisted quenching, involving flooding of the gasifier (Mokhahlane 2019). Following a time delay to ensure cooling of the rock, the UCG verification borehole drilling program commenced in 2015 around the gasification zone to establish the extent of heating and gasification. The aim of the drilling program was to retrieve core samples with residue products. The verification boreholes were drilled at strategic locations within and outside the gasification zone. G1VH3 was drilled within the gasification zone toward the boundary, and the material retrieved from the core is the subject of this study. A discussion around the syngas composition is beyond the scope of this paper.
The targeted coal-bearing formation for the UCG trial was the Permian-age Gus Seam in the Vryheid Formation of the Ecca Group, located in the Ermelo (Majuba) Coalfield of the Main Karoo Basin of South Africa. The Gus Seam varies from 1.8 to 4.5 m in thickness, and in the Majuba UCG site it is found at a depth of around 280 m (Mokhahlane, Gomo, and Vermeulen 2018b). The coal zone bears several thin (5–20 cm) laterally discontinuous bright coal layers above the Gus Seam, termed the Eland and Fritz Seams, which are used as marker seams (de Oliveira and Cawthorn 1999; Hancox and Götz 2014). The main economic Gus Seam is thicker than the other seams, with alternating sequences of bright and dull coal. As stated by Perkins (2018a), most successful UCG demonstrations globally have involved subbituminous coals with relatively high volatile matter contents and moderate ash. The Majuba Gus Seam is Medium Rank C bituminous and has a volatile matter content of 18.39% (db) and an ash yield of 24.06% (Bell and Spurr 1986), and these values provide historic coal parameters of the coal seam. The Majuba Colliery is highly faulted, and the difficult mining conditions resulted in the closure of the coal mine soon after operations commenced in the 1980s. However, the fault blocks provide ideal opportunities for controlled UCG.
Samples were taken from the G1VH3 (Box 41 and 42) core drill from a depth of 281.8–283.9 m (Figure 3). The degasified coal sampling commenced from an ash layer located immediately below the extensively heat-affected sandstone layers and finished at the carbonaceous shale horizon below the coal seam. Core fragments 1–10 cm were handpicked based on distinct changes in lithologies, such as change in grain size, texture, and lithotype. The samples were identified as S9 (ash) to S18 (Figures 3 and 4).
3.2. Analytical methods
The samples were subjected to proximate and ultimate analyses at the coal laboratory at Eskom Research, Training and Development (RT&D) in Johannesburg using standard procedures (Table 2). S9 was omitted from the analyses discussed in this paper, as it is clearly an ash product with no remnant organic matter.
Two sets of epoxy-bound polished blocks were prepared at the University of Johannesburg for each sample (1) to study the in situ lump fragments (prepared perpendicular to the apparent bedding plane); and (2) to determine statistically the reflectance values of the crushed particles and to quantify the maceral composition. Selected fragments from each sample were hammered (where required) to fit into the 30-mm-diameter molds (Figure 5). The rest of the sample was crushed to pass a −2 mm screen (80% −1 mm) and placed in the second mold. The particles were set with epoxy resin and left to cure overnight under vacuum. All samples were polished using a Struers Tegra grinder and polisher with an automated powerhead to achieve a final 0.05-micron alumina polish, following SANS 7404-2 (2015).
A Zeiss Axio Imager M2M reflected light petrographic microscope, with a total magnification of ×500 under oil immersion (refractive index of 1.518), was used for the petrographic assessment of the fragment and grain mounts. Mosaic images of sections of the fragments were stitched together using the Hilgers Diskus system to obtain a composite image). The relative proportion of vitrinite to inertinite was determined, reported as %vol).
The reflectance analysis was conducted on the grain-mount samples to determine the impact of the heat on the coal and to gain insight into the gasification process. The Hilgers Fossil system was calibrated using the 5.37 ST, 3.24 CZ, and 0.900 (with a zero standard) YAG calibration standards, at a wavelength of 546 nm, based on SABS (2016). The system was stable throughout the analyses, but the standards were checked before commencing each sample. The Hilgers system enables the categorization of reflectance readings based on the maceral type, and hence reflectance values were recorded separately for vitrinite and various inertinite macerals occurring in the samples. The reflectance readings were taken on the particle falling immediately under the cross hair (or the closest clean surface) to prevent bias toward a specific maceral type. A total of 300–400 readings were recorded per sample depending on the available organic matter in the sample. The entire surface of the grain mount was assessed using a consistent step size. Only the vitrinite reflectance results are discussed in this paper; the mean value, standard deviation, and range are provided in Table 3.
The mineralogical analyses of the coal and ash samples was undertaken using a FEG QEMSCAN (quantitative evaluation of minerals by scanning electron microscopy) housed at Eskom Research, Testing and Development (RT&D) Laboratory in Johannesburg. QEMSCAN is a scanning electron microscope configured to determine automatically the mineralogical compositions of particulate samples (Nazari, Ghahreman, and Bell 2017). The technique is typically used to determine the mineral and phase proportions in coal, fly ash, and clinkers. Following the Eskom in-house preparation procedure, fragments were mounted using carnauba wax in 30-mm-diameter polished molds. A full account of the operation of QEMSCAN is given by Goodall, Scales, and Butcher (2005). The polished samples were carbon coated to avoid the sample charging in the chamber when struck by the electron beam.
4. Results and Discussion
4.1. Chemical analysis
The chemical properties of coal samples retrieved from core VH3 are displayed in Table 4, along with data from samples obtained from boreholes drilled away from the gasification zone (VH4 and VH6). The latter samples were included as reference points of unheated coal; these boreholes were drilled prior to the gasification trial. All samples originate from the Gus Seam.
The samples from VH3 are depleted in volatile matter compared to the coal samples from outside the gasification zone (Figure 6); the VH3 samples average 3.24 wt% in terms of the volatile matter content, compared to 21.70 wt% average of samples outside the gasification zone. Therefore, the VH3 samples can be classified as devolatilized or degasified coal. The ash content is variable through the VH3 samples, ranging from 20.5 to 89.6 wt%; S18 and S10 represent the carbonaceous shale horizon, as the ash content is more than 50%. The samples from VH4 also indicate a range of ash values (19.9–56.1 wt%).
The hydrogen fraction in the VH3 samples is significantly depleted compared to the unheated samples from VH4 and VH6 (Figure 7), indicating that hydrogen has been dispelled from the coals during gasification. The ungasified samples from VH4 and VH6 reported an average total sulfur value of 1.81%, whereas the gasified samples show depletion in total sulfur, with an average of 0.6% (Figure 7). High sulfur content in coal can be a contributing factor to acid rock drainage (Mokhahlane, Gomo, and Vermeulen 2018a). The chemical results show that the degasified coal left in the gasification chamber is depleted in sulfur and hence is environmentally more stable.
4.2. Petrographic analysis
The VH3 samples are dominated by inertinite (inertodetrinite and semifusinite) (Table 3; Figure 8), which is typical for coals from South Africa. Snyman (1998) documented a total vitrinite content of 38 %vol for the Majuba coal, which is higher than the average of 25 %vol (mineral matter free) reported for these samples, although two samples do report values of 47 %vol. Three samples (S12–S13) report extremely low vitrinite contents, thus reducing the average vitrinite content in this borehole core. The variable maceral composition may influence the total carbon values, as inertinite-rich coals will have a higher carbon content than vitrinite-rich coals of the same rank. Total carbon values reported in Table 4 range from 42.8 to 76.9% (excluding S18 [carbonaceous shale]).
The composite images (Figure 8), derived from the lump VH3 core samples, highlight the banded nature of these Majuba coals, with a predominance of inertinite. The vitrinite bands in S10 show a degree of swelling in terms of vacuole development, which was not observed in the samples further away from the ash layer (S9). In the other samples, the vitrinite bands tend to exhibit cracking due to heat stress. Pore development occurs during pyrolysis, when vitrinite softens, forming gas bubbles from the release of volatile matter. Cracking and pore development will enhance diffusion of the heat and oxidants into the coal seam, and the products out of the coal. Cracking may occur during drying and pyrolysis, with cross-linking of cracks being beneficial to heat and gas transfer through the seam. All the coal in the samples assessed has been affected; no low-reflecting coal was determined.
Snyman (1998) documented a vitrinite reflectance value of 0.82 %RoVmr for Majuba coals, indicating that the coal falls into the medium rank C bituminous range of rank. Unfortunately, the unaltered samples from VH4 and VH6 were not available for petrographic assessment. The reflectance analyses (Table 3) confirm that all the samples from VH3 have been exposed to high temperatures. An average mean random vitrinite reflectance value of 5.95 %RoVmr was determined, with a mean value range of 5.76–6.73 %RoVmr, and readings ranging from 5.06 to 7.24 %RoVmr (Figures 9 and 10).
The higher the vitrinite reflectance value, the higher the temperature encountered by the sample. Temperatures within the gasification zone affect the quality of the syngas produced, as well as the behavior of the surrounding rocks and minerals. Róg (2018) used vitrinite reflectance analysis to estimate gasification temperatures for coal extracted from a UCG site in Poland. The Polish coal is comparable to the Majuba coal in terms of rank, but with a higher vitrinite content (63%) (Róg 2018). In the study by Róg (2018), coals were prepared to 700, 1000, and 1330°C in a laboratory, and the vitrinite reflectance was determined on the resultant chars. A profile of vitrinite reflectance along the heating profile was constructed, where a mean vitrinite reflectance value of 6.70 was correlated to 1300°C (range 5.10–7.60 %RoVmr); and 1.26 %RoVmr was correlated to 1000°C (range 0.55–2.30 %RoVmr) (Róg 2018). Hence, the coals in this borehole coal drilled at the boundary of the UCG cavity at Majuba are likely to have experienced temperatures up to 1300°C, as the vitrinite reflectance readings fall within, or close to, the range documented by Róg (2018). The three samples with vitrinite contents below 7 vol% may produce statistically unreliable reflectance results because of the limited amount of vitrinite particles available; however, the values reported do support our findings.
Róg (2018) documented a mean vitrinite reflectance value of 5.38 %RoVmr, with a range of 3.4–7.10 %RoV, for the sample at the boundary of the cavity, and value of 7.12 %RoVmr (range 5.2 to >8.35) for the sample taken from the UCG cavity in Poland. The Majuba sample at the boundary with the ash layer has a %RoVmr value of 5.76 and a range of readings from 4.98 to 6.80, which is highly comparable to the data produced by Róg (2018). There is a slight decline of reflectance readings through the borehole core as S17 reports the lowest value, indicating that a slight temperature gradient exists. The lower reflecting samples were exposed to slightly lower temperatures, although still well above 1000°C when considering Róg’s data (2018).
There is limited literature pertaining to the temperature of a UCG site based on organic petrography, so one may consider the more prolific literature pertaining to the exposure temperatures of dike-intruded coals (induced contact metamorphism) and regional metamorphism. Rimmer, Yoksoulian, and Hower (2009) concluded that dike-intruded southern Illinois coals may have been exposed to temperatures as high as 500°C based on the observed fine mosaic coke structure, but decreased this to 70–300°C based on the use of Raman parameters (Li et al. 2020) (this low value may be questionable). Quaderer et al. (2016) also provide a temperature around 370°C for intruded Springfield coals (United States), increasing to >500°C based on the presence of fine-grained mosaic texture. These temperatures are lower than that estimated for coals in UCG cavities, and mosaic texture was not observed in the Majuba UCG samples (most likely because of the low rank). In addition, high volatile-matter values were documented for the samples in the Quaderer et al. (2016) study, indicating far lower temperatures of exposure. At temperatures from 375–400°C, vitrinite goes through a plastic phase, and resolidifies in the mid-400°C temperature or possibly in an hour from heating (Hower et al. 2021). Hence, devolatilization pores in vitrinite, as observed in this study, indicate a minimum temperature over 450°C.
In perhaps a more viable comparison to a UCG site, Malumbazo, Wagner, and Bunt (2012) used total reflectance (vitrinite and semifusinite) to determine the thermal alteration of coals in a pipe reactor fed with N2. The total reflectance values ranged between 5.36 and 5.65%Ro, correlating to pipe-reactor temperatures between 1160 and 1225°C. As the reflectance readings are comparable to and higher in the current study, it could be extrapolated that the UCG temperature reached was above 1225°C.
Of interest, the samples closest to the ash layer did not provide the highest reflectance values; S11 has the highest reflectance value. Matlala, Moreong, and Wagner (2021) and Wagner (pers com) noted similar findings in samples taken adjacent to dolerite dikes in the Witbank and Highveld Coalfields, where samples taken slightly away from the zone of contact report the highest vitrinite reflectance readings. Rimmer, Yoksoulian, and Hower (2009) provided a similar finding. It is possible that the evolving volatile gases have precipitated, resulting in the decreased reflectance from the vitrinite-derived particles in the samples closest to the heat source. Alternatively, as the samples closest to the heat source in all three instances (here, and the other two studies conducted on South African dolerite dike devolatilized samples and studies in the United States) have shown pore development, the heat may be dissipated more rapidly with the evolution of the volatile gases. The samples slightly away from the heat source may sustain the heat longer, thus the higher reflectance readings. The sharpest temperature gradient occurs at the contact, and hence quenching may be more rapid here. The UCG process is generally a slow process (weeks to months), providing time for secondary reactions where the heat is high enough (refer to Perkins 2018a, 2018b). Quaderer et al. (2016) conclude that porosity and permeability of the coal enable convection cells and hence variable heat transfer emanating from dike-induced thermal alteration. Coal has a relatively low thermal conductivity, density, and specific heat affecting heat transfer. Dike-induced thermal alteration is likely to occur at a lower temperature, possibly over a longer time period than UCG processes, and heat transfer may be related to hydrothermal fluid injection. But the idea of convection cells may still apply to both thermal maturation environments.
4.3. Mineralogical analysis
The mineral phases were determined using QEMSCAN analysis for samples S10–S18. The QEMSCAN images show widespread cracks within the degasified coal samples. These cracks were possibly escape channels for the syngas produced by the gasification process. There is also noticeable in-filling of the cracks with glass/molten matter, seen more prominently in S11 and S12, where glass (blue) has partially filled the cracks in the degasified coal (Figure 11). The glass profile within the degasified coal section of the core recorded the highest amount of volume of around 10 vol% in S12 and was lowest in S18 (Figure 11). This can be clearly seen in the glass profile in Figure 12C. This further indicates that the lowermost part of the seam experienced the lowest temperature-related changes. The carbonaceous content proportion is lowest in S10 and highest in S18 (Figure 12A), and the inverse is true for metakaolinite (Figure 12B). Kaolinite transforms to metakaolinite at around 500°C (Grapes 2011). This shows that the highest coal conversion occurred in the upper sections of the coal seam next to the ash layer. The lower parts of the core are poorer in metakaolinite at less than 4 vol% for S17 and S18 compared to sample S10 which recorded 35.9 vol%. This supports the vitrinite reflectance data where the temperature of exposure was lower in the deeper part of the core.
Most of the iron sulfide mineralization in the degasified coal samples is pyrrhotite. The pyrrhotite profile shows that S14 has the highest proportion at 6.3 vol% (Figure 12D). A pyrrhotite melt forms as the temperature reaches 1083°C, which subsequently transforms to iron oxide melt during oxidation (Grapes 2011). The devolatilized gasified samples show lower levels of sulfur as compared to the unheated natural coal from this seam. Hence, the remnant organic material in the georeactor chamber has a lower total sulfur content, which implies a lower risk of acid rock drainage from sulfur oxidation (Mokhahlane, Gomo, and Vermeulen 2018a).
A series of samples extracted from a borehole core drilled within the gasification cavity toward the boundary in the Ermelo Coalfield, South Africa, were examined. It is evident from the data presented that the coal in the VH3 core was exposed to high temperatures, indicative that the coal was gasified. The volatile matter content is very low, and the carbon content is high, clearly different from coals of the same seam from the surrounding area. Most of the iron sulfide mineralization in the degasified coal samples was transformed to pyrrhotite. The samples show lower levels of sulfur as compared to natural coal from this seam.
The vitrinite reflectance data indicate that the coal seam was exposed to temperatures well above 1000°C, most likely around 1300°C. The samples closest to the ash layer show devolatilization pores in the vitrinite bands, and the samples further away contain stress cracks in the vitrinite bands. The vitrinite reflectance values recorded in this study are comparable to data generated at a Polish UCG site (Róg 2018) and by Malumbazo, Wagner, and Bunt (2012) on South African coals from a controlled pipe-reactor experiment. All the degasified coal samples taken from the VH3 core reported high vitrinite reflectance values (above 5.5 %RoVmr), although a slight decrease was observed away from the ash layer, indicative of a very slight heat gradient. But, in general, the heat was fairly evenly distributed along the 2-m section sampled as the reflectance readings are fairly constant, which is an interesting finding. Heat was able to diffuse evenly through the seam, and the volatile gases were removed (as evidenced in the very low volatile matter values). It is stated in Perkins (2018a) that coals with a tendency to swell may affect the diffusion of the volatile gases from the pore structure. As the Majuba coals had moderate to very low vitrinite contents, swelling due to vacuole development is not considered to have been a major issue. Minor vacuole development in some vitrinite bands was noted in the samples closest to the ash layer. Hence, it is feasible to use mean vitrinite reflectance data from heat-altered coals to estimate the probable temperatures achieved in and close to the UCG georeactor.
The reviewers are thanked for their valuable comments toward further reference sources.
LSM acknowledges support from Eskom Research, Training and Development (RT&D). Opinions expressed and conclusions arrived at are those of the author(s) and are not necessarily to be attributed to the Eskom’s RT&D. NJW acknowledges support from the Department of Science and Innovation through its funding agency, the National Research Foundation, and the Centre of Excellence for Integrated Mineral and Energy Resource Analysis (DSI-NRF CIMERA, grant number 91487). Opinions expressed and conclusions arrived at are those of the author(s) and are not necessarily to be attributed to the CoE, DSI, or NRF. The authors declare no competing interests. Author contributions are as follows: LSM: core sampling; analysis of chemical and mineralogical data; and writing of manuscript, including review and editing. NJW: petrographic analysis, compilation of the manuscript section on petrography, and reviewing and editing of whole manuscript. DV: Ph.D. supervisor for Mokhahlane and reviewing and editing of the manuscript.