Correlation between the critical viscosity and ash fusion temperatures of coal gasifier ashes

Fuel Processing Technology 142 (2016) 13–26

Contents lists available at ScienceDirect

Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc

Correlation between the critical viscosity and ash fusion temperatures of coal gasifier ashes☆ Peter Y. Hsieh ⁎, Kyei-Sing Kwong, James Bennett National Energy Technology Laboratory, Structural Materials Development Division, 1450 Queen Ave SW, Albany, OR 97321-2198, USA

a r t i c l e

i n f o

Article history: Received 24 July 2015 Received in revised form 14 September 2015 Accepted 15 September 2015 Available online xxxx Keywords: Coal gasification Critical viscosity temperature Ash fusion Rotary viscometry Non-Newtonian flow Slag

a b s t r a c t Coal gasification yields synthesis gas, an important intermediate in chemical manufacturing. It is also vital to the production of liquid fuels through the Fischer-Tropsch process and electricity in Integrated Gasification Combined Cycle power generation. Minerals naturally present in coal become molten in entrained-flow slagging gasifiers. Molten coal ash slag penetrates and dissolves refractory bricks, leading to costly plant shutdowns. The extent of coal ash slag penetration and refractory brick dissolution depends on the slag viscosity, the gasification temperature, and the composition of slag and bricks. We measured the viscosity of several synthetic coal ash slags with a high-temperature rotary viscometer and their ash fusion temperatures through optical image analysis. All measurements were made in a carbon monoxide-carbon dioxide reducing atmosphere that approximates coal gasification conditions. Empirical correlation models based on ash fusion temperatures were used to calculate critical viscosity temperatures based on the coal ash compositions. These values were then compared with those obtained from thermodynamic phase-transition models. An understanding of slag viscosity as a function of ash composition is important to reducing refractory wear in slagging coal gasifiers, which would help to reduce the cost and environmental impact of coal for chemical and electricity production. Published by Elsevier B.V.

1. Introduction Coal gasification is an important industrial chemical process. Synthesis gas, a mixture of hydrogen, carbon monoxide and carbon dioxide, is produced from coal and steam as the net product of several chemical reactions outlined in Table 1 [1]. For over a century, chemical manufacturers have used synthesis gas as a feedstock to produce ammonia, alcohols, and synthetic fuels [2]. More recently, it was found that the amount of sulfur, nitrogen oxides, and particulates emitted by Integrated Gasification Combined Cycle (IGCC) electrical power plants could be reduced significantly by using synthesis gas as a fuel for gas turbines [3]. In addition, carbon dioxide is readily captured from pre-combustion synthesis gas [4,5]; consequently, a net reduction in carbon dioxide emissions by IGCC power plants may be realized through coal gasification and pre-combustion carbon capture. The development of improved carbon capture and storage capabilities for future IGCC power plants continues to be an active topic of research today [6,7]. An improved understanding of coal ash slag formation is a key factor in enabling the design and operation of entrained-flow gasifiers with greater efficiency and reliability.

☆ Contribution of the National Energy Technology Laboratory. Not subject to copyright in the U.S.A. ⁎ Corresponding author. E-mail address: [email protected] (P.Y. Hsieh).

http://dx.doi.org/10.1016/j.fuproc.2015.09.019 0378-3820/Published by Elsevier B.V.

Coal is a complex mixture of organic and inorganic compounds. During gasification, the organic compounds are converted to synthesis gas while the inorganic compounds become coal ash. A survey of bituminous coals produced in the United States reports a range of 1 to 23% (mass/mass) ash content of moisture-free coal, with an average sample containing approximately 10% (mass/mass) ash [8]. Coal ash minerals encompass oxides, sulfides, and carbonates of silicon, aluminum, iron, calcium, magnesium, sodium, potassium, and trace elements such as titanium. While over 120 minerals have been identified in coal, only 8 are common constituents: quartz (SiO2), kaolinite (Al2Si2O5(OH)4), illite ((K,H3O)(Al,Mg,Fe)2(Si,Al)4O10[(OH)2,(H2O)]), montmorillonite ((Na,Ca)0.33(Al,Mg)2(Si4O10) (OH)2·nH2O), chlorite ((Mg,Fe)3(Si,Al) 4O10(OH) 2· (Mg,Fe) 3(OH)6), pyrite (FeS2), calcite (CaCO3), and siderite (FeCO3) [9]. The proportion of these mineral constituents varies considerably for coal ashes, depending on the geographic origin of the coal. The difference in coal ash compositions causes substantial variability in their viscosity as a function of temperature. Coal ash management is an important issue for coal gasifier operation, given the high temperatures needed for efficient synthesis gas production. Gasification temperatures range from 950 to 1100 °C for fluid-bed gasifiers, and 1300 to 1600 °C for entrained-flow gasifiers [10]. Fluidbed gasifiers are suitable for gasifying low-rank coals; however, synthesis gas produced by fluid-bed gasifiers often contains volatile coal tars. Depending on the intended use, an additional clean-up step may be needed to remove the coal tars from the synthesis gas [11]. In

14

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

Table 1 Principal coal gasification chemical reactions [1]. Carbon exists in coal as a solid, while oxygen, steam and all reaction products are gases. Partial coal combustion: Boudouard reaction: Water-gas shift reaction: Carbon monoxide shift reaction

2C + O2 → 2 CO C + CO2 ⇄ 2 CO C + H2O ⇄ CO + H2 CO + H2O ⇄ CO2 + H2

comparison, entrained-flow gasifiers are capable of handling a wider range of coals, have better carbon conversion efficiency, and produce tar-free synthesis gas [1]. Due to the higher operating temperatures found in entrained-flow coal gasifiers, coal ash minerals agglomerate to form a molten slag that must be removed periodically. This is typically accomplished by tapping, where the molten slag flows through a tap hole into an ash hopper; the slag is then quenched and collected for disposal. In a typical entrained-flow gasifier, the optimal slag tapping viscosity is approximately 80 to 150 P (1 P = 0.1 Pa·s), while the maximum slag viscosity for continuous flow is 250 P [12]. Slag viscosity varies as a function of coal ash composition and gasifier operating temperature. Control of the molten slag viscosity is typically achieved by adding fluxing minerals to the slag [13,14], coal blending [15], or by adjusting the gasifier operating temperature [16]. In order to do so effectively, it is necessary to understand how the slag viscosity changes as a function of composition and temperature. 2. Literature review Coal ash slags are similar in their composition and transport properties to magmas, glasses, and slags [17]; therefore, the study of coal ash slags yields data relevant to the study of molten silicates in general. Conversely, several concepts and models from the study of molten silicates are useful in understanding how changes in composition and operating temperature affect coal ash slag viscosity. More specifically, the study of local silicate structure through optical basicity measurements [18] is important in understanding the effect of coal ash slag composition on the viscosity of fully molten slags [19–23]. In addition, the thermodynamics and kinetics of solid phase formation in molten silicates are important factors in understanding the transition from Newtonian to non-Newtonian flow at the critical viscosity temperature. 2.1. Viscosity of fully molten silicates Fully molten silicates, including coal ash slags, contain numerous SiO4 tetrahedrons interlinked by Si–O–Si bonds (Fig. 1). The extent of

SiO4 polymerization has a strong effect on viscosity, due to the fact that larger polymeric structures are more likely to become entangled, thereby impeding the shear flow of the melt [24]. The extent of polymerization in a silicate melt is affected by the presence of network modifiers: alkali metal oxides, alkaline earth metal oxides or transition metal oxides. Network modifiers convert bridging Si–O–Si bonds to nonbridging terminal oxides through Lewis acid–base reactions with the anionic SiO4 network [18]. Increasing the concentration of network modifiers present in the melt decreases its degree of polymerization, thereby decreasing its viscosity. Changes in the electron density around oxygen atoms in the SiO4 anions due to network modifiers can be assessed indirectly through optical basicity measurements [25]. Optical basicity is measured spectroscopically as the ratio of the shift in the ultraviolet absorption maximum of a probe ion (Th+ or Pb2+) in the modified silicate network relative to the shift in a reference oxide (CaO) [26]. An increase in the optical basicity of a slag is associated with a decrease in its viscosity [19,20], consistent with the formation of nonbridging terminal oxides and the breakup of large polymeric structures. The effect of composition on the viscosity of molten silicates has been studied extensively, and several local network structure models have been developed for fully molten silicates [27–29]. These models typically account for the effect of network modifiers by calculating the acid-to-base ratio in the molten silicate, where silica (SiO2), alumina (Al2O3), and iron(III) oxide (Fe2O3) are acidic oxides, and network modifiers (e.g. Na2O, K2O, Li2O, CaO, MgO, MnO, FeO) are basic oxides. The composition of the molten coal ash slag is known to play an important role in its viscosity. Interestingly, coal ash slags can be modified chemically during gasifier operation, resulting in measureable changes in viscosity. The reduction of iron(III) oxide to iron(II) oxide converts an acidic oxide to a basic oxide, which lowers the slag viscosity. The oxygen partial pressure inside a coal gasifier is typically on the order of 10−8 to 10−9 atm (1 to 0.1 mPa), sufficient to reduce most of the iron present in slags to a Fe2+ state. A decrease in slag viscosity was observed by Folkedahl and Schobert in their study on the effect of reducing atmospheres on the viscosity of coal ash slags [30]. The effect of oxygen partial pressure on coal ash slag viscosity highlights the importance of matching experimental parameters to gasifier operating conditions in data collection. While high-temperature rotary viscometry is typically employed to measure the viscosity of fully molten coal ash slags (see Section 2.4), a number of other approaches have been reported in the literature. Stanmore and Budd [31], as well as Buhre et al. [32], proposed the use of thermomechanical analysis as an alternative to rotary viscometry for fully molten coal ash slags; instead of torque, the force required to force the slag up into an annular region of defined geometry is measured instead. Buhre et al. noted that the approach is not suitable for testing subliquidus slags, as non-Newtonian slag flow leads to an erroneous interpretation of the thermomechanical data. 2.2. Critical viscosity temperature

Fig. 1. Covalent polymeric networks found in molten silicates. SiO4 tetrahedrons are linked by bridging oxygens to form a wide range of structures. The presence of cationic network modifiers breaks down the network, forming non-bridging oxygens to maintain overall charge balance. The decrease in degree of polymerization leads to a decrease in viscosity.

The service life of refractory bricks lining the interior of gasifiers can be prolonged by lowering the gasification temperature [33]; however, if the slag is cooled below its liquidus temperature, nucleation and growth of one or more solid phases becomes thermodynamically favorable. The formation of a solid phase can cause the molten slag to become a nonNewtonian fluid. The viscosity of a non-Newtonian fluid is a function of shear rate due to the interactions between solid particles suspended in the liquid phase; moreover, the shear rate dependence may itself vary as a function of time. The critical viscosity temperature (Tcv) is defined as the temperature at which solid phase formation causes a transition from Newtonian to non-Newtonian flow in molten coal ash slags [34]. Johnson reviewed viscosity data for slagging gasifier operations, and proposed a Bingham plastic model to describe the non-Newtonian slag flow [35]. A Bingham fluid is a viscoplastic substance which yields

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

only when the stress acting upon it exceeds a threshold; the yield stress reflects an internal structure formed from the solid phase present in the molten slag. Since the slag viscosity may increase abruptly below the critical viscosity temperature, it sets the lower bound for both the slag flow temperature and the slag removal temperature for an entrainedflow gasifier [36]. The effect of solid phase formation on coal ash slag viscosity has been known for over 70 years. Reid and Cohen measured the viscosity of coal ash slags as a function of temperature, and tested the hypothesis that the critical viscosity temperature is equal to the liquidus temperature (Tliq) of a coal ash slag. They observed no abrupt change in the viscosity of a synthetic slag at its liquidus temperature; instead, they noted that the critical viscosity temperature is marked by the break in the viscosity-temperature curve when the “concentration of solids has reached such a value that the particles begin to interfere with each other in the flow process [37].” Reid and Cohen were the first to note the existence of a positive linear correlation between the critical viscosity temperature and the cone softening temperature, as measured by the ash cone fusion test; however, they further note that ash cone fusion data only serve as a rough guide to slag viscosity. Watt noted that for most slags, the increase in viscosity upon cooling below the critical viscosity temperature is not the same as the decrease in viscosity when the slag is reheated [38]. The viscosity hysteresis is attributed to the difference in crystal formation and dissolution kinetics. He defined the rejoining temperature (Tj) as the temperature at which the slag viscosity during reheating equals the viscosity measured during cooling (see Fig. 2). The rejoining temperature is significant in that it represents the temperature at which the concentration of solids in the subliquidus slag ceases to have an effect on its viscosity. It is important to note that the viscosity-temperature relationship shown in Fig. 2 does not account for the effect of shear rate on viscosity. The determination of Tcv using fixed shear rate viscosity data can be subjective, particularly for coal ash slags that do not exhibit a sharp increase in viscosity during cooling (shown as point Tx, in Fig. 2). In this instance, the rejoining temperature is probably closer to the actual Tcv. It is important to keep this distinction in mind when evaluating published Tcv data. Several methods of estimating or measuring the critical viscosity temperature of molten coal ash slags have been developed over time: empirical modeling based on ash cone fusion temperatures, direct measurement of slag viscosity as a function of temperature, and thermophysical modeling based on coal ash composition.

Fig. 2. Coal ash slag viscosity increases as it is cooled below its liquidus temperature as measured in a fixed shear rate experiment. When reheated in the subliquidus range, the slag viscosity is often higher than the initial value measured during cooling. This is due to the incomplete dissolution of solid crystals formed during cooling. The rejoining temperature marks the intersection of the two viscosity-temperature curves, and can be interpreted as the temperature at which the solid phase no longer has a measureable effect on slag viscosity. A sharp rise in viscosity (Tx) was often reported as the critical viscosity temperature in past studies.

15

2.3. Viscosity of subliquidus molten silicates Once the temperature of a molten coal slag drops below its critical viscosity temperature, its viscosity becomes difficult to predict if only its composition is known. Both the cooling rate and shear rate can affect the crystallization of the molten slag. The effective viscosity (η) of a molten silicate containing a suspension of solid crystals has been described for magmas [39–49] and slags [50] using the Einstein-Roscoe equation: η ¼ η0 ð1–RΦÞ−2:5

ð1Þ

where η0 is the fluid viscosity in the absence of solids, Φ is the solid volume fraction, and R is the inverse of the maximum solid volume fraction (i.e. R = Φ−1). Mueller et al. investigated the effect of particle shape on the effective viscosity using model particles suspended in silicone oil, and found that particle jamming effects can be described by R [51]. Song et al. note that the Einstein-Roscoe model does not account for shear rate dependence, and proposed the use of a modified semi-empirical model that incorporates high- and low-shear rate viscosity values instead [52]. 2.4. Ash cone fusion test The propensity for coal ashes to form molten slags was first studied in a systematic manner through ash cone fusion testing. The ceramics industry has long utilized pyrometric cones as a tool for measuring heat work, or the combined effect of temperature and time, to ensure uniformity and reproducibility over multiple kiln firing cycles [53]. While conceptually similar to the pyrometric cone equivalent test [54], the ash cone fusion test differs in defining four critical temperatures as a function of cone geometry during the test. The method described by Fieldner et al. was adopted by American Society for Testing and Materials in 1919 [55], and remains substantially the same today (ASTM D1857) [56]. Similar ash cone fusion tests have been defined by international (e.g. ISO 540) and national standards organizations (e.g. BS 1016-15, DIN 51730, JIS M 8801, GB/T 219, and GOST 2057). Under ASTM D1857, coal ash samples are first wetted with a dextrin solution or water and molded into triangular pyramidal cones. The cones are then dried, heated to remove the binder, and mounted on a kaolin-alumina base. Changes in cone geometry are observed optically while the cones are heated at 8 °C per minute. The physical and chemical changes that take place as the cones are heated, including the sintering of ash or slag particles, cause them to slump over or fuse down into a lump. The temperature at which the sharp tip of the cone becomes rounded due to sintering is defined as the initial deformation temperature (IDT). The softening temperature (ST) is defined as the temperature at which the cone height is equal to its width, while the hemispherical temperature (HT) is defined as the temperature at which the cone width is equal to twice its height. Finally, the fluid temperature (FT) is defined as the temperature at which the molten cone has formed a flat layer with a maximum height of 1.6 mm. The gas atmosphere may be either reducing or oxidizing, provided that there is constant gas flow over the cones during the test. Ash cone fusion tests are most useful for comparing the slagging behavior for coals from different sources. For this reason, a large body of data on coal ash composition and critical temperatures has been assembled over time for coal samples across the world (see Table 2) [15,57–94]. Several efforts to relate ash cone fusion temperatures to coal ash composition and critical viscosity temperature have been made. Reid and Cohen measured the viscosity of coal ash slags below their liquidus temperatures, and noted a positive linear correlation between their cone softening temperatures with their critical viscosity temperatures. Selvig and Gibson noted that ash fusion temperatures and slag viscosity both depend on the starting coal ash composition [8]. Sage and McIlroy

16

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

Table 2 Coal ash cone fusion data reported for various coal-producing regions across the world, with the national or international standard used in the respective studies. Year

Author

Country

Region

Standard

Ref.

1944 1975 1984 1987 1992 1993 1995 1998 1998 1999 2000 2000 2001 2002 2002 2005 2006 2007 2009 2009 2009 2009 2009 2010 2011 2011 2011 2011 2012 2013 2013 2013 2013 2013 2014 2014 2014 2015

Stansfield Bowden Quon Gray Mukhergee Vorres Vassilev Bragg Van der Veen Qiu Bryant Weeber Alastuey Bradley Lolja Vassileva Vassileva Hackley Akar Van Dyk Wu Maystrenko Mladenovic Tewalt Wu Vamvuka Wang Li Yuan Liu Naveed Huang Chelgani Jing Duchesne Kong Čarnogurská Chakravarty

Canada New Zealand Canada New Zealand India United States Various United States Various China Unspecified South Africa Spain United Kingdom Albania Bulgaria Bulgaria Venezuela Turkey South Africa China Ukraine Various Various China Greece China Various China Various Pakistan China Afghanistan China Canada China Ukraine India

Alberta

ASTM D1857

Various Various Rajasthan Various — Argonne Premium Coal Sample Program

ASTM D1857 Multiple ISO 540a ASTM D1857 JIS M 8801 ASTM D1857 ISO 540 GB/T 219 AS 1038, Part 15

[57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [68] [67] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [82] [83] [93] [84] [85] [86] [87] [88] [89] [90] [91] [15] [94] [92]

a

various - US Coal Quality (COALQUAL) database European Centre for Coal Specimens Various Witbank Castile-La Mancha various - BCURA Coal Sample Bank Various Various Various Zulia Mpumalanga Various Donetsk World Coal Quality Inventory (WoCQI) database Various Macedonia Various Various Punjab Various Various Shanxi Various Various Donetsk Odisha

ASTM D1857 BS 1016, Part 15 ASTM D1857 JIS M 8801 JIS M 8801 ASTM D1857 ASTM D1857 GB/T 219 GOST 2057 SRPS B.H8.325 ASTM D1857 GB/T 219 ASTM D1857* Not specified GB/T 219 GB/T 219 GB/T 219 See reference ASTM D1857 GB/T 219 ASTM D1857 GB/T 219 ISO 540 See reference

Use of heating microscope suggests ISO 540 or equivalent.

suggested the use of the hemispherical temperature as the basis for predicting the critical viscosity temperature of eastern U.S. coals containing iron oxides [95]. Marshak and Ryzhakov also proposed the use of the hemispherical temperature, equivalent to tB as defined in GOST 2057, to estimate Tcv [36]. Seggiani developed a multifactorial regression model based on the coal ash composition of 295 samples to predict both the ash fusion temperatures and the critical viscosity temperature [96]. Though the ash cone fusion test is conceptually simple and readily implemented in the laboratory, the complexity of the physical and chemical changes occurring during the test makes it difficult to relate the observed critical temperatures to measurable thermophysical properties. For example, Hansen et al. applied simultaneous thermogravimetric analysis and differential scanning calorimetry to biomass and coal ash samples and found that the samples started to melt between 50 to 100 °C below the initial deformation temperature as measured through the ash cone fusion test [97]. Clearly, the slumping behavior of coal ash cones as a function of composition remains an interesting unsolved rheological problem that merits further study in the future. 2.5. High-temperature rotary viscometry Viscosity can be defined in terms of momentum transport through a fluid [98]. In rotary viscometry, the fluid under investigation is placed between two concentric cylinders, where one of the cylinders is stationary and the other is rotating. The torque transmitted by the fluid is directly proportional to its viscosity and the angular velocity of the rotating cylinder [99]. By varying the angular velocity and shear rate during stepwise cooling of a molten coal ash slag sample in a rotary

viscometer, the transition from Newtonian to non-Newtonian flow can be measured directly. The first high-temperature rotary viscometer was built and operated by the U.S. Bureau of Mines to investigate the viscosity of blast furnace slags [100] based on principles outlined by Margules in 1881 [101]. Nicholls, Reid and Cohen later extended the study to include coal ash slags [37,102]. The rotary viscometer devised by Reid and Cohen differed from its predecessors in that it utilized a rotating inner cylinder instead of a stationary inner cylinder; the modification enabled the measurement of coal ash slag viscosity under a controlled gas atmosphere. Though torque transducers employed in rotary viscometers have changed over time, the same rotating inner cylinder design remains the most commonly used geometry for high-temperature coal ash slag viscosity measurements today. Earlier rotary viscometry studies on molten coal ash slags often report the critical viscosity temperature as the temperature at which an abrupt increase in viscosity is observed during constant cooling of the slag sample (e.g. point Tx in Fig. 2). Shear rates were typically not varied or reported. For example, Quon et al. used a rotary viscometer to measure the viscosity of western Canadian coal ash slags, noting that a sharp decrease in viscosity occurred around 1450 °C [59]. The shear rate used in the study was not reported. Vorres and Greenberg hypothesized that the critical viscosity temperature may be related to the ternary eutectic temperature of the slag, but did not explore the transition from Newtonian to non-Newtonian behavior by varying the shear rate in their measurements [103]. Schobert et al. investigated the viscosity of lignite coal ash slags in oxidizing and reducing environments relevant to slagging gasifier operation [104]. A fixed rotation rate of 64 rpm was used in their study. They hypothesized that changes in

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

the molten slag composition due to solid phase formation may cause flow problems in slag tapping, even if non-Newtonian effects are absent. Cho et al. described the design and operation of a high-temperature rotary viscometer, which was operated at 5 or 10 rpm as the slag temperature was decreased at 2 °C/min [105]. They reported the temperature at which the viscosity increased sharply for two U.S. coal ashes, and further noted that anorthite grains aligned with the flow direction were observed in the cooled slag. Though constant shear rate and constant cooling rate measurements are easier to implement experimentally in a rotary viscometer, the resultant data are difficult to interpret correctly because the viscosity of non-Newtonian fluids is a function of shear rate and time. Further work is needed to relate critical viscosity temperature reported for constant shear and cooling rate studies to values reported in studies that vary the shear rate and employ stepwise cooling. The importance of observing changes in slag viscosity as a function of both temperature and shear rate was recognized in more recent studies on slag critical viscosity temperature. Nowok systematically varied the rotation rate from 4 rpm to 64 rpm, between 1240 °C and 1380 °C in a study on a western U.S. coal ash slag (Beulah, North Dakota); the critical viscosity temperature was found to be between 1240 and 1280 °C [106]. In the same work, the role of nucleation and spinodal decomposition was explored through microstructural analysis of quenched slags. Spinel, gehlenite, fayalite, and anorthite crystals were observed in the quenched slags through scanning electron microscopy (SEM). Oh et al. studied the effect of crystalline phase formation on coal ash slag viscosity [107]. The rotation rate was varied from 0 to 65 rpm over a period of 3 min, but the resultant shear rate-shear stress curves were not reported; the authors note only that the slag was a Newtonian fluid at high temperatures, and showed non-Newtonian characteristics at low temperatures. The alumina spindle and crucible used in the study were found to dissolve into the slags, increasing their alumina content by 2 to 4%. Anorthite, mullite, hercynite, and corundum crystals were observed in SEM micrographs of the rapidly cooled slags. The crystals were found to be aligned with the slag flow direction, which suggests that they were already present during the viscosity measurement and not formed during cooling. Hurst et al. performed a series of rotary viscometry measurements on Australian bituminous coal ash slags to determine their critical viscosity temperatures and viscosities as a function of composition [108–110]. They recorded the shear rate and torque as a function of temperature, at 30 °C intervals during stepwise cooling. Since the critical viscosity temperature marks a transition from Newtonian to non-Newtonian behavior, the value reported are bounded by the temperature interval used, limiting the accuracy to ±15 °C. Seok et al. investigated the viscosity of basic oxygen furnace slags, which are similar in composition to coal ash slags with very large amounts of CaO and MgO additives [50]. They noted that an abrupt increase in the slag viscosity on cooling at 17 to 25% solid content by mass fraction, as estimated through thermophysical modeling of the slags based on their composition. Song et al. studied the viscosity of Chinese bituminous coals (e.g. Huainan [111], Shen-Fu [112]) under gasification conditions. Both constant rotation rate (20 rpm) [111] and variable rotation rate (10 to 100 rpm) methods were employed [112,113]. Slag samples were quenched periodically and removed for study using SEM analysis. They noted changes in molten slag viscosity as a function of temperature and crystal size distribution, and reported that transition from Newtonian to non-Newtonian flow occurs around 5% (volume/volume) solid phase content in the slag [52]. Ilyushechkin and Hla measured the viscosity of Australian bituminous coals with high iron content [114], over the temperature range 1200 to 1600 °C during stepwise cooling. The shear rate was varied between 5 to 15 s − 1 to detect the transition from Newtonian to non-Newtonian flow. They noted that anorthite, mullite, and spinel

17

constitute the primary phase fields for the slags used in their study. The amount of solid phase observed in the quenched slags was found to be less than the amount predicted by thermophysical modeling, which was attributed to the short equilibration time (0.5 to 1 h) used in the viscosity measurements. Interestingly, the transition from Newtonian to non-Newtonian flow has also been reported in subliquidus magmas [45–47,49,115,116]. Of particular relevance is the observation that shear-thinning tendency increases with solid phase content, and that the increase in viscosity is attributable mainly to the solid phase formation [45]. The critical volume fraction for the Newtonian to non-Newtonian transition was found to be between 6 and 13% for an anisotropic plagioclase primary phase field [47]. Advances in experimental and analytical methods developed for magma rheology may be helpful in understanding the microstructural evolution of molten coal ash slags upon cooling. Compared with the ash cone fusion test, high-temperature rotary viscometry yields data that is directly relevant to coal gasifier design and operation; however, it also more time consuming and resource intensive to perform in the laboratory. While the measurement of viscosity as a function of temperature and shear rate forms the basis for accurate determination of critical viscosity temperatures, empirical models that rely solely on coal ash composition and slag viscosity data remain constrained by the range of compositions used in their development. In contrast, thermophysical models based on Gibbs energy minimization can be applied to the critical viscosity temperature problem, if the coal ash composition and the primary phase field (i.e. the first crystal phase to form as the molten slag is cooled below its liquidus temperature) are known. 2.6. Thermophysical modeling of coal ash slags Crystallization is a complex phenomenon. While thermodynamics dictates whether solid phase formation is energetically favorable, nucleation and growth kinetics can have a profound effect on the size, number, and shape of the crystals that are formed. In principle, it is possible to calculate the liquidus temperature of a molten coal ash slag as a function of its composition through thermophysical modeling. The critical viscosity temperature can then be estimated using the liquidus temperature and the solid phase fraction present in the molten slag as a function of temperature. Early attempts to apply thermophysical modeling to the prediction of the critical viscosity temperature were hindered by the paucity of reliable thermodynamic data on the activity of trace oxides (e.g. TiO2) in coal ash slags. Trace oxides were thought to affect the nucleation and growth of solid phases in molten slag [106]. This limitation was noted by Jung and Schobert in their early work on coal ash slag viscosity prediction using the SOLGASMIX program [117]. In the last three decades, thermophysical modeling of molten oxide systems has been improved significantly by the introduction of new software tools based on the CALPHAD method [118]. For example, FactSage, an integrated software package that combines thermochemical databases with Gibbs energy minimization software [119,120], is widely used to calculate chemical equilibria in slag systems. It has been used to predict the formation of solid phases in basic oxygen furnace slag [50], estimate coal ash fusion temperatures [121], model the effect of solid phase formation on viscosity in coal ash slags [52,85, 111,112,114,122–124] and petroleum coke-coal blend slags [125], and assess the effect of additives on coal ash slag viscosity [15,126]. MTDATA is another CALPHAD-based software tool that has been used to estimate coal fusibility [127] and ash fusion temperatures [128]. The critical transition temperature from Newtonian to nonNewtonian flow for a subliquidus coal ash slag is intrinsically linked to its liquidus temperature, which is a function of its composition. In the past, ash cone fusion temperatures were used as proxy variables for coal ash liquidus temperatures, which were difficult to measure experimentally and challenging to calculate through Gibbs energy

18

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

minimization. The availability of CALPHAD-based software tools today means the liquidus temperature of a slag can be calculated readily from its composition, as well as its subliquidus solid fraction as a function of temperature. Song et al. have developed a model that relate the critical transition temperature to the liquidus temperature, as calculated from the coal ash slag composition [113]. The approach offers greater flexibility in the range of coal ash compositions over which the model is applicable, compared with earlier approaches based on proxy variables. 3. Materials and methods 3.1. Synthetic coal ash preparation Six synthetic coal ash slag samples with compositions drawn from the U.S. COALQUAL database [64] were prepared from minerals and chemically pure oxides from commercial suppliers (see Table 3). First, red mica (New Mexico Clay), kaolin (Sheffield Pottery), and MgO (ACS reagent grade, Matheson Coleman & Bell) were calcined separately at 800 °C for 4 to 6 h in air to ensure that they are fully dehydrated. Next, dolomite (Sheffield Pottery), feldspar (G-200EU potash, Sheffield Pottery), and microcrystalline silica (IMSIL, Unimin Specialty Minerals) were combined and mixed for 20 min in a tubular mixer. The calcined mica, kaolin, and MgO were then added to the synthetic coal ash mixture, along with red iron oxide (R2501, Nubiola). The mixture was then ball milled for 6 h. A smear test was performed on each of the milled powders to ensure uniformity. Synthetic slag compositions were verified using wavelength dispersive X-ray spectroscopy. An additional synthetic coal ash sample with lower iron oxide content was prepared to expand the composition range of samples tested in this study. It was prepared by mixing olivine (Twin Sisters mine, WA), feldspar, silica, iron oxide, wollastonite (NYCO Minerals), and Al2O3 (99.9%, AEE/Micron Metals), then ball milling the mixture for 6 h. The synthetic ash samples were melted in high-density (99.9%, mass/mass) alumina crucibles in electric furnaces, at temperatures ranging between 1400 to 1535 °C, based on the estimated liquidus temperatures for each composition. The following temperature program was used for all samples: 1. Heating to 1000 °C at a rate of 150 °C/hour, with a dwell time of 2 hours. 2. Heating to 1350 °C at a rate of 50 °C/hour, with a dwell time of 2 hours. 3. Heating to 1400 to 1535 °C, at a heating rate of 50 °C/hour, with a dwell time of 1 hour. 4. Cooling to ambient temperature, at a cooling rate of 150 °C/hour. The resultant synthetic slags were mechanically separated from the alumina crucibles, crushed, and ground into irregular particles approximately 2 mm in diameter and finer. The ground slag samples were used directly for high-temperature rotary viscosity measurements. The slag particles were further reduced to a fine powder with a shatterbox in

the preparation of ash cone samples. A dextrin binder was used in the fabrication of ash cones, which were air-dried prior to testing. 3.2. Ash fusion temperature measurement Ash cones prepared according to ASTM D1857 were mounted on a prefabricated kaolin-alumina base with alumina-based cement. Samples were heated at 8.7 °C/min, with an expanded uncertainty of 0.8 °C/min, under a reducing atmosphere (0.799 standard liters per minute CO2 and 1.200 slpm CO, with an expanded uncertainty of 0.007 slpm) to 1500 °C in a commercial ash fusibility determinator. Expanded uncertainties are calculated with a coverage factor of 2, which approximates 95% coverage of the true measurand value. The platinum/platinum-rhodium thermocouple in the ash fusibility determinator was calibrated against the melting points of gold and nickel wires. Changes in the cone geometry were monitored with a digital camera and recorded at a rate of 1 frame/°C. Resultant images were analyzed using a public domain image processing program (NIH ImageJ, version 1.49t) [129]. The critical temperature points (e.g. IDT, ST, HT, and FT) were identified and recorded by the operator. 3.3. High-temperature rotary viscometry High-temperature rotary viscometry of molten synthetic coal ash slags was performed under a reducing atmosphere (19.3 standard cubic centimeter per minute CO and 10.7 sccm CO2, with an expanded uncertainty of 0.2 sccm), over a temperature range spanning 1296.5 to 1530.3 °C, and a shear rate range spanning 0.13 to 25.3 s−1. The oxygen activity (≈ 10− 8) reflects conditions typically encountered in coal gasification. The sample temperature was measured using an S-type (platinum/platinum–rhodium) thermocouple inside an alumina protection tube placed next to the alumina crucible containing the molten slag. The sample temperature measured with the S-type thermocouple was corrected using previously determined constants derived from calibration measurements made with a K-type (chromel/alumel) thermocouple placed inside a crucible filled with granular alumina. The expanded uncertainty of the sample temperature ranges from 7.2 to 7.8 °C, with the standard thermocouple limit of error (1.5 °C or 0.25%) and measurement repeatability as the principal sources of uncertainty. The expanded uncertainty of the shear rate is estimated to be on the order of 0.1 s−1, and reflects uncertainty due to thermal expansion of the high-density alumina spindle (rotor) and crucible (see Fig. 3 for dimensions). Rotational eccentricity due to curvature of the hollow alumina tube used to connect the spindle to the torque transducer appears as a sinusoidal oscillation in the torque data; the effect is more noticeable at low shear rates. The uncertainty contribution due to eccentricity

Table 3 USGS COALQUAL database identifiers and compositions (%, mass/mass) of synthetic coal ashes used in this study. Sample 7 is not a part of the USGS COALQUAL database; it is included in this study to expand the composition range to include slags with lower iron(III) oxide content. Sample number

COALQUAL identifier

SiO2

Al2O3

CaO

MgO

Na2O

K2O

TiO2

Fe2O3

1 2 3 4 5 6 7

W236240 W201717 WAVE 2 W233566 W188929 W216040

40.8 38.1 36.7 35.7 44.8 45.1 49.6

20.4 25.4 29.3 20.4 24.0 20.5 26.3

0.6 2.6 0.9 2.0 1.0 1.1 2.1

0.8 0.7 0.3 0.5 0.6 0.7 1.1

0.35 0.3 0.1 0.2 0.1 0.1 0.64

2.45 1.6 0.7 1.6 1.5 2.5 2.4

0.8 1.0 0.7 0.9 1.0 1.3 1.2

32.6 27.5 31.4 36.7 27.1 28.7 15.4

Fig. 3. Dimensions of high-density alumina crucibles and spindles (rotors) used in hightemperature rotary viscometry of molten synthetic coal ash slags. The top of the molten slag (shown in dark gray) is located at the start of each measurement by monitoring the change in torque as the spindle is lowered into the crucible. The spindle is immersed to a depth of 18 mm, as measured from the top of the slag. See text for discussion of uncertainty due to variations in immersion depth.

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

is minimized during data analysis by averaging over multiple rotation cycles. The viscometer was calibrated against 1-decene (Cannon N2500) and polybutene reference standards (Cannon N18000, N190000) from 22 to 100 °C, and borosilicate glass (NIST Standard Reference Material 717a) from 1150 to 1250 °C. Fig. 4 plots the standard uncertainty of the calibration curve (coverage factor k = 1)along with the standard error ranges of the viscosity standards. For each viscosity measurement, 50 g of synthetic coal ash slag is first added to a high-density alumina crucible. The crucible is then loaded into the viscometer, supported by three machined alumina support rods. The alumina spindle is then lowered to the top of the slag-filled crucible without coming into contact with the contents. An alumina protection tube is then securely attached to a stainless steel cooling plate with bolts; a rubber gasket between the tube and plate ensures a gas-tight seal that prevents the carbon monoxide-carbon dioxide mix used in the experiment from escaping. The flow of the reducing gas mix was started 10 to 12 h prior to the start of the heating program, and continued until the viscosity measurements were complete. Given the oxygen activity (≈ 10− 8) associated with the carbon monoxidecarbon dioxide mix within the temperature range of the measurements, all of the iron present in the molten slags can be expected to exist in the Fe2+ state. The slag sample was heated at a rate of 5 °C per minute until the target temperature, typically 1530 °C, was reached. The molten slag sample was then allowed to equilibrate for 30 min before the alumina spindle was slowly lowered towards the slag surface while rotated at a constant rotation rate of 3 rpm. The spindle insertion depth at which the viscometer registered a torque greater than 1% was used as the reference point for initial spindle-slag contact; the spindle was then lowered an additional 18 mm into the molten slag. The choice of slag viscosity measurement depth was based on the change in torque as a function of spindle depth measured using 1-decene standard fluid; uncertainty in the spindle depth is expected to contribute no more than 1% to the uncertainty of the measured torque value. Slag viscosity was measured in accordance with ASTM D2196, test method B (viscosity under changing speed conditions) [130], to assess whether the slag has transitioned from Newtonian to non-Newtonian flow at the measurement temperature. Each viscosity measurement was followed by a 20 to 30 °C decrease in slag temperature. The spindle rotation was stopped, and the slag allowed to equilibrate at the new temperature for 60 min prior to the next slag viscosity measurement. Measurements were stopped either when one or more viscosity measurements have been made below the critical viscosity temperature,

19

or when the viscosity exceeded the measurement range of the viscometer. The slag was then reheated above its liquidus temperature to facilitate the extraction of the spindle. Once the spindle has been raised above the top of the crucible, the slag and crucible were allowed to cool at a rate of 5 °C per minute to ambient temperature. The total time from the start of reducing gas injection to the start of the furnace cooling step ranges from 24 to 30 h, depending on the number of viscosity measurements made and the number of shear rates tested. Lower shear rates increase the measurement time considerably, since the viscosity is measured after ten spindle rotations to allow for torque stabilization. 4. Results and discussion 4.1. Scanning electron microscopy and energy-dispersive X-ray spectroscopy The microstructures and compositions of the post-test synthetic slags were examined using scanning electron microscopy and energydispersive X-ray spectroscopy. Table 4 summarizes the post-test sample compositions. Mullite, spinel, and cordierite phases present in the slags were identified based on their compositions. Fig. 5 shows the microstructure of slag sample 4, with an iron-magnesium spinel primary phase field; Fig. 6 shows the microstructure of slag sample 5, with a mullite primary phase field. Due to the slow cooling rate, the solid phases and fractions observed in the post-test slags do not accurately reflect the slag composition during high-temperature rotary viscometry. 4.2. Repeatability and reproducibility of ash fusion temperatures Table 5 summarizes the ash fusion temperatures associated with the six sample compositions drawn from the U.S. COALQUAL database [64], as well as values measured in the present study. According to ASTM D1857, the difference in ash fusion temperatures should not exceed 30 °C for repeated measurements by the same operator, using the same apparatus; for different operators and apparatuses, differences in ST and HT should not exceed 55 °C, and differences and FT should not exceed 85 °C for reducing atmosphere measurements [56]. While the repeatability criterion was satisfied in the present study, the reproducibility criterion was met in only half of the measurements. Only sample 1 met reproducibility limits for both ST and FT. Samples 3 to 6 failed either the ST or FT reproducibility limit, while sample 2 failed both. The COALQUAL database contains ash fusion temperatures from measurements on real coal ashes, while the present study used coal ash slags synthesized to match the reported composition of those coal ashes. Slight differences in composition, introduced through measurement uncertainty and trace elements not reported, may account for some of the observed differences between the two ash fusion temperature data sets. Moreover, the synthetic coal ash slag samples used in the present study were homogenized during sample preparation through melting and crushing; the COALQUAL data is derived from real coal ashes that have not been homogenized in this manner. The choice to

Table 4 Post-test synthetic slag compositions, by mole fraction (atomic per cent), measured using energy-dispersive X-ray spectroscopy. The composition of sample 2 was measured by an independent laboratory using X-ray fluorescence spectroscopy. Percentages may not sum to 100 due to rounding.

Fig. 4. The rotary viscometer was calibrated against four reference fluids: 1-decene (N2500), polybutene (N18000, N190000), and borosilicate glass (SRM717a). The measured torque, normalized for rotation rate, is proportional to the dynamic viscosity of the fluid. Error bars show the published expanded uncertainties in viscosity for the reference fluids. The dashed lines show the standard uncertainty (k = 1) for the torqueviscosity calibration curve.

Sample number

O

Na

Mg

Al

Si

K

Ca

Ti

Fe

1 2 3 4 5 6 7

70.7 62.5 69.7 53.4 71.1 70.4 73.8

0.1 0.1 0.1 0.1 0.1 0.1 0.1

0.3 0.5 0.1 0.2 0.2 0.3 0.4

7.1 11.1 8.9 9.8 7.7 7.2 8.8

14.0 15.9 13.5 20.4 14.7 14.8 12.6

0.7 0.7 0.3 1.0 0.4 0.7 1.0

0.1 0.9 0.3 1.1 0.3 0.3 0.5

0.1 0.6 0.2 0.3 0.2 0.3

6.8 7.7 6.8 13.7 5.4 5.9 2.7

20

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26 Table 5 USGS COALQUAL ash fusion temperature data and values measured in the present study. The expanded uncertainty (U) reflects a coverage factor of 2, corresponding to 95% coverage of the measurand. Measurement repeatability is the largest source of uncertainty in the ash fusion temperature values.

Fig. 5. Electron micrograph of slag sample 4. Brightness and contrast of the whole image were enhanced linearly to improve the clarity of fine details. Iron-magnesium spinel (hercynite) constitute the primary phase present in the cooled slag sample.

homogenize the composition of the synthetic slag samples may also account for some of the observed differences in ST and FT. The difficulty of reproducing ash fusion temperatures for coal ashes based on published composition values highlights the need for accuracy in ash fusion and composition measurements. 4.3. Molten slag viscosity as a function of temperature The synthetic slag compositions were selected from the COALQUAL database initially to investigate the effect of the silica-to-alumina ratio and iron(III) oxide content on molten slag viscosity. Table 6 provides a summary of the theoretical and experimental dynamic viscosities at the respective slag liquidus temperatures. Theoretical dynamic viscosities were calculated using a Modified Quasichemical Model approach (Viscosity module, Melts database, FactSage version 6.4), which estimates slag viscosity based on the fraction of free, bridging, and non-bridging oxygen atoms as a function of composition of temperature (see Section 2.1) [131]. Experimental values were calculated by interpolating measured viscosities (samples 1, 4, and 6) using linear or polynomial least-squares fits, or by extrapolation where the liquidus temperature exceeds the maximum temperature for which dynamic viscosity data is available. The theoretical and experimental singlephase viscosity values are in good agreement (R2 = 0.88). The dynamic viscosity of samples 1–7 are plotted as functions of temperature or shear rate in Figs. 7–13; the tabulated form of the data may be found in the supplemental material section. Figs. 8–10 illustrate the Ostwald-de Waele power law dependence of apparent viscosity on shear rate (see Section 4.3). The transition from Newtonian to nonNewtonian flow is evident in Figs. 11–13: above the critical viscosity

sample number

USGS

NETL

ST, °C

FT, °C

ST, °C

U(ST)

HT, °C

U(HT)

FT, °C

U(FT)

1 2 3 4 5 6 7

1182 1410 1284 1132 1166 1154

1304 1466 1367 1243 1221 1182

1187.5 1249.8 1202.0 1235.0 1184.0 1194.0 1211.6

36.3 33.7 19.4 20.8 4.0 38.3 23.3

1234.8 1279.8 1249.0 1267.3 1213.0 1234.0 1250.8

1.9 15.9 28.6 12.9 28.3 27.8 46.4

1302.0 1342.5 1317.4 1306.3 1316.3 1305.7 1407.2

21.9 19.4 34.1 17.0 37.0 13.3 28.3

temperatures, the measured viscosities were not affected by changes in the shear rate. The accuracy of the viscosity measurement is affected by the magnitude of the torque exerted on the torsion spring in the rotary viscometer. The uncertainty relative to the measurand is comparatively lower at higher torque values; consequently, a spindle with an appropriate diameter is typically chosen to yield the greatest accuracy for the fluid of interest over the temperature range of the measurement. In this experiment, only one spindle diameter was used over a wide range of viscosity values due to experimental constraints. More specifically, it was not possible to replace the spindle in the middle of the experiment out of concern for laboratory safety. Consequently, the accuracy of the viscosity measurement relative to the expanded uncertainty is lower at higher temperatures where the molten slag is more fluid. At the other end of the temperature range, the rise in the coal ash slag viscosity means that the shear rate must be decreased to keep the torque within the instrument limit. For this reason, the shear rate measurement range decreases while the measurement time increases. The increase in measurement time is a potential source of error from composition drift. The high-density alumina spindles and crucibles used in this study are known to dissolve in molten coal ash slags over time [132,133]. Kim and Oh reported that the alumina content of coal ash slags measured with alumina spindles and crucibles increased by 2 to 5% compared to slags measured with platinum-lined alumina crucibles [134]; they noted that a 2.1% increase in alumina raised the liquidus temperature by 32 °C for an anorthite primary phase field. Though problematic for the accuracy of the Tcv measurement, the systematic error introduced by alumina dissolution leads to a more conservative measure of the lower temperature limit for entrained-flow gasifier operation. The dissolution of alumina in coal ash slags highlights a key challenge for high-temperature rotary viscometry of molten silicates. At temperatures of interest, few sensor materials are truly inert. Mills and Rhine noted that graphite crucibles reacted with iron oxides present in the slag to form carbon monoxide bubbles and metallic iron [135]. French et al. observed that molybdenum also reacts with slags, Table 6 Liquidus temperatures, theoretical dynamic viscosities (ηMQM, see text for discussion), measured dynamic viscosities, and critical viscosity temperatures (Tcv) of synthetic coal ash slags. The expanded uncertainty (U) reflects a coverage factor of 2, corresponding to 95% coverage of the measurand. The Tcv value reported for slag 3 is an estimate based on data from a related study on coal ash slag additives, and is for a slag with 10% (mass/ mass) MgO addition.

Fig. 6. Electron micrograph of slag sample 5. L Brightness and contrast of the whole image were enhanced linearly to improve the clarity of fine details. Mullite (dark gray, acicular) constitutes the primary phase, as predicted through thermophysical modeling; cordierite (white, dendritic) is formed later during final cooling of the sample.

sample number

Tliq, °C

ηMQM, P

ηmeas, P

U(ηmeas)

Tcv, °C

U(Tcv)

1 2 3 4 5 6 7

1356 1444 1501 1375 1454 1374 1541

37 18 9 10 38 70 81

46 15 11 14 18 77 72

26 26 26 26 26 26 26

1355 1436 1443 1338 1391 1349 1504

15 8 15 5 5 6 7

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

Fig. 7. Select dynamic viscosity data as a function of temperature and shear rate for slag sample 1 (COALQUAL identifier: W236240); additional data may be found in the supplemental material section. Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value.

albeit to a smaller extent; however, the components could not be reused due to the loss of ductility to recrystallization [136]. In addition to providing data useful for developing viscosity models for fully molten silicates, the present study also highlights the importance of measuring slag viscosity as a function of shear rate in addition to temperature. For single shear rate measurements, only samples 1, 4, 5 and 6 exhibited the sharp increase in viscosity that is typically associated with Tcv in the early literature (e.g. Fig. 2). Moreover, the Tcv value determined from a constant cooling rate experiment would depend on the choice of shear rate used in the measurement. Samples 2 and 3 exhibit non-Newtonian flow behavior over much of the temperature range measured in this study as shown by the shear rate dependence plotted in Figs. 8 and 9. A single shear rate viscosity measurement would not provide sufficient data to show that the Tcv falls outside the measured temperature range for sample 3. 4.4. Critical viscosity temperatures of synthetic coal ash slags The critical viscosity temperatures of samples 1–7 are also reported in Table 6. A transition from Newtonian to non-Newtonian flow within the temperature range studied for all samples except sample 3; only non-Newtonian flow was observed for this sample up to 1427 °C. Since the change in viscosity as a function of shear rate drops to zero

Fig. 8. Select dynamic viscosity data as a function of temperature and shear rate for slag sample 2 (COALQUAL identifier: W201717); additional data may be found in the supplemental material section. Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value. The molten slag clearly exibits non-Newtonian behavior at 1403 °C and lower temperatures, following a pseudoplastic power-law relationship. See text for further discussion.

21

Fig. 9. Dynamic viscosity data as a function of temperature and shear rate for slag sample 3 (COALQUAL identifier: WAVE 2). Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value. The molten slag exibits nonNewtonian behavior over the entire temperature range measured in this study, as shown by the Ostwald-de Waele power-law relationship between dynamic viscosity and shear rate.

at the critical viscosity temperature, we estimate the upper bound of the critical viscosity temperature for sample 3 to be 1604 °C based on the intersection of the 0.13 cm− 1 (0.50 rpm) and 25.3 cm− 1 (100 rpm) linear regression lines. The actual value is expected to be lower, due to the sharp increase in dynamic viscosity at the critical viscosity temperature. In a related study on the effect of additives, we observed a critical viscosity temperature between 1429 and 1457 °C (with an expanded uncertainty of 7.2 and 7.4 °C, respectively) for sample 3 when 10% (mass/mass) MgO was added. Assuming that MgO additives do not have a significant effect on solid phase formation, we estimate the critical viscosity temperature to be 1443 °C, bounded by the observed viscosity values. This is consistent with the upper bound on the critical viscosity temperature noted above. Pseudoplastic flow is observed for samples 1–6, where the apparent viscosity of the fluid decreases as a function of increased shear rate. The change in viscosity as a function of shear rate can be described by the Ostwald-de Waele power law [137]: μ ¼ Kð∂u=∂rÞ

n−1

ð2Þ

Fig. 10. Select dynamic viscosity data as a function of temperature and shear rate for slag sample 4 (COALQUAL identifier: W233566); additional data may be found in the supplemental material section. Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value. The molten slag shows a transition from Newtonian to non-Newtonian flow between 1341 °C and 1331 °C during cooling.

22

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

Fig. 11. Select dynamic viscosity data as a function of temperature and shear rate for slag sample 5 (COALQUAL identifier: W188929); additional data may be found in the supplemental material section. Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value. The molten slag shows a transition from Newtonian to non-Newtonian flow between 1393 °C and 1388 °C during cooling, accompanied by a discontinuous increase in viscosity.

where μ is the dynamic viscosity (Pa·s), K is the flow consistency index (Pa·sn), ∂u/∂r is the shear rate (s−1), and n is the average flow behavior index (dimensionless). The flow behavior indices for samples 1–6 range from 0.4 to 0.8; for contrast, a Newtonian fluid has a flow behavior index of 1. The pseudoplastic response to shear can be understood in terms of the breakup of the internal structure formed from the solid phase present in the molten slag. 4.4.1. Correlation with ash fusion temperatures Estimated critical viscosity temperatures based on ash fusion data are reported in Table 7. Reid and Cohen hypothesized that the Tcv = ST [37], noting that “[t]he softening temperature is a fair measure of the temperature of critical viscosity, although large unexplained differences can occur.” Sage and McIlroy proposed the use of a relationship based on the hemispherical temperature consistent with the ASTM definition, offset by 200 °F (111 °C) [95]: Tcv ð CÞ ¼ HT þ 111 C:

ð3Þ

Marshak and Ryzhakov proposed a similar approach, based on t2 as defined in GOST 2057 [36]; it should be noted that t2 is equivalent to

Fig. 13. Select dynamic viscosity data as a function of temperature and shear rate for slag sample 7; additional data may be found in the supplemental material section. Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value. The molten slag shows a transition from Newtonian to non-Newtonian flow between 1506 °C and 1495 °C during cooling, accompanied by a discontinuous increase in viscosity. Open circles (○) are from a replicate study under the same testing conditions.

the hemispherical temperature (tB) [78], and not the ASTM softening temperature (ST): Tcv ðKÞ ¼ 0:75 HT þ 548K:

ð4Þ

The correlation between the estimated critical viscosity temperatures and the values measured in the present study is poor (R2 = 0.04 to 0.06). We observed a positive linear correlation between the fluid temperature (FT) and Tcv, as shown in Fig. 14. This is consistent with findings reported by Huggins et al. [138], who noted a correlation between FT and the liquidus temperature of a Al2O3-SiO2-basic oxide pseudoternary system. The finding is reasonable, provided that the variability of the solid fraction needed to cause a transition from Newtonian to non-Newtonian flow is small for the synthetic slags used in this study. Further study is needed to determine whether this correlation is valid over a wider range of coal ash compositions. 4.4.2. Correlation with composition proxy variables The estimated critical viscosity temperatures based on composition proxy variables are reported in Table 8. Watt and Fereday proposed the following model [139]: Tcv ¼ 3263–1470ðS=AÞ þ 360ðS=AÞ2 –14:7ð F þ C þ MÞ þ 0:15ð F þ C þ MÞ2

ð5Þ

Table 7 Critical viscosity temperatures of synthetic coal ash slags predicted from ash cone fusion temperature data using relationships developed by Reid and Cohen, Sage and McIlroy, and Marshak and Ryzhakov. Expanded uncertainties in the predicted values are carried over from the uncertainties reported for the ash cone fusion data. Coefficients of determination (R2) between the predicted values and measured values are also reported for comparison.

Fig. 12. Select dynamic viscosity data as a function of temperature and shear rate for slag sample 6 (COALQUAL identifier: W216040); additional data may be found in the supplemental material section. Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value. The molten slag shows a transition from Newtonian to non-Newtonian flow between 1355 °C and 1346 °C during cooling, accompanied by a discontinuous increase in viscosity.

sample number

Reid and Cohen

Sage and McIlroy

Marshak and Ryzhakov

Tcv, °C

U(Tcv)

Tcv, °C

U(Tcv)

Tcv, °C

U(Tcv)

1 2 3 4 5 6 7 R2

1188 1250 1202 1235 1184 1194 1212 0.04

36 34 19 21 4 38 23

1346 1391 1360 1378 1324 1345 1362 0.06

12 16 29 13 28 28 46

1418 1406 1440 1417 1430 1390 1405 0.06

12 16 29 13 28 28 46

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

Fig. 14. Correlation between Tcv and FT. Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value. The coefficient of determination (R2) for the two variables is 0.77, indicating a strong correlation between the two over the range of compositions tested in this study.

where the sum of masses of silica (S), alumina (A), iron(III) oxide equivalent (F), calcium oxide (C), and magnesium oxide (M) is normalized to 100% (mass/mass). The iron(III) oxide equivalent was initially defined by Reid and Cohen [37]: F ¼ Fe2 O3 þ 1:11FeO þ 1:43Fe

ð6Þ

where all variables have units of mass. The coefficient of determination (R2) between estimated critical viscosity temperatures and the values measured in the present study is 0.14. We fitted a plane to the critical viscosity temperatures using the silica-to-alumina ratio (S/A) and the iron(III) oxide equivalent (F): Tcv ¼ 1900−148:3ðS=AÞ–8:04F:

ð7Þ

While a strong correlation exists between the fitted-plane and the measured Tcv values (R2 = 0.96), the model is based on a sparse data set with only 5 degrees of freedom. The finding suggests that Tcv may be modeled accurately as a function of coal ash composition, provided that relevant viscosity and temperature data are available. 4.4.3. Correlation with liquidus temperatures Song et al. related the critical viscosity temperature of coal ash slags based on their liquidus temperature (Tliq), as calculated using a CALPHAD-based software tool (FactSage) [113]: Tcv ¼ 300 þ 0:768Tliq ðin CÞ:

23

liquidus temperatures, calculated with the FactSage integrated software package (version 6.4, with the FACT pure substances and oxide databases). In our analysis, leucite was ruled out as a primary phase field constituent because it has not been observed in coal ash slags [140]. Mullite and spinel were the principal solid phases predicted to be present in subliquidus coal ash slags with the compositions in this study, and were confirmed to be present in the post-test slags. Fig. 15 plots Tcv values measured through rotary viscometry against liquidus temperatures based on the synthetic coal ash compositions in this study, and the relationship reported by Song et al. (shown in Fig. 15 as a line). Error bars for the predicted Tcv values reflect the uncertainties in the coal ash composition due to alumina dissolution, up to 5% (mass/mass). Critical viscosity temperatures from four previously published studies, as well as liquidus temperatures calculated from corresponding coal ash slag compositions, are also plotted for comparison [106,108,110,114]. The relationship between the critical viscosity temperature and the liquidus temperature reflects a solid phase volume threshold, above which the flow behavior undergoes a transition to non-Newtonian flow. Interestingly, the relationship should be generalizable to any silicate fluid undergoing crystallization in the subliquidus temperature range. Further work to identify the effect of the primary phase field, in terms of particle size and morphology, on slag rheology may yield further insights into the critical viscosity temperature for coal ash slags. 5. Conclusions Coal gasification is an important industrial process for bulk chemical production and is a key enabling technology for pre-combustion carbon capture in IGCC electrical power plants. Accurate modeling of coal ash slag viscosity as a function of temperature, including the transition from Newtonian to non-Newtonian flow, is essential to successful slag management for coal gasification. Experimental data from hightemperature rotary viscometry, combined with thermophysical modeling of coal ash slags, provide the best approach towards improving the accuracy of molten coal ash slag models. Correlation between ST and HT with Tcv was found to be poor, with R2 values between 0.04 and 0.06. We observed a positive linear correlation between FT and Tcv, with a R2 value of 0.77. We also found that it was possible to fit a plane to the Tcv data, using the silica-to-alumina ratio (S/A) and the iron(III) oxide equivalent (F) as independent variables (R2 = 0.96). Further work is needed to understand the slumping behavior of coal ash cones as a function of composition, as it may provide further insights

ð8Þ

We found a strong correlation (R2 = 0.88) between the critical viscosity temperature of the synthetic slags in this study and their Table 8 Critical viscosity temperatures of synthetic coal ash slags predicted from composition data using the Watt and Fereday model. The uncertainty due to the model was not available; the expanded uncertainties are based on the uncertainty of the coal ash compositions (0.01%, mass/mass) as propagated through the model. The coefficient of determination (R2) between the predicted values and measured values is also reported for comparison. sample number

Tcv, °C

U(Tcv)

1 2 3 4 5 6 7 R2

1485 1590 1710 1514 1495 1493 1495 0.14

59 44 135 61 60 50 29

Fig. 15. Correlation between Tcv and Tliq, where Tliq is calculated from reported coal ash composition using a CALPHAD-based software tool (FactSage, version 6.4). Error bars represent the expanded uncertainty (k = 2), and approximate 95% coverage of the true measurand value. The relationship between Tcv and Tliq published by Song et al. is shown as a solid line. Data from the present study (●), Nowok (Δ) [106], Hurst et al. (□) [108,110], and Ilyushechkin and Hla (○) [114] are plotted for comparison.

24

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

on the agglomerations of coal ash particles in a boiler or gasifier. The correlation between the fluid temperature and the critical viscosity temperature identified in this study also merits additional investigation, as the ash cone fusion test is widely employed in industry. Shear rate is an important variable in non-Newtonian flow; accuracy of critical viscosity temperature measurements is compromised if the shear rate is not investigated in a systematic manner during hightemperature rotary viscometry of coal ash slags. Further testing will be needed to elucidate the difference in viscosity values measured at constant shear and cooling rates, compared with those measured in variable shear rate and stepwise cooling experiments. Our study supports the use of CALPHAD-based methods to predict the liquidus temperature and critical viscosity temperature based on coal ash composition, given the strong correlation between the two variables (R2 = 0.88). The approach permits the estimation of the lower operating temperature limit for a given coal sample, provided that its ash composition is known. Additional work to determine the effect of the primary phase on the shape factor in an Einstein-Roscoe model may prove to be useful in improving the accuracy of the model. Disclaimer of liability or endorsement This presentation was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. Acknowledgments This technical effort was performed in support of the National Energy Technology Laboratory's Advanced Gasification Program. The authors thank Rick Krabbe for the synthesis of the slag samples used in this study, Richard Chinn for ceramographic preparation, and Kyle Rozman for the scanning electron microscopy and energy-dispersive X-ray spectroscopy. X-ray fluorescence analysis of slag composition was performed by Clark Laboratories. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.fuproc.2015.09.019. References [1] C. Higman, M. van der Burgt, Gasification, Gulf Professional Publishing, San Francisco, 2003. [2] C. Higman, S. Tam, Advances in coal gasification, hydrogenation, and gas treating for the production of chemicals and fuels, Chem. Rev. 114 (2014) 1673–1708. [3] I. Wender, Reactions of synthesis gas, Fuel Process. Technol. 48 (1996) 189–297. [4] T.F. Wall, Combustion processes for carbon capture, Proc. Combust. Inst. 31 (2007) 31–47. [5] C. Kunze, H. Spliethoff, Assessment of oxy-fuel, pre- and post-combustion-based carbon capture for future IGCC plants, Appl. Energy 94 (2012) 109–116. [6] M.M. Majoumerd, M. Assadi, Techno-economic assessment of fossil fuel power plants with CO₂ capture — results of EU H₂ -IGCC project, Int. J. Hydrog. Energy 39 (2014) 16771–16784. [7] Z.J. Zhou, Z. You, Z.H. Wang, X. Hu, J.H. Zhou, K.F. Cen, Process design and optimization of state-of-the-art carbon capture technologies, Environ. Prog. Sustainable Energy 33 (2014) 993–999.

[8] W.A. Selvig, F.H. Gibson, Analyses of Ash from United States Coals, Bulletin 567, U.S. Government Printing Office, Washington DC, 1956. [9] S.P. Schweinfurth, An introduction to coal quality, in: B.S. Pierce, K.O. Dennen (Eds.), The National Coal Resource Assessment Overview, U.S. Geological Survey, Reston, Virginia, 2009. [10] K.-S. Kwong, A. Petty, J. Bennett, R. Krabbe, H. Thomas, Wear mechanisms of chromia refractories in slagging gasifiers, Int. J. Appl. Ceram. Technol. 4 (2007) 503–513. [11] R.A. Newby, R.L. Bannister, Advanced hot gas cleaning system for coal gasification processes, J. Eng. Gas Turbines Power Trans. ASME 116 (1994) 338–344. [12] P. Wang, M. Massoudi, Slag behavior in gasifiers. Part I: Influence of coal properties and gasification conditions, Energy 6 (2013) 784–806. [13] J.H. Patterson, H.J. Hurst, Ash and slag qualities of Australian bituminous coals for use in slagging gasifiers, Fuel 79 (2000) 1671–1678. [14] M.J. Brannon, G.W. Ollis, G.L. Shoaf, Expanding the range of suitable gasification fuels, Fuel Process. Technol. 99 (2012) 1–5. [15] L. Kong, J. Bai, Z. Bai, Z. Guo, W. Li, Improvement of ash flow properties of low-rank coal for entrained flow gasifier, Fuel 120 (2014) 122–129. [16] M. Grabner, Industrial Coal Gasification Technologies Covering Baseline and HighAsh Coal, Wiley-VCH Verlag GmbH, Weinheim, Germany, 2014. [17] K.C. Mills, Estimation of Physicochemical Properties of Coal Slags and Ashes, in: Mineral Matter and Ash in Coal, American Chemical Society, Washington, DC, 1986 195–214. [18] J.A. Duffy, M.D. Ingram, I.D. Sommerville, Acid–base properties of molten oxides and metallurgical slags, J. Chem. Soc. Faraday Trans. 1 (74) (1978) 1410–1419. [19] K. Seki, F. Oeters, Viscosity measurements on liquid slags in the system CaO-FeOFe₂O₃-SiO₂, Trans Iron Steel Int. Jpn. 24 (1984) 445–454. [20] K.C. Mills, The influence of structure on the physico-chemical properties of slags, ISIJ Int. 33 (1993) 148–155. [21] K.C. Mills, S. Sridhar, Viscosities of ironmaking and steelmaking slags, Ironmak. Steelmak. 26 (1999) 262–268. [22] D. Ghosh, V.A. Krishnamurthy, S.R. Sankaranarayanan, Application of optical basicity to viscosity of high alumina blast furnace slags, J. Miner. Metall. B 46 (2010) 41–49. [23] K. Schumacher, J. White, J. Downey, Viscosities in the calcium–silicate slag system in the range of 1798 K to 1973 K (1525 °C to 1700 °C), Metall. Mater. Trans. B Process Metall. Mater. Process. Sci. (2014) 1–6. [24] R. Gill, Chemical Fundamentals of Geology and Environmental Geoscience, WileyBlackwell, Hoboken, NJ, 2015. [25] J.A. Duffy, M.D. Ingram, Establishment of an optical scale for Lewis basicity in inorganic oxyacids, molten salts, and glasses, J. Am. Chem. Soc. 93 (1971) 6448–6454. [26] J.A. Duffy, A review of optical basicity and its applications to oxidic systems, Geochim. Cosmochim. Acta 57 (1993) 3961–3970. [27] S. Vargas, F.J. Frandsen, K. Dam-Johansen, Rheological properties of hightemperature melts of coal ashes and other silicates, Prog. Energy Combust. 27 (2001) 237–429. [28] G.J. Browning, G.W. Bryant, H.J. Hurst, J.A. Lucas, T.F. Wall, An empirical method for the prediction of coal ash slag viscosity, Energy Fuel 17 (2003) 731–737. [29] J.W. Nowok, J.P. Hurley, D.C. Stanley, Local-structure of a lignitic coal ash slag and its effect on viscosity, Energy Fuel 7 (1993) 1135–1140. [30] B.C. Folkedahl, H.H. Schobert, Effects of atmosphere on viscosity of selected bituminous and low-rank coal ash slags, Energy Fuel 19 (2005) 208–215. [31] B.R. Stanmore, S. Budd, Measuring the viscous flow of molten coal ash, Fuel 75 (1996) 1476–1479. [32] B.J.P. Buhre, G.J. Browning, R.P. Gupta, T.F. Wall, Measurement of the viscosity of coal-derived slag using thermomechanical analysis, Energy Fuel 19 (2005) 1078–1083. [33] J.P. Bennett, K.-S. Kwong, Failure mechanisms in high chrome oxide gasifier refractories, Metall. Mater. Trans. A 42A (2011) 888–904. [34] M. Massoudi, P. Wang, A Brief Review of Viscosity Models for Slag in Coal Gasification, DOE/NETL-2012/1533, National Energy Technology Laboratory, Pittsburgh, PA, 2011. [35] E.K. Johnson, A non-Newtonian flow model for coal-ash slag, J. Eng. Gas Turbine Power Trans. ASME 106 (1984) 777–781. [36] R. Marshak, Determination of coal-ash properties, in: J.G. Singer (Ed.), Combustion, Fossil Power Systems: A Reference Book on Fuel Burning and Steam Generation, Combustion Engineering, Windsor, CT 1991, pp. C1–C18. [37] W.T. Reid, P. Cohen, The flow characteristics of coal-ash slags in the solidification range, Trans. ASME 66 (1944) 83–97. [38] J.D. Watt, F. Fereday, Flow properties of slags formed from ashes of British coals. 2. The crystallizing behaviour of the slags, J. Int. Fuel 42 (1969) 131–134. [39] A.R. McBirney, T. Murase, Rheological properties of magmas, Annu. Rev. Earth Planet. Sci. 12 (1984) 337–357. [40] H. Pinkerton, R.J. Stevenson, Methods of determining the rheological properties of magmas at sub-liquidus temperatures, J. Volcanol. Geotherm. Res. 53 (1992) 47–66. [41] D.J. Stein, F.J. Spera, Rheology and microstructure of magmatic emulsions — theory and experiments, J. Volcanol. Geotherm. Res. 49 (1992) 157–174. [42] H. Pinkerton, G. Norton, Rheological properties of basaltic lavas at sub-liquidus temperatures: laboratory and field measurements on lavas from Mount Etna, J. Volcanol. Geotherm. Res. 68 (1995) 307–323. [43] R.J. Stevenson, D.B. Dingwell, S.L. Webb, T.G. Sharp, Viscosity of microlitebearing rhyolitic obsidians: an experimental study, B, Volcanology 58 (1996) 298–309. [44] J.L. Vigneresse, P. Barbey, M. Cuney, Rheological transitions during partial melting and crystallization with application to felsic magma segregation and transfer, J. Petrol. 37 (1996) 1579–1600.

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26 [45] H. Sato, Viscosity measurement of subliquidus magmas: 1707 basalt of Fuji volcano, J. Mineral. Petrol. Sci. 100 (2005) 133–142. [46] H. Ishibashi, H. Sato, Viscosity measurements of subliquidus magmas: alkali olivine basalt from the Higashi-Matsuura district, Southwest Japan, J. Volcanol. Geotherm. Res. 160 (2007) 223–238. [47] H. Ishibashi, Non-Newtonian behavior of plagioclase-bearing basaltic magma: subliquidus viscosity measurement of the 1707 basalt of Fuji volcano, Japan, J. Volcanol. Geotherm. Res. 181 (2009) 78–88. [48] D. Picard, L. Arbaret, M. Pichavant, R. Champallier, P. Launeau, The rheological transition in plagioclase-bearing magmas, J. Geophys. Res. Solid Earth 118 (2013) 1363–1377. [49] F. Vetere, H. Sato, H. Ishibashi, R. De Rosa, P. Donato, Viscosity changes during crystallization of a shoshonitic magma: new insights on lava flow emplacement, J. Mineral. Petrol. Sci. 108 (2013) 144–160. [50] S.-H. Seok, S.-M. Jung, Y.-S. Lee, D.-J. Min, Viscosity of highly basic slags, ISIJ Int. 47 (2007) 1090–1096. [51] S. Mueller, E.W. Llewellin, H.M. Mader, The Effect of Particle Shape on Suspension Viscosity and Implications for Magmatic Flows, Geophys. Res. Lett. 38 (2011). [52] W. Song, L. Tang, Z. Zhu, Y. Ninomiya, Rheological evolution and crystallization response of molten coal ash slag at high temperatures, AICHE J. 59 (2013) 2726–2742. [53] D. Green, A Handbook of Pottery Glazes, Watson-Guptill Publications, New York, 1979. [54] A.C. Fieldner, A.E. Hall, A.L. Feild, The Fusibility of Coal ash and the Determination of the Softening Temperature, Bulletin 129, U.S. Government Printing Office, Washington DC, 1918. [55] R.C. Corey, Measurement and Significance of the Flow Properties of Coal-ash Slag, Bulletin 618, U.S. Government Printing Office, Washington DC, 1964. [56] D1857/D1857M-04 standard test method for fusibility of coal and coke ash, Annual Book of ASTM Standards, ASTM International, West Conshohocken, PA, 2010. [57] E. Stansfield, W.A. Lang, Coals of Alberta: their occurrence, analysis and utilization, report no. 35, Research Council of Alberta, A. Shnitka (King's Printer), Edmonton, Alberta, 1944. [58] D.N. Bowden, H.S. Roberts, Analyses and fusion characteristics of New Zealand coal ashes, N. Z. J. Sci. 18 (1975) 119–130. [59] D.H.H. Quon, S.S.B. Wang, T.T. Chen, Viscosity measurements of slags from pulverized western Canadian coals in a pilot-scale research boiler, Fuel 63 (1984) 939–942. [60] V.R. Gray, Prediction of ash fusion temperature from ash composition for some New Zealand coals, Fuel 66 (1987) 1230–1239. [61] A.K. Mukherjee, M.M. Alam, S.K. Mazumdar, R. Haque, S. Gowrisankaran, Physicochemical properties and petrographic characteristics of the Kapurdi lignite deposit, Barmer Basin, Rajastan, India, Int. J. Coal Geol. 21 (1992) 31–44. [62] K.S. Vorres, Users Handbook for the Argonne Premium Coal Sample Program, ANL/PCSP–93/1, Argonne National Laboratory, Argonne, Illinois, 1993. [63] S.V. Vassilev, K. Kitano, S. Takeda, T. Tsurue, Influence of mineral and chemical composition of coal ashes on their fusibility, Fuel Process. Technol. 45 (1995) 27–51. [64] L.J. Bragg, J.K. Oman, S.J. Tewalt, C.L. Oman, N.H. Rega, P.M. Washington, R.B. Finkelman, U.S. Geological Survey coal quality (COALQUAL) database; version 2.0, Analytical data, sample locations, and descriptive information, analytical methods and sampling techniques, database pespective, and bibliographic references for selected U.S. coal samples, U.S. Geological Survey, Denver, Colorado, 1998. [65] A.M.H. Van der Veen, A. Broos, Preparation and Characterisation of Coal Samples and Maceral Concentrates for Studies on Gasification and Combustion Reactivity of Coals in Combined Cycle Processes : Final Report, European Commission, Directorate-General for Energy, Luxembourg, 1998. [66] J.R. Qiu, F. Li, Y. Zheng, C.G. Zheng, H.C. Zhou, The influences of mineral behaviour on blended coal ash fusion characteristics, Fuel 78 (1999) 963–969. [67] G.W. Bryant, G.J. Browning, H. Emanuel, S.K. Gupta, R.P. Gupta, J.A. Lucas, T.F. Wall, The fusibility of blended coal ash, Energy Fuel 14 (2000) 316–325. [68] S.L. Weeber, B. Cairncross, R.M.S. Falcon, Mineralogical, petrographic and geological controls on coal ash fusion temperature from New Clydesdale Colliery, Witbank Coalfield, South Africa, J. S. Afr. Int. Miner. Metall. 100 (2000) 181–190. [69] A. Alastuey, A. Jimenez, F. Plana, X. Querol, I. Suarez-Ruiz, Geochemistry, mineralogy, and technological properties of the main Stephanian (Carboniferous) coal seams from the Puertollano Basin, Spain, Int. J. Coal Geol. 45 (2001) 247–265. [70] The BCURA coal sample bank: a users handbook, Environmental Scientifics Group Ltd., (www.bcura.org/coalbank.html, accessed on 6/4/2015). [71] S.A. Lolja, H. Haxhi, R. Dhimitri, S. Drushku, A. Malja, Correlation between ash fusion temperatures and chemical composition in Albanian coal ashes, Fuel 81 (2002) 2257–2261. [72] C.G. Vassileva, S.V. Vassilev, Behaviour of inorganic matter during heating of Bulgarian coals - 1, Lignites Fuel Process. Technol. 86 (2005) 1297–1333. [73] C.G. Vassileva, S.V. Vassilev, Behaviour of inorganic matter during heating of Bulgarian coals — 2. Subbituminous and bituminous coals, Fuel Process. Technol. 87 (2006) 1095–1116. [74] P.C. Hackley, M. Martinez, Organic petrology of Paleocene Marcelina Formation coalsPaso Diablo mine, western Venezuela: tectonic controls on coal type, Int. J. Coal Geol. 71 (2007) 505–526. [75] G. Akar, V. Arslan, M.E. Ertem, U. Ipekoglu, Relationship between ash fusion temperatures and coal mineral matter in some Turkish coal ashes, Asian J. Chem. 21 (2009) 2105–2109. [76] J.C. van Dyk, S.A. Benson, M.L. Laumb, B. Waanders, Coal and coal ash characteristics to understand mineral transformations and slag formation, Fuel 88 (2009) 1057–1063.

25

[77] X. Wu, Z. Zhang, G. Piao, X. He, Y. Chen, N. Kobayashi, S. Mori, Y. Itaya, Behavior of mineral matters in Chinese coal ash melting during char-CO₂/H₂O gasification reaction, Energy Fuel 23 (2009) 2420–2428. [78] A.Y. Maystrenko, A.I. Topal, A.V. Kritskiy, V.P. Patskov, I.V. Bestsennyy, The effect of physical and chemical properties of ash on the parameters of liquid slag removal operation for Donetsk coal, Tech. Therm. Phys. Ind. Power Syst. 1 (2009) 129–148. [79] M. Mladenovic, R. Mladenovic, V. Manovic, M. Radovanovic, D. Stojiljkovic, Criteria selection for the assessment of Serbian lignites tendency to form deposits on power boilers heat transfer surfaces, Therm. Sci. 13 (2009) 61–78. [80] S.J. Tewalt, H.E. Belkin, J.R. SanFilipo, M.D. Merrill, C.A. Palmer, P.D. Warwick, A.W. Karlsen, R.B. Finkelman, A.J. Park, Chemical analyses in the world coal quality inventory; version 1, World Coal Quality Inventory (WoCQI), U.S. Geological Survey, Reston, Virginia, 2010. [81] H. Wang, P. Qiu, X. Shi, J. Zhang, Y. Chen, S. Wu, Effect of key minerals on the ash melting behavior in a reducing atmosphere, Energy Fuel 25 (2011) 3446–3455. [82] X. Wu, T. Zhou, Y. Chen, Z. Zhang, G. Piao, N. Kobayashi, S. Mori, Y. Itaya, Mineral melting behavior of Chinese blended coal ash under gasification condition, Asia Pac. J. Chem. Eng. 6 (2011) 220–230. [83] D. Vamvuka, E. Mistakidou, S. Drakonaki, A. Foscolos, K. Kavouridis, Ash quality of a beneficiated lignite from ptolemais basin, northern Greece, Energy Fuel 15 (2001) 1181–1185. [84] F.-h. Li, J.-j. Huang, Y.-t. Fang, Y. Wang, The effects of leaching and floatation on the ash fusion temperatures of three selected lignites, Fuel 90 (2011) 2377–2383. [85] H. Yuan, Q. Liang, X. Gong, Crystallization of coal ash slags at high temperatures and effects on the viscosity, Energy Fuel 26 (2012) 3717–3722. [86] B. Liu, Q. He, Z. Jiang, R. Xu, B. Hu, Relationship between coal ash composition and ash fusion temperatures, Fuel 105 (2013) 293–300. [87] S. Naveed, A. Malik, M. Akram, M. Deary, A.P. Roskilly, Gasification Perspective of Pakistani Coal, J. Energy Inst. 86 (2013). [88] Z. Huang, Y. Li, D. Lu, Z. Zhou, Z. Wang, J. Zhou, K. Cen, Improvement of the coal ash slagging tendency by coal washing and additive blending with mullite generation, Energy Fuel 27 (2013) 2049–2056. [89] S.C. Chelgani, S. Makaremi, Explaining the relationship between common coal analyses and Afghan coal parameters using statistical modeling methods, Fuel Process. Technol. 110 (2013) 79–85. [90] N. Jing, Q. Wang, L. Cheng, Z. Luo, K. Cen, D. Zhang, Effect of temperature and pressure on the mineralogical and fusion characteristics of Jincheng coal ash in simulated combustion and gasification environments, Fuel 104 (2013) 647–655. [91] M.A. Duchesne, A.D. Hall, R.W. Hughes, D.J. McCalden, E.J. Anthony, A. Macchi, Fate of inorganic matter in entrained-flow slagging gasifiers: fuel characterization, Fuel Process. Technol. 118 (2014) 208–217. [92] S. Chakravarty, A. Mohanty, A. Banerjee, R. Tripathy, G.K. Mandal, M.R. Basariya, M. Sharma, Composition, mineral matter characteristics and ash fusion behavior of some Indian coals, Fuel 150 (2015) 96–101. [93] C.a. Wang, Y. Liu, X. Zhang, D. Che, A study on coal properties and combustion characteristics of blended coals in northwestern China, Energy Fuel 25 (2011) 3634–3645. [94] M. Čarnogurská, M. Příhoda, R. Pyszko, Ľ. Širillová, J. Palkóci, The influence of calcite on the ash flow temperature for semi-anthracite coal from Donbas district, Chem. Process. Eng. Inz. 35 (2014) 515–525. [95] W.L. Sage, J.B. McIlroy, Relationship of coal-ash viscosity to chemical composition, J. Eng. Gas Turbine Power Trans. ASME 82 (1960) 145–153. [96] M. Seggiani, Empirical correlations of the ash fusion temperatures and temperature of critical viscosity for coal and biomass ashes, Fuel 78 (1999) 1121–1125. [97] L.A. Hansen, F.J. Frandsen, K. Dam-Johansen, H.S. Sorensen, Quantification of fusion in ashes from solid fuel combustion, Thermochim. Acta 326 (1999) 105–117. [98] F.M. White, Viscous Fluid Flow, 2nd ed. McGraw-Hill, New York, 1991. [99] P. Dontula, C.W. Macosko, L.E. Scriven, Origins of concentric cylinders viscometry, J. Rheol. 49 (2005) 807–818. [100] A.L. Feild, The viscosity of blast-furnace slag and its relation to iron metallurgy, including a description of a new method of measuring slag viscosity at high temperatures, Trans. Faraday Soc 13 (1917) 3–35. [101] M. Margules, On the determination of the coefficients of friction and of gliding in the plane motions of a fluid, Sitzsungsber. Akad. Wiss. Wien. 83 (1881) 588–602. [102] P. Nicholls, W.T. Reid, Viscosity of coal ash slags, Trans. ASME 62 (1940) 141–153. [103] K.S. Vorres, S. Greenberg, Viscosity of Synthetic Coal Ashes, in: Symposium on Chemistry of Mineral Matter and Ash in Coal, American Chemical Society, Philadelphia, PA, 1984. [104] H.H. Schobert, R.C. Streeter, E.K. Diehl, Flow properties of low-rank coal ash slags implications for slagging gasification, Fuel 64 (1985) 1611–1617. [105] D. Cho, I. Moon, S. Whang, M. Oh, High temperature slag viscometry, J. Ind. Eng. Chem. 7 (2001) 30–37. [106] J.W. Nowok, Viscosity and phase-transformation in coal ash slags near and below the temperature of critical viscosity, Energy Fuel 8 (1994) 1324–1336. [107] M.S. Oh, D.D. Brooker, E.F. Depaz, J.J. Brady, T.R. Decker, Effect of crystalline phaseformation on coal slag viscosity, Fuel Process. Technol. 44 (1995) 191–199. [108] H.J. Hurst, F. Novak, J.H. Patterson, Viscosity measurements and empirical predictions for some model gasifier slags, Fuel 78 (1999) 439–444. [109] H.J. Hurst, F. Novak, J.H. Patterson, Viscosity measurements and empirical predictions for fluxed Australian bituminous coal ashes, Fuel 78 (1999) 1831–1840. [110] H.J. Hurst, J.H. Patterson, A. Quintanar, Viscosity measurements and empirical predictions for some model gasifier slags - II, Fuel 79 (2000) 1797–1799. [111] W. Song, L. Tang, X. Zhu, Y. Wu, Y. Rong, Z. Zhu, S. Koyama, Fusibility and flow properties of coal ash and slag, Fuel 88 (2009) 297–304.

26

P.Y. Hsieh et al. / Fuel Processing Technology 142 (2016) 13–26

[112] W. Song, L. Tang, X. Zhu, Y. Wu, Z. Zhu, S. Koyama, Flow properties and rheology of slag from coal gasification, Fuel 89 (2010) 1709–1715. [113] W. Song, Y. Dong, Y. Wu, Z. Zhu, Prediction of temperature of critical viscosity for coal ash slag, AICHE J. 57 (2011) 2921–2925. [114] A.Y. Ilyushechkin, S.S. Hla, Viscosity of high-iron slags from Australian coals, Energy Fuel 27 (2013) 3736–3742. [115] A. Vona, C. Romano, D.B. Dingwell, D. Giordano, The rheology of crystal-bearing basaltic magmas from Stromboli and Etna, Geochim. Cosmochim. Acta 75 (2011) 3214–3236. [116] M.O. Chevrel, C. Cimarelli, L. deBiasi, J.B. Hanson, Y. Lavallée, F. Arzilli, D.B. Dingwell, Viscosity measurements of crystallizing andesite from Tungurahua volcano (Ecuador), Geochem. Geophys. Geosyst. 16 (2015) 870–889. [117] B.J. Jung, H.H. Schobert, Improved prediction of coal ash slag viscosity by thermodynamic modeling of liquid-phase composition, Energy Fuel 6 (1992) 387–398. [118] P.J. Spencer, A brief history of CALPHAD, Calphad 32 (2008) 1–8. [119] C. Bale, P. Chartrand, S.A. Degterov, G. Eriksson, K. Hack, R. Ben Mahfoud, J. Melancon, A.D. Pelton, S. Petersen, FactSage thermochemical software and databases, Calphad 26 (2002) 189–228. [120] C.W. Bale, E. Belisle, P. Chartrand, S.A. Decterov, G. Eriksson, K. Hack, I.H. Jung, Y.B. Kang, J. Melancon, A.D. Pelton, C. Robelin, S. Petersen, FactSage thermochemical software and databases — recent developments, Calphad 33 (2009) 295–311. [121] E. Jak, Prediction of coal ash fusion temperatures with the F*A*C*T thermodynamic computer package, Fuel 81 (2002) 1655–1668. [122] J.C. van Dyk, F.B. Waanders, S.A. Benson, M.L. Laumb, K. Hack, Viscosity predictions of the slag composition of gasified coal, utilizing FactSage equilibrium modelling, Fuel 88 (2009) 67–74. [123] A.Y. Ilyushechkin, S.S. Hla, D.G. Roberts, N.N. Kinaev, The effect of solids and phase compositions on viscosity behaviour and TCV of slags from Australian bituminous coals, J. Non-Cryst. Solids 357 (2011) 893–902. [124] J. Zhu, J. Nakano, T.K. Kaneko, H. Mu, J.P. Bennett, K.-S. Kwong, P. Rozelle, S. Sridhar, Viscosity determination of molten ash from low-grade US coals, High Temp. Mater. Process. Isr. 31 (2012) 569–580. [125] M.A. Duchesne, A.Y. Ilyushechkin, R.W. Hughes, D.Y. Lu, D.J. McCalden, A. Macchi, E.J. Anthony, Flow behaviour of slags from coal and petroleum coke blends, Fuel 97 (2012) 321–328. [126] L. Kong, J. Bai, Z. Bai, Z. Guo, W. Li, Effects of CaCO₃ on slag flow properties at high temperatures, Fuel 109 (2013) 76–85.

[127] C. Goñi, S. Helle, X. Garcia, A. Gordon, R. Parra, U. Kelm, R. Jimenez, G. Alfaro, Coal blend combustion: fusibility ranking from mineral matter composition, Fuel 82 (2003) 2087–2095. [128] D. Thompson, Predicted mineral melt formation by BCURA Coal Sample Bank coals: variation with atmosphere and comparison with reported ash fusion test data, Fuel 89 (2010) 2062–2071. [129] W.S. Rasband, Image J, Image Processing and Analysis in Java, U. S. National Institutes of Health, Bethesda, Maryland, USA, 2014. [130] D2196-15 Standard Test Methods for Rheological Properties of Non-Newtonian Materials by Rotational (Brookfield type) Viscometer, Annual Book of ASTM Standards, ASTM International, West Conshohocken, PA, 2015. [131] S.A. Decterov, A.N. Grundy, I.H. Jung, A.D. Pelton, Modeling the Viscosity of Aluminosilicate Melts, Computation in Modern Science and Engineering, AIP, Corfu, Greece 2007, pp. 404–407. [132] W. Song, Y. Sun, Y. Wu, Z. Zhu, S. Koyama, Measurement and simulation of flow properties of coal ash slag in coal gasification, AICHE J. 57 (2011) 801–818. [133] J. Zhu, T.K. Kaneko, H. Mu, J.P. Bennett, S. Sridhar, Effects of measurement materials and oxygen partial pressure on the viscosity of synthetic eastern and western United States coal slags, Energy Fuel 26 (2012) 4465–4474. [134] Y. Kim, M.S. Oh, Effect of cooling rate and alumina dissolution on the determination of temperature of critical viscosity of molten slag, Fuel Process. Technol. 91 (2010) 853–858. [135] K.C. Mills, J.M. Rhine, The measurement and estimation of the physical properties of slags formed during coal gasification: 1. Properties relevant to fluid flow, Fuel 68 (1989) 193–200. [136] D. French, H.J. Hurst, P. Marvig, Comments on the use of molybdenum components for slag viscosity measurements, Fuel Process. Technol. 72 (2001) 215–225. [137] R.P. Chhabra, J.F. Richardson, Non-Newtonian Flow and Applied Rheology: Engineering Applications, Butterworth-Heinemann, Elsevier, Amsterdam [etc.], 2008. [138] F.E. Huggins, D.A. Kosmack, G.P. Huffman, Correlation between ash-fusion temperatures and ternary equilibrium phase diagrams, Fuel 60 (1981) 577–584. [139] J.D. Watt, F. Fereday, Flow properties of slags formed from ashes of British coals. 1. Viscosity of homogeneous liquid slags in relation to slag composition, J. Int. Fuel 42 (1969) 99–103. [140] W. Park, M.S. Oh, Slagging of petroleum coke ash using Korean anthracites, J. Ind. Eng. Chem. 14 (2008) 350–356.