Role of condensed phases in the agglomeration of low rank coal ash in fluidized beds

Fuel 232 (2018) 1–11

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Role of condensed phases in the agglomeration of low rank coal ash in fluidized beds Javier Pallarésa, Aditi B. Khadilkarb, Brant Simcock-Baileyb, Sarma V. Pisupatib,

T

⁎

Universidad de Zaragoza-Research Centre for Energy Resources and Consumption (CIRCE) – Campus Río Ebro, Mariano Esquillor Gómez, 15, 50018 Zaragoza, Spain John and Willie Leone Family Department of Energy and Mineral Engineering and The EMS Energy Institute, The Pennsylvania State University, University Park, PA 16802, United States

a

b

A R T I C LE I N FO

A B S T R A C T

Keywords: Agglomeration Low rank coals Biomass Combustion Fluidized bed reactors FactSageTM

Fuel diversity is critical in the utilization of fluidized bed combustors. To support the use of diverse fuels including low rank coals and biomass, interactions of chemical components that lead to operational issues such as agglomeration need to be studied. Low rank coals and biomass are generally rich in alkali and alkaline earth metals and their role in the initiation of agglomeration is investigated in this work. A novel modeling methodology to quantify the contribution of these components to the initiation of agglomeration via the condensation of gaseous species was developed. This condensation modeling methodology complements thermodynamic simulations such as those using FactSage and agglomeration models used for high rank coals. The study shows that at temperatures below 800 °C condensed phases play a critical role in the initiation of agglomerate growth. The condensation temperature is computed to be about 30–50 °C lower than the predicted molten slag-liquid formation temperature (740 °C for the lignite and 810 °C for the subbituminous coals studied). The extent of condensation correlates to the amount of alkali metals in the ash. Sodium sulfate forms a major component of condensates for low rank coals with condensation occurring over a larger temperature range for lignite than the sub-bituminous coals studied.

1. Introduction In the last decade, the search for alternative and economically attractive solutions in the combustion of traditional fossil fuels (anthracites, bituminous and subbituminous coals, natural gas and oil) in thermal or power production energy systems has promoted research interest and development of new technologies that allow the energy valorization of low rank coals, biomass and other waste fuels (culm, plastic residues, municipal solid waste, etc.). Within existing combustion technologies, fluidized beds have aroused high interest due to the numerous advantages that they show. In spite of the maturity of fluidized bed technologies in coal combustion processes, the heterogeneity and variability in fuel and their availability often leads to changes in the fuel supply including fuel blends. This makes it necessary to continue studying in depth the fundamental aspects of these technologies, especially with respect to understanding the effect of these new fuel types on operating issues. One of the issues in the operation of fluidized bed is that of ash agglomeration in the reactor. Agglomeration occurs when the particles that form the dense phase of the bed start to stick to each other giving rise to an increase in particle size, and thus modifying the fluid ⁎

Corresponding author. E-mail address: [email protected] (S.V. Pisupati).

https://doi.org/10.1016/j.fuel.2018.05.098 Received 6 February 2018; Received in revised form 17 May 2018; Accepted 19 May 2018 0016-2361/ © 2018 Elsevier Ltd. All rights reserved.

dynamics characteristics in the bed and consequently affecting the heat transfer and chemical reactions inside the reactor: uneven temperature distribution, lower conversion efficiency and ultimately abrupt defluidization. Agglomeration typically occurs in the dense phase of a fluidized bed where, due to the mixing of solids promoted by the fluidizing agent, solid particles collide against each other. During the collision, the impact force due to the kinetic energy of the particles is dissipated in the form of viscous dissipative forces, cohesive forces, capillarity forces and superficial tension forces. If the sum of the impact forces and the elastic repulsive forces is lower than the sum of dissipative forces, then the particles remain stuck after the collision, initiating the agglomeration process [1]. A detailed discussion of initiation of agglomerate growth in fluidized beds using high rank coals has been made by Khadilkar et al. [2,3]. The model proposed in these earlier studies was also validated with experiments involving agglomerates obtained from fluidized beds [4]. In addition to high rank coals, numerous industrial and laboratory-scale issues have been repeatedly reported in the literature during the utilization of low rank coals as well [5–8]. JEA’s circulating fluidized bed boilers clearly demonstrated the industrial relavance of agglomeration forcing frequent shutdowns resulting form alkali-rich fuels [9]. Additionally, an U.S. DOE study

Fuel 232 (2018) 1–11

J. Pallarés et al.

showed agglomeration in 5 out of 13 industrial fluidized bed boilers operating with low rank coals (North Dakota lignites and Montana subbituminous coals) [10]. Goblirsch et al. [12] presented experimental work conducted in four fluidized bed facilities of different sizes (nominal fuel rate: 15 lb/hr, 110 lb/hr, 500 lb/hr, and 800 lb/hr) with four North Dakota lignites (including Beulah-Zap), six Montana and Wyoming subbituminous coals (including Wyodak), and one Utah bituminous coal. The post-mortem analysis of the agglomerates formed in these reactors showed that the condensation of sodium and sulfur as sodium sufate caused the increased sodium levels in the bed material and agglomeration [11,12]. The works of Manzoori et al., Vuthaluru et al. and Bhattacarya et al. [13–15] elaborated on similar problems and analyzed the role of inorganic matter in the formation of agglomerates from South Australian brown coals (Lochiel and Bowmans) with high alkali content in fluidized bed combustion and gasification processes. They also referenced that the melt-forming inorganic elements and mineral inclusions in the ash have a pronounced effect on agglomeration and defluidization tendencies and that the release of compounds (mainly sulfates) may cause growth of particles, poor fluidization, and bed agglomeration. Given the importance of different fuel types discussed above and the agglomeration issues seen with low rank coals, this study attempts to extend the agglomeration model developed by Penn State for high rank coals to the fuel chemistry of low rank coals. Our modeling approach in the current study, is a step towards quantification of the alkali, alkalineearth metals from fuel and resulting fluidized bed condensates that have been observed to be the culprits causing agglomeration when using low rank coals.

pressure. For this reason, condensation problems over heat exchange surfaces usually occur when the gases leave the reaction zone (typically in the convective zone of industrial boilers), since the gas temperature decreases, thus decreasing the saturation pressure. Inside the dense phase of fluidized bed reactors, the presence of an enormous effective surface for deposition (all the particles inside the bed and the presence of internals), creates the possibility of the condensation of species inside the bed due to temperature differences between burning char particles and inert bed particles at bed temperatures. There are two main routes for the condensation of gaseous species: homogeneous nucleation (or aerosols formation) and heterogeneous condensation over other particles or surfaces. Homogeneous nucleation takes place when the gas nucleates to form a new phase, solid or liquid, without the presence of any surface. In this case, the gas molecules converge to form a nucleus, which then forms a particle. Homogeneous nucleation is expected to occur in systems when the saturation ratio, defined as the ratio of the partial pressure of the condensing gas to the vapor pressure of the condensing gas, is much higher than unity. McNallan et al. [21] summarize that the occurrence of homogeneous nucleation depends on the cooling rate of the combustion products, which determines the amount of condensable vapor species in the gas phase, and number and size of existing sites available for condensation, concluding that if surface areas of particles are already present in the system (for example in a fluidized bed), then it is more likely that heterogeneous condensation will occur. Although thermodynamic equilibrium models are capable of predicting the formation of melt phase by chemical reactions of solid mineral matter in fuel and bed material, their capability in the prediction of condensation is limited to the Scheil-Gulliver cooling model calculations of the condensation of gaseous species. In this work, a condensation model that allows determining the heterogeneous condensation rate of inorganic gaseous species on the bed particles was developed. This enables a study of the role of the condensation mechanism in the initiation and propagation of agglomeration. The implementation of this sub-model in an integrated ash agglomeration model will allow modeling the change in the average diameter of particles with time in the reactor as a result of both agglomeration mechanisms: coating-induced agglomeration and condensation. A model to predict agglomerate growth in high rank coals was presented earlier by Khadilkar et al. [22]. Moreover, the chemistry of low rank coals makes extension of the model critical in the understanding and prediction of fluidized bed agglomeration for these fuel types. Therefore, specific objectives of this study are:

2. Background The difference in the agglomeration behavior of low-rank coals from bituminous coals and anthracite comes from the composition of the inorganic matter of the fuel. Low rank coals have higher amounts of alkali metals (Na, K), alkaline earth metals (Ca, Mg), as well as Si, Cl, S, P, Fe, that can interact with the bed material (generally silica) forming low melting compounds such as alkali-silicates. Deposition of liquid phase on the surface of the bed particles can occur by melting, condensation or chemical reaction [16,17]. These compounds have adhesive properties thus initiating agglomeration. A discussion of differences in mechanisms due to differences in chemical composition based on a review of the literature has been presented by Khadilkar et al. [18]. In the case of low rank coals these chemical interactions become more critical considering the high content of alkali, alkali-earth metals and sulfur of these fuels and thereby warrant a detailed study, especially under the lower temperatures at which fluidized bed combustors operate. Agglomeration phenomena may be initiated by two mechanisms. The first mechanism is the melting of fuel mineral matter and this molten phase acts as a glue to adhere to other particles in the bed. This is a non-reactive mechanism, wherein some inorganic compounds melt at the bed operating temperature. During char combustion phase, the surface of the particle exceeds the bulk temperature of the bed by about 100–200 °C due to exothermic reactions depending on the particle size and oxygen concentration at the surface [19]. In the second mechanism, vaporized inorganic species (NaOH, KOH, NaO, KO, Na2SO4, K2SO4) from the fuel particles condense over cooler bed particles covering their surface and leading to reactions that form low melting point eutectics, which are adhesive. In parallel, these vaporized species may also directly react with the particle surface leading to the same phenomena [20]. Condensation of gaseous species downstream onto heat exchangers and in the loop seal region has led to industrial deposition issues. The condensation of inorganic species from the gaseous phase onto cooler surfaces,/particles, takes place when the vaporized compounds are oversaturated i.e. their partial pressure is greater than the vapor

1) To incorporate the effect of the condensation of gaseous species in the prediction of slag-liquid formation in fluidized bed combustors 2) To use FactSage thermodynamic simulations with condensation modeling to quantify the occurrence of condensation during low rank coal ash agglomeration 3) To identify the components that initiate ash agglomerate growth in low rank coals 3. Materials and methodology Two low rank coals, Wyodak-Anderson and Beulah-Zap (sub-bituminous and lignite respectively according to ASTM coal classification standards) from the Northern Great Plains province in USA have been selected for the study. The coal samples for this study were obtained from the Penn State Coal Sample Bank (samples APCS-8 – Beulah-Zap and APCS-2 – Wyodak-Anderson). These coals were subjected to chemical fractionation in order to obtain fractions rich in alkali and alkaline earth metals, for this study. The chemical composition of ash (500 °C) was obtained using ICP-AES. FactSageTM version 5.2 was used to determine the slag-liquid formation and the quantity and mass composition of final stable compounds under equilibrium, thereby identifying the components that initiate agglomerate growth in low 2

Fuel 232 (2018) 1–11

J. Pallarés et al.

Table 1 Proximate, ultimate, and ash composition of coals used in this study (% wt.) Coals:

Subbituminous (WyodakAnderson)

Lignite (BeulahZap)

Moisture C* H* N* S* O (difference)* Ash* Volatile matter* Fixed carbon (by difference)* High Heating Value (kJ/kg) SiO2 Al2O3 Fe2O3 CaO MgO Na2O K2O SO3 TiO2

27.42 69.21 4.70 0.90 0.30 18.70 6.20 43.61 50.19

36.08 61.87 4.29 0.99 0.99 16.44 15.43 41.49 43.09

28,202 22.57 12.61 4.53 22.45 4.73 1.91 0.42 11.70 1.17

24,079 22.19 8.71 7.64 21.11 6.65 6.86 0.20 16.27 0.58

* Dry basis.

cases the reduced phases of the Fe can give rise to several eutectics with low melting points hence promoting agglomeration. The local atmosphere in which the reactions takes place and Fe2O3 ratio are important to evaluate the effect that the presence of Fe has on the fuel ash. As a preliminary indication of the fuel composition influence, different indexes (Alkali index, Bed agglomeration Index (BAI), Base to acid ratio (Rb/a)) were determined to forecast agglomeration problems and their severity [25,26]. Since these coals contain large amounts of alkali-alkaline earth metal components, characterization of the studied coals included chemical fractionation analysis [27], based on the varying solubility of the elements as a result of their occurrence in the ash. These were separated and the effect of these components on slag formation tendencies was studied. Chemical fractionation of fuels ashed at 500 °C, was performed conducting three consecutive separations using deionized water, ammonium acetate and hydrochloric acid as follows [28]. In the first step, 4 ml of deionized water for every gram of sample was used and was constantly stirred for 24 h on a magnetic stir plate to remove elements that are in a water-soluble form. After the 24 h, the mixture was vacuum filtered to obtain a solid residue and a liquid extract in deionized water. The remainder of the solid residue was subjected to a second leaching using 4 ml of 1 M ammonium acetate solution for every gram of sample to remove ion-exchangeable elements that are bound loosely to organic matter such as potassium, calcium, sulfur and sodium. It was sealed, heated at 70 °C and constantly stirred for 24 h, and finally the mixture was vacuum filtered to separate the solid residue and the liquid extract in ammonium acetate. Following the same procedure, the final step of leaching used 4 ml of 1 M hydrochloric acid to remove elements that exist as acid-soluble salts such as carbonates, sulfates, and simple oxides. Each leaching results in a liquid and solid residue sample, which were analyzed for the following major and minor elements: Al, Ba, Ca, Fe, K, Mn, Mg, Na, P, Si, Sr, S and Ti using inductively coupled plasma atomic emission spectroscopy (ICP-AES). In order to validate the predictions made using FactSage simulations, High temperature X-ray diffraction (HT-XRD; PANalytical XPert Pro MPD theta-theta Diffractometer N-008 MSC Bldg) and Hot stage scanning electron microscopy (SEM; FEI Quanta 200 Environmental SEM) in the temperature range of 500 °C to 1000 °C were used to observe in-situ changes in the sample.

Fig. 1. Methodology used in study of agglomeration in low rank coals.

rank coals. The FACT Slag A database was used for the slag phase thermodynamic properties. All the phases available in this Slag A database and phases available in the FACT database for pure solids, liquids and gases were used. Then, the condensation process of inorganic gas species was modeled, so that starting from the gas species calculated by the chemical equilibrium software, the condensation rates of the main species related to agglomeration processes over bed particles were determined to understand the significance of the condensation phenomenon on low rank coal agglomeration. High temperature X-ray diffraction and hot stage microscopy was used to experimentally validate the modeling results. Fig. 1 summarizes the methodology and experimental techniques utilized in the study. 3.1. Experimental characterization techniques Table 1 presents proximate, ultimate and ash composition analysis (obtained using Inductively coupled plasma- Atomic emission spectrometry) of the studied coals. It can be observed that both ashes present a similar content of Si, but the lignite (Beulah-Zap) presents a higher content of alkali metals, especially of Na. High sodium contents is one of the primary contributors to formation of low melting point eutectics, and thus the initiation of agglomeration. The subbituminous coal (Wyodak-Anderson) has a slightly higher CaO content, but a slightly lower alkaline-earth oxide (MgO + CaO) content. Al is present to a greater extent in the subbituminous, Wyodak Anderson coal. It can lead to formation of alumino-silicates with high melting points, which therefore decrease the propensity to form agglomerates [20,23]. However, it can also appear in the form of alkali metal-alumino silicates (Na2O-Al2O3-SiO2) that together with alkali silicates can give rise to low temperature eutectic points. The Fe content of Beulah-zap is higher than the sub-bituminous Wyodak Anderson. The presence of Fe (III) in the form of Fe2O3, which is characteristic under oxidizing conditions, reduces the risk of agglomeration, since they react with alkali K2O/Na2O and K2CO3/Na2CO3 species, forming mixtures K2Fe2O4/Na2Fe2O4 with high melting points (> 1135 °C), and leaving at the same time less alkali available for reaction with SiO2 [20,24]. On the other hand, at certain ratios of Fe2O3 or under intermediate reducing conditions Fe (III) is partially reduced to Fe (II) in the form of FeO and metallic Fe in strongly reducing conditions. In these 3

Fuel 232 (2018) 1–11

J. Pallarés et al.

3.2. Computation of slag formation

Dij =

Slag formation contributes to the sticky material that leads to particle-to-particle adhesion and hence becomes an important parameter in the estimation of agglomeration tendencies.

−1

(4)

σij = (σi·σj )1/2

(5)

εij = (εi·εj )1/2

(6)

0.19·δij2 A C E G + + + + B (T ) exp (D ·T ) exp (F ·T ) exp (H ·T ) T

(7)

A = 1.06036 B = 0.15610 C = 0.19300 D = 0.47635 E = 1.03587 F = 1.52996 G = 1.76474 H = 3.89411

κT εij

(8)

δij = (δi·δj )1/2

(9)

T=

δ=

1.94·103·μp2 Vb·Tb

(10)

Values for the characteristic Lennard Jones energy and length and for the dipolar moment (ε, σ, µp) were determined based on the viscous theory of pure gases, tabulated in many textbooks for many gases (see for example [32]). Tb is the normal boiling temperature and Vb the liquid molar volume at the normal boiling point. For gases for which ε and σ are not tabulated, they can be obtained respectively by Eqs. (11) and (12) [32].

ε = 1.18·(1 + 1.3·δ 2)·Tb κ

(11)

1.585·Vb ⎞1/3 σ=⎛ ⎝ 1 + 1.3·δ 2 ⎠

(12)

The gaseous species composition was calculated by the chemical equilibrium software FactSageTM, based on the ash composition and ash/char temperature. Thereafter the condensation rate of main species (KO, NaO, Na2SO4, K2SO4, KOH, NaOH) over bed particles (assuming a mean particle size of 100 µm) was determined. Saturation pressure for the involved species was computed assuming a temperature delta of 20 °C between burning char surface temperature and bed bulk temperature for the combustion conditions considered in the simulations.

(1)

where dp is the diameter of the particle available for condensation, Di,m is the diffusion coefficient of condensing gas in flue gas, Mm is the molecular weight of gas species, and Pi and Pi,s are respectively the partial and vapor pressure of the condensing gas in the system. The diffusion coefficient of gas ‘i’ in a homogenous mixture of gases was estimated using the de Blanc law (Eq. (2)) [32].

4. Results and discussion 4.1. Indices used in the literature Conventionally, indices calculated based on chemical composition have been frequently used as an estimate of the agglomeration potential of fuel ashes. Hence, alkali index (AI), bed agglomeration index (BAI) and base to acid ratio (Rb/a) were first calculated to establish the baseline. Table 2 presents the calculated indices for all the coals in the study, highlighting the values that can be problematic. Fuels with an alkali index over 0.17 kJ/kg are susceptible to agglomeration, while values over 0.34 kJ/GJ indicates severe fouling and slagging problems. According to this, Beulah-Zap present a mediumlevel tendency of agglomeration. This result is confirmed by the base to acid ratio, which specifies that the agglomeration tendency is medium with values between 0.5 and 1, and severe with values over 1, which

−1

⎛ n ⎞ Xj ⎟ Di,m = ⎜ ∑ ⎜ j = 1 Dij ⎟ ⎜ ⎟ ⎝ i≠j ⎠

1 1 ⎤ Mij = 2 ⎡ + ⎢ Mi ⎥ M j⎦ ⎣

ΩD =

3.2.2. Development of modeling methodology to incorporate condensation of gaseous species onto particles In order to determine condensation of gaseous species over the particles, only the heterogeneous condensation mechanism was considered. Heterogeneous condensation takes places when the gaseous species, with a saturation ratio just over unity, condense on foreign particles or other surfaces. This means that it is more thermodynamically possible for heterogeneous condensation to occur rather than homogenous nucleation, as the free energy required for the former is much lower than that required for the latter. Assuming steady diffusion of molecules in the continuous regime (Kn < < 1), that is when the particle is sufficiently large compared to the mean free path of the diffusing vapor molecules, the heterogeneous condensation rate (Jhet) of the gaseous species on a spherical surface is given by Maxwell’s equation (Eq. (1)) [30,31].

Pi−Pi,s RT

(3)

where Mi and Mj are the molecular weight of each species, σi and σj are the Lennard Jones characteristic lenght, εi and εj the Lennard Jones characteristic energy, and ΩD is the collision integral function for the diffusion, which depends with the temperature on the intermolecular forces of the colliding molecules (Eqs. (410)).

3.2.1. By chemical reaction and melting of solid components Gaseous yield and composition and slag-liquid formation was determined based on chemical equilibrium calculations using the commercial code FactSageTM. FactSageTM calculates the stable phases under thermodynamic equilibrium based on Gibbs free energy minimization [29]. The inputs to the code are the elemental composition of the fuel ashes, the composition of the gaseous atmosphere in the reactor, and the temperature and pressure conditions of the process. The program determines the quantity and mass composition of stable compounds under equilibrium for each phase: solid, liquid, gas and slag. This way, the amount of slag and its composition is determined for any equilibrium condition considered. Simulations for the two studied coals, under typical combustion conditions in the temperature range of fluidized bed reactor operation (700–1100 °C) were carried out. The gaseous atmosphere was determined based on the amount of carbon dioxide formed on complete combustion of whole coal and a typical flue gas composition containing 3% oxygen, 15% carbon dioxide, 10% water and 72% nitrogen. In order to study agglomeration of low rank coal ashes, the following modifications were incorporated to extend the ash agglomeration modeling methodology that was developed at PSU for high rank coals [22].

Jhet = 2πdp Di,m Mm

0.00266·T 3/2 P·Mij1/2·σij2·ΩD

(2)

where Xj is the ratio of the species j concentration and the mixture concentration (cj/c), and Dij is the binary diffusion coefficient of species i in species j. The binary diffusion coefficients for the species involved in the condensation problem were calculated according to the kinetic theory of Chapman-Enskog, modified by Brokaw for polar gases (Eq. (3)) [33]. 4

Fuel 232 (2018) 1–11

J. Pallarés et al.

Table 2 Values of indices used in the literature for Beulah-Zap (BZ) and WyodakAnderson (WY) coals. Index 1 (wt. %(K2 O+ HHV wt . % (Fe2 O3) = wt . % (K2 O+ Na2O)

AI =

BAI

Rb/a =

Na2O))

wt . % (Fe2 O3 + CaO + MgO + K2 O+ Na2 O+ P2 O5) wt . % (SiO2 + Al2O3 + TiO2)

WY

BZ

0.08

0.29

1.94

1.08

0.97

1.36

are the cases presented by Wyodak-Anderson and Beulah-Zap respectively. On the other hand, according to the bed agglomeration index none of the coals presents a high probability of agglomeration (< 0.15). From this preliminary analysis, the significance of high alkali, alkaline-earth metals is evident. However, a clear estimate of the bed agglomeration tendency is not obtained. It has also been stated in the literature that these indexes are not accurate and in general they are not recommended as a prediction tool for agglomeration. As an example, Mac. et al. [25] carried out an experimental study on agglomeration for nine different fuels, and in spite of the fact that the alkali index was useful in predicting the existence of agglomeration problems (succeeded in the prediction of 7 out of 9 fuels under different conditions), they concluded that the indices are only useful as a preliminary indication of the influence of fuel composition on agglomeration. Hence, it is important to understand the influence of alkali/alkalineearth metals on agglomerate growth further and develop methods to quantify this influence.

Fig. 3. Chemical fractionation analysis Wyodak-Anderson coal: concentration of major and minor elements in the liquid phase (DI: deionized water, AmAc: ammonium acetate, HCl: hydrocloric acid).

4.2. Chemical fractionation analysis for the separation of chemical components for study In order to study the influence of the basic components and their effect on agglomerate growth further, fuel separation into distinct fractions rich in these different components was desired. Chemical fractionation was chosen to obtain this segregation. Miller and Miller [27] reported that the higher the percentage of the alkali and alkaline earth elements occur in the water-soluble and ion-exchangeable portion of the fuels, the higher is the potential for forming molten phases in the bed during combustion. Figs. 2 and 3 present the concentration of major and minor elements in the liquid phase after each leaching step in the chemical fractionation procedure: Deionized water, Ammonium Acetate and Hydrochloric acid. Figs. 4 and 5 show the composition of the insoluble fraction of the solid remaining after the three leaching processes. For the purpose of

Fig. 4. Chemical fractionation analysis Beulah-Zap coal: concentration of major and minor elements in the solid phase: initital ashed and insoluble phase.

Fig. 5. Chemical fractionation analysis Wyodak-Anderson coal: concentration of major and minor elements in the solid phase: initital ashed and insoluble phase.

comparison, the initial ash composition has been included in these figures. Sodium and Potassium occur predominantly as water soluble and ion-exchangeable species. This result is even more distinct in the case of Beulah-Zap coal, which has a higher Na content in both fractions, especially in the case of the ammonium acetate solution. The alkali metal content in the insoluble fraction of both coals is practically negligible. As for alkaline earth metals, these are mostly present in the ion exchangeable fraction but unlike alkali metals, also appear in the acid soluble fraction. Again the presence of Mg and Ca in the insoluble

Fig. 2. Chemical fractionation analysis Beulah-Zap coal: concentration of major and minor elements in the liquid phase (DI: deionized water, AmAc: ammonium acetate, HCl: hydrocloric acid). 5

Fuel 232 (2018) 1–11

J. Pallarés et al.

fraction is very small. Conversely, aluminium and silica are concentrated in the acid soluble fractions and mainly in the insoluble fraction, pointing to the formation of aluminosilicates with high melting points. On the other hand, iron has a greater presence in phases soluble in acid. It is also present in insoluble phases in the case of Beulah-Zap. These aluminosilicate and iron rich phases are also found in large concentrations in high rank coals. Finally, sulfur, which is very reactive, appears predominantly in the water soluble and ion-exchangeable form, and in a lower proportion in the acid soluble phase. Hence, chemical fractionation has enabled clear separation of the phases.

Fig. 7. Comparison of calculated condensates for coals studied.

4.3. Significance of gaseous component condensation on agglomerate growth

0.915% of K). The modeling results presented in this figure are consistent with observations made in the literature. Guo et al. [34] found that sodium release from coals during combustion shows a strong positive correlation to the Na/(Si + Al) ratio through their experimental study of six sodium-rich coals. From the composition of the coals provided, in Table 1, Beulah-zap clearly has a significantly higher Na/ (Si + Al) ratio. This correlates well with the results showing sodium release almost twice as much for this coal over Wyodak-Anderson. The forms of alkali metal compounds in the raw coal also influence this release into the gaseous components. Naruse et al. [35] studied the influence of coal characteristics on the evolution of alkali metal compounds in the combustion of four coals, and concluded that the watersoluble and/or ion-exchangeable fractions are more likely to evolve than other alkali forms. Accordingly, the gas species composition presented in Fig. 6 are consistent with the chemical fractionation analysis depicted in Figs. 2 and 3, wherein sodium and potassium were the water soluble and ion-exchangeable species for Beulah Zap and Wyodak Anderson respectively. The weights of condensates obtained using the developed condensation model are presented in Fig. 7. They show that sodium sulfate is the only gaseous species that can condense over the bed particles for both coals under the oxidizing conditions considered in the simulations, reacting with silica, for the most common case of a silica sand bed (SiO2 + Na2SO4 → Na2O-SiO2 + SO2 + 0.5 O2), and forming low temperature melting points silicates. Beulah-Zap coal exhibits increasing condensation of Na2SO4 with temperature within the temperature range analyzed (700–1100 °C). In the case of Wyodak-Anderson, saturation pressure is not achieved for K2SO4 condensation while the temperature range at which condensation of Na2SO4 can occur is restricted to 750–860 °C. Fig. 8 shows that although the release of Na2SO4

Focusing now on the gaseous yield of inorganic species, during the conversion of a fuel particle, some of the alkali metals are in form of gaseous species. Mass balance distribution of predominant elements (Na, K, Ca, Mg, Si, Al, Fe, S) among the three phases (slag, gas and solid) was performed. While all the K is mainly transferred to the slag phase, in the case of Na, depending on the Na2O-Al2O3-CaO distribution in the phase diagram, it contributes to the slag formation or tends to stabilize in form of sodium aluminium-silicates in the solid phase. On the other hand, while the alkaline earth metals, aluminium and silica progressively goes through to the slag phase in form of silicates as temperature is increased, the iron content in the slag phase is negligible as has been previously discussed, and sulfur is predominantly evolved in form of SO2 and SO3. Fig. 6 presents the gaseous yield and composition, on a SO2 and SO3 free basis, for both coals at 900 °C under the same input energy basis (1 MJ). For both coals, alkali oxides, hydroxides and sulfates are predominant. These results reveal that the yield of gaseous inorganic species from low rank coals, such in the case of Wyodak-Anderson and Beulah-Zap, leads to the possibility for a heterogeneous condensation route for agglomeration over the bed particles covering its surface and giving rise to reactions that form low melting point eutectics when the gaseous compounds (KO, NaO, Na2SO4, K2SO4, KOH, NaOH) are oversaturated. In the case of Beulah-Zap, which presents much higher content of Na than K in the ash composition analysis (Table 1), the predominant gaseous species are NaOH, Na2SO4 and KOH while K2SO4 is released to a lower extent. In the case of Wyodak-Anderson, the contribution of K2SO4 to the gaseous components is greater than that of Na2SO4 because the content of Na and K in the parent coal is similar (Table 1) and a higher percentage of K is transferred to the gas phase (0.04% of Na vs.

Fig. 8. Na2SO4 gas yield of Wyodak-Anderson coal under the same input energy basis (1 MJ). Saturation moles were computed assuming a temperature delta of 20 °C.

Fig. 6. Gas yield and composition of coals at 900 °C under the same input energy basis (1 MJ) on a SO2 and SO3 free basis. 6

Fuel 232 (2018) 1–11

J. Pallarés et al.

Beulah-Zap takes place at a temperature of only 740 °C, compared to 810 °C in the case of Wyodak-Anderson. In addition, in the temperature range studied, the percentage of this molten phase is much greater for the former increasing to 32% at 900 °C and above 40% at temperatures over 1000 °C. Further analysis on the different phases and their composition was performed based on characteristics eutectics points for the involved binary and ternary systems. Table 3 shows the slag-liquid composition for both coals in the temperature range from 750 to 950 °C. According to the results presented in Table 3, the presence of Si, Ca, Al and Na is responsible for the initiation and progression of the slag phase. Alkali metals, especially Na which is the dominant alkali element in both fuels, react forming new compounds, some of them with low melting temperatures such as silicates (Na2SiO3, Na2SiO5, Na4SiO4, Na6Si2O7). According to the work of Kracek [36], the most influential compounds in slag formation from the binary Na2O-SiO2 system are sodium metasilicate (Na2SiO3) and sodium disilicate (Na2SiO5) having melting temperatures of 1088 °C and 874 °C respectively. These two compounds give rise to a first eutectic point at 840 °C with 62.1% SiO2. The second eutectic point, between the disilicate and SiO2, takes place at 793 °C with 73.9% SiO2. Including sodium orthosilicate (Na4SiO4) in the analysis, a new eutectic point appears between it and the sodium metasilicate at 1022 °C with 43.1% SiO2. Finally, from the works of Loeffler and D’Ans [37], the formation of a new compound Na6Si2O7 from Na4SiO4-Na2SiO3 is described, giving rise to two new eutectic points between Na4SiO4 and Na6Si2O7 at 962 °C and 37.2% SiO2 and between Na6Si2O7 and Na2SiO3 at 1015 °C and 41.2% SiO2. In the analysis of the binary system Na2O-Al2O3, according to the work of De Vries and Roth [38], the formation of different phases of sodium aluminates (NaAlO2, NaAl11O17 and NaAl5O8) were identified. All the phases had melting points over 1800 °C (NaAlO2 at 1867 °C, and β and β” have even higher melting points because of higher inclusion of Al2O3). A eutectic point was determined at 1595 °C. Since the temperature of these events is greater than the typical operating temperatures of a fluidized bed, its contribution to the beginning and progress of the agglomeration in this study was not considered. The relationship of the ternary system Na2O-Al2O3-SiO2 is analyzed below. In the literature, numerous binary and ternary subsystems are identified that give rise to the formation of eutectics with low melting points. Among the first are emphasized the eutectic points corresponding to the binary subsystems sodium disilicate-albite (Na2Si2O5NaAlSi3O8) at 767 °C (25.6% Na2O; 67% SiO2; 7.4% Al2O3), sodium disilicate-nepheline/carnegieite (Na2Si2O5-NaAlSiO4) at 768 °C (30.5% Na2O; 59.1% SiO2; 10.4% Al2O3) and sodium metasilicate-nepheline/ carnegieite (Na2SiO3-NaAlSiO4) at 906 °C (37.2% Na2O; 46% SiO2; 16.8% Al2O3) [25]. Within the ternary systems, are highlighted the eutectics formed by nepheline/carnegieite-albite-sodium disilicate (NaAlSiO4-NaAlSi3O8-Na2Si2O5) at 732 °C (26% Na2O; 61.5% SiO2; 12.5% Al2O3), albite-sodium disilicate-silica (NaAlSi3O8- Na2Si2O5SiO2) at 740 °C (21.5% Na2O; 73.8% SiO2; 4.7% Al2O3) and sodium metasilicate-sodium disilicate-nepheline/carnegieite (Na2SiO3Na2Si2O5-NaAlSiO4) at 760 °C (32% Na2O; 57.9% SiO2; 10.1% Al2O3) [39]. The influence of alkali earth metals (in particular of Ca due to its greater presence in both coals), is analyzed by the ternary system CaOSiO2-Na2O. The presence of calcium monosilicate, calcium disilicate and calcium orthosilicate together with sodium silicates leads to the formation of new compounds (Na2Ca2Si3O9, Na4CaSi3O9, and Na2Ca3Si6O16, Na6Si8O19, Na2CaSi5O12, Na8Ca3Si5O17) and to the appearance of eutectics with low melting points. In the case of systems with high Si content (> 50% wt.), predominate the eutectics formed by SiO2-Na2Ca3Si6O16-Na2Si2O5 at 725 °C (73.5% Na2O; 21.3% SiO2; 5.2% CaO), Na4CaSi3O9-Na2Si2O5-Na2SiO3 at 821 °C (37.5% Na2O; 60.7% SiO2, 1.8% CaO) and Na2SiO3-Na2Ca2Si3O9 at 840 °C (46.4% Na2O; 49.4% SiO2; 4.2% CaO) [40]. As the Si content decreases the formation

Fig. 9. Gas yield of inorganic species for different solid fractions of the chemical fractionation analysis.

for Wyodak-Anderson extends up to 1000 °C, due to the lower Na presence in relation to other elements its partial pressure decreases below the saturation pressure as it is released, restricting the condensation route to below 860 °C. The occurrence of condensation in these coals below 1000 °C shows that it is likely to be the cause of initiation of agglomerate growth at these low temperatures. Extending the study to the different solid fractions obtained with the chemical fractionation analysis (after DI water, AmAc and HCl leaching steps), Beulah-Zap samples show that for all the cases the yields of the gaseous inorganic species are lower than that corresponding to the residue from the previous leaching step (Fig. 9). Condensation of Na2SO4 does not take place under 1000 °C, except for the whole coal ash and residue obtained after DI water extraction. For the case of WyodakAnderson, Na2SO4 is released between 750 and 1000 °C but saturation is not achieved and no condensation takes place for any of the remaining solid fractions of the leaching process. These results are consistent with the elimination of the alkali elements during the progressive leaching steps as shown in Figs. 2 and 3, and confirm the relevance of these elements in the formation of agglomerates by means of the condensation route.

4.4. Chemical equilibrium calculations to obtain slag-liquid formed by the process of melting First, the formation of the slag phase in both coals is analyzed. As the fuel particle temperature increases, generally over the surrounding bed media temperature, part of ash mineral compounds start to melt. It is then that by collisions with other particles, these can coalescence or part of the molten material can be transferred to the bed material coating it and starting further reaction processes. From the equilibrium calculations obtained with FactSageTM, Fig. 10 shows the results of the liquid slag formation for the two coals for the temperature range 700–1100 °C. These results show that the appearance of a slag phase for

Fig. 10. Mass fraction of slag as a function of temperature. 7

Fuel 232 (2018) 1–11

J. Pallarés et al.

Table 3 Slag phase composition (% wt.) at different temperatures (WY: Wyodak-Anderson; BZ: Beulah-Zap).

SiO2 CaO MgO Al2O3 Na2O TiO2 K2O MnO FeO

750 °C WY

BZ

800 °C WY

BZ

850 °C WY

BZ

900 °C WY

BZ

950 °C WY

BZ

– – – – – – – – –

0.932 22.88 6.893 55.38 12.44 0.025 0.8609 0.5945 0.00031

– – – – – – – – –

23.49 12.06 2.961 36.82 20.21 0.6488 2.819 0.9858 0.00017

41.04 18.69 1.869 20.46 12.15 2.09 3.56 0.1374 0.00159

34.16 24.97 1.876 19.68 16.47 1.916 0.6949 0.2416 0.00088

41.34 20.71 2.466 20.2 10.37 2.537 2.282 0.08841 0.00589

34.43 26.04 2.274 19.66 15 1.74 0.6289 0.2194 0.00242

40.96 22.47 3.085 19.76 8.663 3.076 1.902 0.07385 0.01713

33.55 27.05 3.007 20.17 13.93 1.54 0.5508 0.1942 0.00567

of eutectic points gives higher temperatures (> 1200 °C). Similarly, the ternary system CaO-SiO2-Al2O3 has several eutectics in the 1200–1300 °C range. Finally, the influence of iron in the ash melting behavior is controlled by the Fe2O3-Al2O3-SiO2 and FeO-SiO2-CaO phase diagrams. However, the presence of Fe (III) in form of Fe2O3 reduces the risk of agglomeration since it reacts with alkali species, giving rise to the formation of eutectics with higher melting temperatures (> 1135 °C), and leaving, at the same time, less alkali metals available for the reaction with the silica oxide. In order to identify eutectics formed at low temperatures, Table 4 shows the proportions of the oxides corresponding to the Na2O-SiO2, Na2O-SiO2-Al2O3 and Na2O-SiO2-CaO systems, in the slag phase in the range 750–950 °C. The analysis of the Beulah Zap composition suggests the formation in the binary system Na2O-SiO2 of a sodium metasilicate and sodium disilicate eutectic at 840 °C, while Wyodak Anderson would not lead to this eutectic supporting a lower initiation temperature for the melting process of the former. No eutectics have been identified corresponding to the ternary systems composition. Notwithstanding the foregoing and the probable local formation of other eutectics that cannot be identified through this global analysis, these results show the presence of one of the most relevant compounds in slag formation at low temperatures responsible of the initiation of the agglomeration. From a general analysis of these results it can be concluded that the greater tendency of ash melting processes is primarily due to the formation of alkali metals silicates with low melting points. This is more critical in the case of Beulah-Zap which has a higher content of alkali species and lower content of alumina and iron oxides. These results are in agreement with previous studies on low rank coal agglomeration in FBC, which identify alkali and fluxing reactions between iron from pyrites and aluminosilicates from clays as the main mechanisms in the initiation of the agglomeration. Finally, in order to show the effect on agglomeration of the studied coals in a real thermochemical conversion process in a fluidized bed, the amount of ash and the high heating value of the coals were taken into account. Fig. 11 shows the slag-liquid formation and composition for both coals at 750 °C and 900 °C under the same ash mass basis (1 kg)

Fig. 11. Slag-liquid formation and composition of coals at 750 °C and 900 °C under the same ash mass basis (1 kg) and energy basis (1MJ).

and input energy basis (1 MJ). These figures really indicate the slagliquid formation tendency showing a clear comparison of the severity of the agglomeration of the coals. These results reveal that the Beulah-Zap lignite is more than five times predisposed to develop agglomeration problems than Wyodak-Anderson for the same energy output production.

Table 4 Proportions of the oxides corresponding to the Na2O-SiO2, Na2O-SiO2-Al2O3 and Na2O-SiO2-CaO in the slag phase (wt.%) in the range 750 – 950 °C.

Na2O SiO2 Na2O SiO2 Al2O3 Na2O SiO2 CaO

750 °C WY BZ

800 °C WY BZ

850 °C WY BZ

900 °C WY BZ

950 °C WY BZ

–

–

4.85 95.15 3.29 64.54 32.17 3.38 66.39 30.20

6.98 93.02 4.82 64.20 30.97 4.62 61.59 33.78

8.39 91.61 5.86 63.97 30.17 5.36 58.52 36.12

–

–

93.03 6.9 18.09 1.36 80.56 34.31 2.57 63.12

–

–

46.26 53.74 25.11 29.17 45.72 36.26 42.12 21.63

32.52 67.48 23.42 48.59 28.00 21.78 45.19 33.02

30.36 69.64 21.72 49.82 28.46 19.88 45.61 34.50

4.5. Comparison of role of condensation and melt formation in agglomerate growth Comparing the results of the condensation (Fig. 7) with those obtained from the analysis of the formation of the slag phase (Fig. 11), we conclude that the amount of slag-liquid formed from condensation is lower than that corresponding to the fusion of mineral species. However, when starting at relatively low temperatures of 750 °C, its contribution in the very initial stages of agglomeration is not negligible and should be taken into account. Thus, for example, in the case of Wyodak-Anderson, from 750 °C

29.34 70.66 20.60 49.59 29.81 18.69 45.01 36.30

8

Fuel 232 (2018) 1–11

J. Pallarés et al.

crystalline peak decreased and it fully decomposed at 1000 °C. Microscopic imaging of Beulah Zap coal shows considerable binding at the end of the process (Fig. 14c) compared to the sample at room temperature (Fig. 14a). A video of the entire process indicated that deformation began at about 700 °C. Fusion with bridge formation between particles was distinctly seen at around 870 °C as shown in Fig. 14b. This onset temperature also supports that predicted using Factsage simulations.

condensation processes can occur which in turn can react with the species on the surface of the particle facilitating the formation of lowmelting eutectics and provide a route for the onset of agglomeration, while the onset temperature of the formation of the slag phase has been determined to occur at 840 °C. Based on the quantification obtained from the simulations, it is likely that although the melt formation process is critical to the formation of the bulk of sticky material that binds ash particles, the low temperature condensation of alkali-, alkaline-earth metal phases leads to the initiation. This is further supported by literature such as microscopic analyses of ash samples from JEA’s CFB which showed that the primary bonding mechanism resulted from the volatilizing of alkali compounds containing sodium, potassium and chlorides at relatively low temperatures. The dominant mechanism of agglomeration in low rank coals will be driven by specific temperature, gaseous conditions and exact coal composition. However, for most cases from the literature in which agglomeration was witnessed, the condensation mechanism contributed to low temperature initiation of the process and alumino-silicates thereafter resulted in bulk melt formation. This study presents a means to quantify and predict the formation of these initiators and suggests their significance in agglomerate formation. This is also supported by the validation of similar mechanisms in the case of high rank coals, wherein iron or calcium content of the coal contributed to the initiation of agglomerate growth. Thereafter, agglomeration continued due to extensive melt phase formation of the alumino-silicates. Extending the study to the different solid fractions obtained with the chemical fractionation analysis, results of the slag onset temperature and the temperature range in which condensation occurs are presented in Table 5. These results are consistent with the progressive elimination of alkali and alkaline earth metals elements during the different leaching steps (Figs. 2 and 3). Alkaline earth metals rich fractions present both lower slag onset temperatures and have condensation occurring at lower temperatures. On the contrary, the remaining solid fractions after the HCl leaching, where the alkali metals content is practically negligible, correspond to those with a higher slag onset temperature and no condensation takes places. These results confirm an evident relation of the alkali content of the fuels with the occurrence of agglomeration initiation and growth at low temperatures as a result of both condensation and melt formation mechanisms. Moreover, at temperatures below 800 °C condensed phases play a critical role in the initiation of agglomerate growth for both fuels. This initiation temperature is about 30–50 °C lower than the predicted molten slag phase formation temperature.

5. Conclusions A detailed study of agglomerate initiation and growth behavior of low-rank coals has been carried out. Low rank coals when compared to high-rank coals present a higher tendency to agglomerate since they have higher amounts of alkali metals (Na, K), alkaline earth metals (Ca, Mg), as well as Si, Cl, S, P, Fe, that can interact with the bed material forming low melting compounds such as silicates. This melt formation mechanism is the main route for the agglomerate growth. However, a second mechanism due on the condensation of the vaporized inorganic species giving rise to low melting point eutectics, also plays a role in the initiation of agglomeration. Consequently, a novel condensation model has been developed to extend agglomeration models developed for the study of high rank coal agglomeration to low rank coals and biomass that contain alkali and alkaline earth species in the ash. Assuming that heterogeneous condensation is more likely to occur rather than homogenous nucleation, the model, starting from chemical equilibrium calculations using the software FactSageTM, predicts the condensation rate of main species related to agglomeration processes (KO, NaO, Na2SO4, K2SO4, KOH, NaOH) over bed particles. Results for two low-rank coals (WyodakAnderson and Beulah-Zap, sub-bituminous and lignite respectively) show that sodium sulfate is the only gaseous species that condenses for both coals. Condensation starts taking place at low temperature (750 °C) but occurs over a larger temperature range for the lignite, which has a higher Na content. Extending the study to different solid fractions of the coals obtained with a chemical fractionation analysis shows that the yields of the gaseous inorganic species are lower for the alkali-poor fractions for which saturation is not achieved and no condensation takes place. Slag formation onset temperatures were also observed by HT-XRD analysis and HT-SEM results. Results show that Beulah-Zap has a slag onset temperature of only 740 °C, compared to 810 °C in the case of Wyodak-Anderson. Morevoer, a low temperature eutectic of sodium metasilicate and sodium disilicate has been identified for Beulah-Zap supporting the lower initiation temperature of the melting process. Comparing the results of the condensation rate with those of the melting process, we conclude that the contribution of condensation to the agglomeration phenomenon is lower than that corresponding to the fusion of mineral species. However, condensation can start at 30 to 50 °C lower temperatures than those predicted for the molten slag phase formation indicating that condensed phases play a critical role in the initiation of agglomerate growth for both fuels at temperatures below 800 °C. The computational method developed for the mathematical estimation of condensates can be incorporated into models for the prediction of agglomeration for alkali/alkaline-earth rich fuels such as low rank coals and biomass.

4.6. Experimental results to support initiation of agglomerate growth The slag formation onset temperature results, presented in Table 5, obtained using FactSage simulations are supported by the HT-XRD analysis and HT-SEM results, performed in the temperature range of 500 to 1000 °C. Differences were observed in the behavior of Ca-based phases during slag-liquid formation. In the Wyodak-Anderson sample, CaSO4 was seen to convert to CaO at about 700 °C, as seen in Fig. 12. Similar transformations were seen in HT-XRD of Beulah-zap coal (Fig. 13). Also, CaO began to form slag-liquid, and the intensity of its Table 5 Slag onset temperature (°C) and condensation temperature range (°C) corresponding to the chemical fractionation analysis solid fractions. Fuel

Slag onset T (°C)

Condensation T (°C)

Beulah-Zap Beulah-Zap DI Beulah-Zap HCl Wyodak Wyodak HCl

740 °C 800 °C 1020 °C 810 °C 890 °C

750–1150 °C 770–1000 °C – 750–860 °C –

Acknowledgements Javier Pallarés gratefully acknowledges the José Castillejo/ Fulbright Fellowship Program (Reference CAS10/00329) sponsored by the Government of Spain for their support to visitor research exchange programs to conduct this research at the Pennsylvania State University.

9

Fuel 232 (2018) 1–11

J. Pallarés et al.

Fig. 12. HT-XRD Wyodak-Anderson.

Fig. 13. HT-XRD Beulah-Zap. 10

Fuel 232 (2018) 1–11

J. Pallarés et al.

Fig. 14. HT-SEM Beulah-Zap: (a) sample before heating (b) 870 °C (c) 1000 °C.

References [1] Khadilkar AB, Rozelle PL, Pisupati SV. Models of agglomerate growth in fluidized bed reactors: critical review, status and applications. Powder Technol 2014;264:216–28. [2] Khadilkar AB, Rozelle PL, Pisupati SV. Study of initiation of ash agglomeration in fluidized bed gasification systems. Fuel 2015;152:48–57. [3] Khadilkar AB, Rozelle PL, Pisupati SV. Effect of heterogeneity in coal ash chemical composition on onset of conditions favorable for agglomeration in fluid beds. Energies 2015;8:12530–45. [4] Khadilkar AB, Rozelle PL, Pisupati SV. Investigation of fluidized bed agglomerate growth process using simulations and SEM-EDX characterization of laboratorygenerated agglomerates. Chem Eng Sci 2018;184:172–85. [5] Rice RL, Shang JY, Ayers WJ. Fluidized-bed combustion of North Dakota lignite. in: Proceedings 6th International Conference on Fluidized Bed Combustion (ASME) 1980: p. 863–871. [6] Bobman MH, Hajicek DR, Zobeck BJ. A study of bed agglomeration resulting from the AFBC of low-rank coals. in: Proceedings 8th International Conference on Fluidized Bed Combustion (ASME) 1985: p. 1399–1407. [7] Rizeq RG, Shadman F. Alkali-induced agglomeration of solid particles in coal combustors and gasifiers. Chem Eng Commun 1989;81:83–96. [8] Atakül H, Ekinci E. Agglomeration of Turkish lignites in fluidised-bed combustion. J Inst Energy 1989;62:56–61. [9] Kang J, Beverly R, Porter D, Leach B. Reducing ash agglomeration in JEA’s CFB boilers. Power 2012. [10] Brown RC, Dawson R, Smeek JL, Bed material agglomeration during fluidized bed combustion, Final report for U. S. Department of Energy, No. DE-FG22-92PC92530; 1996. [11] Goblirsch GM, Vander Molen RH, Wilson K, Hajicek D. Atmospheric fluidized bed combustion testing of North Dakota lignite. in: Proceedings 6th International Conference on Fluidized Bed Combustion (ASME) 1980: p. 850–862. [12] Goblirsch GM, Benson SA, Hajicek DR, Cooper JL. Sulfur control and bed material agglomeration experience in low-rank coal AFBC testing. in: Proceedings 7th International Conference on Fluidized Bed Combustion (ASME) 1982: p. 1107–1120. [13] Manzoori AR, Agarwal PK. Agglomeration and defluidization under simulated circulating fluidized-bed combustion conditions. Fuel 1994;73:563–8. [14] Vuthaluru HB, Zhang DK. Control methods for remediation of ash-related problems in fluidised-bed combustors. Fuel Process Technol 1999;60:145–56. [15] Bhattacharya SP, Harttig M. Control of agglomeration and defluidization burning high-alkali, high-sulfur lignites in a small fluidized bed combustors. Effect of additive size and type, and the role of calcium. Energy Fuels 2003;17:1014–21. [16] Bartels M, Lin W, Nijenhuis J, Kapteijn F, van Ommen JR. Agglomeration in fluidized beds at high temperatures: Mechanisms, detection and prevention. Prog Energy Combust Sci 2008;34:633–66. [17] Chaivatamaset P, Tia S. The characteristics of bed agglomeration during fluidized bed combustion of eucalyptus bark. Appl Therm Eng 2015;75:1134–46. [18] Khadilkar AB, Rozelle PL, Pisupati SV. Review of particle physics and chemistry in fluidized beds for development of comprehensive ash agglomeration prediction models. Energy Fuels 2016;30. 3714–373. [19] Lin W, Dam-Johansen K, Frandsen F. Agglomeration of bio-fuel fired fluidized bed combustors. Chem Eng J 2003;96:171–85. [20] Elled AL, Amand LE, Steenari BM. Composition of agglomerates in fluidized bed

[21]

[22]

[23]

[24]

[25]

[26] [27] [28]

[29] [30]

[31] [32] [33] [34] [35]

[36] [37] [38]

[39] [40]

11

reactors for thermochemical conversion of biomass and waste fuels. Experimental data in comparison with predictions by a thermodynamic equilibrium model. Fuel 2013;111:696–708. McNallan MJ, Yurek GJ, Elliott JF. The formation of inorganic particulates by homogeneous nucleation in gases produced by the combustion of coal. Combust Flame 1981;42:45–60. Khadilkar AB. Development of a fluidized bed agglomeration modeling methodology to include particle-level heterogeneities in ash chemistry and granular physics. Dissertation in Energy and Mineral Engineering, The Pennsylvania State University; 2016. Chou J-D, Lin Ch-L, Yang Y-K, Cheng L-M, Lua Z-Y, Cen K. Inhibition of agglomeration/defluidization by different calcium species during fluidized bed incineration under different operating conditions. Powder Technol 2012;219:165–72. Jing N-J, Wang Q-H, Yang Y-K, Cheng L-M, Lua Z-Y, Cen K. Influence of ash composition on the sintering behavior during pressurized combustion and gasification process. Appl Phys Eng 2012;13:230–8. Mac an Bhaird ST, Walsh E, Hemmingway P, Maglinao AL, Capareda SC, McDonnell KP. Analysis of bed agglomeration during gasification of wheat straw in a bubbling fluidised bed gasifier using mullite as bed material. Powder Technol 2014;254:448–59. Vamvuka D, Zografos D, Alevizos G. Control methods for mitigating biomass ashrelated problems in fluidized beds. Bioresour Technol 2008;99:3534–44. Miller BG, Miller SF. Combustion Engineering Issues for Solid Fuel Systems. 1st ed Elsevier; Academic Press; 2008. O'Keefe CA, Eylands KE, Pflughoeft-Hassett DF. Application of selective leaching technique for major, minor and trace element analysis in coal. Am Chem Soc Div Fuel Chem Preprints 1993;38:966–71. FactSage Thermodynamic Software. http://www.factsage.com/; 2018 (accessed 30 January 2018). Doshi V, Vuthaluru HB, Korbee R, Kiel JHA. Development of a modeling approach to predict ash formation during co-firing of coal and biomass. Fuel Process Technol 2009;90:1148–56. Seinfield JH, Pandis SN. Atmospheric Chemistry and Physics: From Air Pollution to Climate Change. 2nd ed. John Wiley & Sons; 2006. Poling BE, Prausnitz JM, O'Connell JP. The Properties of Gases and Liquids. 5th ed. Mc-Graw Hill; 2001. Brokaw RS. Predicting transport properties of dilute gases. Ind Eng Chem Process Des Dev 1969;8:240–53. Guo S, Jiang Y, Yu Z, Zhao J, Fang Y. Correlating the sodium release with coal compositions during combustion of sodium-rich coals. Fuel 2017;196:252–60. Naruse I, Murakami T, Noda R, Ohtake K. Influence of coal type on evolution characteristics of alkali metal compounds in coal combustion. in: 27th Symposium (International) on combustion. The Combustion Institute; 1998. p. 1711–1717. Kracek FC. The system sodium oxide-silica. J Phys Chem 1930;34:1583–98. D’Ans J, Loeffler J. Untersuchungen im System Na2O-SiO2-ZrO2. J Inorgan Gen Chem 1930;191:1–35. de Vries RC, Roth WL. Critical evaluation of the literature data on beta alumina and related phases: I, phase equilibria and characterization of beta alumina phases. J Am Ceram Soc 1969;52:364–9. Rys M. Investigation of Thermodynamic Properties of Alkali Metals in Oxide Systems Relevant to Coal Slags. Doctoral Thesis RWTH Aachen University; 2007. Zhang Z, Xiao Y, Voncken J, Yang Y, Boom R, Wang N, et al. Phase equilibria in the Na2O–CaO–SiO2 system. J Am Ceram Soc 2011;94:3088–93.