Comprehensive coal quality index for evaluation of coal agglomeration characteristics

Comprehensive coal quality index for evaluation of coal agglomeration characteristics

Fuel 231 (2018) 379–386 Contents lists available at ScienceDirect Fuel journal homepage: www.elsevier.com/locate/fuel Full Length Article Comprehe...

2MB Sizes 0 Downloads 132 Views

Fuel 231 (2018) 379–386

Contents lists available at ScienceDirect

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

Full Length Article

Comprehensive coal quality index for evaluation of coal agglomeration characteristics

T

⁎,1

Li Guo1, Ming Zhai

, Zhentong Wang, Yu Zhang, Peng Dong

School of Energy Science and Engineering, Harbin Institute of Technology, Harbin, Heilongjiang 150001, China

A R T I C LE I N FO

A B S T R A C T

Keywords: Coal quality characteristics Coal agglomeration characteristics Entropy method Weight analysis Discriminant index

A comprehensive analysis of coal quality factors affecting coal agglomeration characteristics was carried out. The influence weight of seven main coal quality indices on agglomeration was calculated by the entropy method. Furthermore, a reliable and comprehensive coal quality index was obtained to evaluate the characteristics of coal agglomeration by the agglomeration experiment results. The proximate analysis, ultimate analysis, and petrographic analysis results were applied for the analysis of organic constituents. The XRF and ash fusion temperatures were applied for the analysis of inorganic constituents. Results show that there is an absolute positive correlation between the C/H, FC/V and the characteristics of coal agglomeration. Both of the vitrinite content and the average reflectivity of the vitrinite have a positive effect on the coking ability of coal. Alkali oxides not only have a low melting point but also can react with Si and Al to form low melting point eutectics. Both of these characteristics result in coal ash slagging. The entropy method was adopted to analyze the effect weight of C/H, FC/V, vitrinite content Vi, vitrinite average reflectivity Rave, ash deformation temperature (DT), Base/Acid ratio (B/A) and Silicon ratio (G) on the agglomeration characteristics of coal. The weights are 11%, 10%, 26%, 12%, 9%, 14% and 18%, respectively. Among them, vitrinite content had the largest effect, and DT had the least effect. In addition, an empirical index of coal agglomeration characteristics is proposed, and the discriminant limits are provided. The results are helpful to the evaluation of coal agglomeration, and useful to the safe and efficient operation of boilers.

1. Introduction Coal is the dominant energy source in China, and about 70% of total electricity generation is from coal. During the process of coal-fired power generation, the agglomeration of fuel often occurred in boilers, especially the fluidized bed combustor [1]. The agglomeration will seriously reduce the combustion efficiency and lead to waste of fuels. Furthermore, the large slag block formed on the heating surface of boilers will not only worsen the heat transfer of the heating surface but also cause safety accidents [2]. Therefore, it is necessary and significant to study and analyze the coal agglomeration in the combustion process for the safe and efficient operation of boilers in industry. At present, most studies of the agglomeration problem focus on the slagging which is caused by the fusion of ash in coal. However, in fact, a great many coals used for combustion have a degree of coking ability. In coking process, there also will be molten materials formed, which can cause the agglomeration phenomenon. Therefore, the agglomeration phenomenon indeed includes both of the coking and the slagging. For the coal with coking ability, the agglomeration phenomenon caused ⁎

1

Corresponding author. E-mail address: [email protected] (M. Zhai). These two authors contributed equally to this work.

https://doi.org/10.1016/j.fuel.2018.05.119 Received 3 January 2018; Received in revised form 22 March 2018; Accepted 23 May 2018 0016-2361/ © 2018 Elsevier Ltd. All rights reserved.

by the organic components in the combustion process cannot be neglected. Therefore, it is necessary to make a comprehensive analysis of coking and slagging in the combustion process for predicting the potential agglomeration possibility of coking coal. It is known that coking and slagging characteristics are determined by the organic and inorganic components of the coal, respectively [3,4]. Currently, the research on coking ability mainly concentrated in the field of coal chemical industry. For coking coals, when the temperature increases, the coals always go through a specific process: thermoplastic properties soften, swell, and eventually re-solidify [5]. According to many researchers, there are many parameters which affect coke performance. Precisely, the coal rank parameters (moisture, volatile matter, carbon, etc.), petrographic composition, and impurities (ash, sulfur, and alkali contents) together control the quality of coals and fundamentally affect the coking coal quality [6,7]. Zhang [8] conducted coking experiments on 19 individual coals and 64 coal blends and found that coal with volatile content has a 22% −26% better coking ability. Speight [9] and Ghosh [10] found that the caking tendency of coals rises dramatically between 25% and 35% (by weight) volatile

Fuel 231 (2018) 379–386

L. Guo et al.

volatile content was the reduced mass minus the moisture content. For the ash content, the coal sample was put into a muffle furnace by raising the temperature up to 500 °C in more than 30 min and kept that temperature for 30 min and rose up to 815 ± 10 °C; then after 1 h, the residue mass was the ash content. The ultimate analysis of coals was measured by the Ultimate Analyzer (VARIO MACRO CUBE, Elementar, Germany). The ultimate analysis was to burn or decompose the coal sample at high temperature and convert the test elements into gas for analysis. Analytical gases were analyzed by purge-trap adsorption and desorption on three special columns and then detected by a thermal conductivity detector (TCD) to determine the content of each element. The petrographic analysis was performed with the microscope (LEICA DM2500P, LEICA, Germany) and the photometer (MSP200, J& M, Germany), following the procedures in the Chinese standards (GB/ T6948-2008 and GB/T8899-2013). About 2 g of the coal sample was mixed with shellac (at a volume ratio of 2:1) and heated to 50 °C to prepare a briquette, and then the briquette was polished. The polished briquette was analyzed by microscope for its maceral compositions, and the same briquette was used to analyze the vitrinite reflectance in oil.

matter content (or between 81% and 92% (by weight) of carbon in coals, with a maximum at 89% carbon) and then decreases. Hower et al. [11] studied the coking characteristics of a number of eastern Kentucky coals and found that the coking ability increased with increased vitrinite maximum reflectance. Miller [12] analyzed the effect of petrographic composition on coking and considered the vitrinite as the most critical component of coal for coking. Moreover, there are a large number of appraising methods to evaluate the coking property, using indices such as Roga Index (RI), Caking Index (GR.I.), Plastic Layer Index (Y) and other parameters. Meanwhile, there are a large number of inorganic elements in coal, formed kinds of inorganic oxides known as mineral compositions such as SiO2, TiO2, Al2O3, CaO, MgO, Na2O, K2O, P2O5, Mn3O4, SO3, Fe2O3, etc [10]. The minerals undergo a series of changes at high temperatures, and the slagging phenomenon occurred when the temperature reaches a certain value. Liu [13] studied the melting behavior of seventeen representative coal samples using SEM and XRD, and found that the ash fusion temperatures (AFTs) decreased with increasing Fe2O3 content. For the effect of the CaO content, the AFTs reached a minimum value when the CaO content approaches 30% and then increases once again. Shao et al. [14] reported phosphates together with the eutectics of Fe2O3 and SiO2 might play the most important role in bed agglomeration during sludge combustion by forming low melting point compounds with alkalis. Additionally, some publications indicated that increased P2O5 content enhances the development of lowmelting-point phases in the ash [15–17]. Base on much such research on ash compositions, there are some predictive and empirical indices introduced to evaluate the slagging tendency—such as Base/Acid ratio (B/A), Silica/Alumina ratio (SiO2/Al2O3), Silica ratio (G), Iron/Calcium ratio (Fe2O3/CaO), and Composite Index R. The present study made a relatively comprehensive analysis including proximate analysis, ultimate analysis and petrographic analysis of the coal quality characteristics for the five coals. The combustion experiment with muffle furnace and coal ash preparation experiment were carried out. In addition, SEM was used to analyze the morphology of the five coal coking samples. The composition of five coal ash and the ash fusion temperature were analyzed by XRF and triangular pyramid method. Based on the test and analysis of coal quality, an objective weight analysis method (Entropy Method) in the field of statistics [18] was used to analyze the effect weight of seven indices (C/H, FC/V, vitrinite content Vi, vitrinite average reflectance Rave, ash deformation temperature (DT), B/A and G) on the degree of agglomeration of coal in combustion process. Furthermore, combined with the combustion results of five coals in a muffle furnace, an empirical formula for the evaluation of a coal agglomeration in combustion process was obtained, and the corresponding evaluation limits were given.

2.1.2. Combustion experiment and agglomeration analysis The combustion experiment was conducted in a muffle furnace. Five coal samples were milled and sieved to about 0.2 mm in size, and each sample weighing 2 g was put in a crucible, then the crucible was placed directly into the high-temperature (1000 °C) muffle furnace for 7 min. After the combustion, the degree of agglomeration was evaluated according to the Char Residue Characteristic (CRC) value, which was used to evaluate the potential coking ability of coal in 8 ranks by the shape and strength of the coal char in the proximate analysis, just like the Free Swelling Index (FSI) in America. It is classified into weakly (Rank 1–2), medium (Rank 3–5), strongly (Rank 6–8). Rank 1–2 refers the coal char is still in powder without agglomeration. Rank 3–5 refers the coal char is agglomerated but can be crushed by finger press. Rank 6–8 refers the agglomerated coke with silver-white luster on its surface is formed, and it cannot be crushed by a finger press. Moreover, the micro-morphology of char was observed by scanning electron microscopy (SEM) (EVO18, produced by Carl Zeiss, Jena, German). X-ray diffraction (XRD) (Empyrean, produced by Panalytical, Netherlands) and X-ray fluorescence (XRF) (PW4400, produced by Panalytical, Netherlands) were used to analyze the mineral compositions and element contents. 2.1.3. Ash preparation and the melting characteristics analysis The ash of coal samples was prepared in a muffle furnace based on the Chinese standard GB/T212-2008. The specific process of ash preparation is the same as the determination of ash content in the proximate analysis, which has been introduced in Section 2.1.1. The elements of the ash were measured by XRF. Moreover, several commonly used indices for the prediction of slagging were evaluated, including the Base/Acid ratio (B/A), Silicon ratio (G), Silicon/Calcium ratio (SiO2/ CaO), and Iron/Calcium ratio (Fe2O3/CaO). Finally, The characteristic temperatures of coal ash fusion were measured by using the triangular pyramid method.

2. Experiments and methods 2.1. Experiments The experimental procedures involve sample preparation and coal analysis, combustion experiment and agglomeration analysis, ash preparation and the melting characteristics analysis. 2.1.1. Sample preparation and coal analysis Five coal samples which were frequently used as a source of power in the Heilongjiang Province were selected as the experimental materials. The proximate analysis of coals was measured by the Proximate Analyzer (5E-MAG6700, Changsha Kaiyuan Instruments, PRC). The proximate analysis was to measure the moisture, volatile matter, and ash content of the coal samples successively at different temperatures and residence time. The remaining part defaulted to the fixed carbon content. The coal sample was dried to constant quality at 105 °C–110 °C and the reduced mass was the moisture content. The coal sample was insulated from the air and heated for 7 min at 900 ± 10 °C, and the

2.2. Methods The entropy method firstly appeared in thermodynamics and was introduced into the information theory later by Shannon [19]. Nowadays, it has been widely used in engineering, economy, finance, etc. The entropy theory is an objective way for weight determination. Tian and Du [20] used the entropy method to evaluate the performance of mechanical products, and the results were in good agreement with the actual conditions. Fang [21] et al. used entropy method in the field of financial investment, which proved the simplicity and scientificity of entropy method in market state forecasting. Xu [22] et al. In the field of 380

Fuel 231 (2018) 379–386

L. Guo et al.

1) Establish the evaluation matrix based on original data. 2) Normalize the evaluation matrix. 3) Calculate the information entropy ej and the information utility value dj..

water conservancy engineering, established a multi-objective fuzzy optimization model of diversion criteria based on the entropy method, and the model has been applied in practical projects. In short, a large number of studies have demonstrated the reliability and practicality of the weight analysis by entropy method in various fields. This paper uses the entropy method to analyze the coal agglomeration, objectively obtained the influence weight of various factors on coal agglomeration. The entropy method is a weight analysis method for multi-objects and multi-index systems, and it corresponds exactly to the multi-coals and multi-indices of agglomeration studied in this paper. It is believed that the entropy method can deeply reflect the credit value of the index, and the index weight value given by the entropy has higher credibility than the general objective weight method [23].

n

ej = −(lnn)−1

pij =

(1)

yij n



yij

i=1

(2)

dj = 1−ej

(3)

It is prescribed that pij lnpij = 0(i = 1, 2, …, n;j = 1, 2, …m)

when

pij = 0,

4) Calculate the weight of the indices, wj .

wj =

dj n



dj (4)

j=1

3. Results and discussion As mentioned above, the agglomeration is a complex process, such that neither indicator alone can reflect the agglomeration of the coal comprehensively. Therefore, it would be best to build an accurate model by combining different coal properties in both organic and inorganic aspects. In this paper, according to the results of the combustion experiment, the factors that affect the agglomeration were analyzed from both organic indicators (C/H, FC/V, vitrinite content Vi, vitrinite average reflectance Rave), and inorganic indicators (B/A, G, and DT). 3.1. Combustion experiment results Fig. 1(a–e) shows the agglomeration of five coal samples during the 1000 °C crucible combustion experiment. Fig. 1 shows the degree of agglomeration of the five coal samples, where FY, ML, and MD obviously agglomerate. The surface of FY is the densest with silvery-white luster which means the severity agglomeration of FY. SX and BR are still in powder, and agglomeration of BR is the weakest. Fig. 2(a–e) is the SEM images of five coal agglomerates and shows that the FY has an observable bonding phenomenon because a large coke block formed. ML and MD have a certain degree of coking, due to the higher volatility, and the formation of the MD coke block is filled with a variety of pore structures, which may affect the degree of agglomeration [26]. SX is slightly coked, but not obvious and the BR has almost no adhesion. According to the CRC criteria, the CRC value of the five coal samples was obtained, as shown in Table 1. The CRC value of FY is 6. It is the highest among the five coal samples, indicating that the agglomeration ability is the strongest, BR has the lowest CRC value, which is 1, and the agglomeration ability is the weakest.

2.2.2. The principle of entropy method Entropy can measure the amount of valid information contained in the known data and determine the weight of the indicator [24]. The weight of the index is calculated by the entropy method. The weight is determined according to the degree of difference of each index value. When the difference of the evaluation index is large, the entropy is small, indicating that the effective information provided by the index is large and its weight corresponds to the larger index. The indicator value states that the larger the entropy, the corresponding weight will also be relatively small. When the value of an evaluation index is the same, the entropy is the largest, which means that the index basically provides no useful information, and its weight is zero [25]. The main steps to determine the weight using the entropy method are as follows [18]:

b) ML

pij lnpij

i=1

2.2.1. Establishment and normalization of the evaluation matrix The evaluation matrix is established by assuming the number of multi-indicator experiments is n, marked as I = {1, 2, …, n}, and the number of the indicators is m, marked as J = {1, 2, …, m}. The experimental value of the indicator j corresponding to experiment i is x ij (i = 1, 2, …, n; j = 1, 2…m) , and the matrix (x ij )n × m is called the evaluation matrix. Due to the differences in the dimensions and units of the different coal characteristic evaluation indicators, the direct addition of the indicators of different characteristics cannot correctly reflect the comprehensive results of different indicators. Therefore, before the calculation of the evaluation matrix, the normalization of the matrix is essential. There are several commonly used normalization methods including the range method, Z-score normalization method, and normalization by decimal scaling. In this paper, the range method was adopted for the matrix normalization. The principle is as follows: When the indicators require “the bigger, the better”, the upper-limit effect measurement was used: New data = (original data − minimum)/ (maximum-minimum). When the indicators require “the smaller, the better”, the lowerlimit effect measurement was used: New data = (maximum − original data)/(maximum-minimum). After normalization of the matrix, the normative matrix Y = (yij ) was obtained.

a) FY



c) MD

d) SX

Fig. 1. The degree of agglomeration of the coal samples. 381

e) BR

Fuel 231 (2018) 379–386

L. Guo et al.

a) FY

b) ML

c) MD

d) SX

e) BR Fig. 2. The SEM images of the coal agglomerations.

382

Fuel 231 (2018) 379–386

L. Guo et al.

Table 1 The CRC value of five coal samples.

Table 3 Ultimate analysis of coal samples.

Sample

FY

ML

MD

SX

BR

Sample

Car (%)

Har (%)

Oar (%)

Nar (%)

Sar (%)

Car/Har

CRC value

6

5

5

3

1

FY ML MD SX BR

50.62 49.48 68.89 44.06 39.91

3.20 3.40 4.61 3.18 2.98

4.39 1.12 10.65 4.90 10.29

0.62 0.32 0.48 0.30 0.28

0.29 0.27 0.51 0.27 0.49

15.80 14.56 14.94 13.85 13.38

Table 2 Proximate analysis of coal samples. Sample

Mad (%)

Aad (%)

Vad (%)

FCad (%)

Qnet,ad (MJ/kg)

FCad/Vad

FY ML MD SX BR

1.1 0.9 6.1 1.8 6.8

39.78 44.52 8.76 45.49 39.25

19.38 21.83 33.17 21.82 25.15

39.74 32.75 51.97 30.89 28.80

19.33 18.39 26.85 16.71 15.47

2.05 1.57 1.14 1.50 1.42

content of C in BR is the lowest. The H contents of the five coal samples are basically the same. Among these five elements, the carbon content has a positive effect on the coking ability just like the fixed carbon. When the carbon content is between 81% and 92%, the coking ability has a significant increase [9]. The hydrogen content also has a significant effect on the coking ability due to the hydrogen content decrease with an increase in rank [7]. The sulfur content has a complex effect on the coking ability. The sulfur in coal can be classified into two categories: organic sulfur and inorganic sulfur. The organic sulfur will reduce the fluidity of the colloid layer during coking process and weaken coking performance [29]. The inorganic sulfur will easily react with Fe to form low melting point eutectics which tend to slag [30]. However, the sulfur released from coal during combustion like H2S does not affect coking and slagging [29]. Compared with these elements, the effects of the oxygen and nitrogen content on the coking ability are insignificant. Therefore, the ratio of carbon to hydrogen C/H was selected among the ultimate analysis for predicting the coking ability. Fig. 4 shows the correlation between the C/H and CRC value. Comparing Fig. 3 with Fig. 4, the influence of C/H on CRC value is basically the same as that of FC/V on CRC value. The higher the content of C element, the stronger the coking ability of the coal sample, but when the C element content of the coal sample reaches a certain value, the influence of the C element content on the coking of the coal sample becomes smaller.

3.2. Proximate and ultimate analysis As listed in Table 2, the proximate analysis determines the content of moisture, volatile matter, fixed carbon, and ash. The fixed carbon content of MD is the highest (51.87%), and the fixed carbon content of BR is the lowest (28.80%). The volatile contents of the five coal samples are in the range of 20%–30%. The ash content of SX is as high as 45.49% while the ash content of MD is only 8.76%. Many investigations have studied the effects of the proximate variables of coal on its coking ability and reported that the coking ability decreased with the increased moisture content of the coal samples [27]. Fixed carbon can reflect the coal rank in the range of lignite and bituminous coal. Additionally, the coking of coal increases with the increase of the fixed carbon content [28]. In this study, the ratio of fixed carbon to volatile FC/V was selected among the proximate parameters for predicting the coking ability. Fig. 3 shows the correlation between FC/V and CRC value. The CRC value increases with the increase of FC/V, but the growth rate becomes smaller. The higher the fixed carbon content, the stronger the coking property of the coal sample. However, when the fixed carbon content of the coal sample reaches a certain value, the influence of the fixed carbon content on the coking of the coal sample becomes smaller. As listed in Table 3, the ultimate analysis determines the content of carbon, hydrogen, oxygen, nitrogen, and sulfur. The results of the ultimate analysis are consistent with the results of the proximate analysis in Table 2. The contents of C and H in MD are the highest while the

3.3. Petrography analysis The petrography analysis focuses on maceral compositions of coal, mainly including vitrinite, liptinite, and inertinite. The vitrinite is a component formed by the gelation of ancient plants due to swelling of the water, together with the biological and chemical reactions, in the presence of anaerobic bacteria. The vitrinite is the primary raw material of the coke. During the process of coke formation, non-volatile liquid components are formed on the particle surface. The presence of the

Fig. 3. The relationship between FC/V and CRC value.

Fig. 4. The relationship between C/H and CRC value. 383

Fuel 231 (2018) 379–386

L. Guo et al.

Table 4 The petrography analysis of five coal samples.

Table 5 Coal ash composition.

Sample

Vitrinite (%)

Inertinite (%)

Liptinite (%)

Mineral (%)

Rave

Sample

SiO2

Al2O3

TiO2

Fe2O3

K2O

CaO

MgO

Na2O

FY ML MD SX BR

67.2 59.8 88.8 58.2 59.1

11.6 6.2 1.2 7.2 13.9

0.0 0.0 1.0 0.6 0.0

21.2 34.0 9.0 34.1 27.0

0.97 0.8 0.5 0.81 0.56

FY ML MD SX BR

68.59 70.29 69.49 65.53 57.74

20.65 20.95 18.59 24.74 21.87

1.17 1.46 0.89 1.04 1.14

2.65 1.65 5.99 2.49 5.73

3.37 2.57 1.45 3.76 2.10

1.09 1.18 1.23 0.51 4.28

0.97 0.54 0.62 0.78 1.47

0.53 0.64 0.28 0.46 0.71

volatile liquid on the surface of the coal particles contributes to the formation of the coal coke. In the inertinite group, the carbon content is high, and the oxygen and hydrogen content is low. In the heating process, the inertinite group thermally decomposes but does not soften or generate the mesosphere, which is an inert component in the process of coke formation. The liptinite has a high hydrogen content, low decomposition temperature, high volatility, and low residual carbon. The coking ability of the liptinite is not good, but it has more gaseous and liquid products, which have a positive effect on improving the coking ability of the vitrinite, except that the amount of liptinite is small, and the improvement in the process is also limited. Meanwhile, the vitrinite average reflectance (R) is an excellent parameter for the coal rank measurement. Many studies found that the coking ability depended on the petrographic composition and the coal rank. The results in Table 4 show that vitrinite is the major maceral component of all the five coal samples, but there is a significant gap in the vitrinite content between the different coals, such that the maximum is 88.8% and the minimum is 58.2%. For all of the five coal samples, the content of inertinite is small, and the content of liptinite almost is zero. The distribution of the vitrinite average reflectance is 0.5–0.97. Combined the petrography analysis and the combustion experiment results, it is inferred that FY has the highest CRC value, which is 6, mainly due to its high vitrinite content and high vitrinite average reflectance. BR has the lowest vitrinite and the smallest vitrinite average reflectance. Thus the CRC value is only 1. Interestingly, even though the MD has the highest vitrinite content, its CRC value is 4, which means that the coking ability of MD is just medium. Most likely, this is because its vitrinite average reflectance is too small, as listed in Table 4, and it is just 0.56. This phenomenon also indicates that both the rave and vitrinite content have a significant effect on the coking ability. Among five coal samples, the FY has the highest grade of agglomeration, mainly due to its high content of vitrinite and the largest average vitrinite reflectance. Although the vitrinite content of the MD is the highest, the vitrinite average reflectivity is only 0.5, resulting in moderate agglomeration. The vitrinite content and vitrinite average reflectance are very small, so its agglomeration is the weakest.

layer will protect the inside metallic Al from oxidation because the metallic Al has a low melting temperature (around 600 °C) and it will easily lead to the slagging. The content of TiO2 in coal ash is low, generally not more than 5%. In the fly ash, the TiO2 always plays the role of increasing the ash fusion temperature. For Fe, the melting point of Fe2+ is much lower than Fe3+, so Fe is more likely to cause slagging in a reducing atmosphere [13]. The low melting temperature minerals in Fe, such as wustite (FeO), almandite (3FeO_Al2O3_3SiO2) and fayalite (2FeO_SiO2) readily melt and agglomerate with one another to form large particles, which are sticky in combustion process [33]. CaO itself has a high melting point (2610 °C), but it is an alkaline oxide. Therefore, the effect of CaO on the fusibility of coal ash is rather complicated and requires specific analysis of specific issues. Mg is chemically similar to calcium but in much lower concentrations than Ca [34]. K, Na are very important alkaline elements in coal ash and have a low melting point. They are easy to react with a variety of other substances to produce a low melting point material in the combustion process [35]. Therefore, these elements are the key factors that lead to a low melting point of the coal ash. Table 5 shows the eight main element compositions of the five coal samples as oxides. The main components of all five coal ashes are SiO2 and Al2O3 which account for more than 80% of the total. MD has the highest Fe2O3 content, SX has the highest K2O content, and BR has the highest CaO and MgO content. All of the five coal ashes have very low Na2O content. According to the data in Table 5, the slagging characteristics of five coal ashes were predicted using the four most commonly used discriminant indices, namely Base/acid ratio, Silica ratio, Silica/alumina ratio and Iron/calcium ratio, and each discriminant index is defined as follows:

Base/Acid ratio, or B / A = (Fe2 O3 + CaO + MgO + Na2 O+ K2O)/(SiO2 + Al2O3 + TiO2) Silica ratio, G = 100·SiO2 /(SiO2 + Fe2 O3 + CaO + MgO)

Silica/Alumina ratio = SiO2 /Al2O3

3.4. Ash melting characteristics analysis

Iron/Calcium ratio = Fe2 O3 /CaO

The ash melting characteristics are determined by the ash-forming elements, which mainly include Si, Al, Ti, Fe, K, Ca, Mg, Na and so on. Many papers have analyzed the effect of these elements on the ash melting characteristics. Si is always the element with the highest content and is present as silica (SiO2) or various silicate minerals. They are relatively inert under combustion conditions, and the ash formed from them has mostly high melting properties [31]. However, one important exception is the silicates that can easily react with the alkali elements (K, Na and Ca), and form low melting eutectics. The most important reactions include 2KOH + SiO2 → K2SiO3 + H2O, Na2O + SiO2 + H2O → Na2SiO3 + H2O. These reactions modify the silicate into low melting eutectics and reduce the evaporation of alkali elements, which leads to stronger slagging [32]. Al is the second largest element, in the combustion process. The metallic aluminum will be oxidized to alumina (Al2O3), and the characteristics of Al2O3 are just like SiO2. It has a high melting point but can react with alkali metals to form a low melting point eutectic. At the same time, in the oxidation process, the Al2O3

The Base/Acid ratio or B/A is a comprehensive index which contains all the eight main components in coal ash. The basic oxides will reduce the ash fusion temperature and improve the ash fluidity, but the acidic oxides have exactly the opposite characteristics of the basic oxides. Therefore, the tendency of slagging increases with higher Base/ Acid ratio. Silica ratio or G mainly focuses on the silica content which is the highest content in coal ash. The ash fusion temperatures increase with silica ratio value. Silica ratio is one of the important indices for coal slagging evaluation, but it does not consider the influence of Al2O3 on slagging. Thus, Silica/Alumina ratio is considered. Al2O3 is the ‘support skeleton’ that inhibits the deformation of the ash, while SiO2 tends to form low melting point eutectics with basic oxides. The increase of the Silica/Alumina ratio promotes slagging. Iron/Calcium ratio was used firstly to evaluate bituminous coal ash slagging. It has an advantage in evaluating slagging caused by the iron content. Based on the discriminant limits in Table 6, using the data of coal 384

Fuel 231 (2018) 379–386

L. Guo et al.

results are shown in Table 8. The ash melting temperature is consistent with the discriminant results of B/A and G.

Table 6 The limits of the discriminant index for slagging. Index

Tendency of slagging

B/A G SiO2/Al2O3 Fe2O3/CaO

Weakly

Medium

Strongly

< 0.206 > 78.8 < 1.87 Out of 0.3–3

0.206–0.4 78.8–66.1 1.87–2.65 0.3–3

> 0.4 < 66.1 > 2.65 Near 1

3.5. The index of agglomeration Through the above comprehensive analysis of the coal quality characteristics, the factors that affect the coal agglomeration were understood more clearly. To have a unified evaluation standard, seven important indicators regarding the coal quality were selected. The agglomeration phenomenon includes both of the coking and the slagging aspects. In the aspect of coking, according to the ultimate analysis and the proximate analysis, C and H are the main elements of the organic components in coal. They have a great effect on coking ability. FC is the base material for coking, and V has great influence on the coking intensity. In the petrography analysis, the vitrinite Vi is the main component for coking, and the vitrinite reflectivity Rave is the most effective indicator reflecting the degree of coal metamorphism. As for the slagging index, DT can be experimentally measured, and the reflection on ash fusion characteristics is intuitive and reliable. B/A ratio contains the most important eight components in coal ash and is relatively comprehensive. G ratio focuses on the content of SiO2 which is the highest content in coal ash, and several components that can react with SiO2 to generate low melting point eutectics. G ratio is a very representative slagging discriminant index. Based on the above analysis, C/H, FC/V, vitrinite content Vi, vitrinite average reflectivity Rave, ash deformation temperature (DT), B/A and G are very representative indicators of agglomeration. Therefore, the present study finally chooses them for obtaining a comprehensive discriminant index. Moreover, the comprehensive coal quality index to evaluate the characteristics of coal agglomeration is put forward through the mathematical methods such as the entropy method and linear fitting. The values of the seven major indicators and the agglomeration degree of the five coal samples are summarized in Table 9. The data in Table 9 and the calculation process of the entropy method were introduced. First, the evaluation matrix was set up as shown below:

Table 7 Comparison of four slagging discriminant indices. Sample

B/A

FY ML MD SX BR

0.10 0.07 0.11 0.09 0.18

G (weakly) (weakly) (weakly) (weakly) (weakly)

93.57 95.43 89.86 94.53 83.42

(weakly) (weakly) (weakly) (weakly) (weakly)

SiO2/Al2O3

Fe2O3/CaO

3.32 3.35 3.74 2.65 2.64

2.42 1.40 4.87 4.87 1.34

(strongly) (strongly) (strongly) (Medium) (Medium)

(Medium) (strongly) (weakly) (weakly) (strongly)

Table 8 The ash melting temperature of coal samples. Sample

DT

ST

HT

FT

FY ML MD SX BR

1500 °C 1500 °C 1390 °C 1453 °C 1390 °C

> 1500 °C > 1500 °C > 1500 °C > 1500 °C 1418 °C

> 1500 °C > 1500 °C > 1500 °C > 1500 °C 1490 °C

> 1500 °C > 1500 °C > 1500 °C > 1500 °C > 1500 °C

Table 9 Seven major indicators and the agglomeration levels of five coal samples. Sample

C/H

FC/V

Vi

Rave

DT

B/A

G

Agglomeration degree

FY ML MD SX BR

15.80 14.56 14.94 13.85 13.38

2.05 1.50 1.57 1.42 1.14

67.2 59.8 88.8 58.2 59.1

0.97 0.80 0.50 0.81 0.56

> 1500 > 1500 1410 1453 1390

0.10 0.07 0.11 0.09 0.18

93.57 95.43 89.86 94.53 83.42

Strongly Medium Strongly Weakly Medium

⎧15.80 ⎪14.94 X = 13.38 ⎨14.56 ⎪ ⎩13.85

Table 10 The agglomeration evaluation results of the coal samples. Sample

C/H

FC/V

Vi

Rave

DT

B/A

G

Evaluation value

FY ML MD SX BR

15.80 14.56 14.94 13.85 13.38

2.05 1.50 1.57 1.42 1.14

67.2 59.8 88.8 58.2 59.1

0.97 0.8 0.5 0.81 0.56

1500 1350 1410 1453 1390

0.10 0.07 0.11 0.09 0.18

93.57 95.43 89.86 94.53 83.42

0.37 0.18 0.45 0.11 0.31

67.2 88.8 59.1 59.8 58.2

0.97 0.5 0.56 0.8 0.81

1500 1410 1390 1350 1453

0.10 0.11 0.18 0.07 0.09

93.57 ⎫ 89.86 ⎪ 83.42 95.43 ⎬ ⎪ 94.53 ⎭

(5)

Then, the range method was used to normalize the formula X, and the normalization matrix was obtained as:

1 0.29 1 0 0.27 0.15 ⎫ ⎧ 1 0 0.6 0.36 0.46 ⎪ ⎪ 0.64 0.47 1 Y= 0 0 0.03 0.13 0.73 1 1 ⎨ 0.49 0.39 0.05 0.64 1 0 0 ⎬ ⎪ ⎪ ⎩ 0.19 0.31 0 0.66 0.31 0.18 0.07 ⎭

(6)

Finally, based on the formula (1)–(3), the weights of C/H, FC/V, Vi, Rave, DT, B/A and G on the agglomeration were calculated:

Table 11 The limits of the discriminant index.

Rx

2.05 1.57 1.14 1.50 1.42

Weakly

Medium

Strongly

≤0.15

0.15–0.35

≥0.35

wk = (0.11 0.10 0.26 0.12 0.09 0.14 0.18)

(7)

The effect degree of seven indicators on coal agglomeration can be discerned from the calculated weight value. The value of vitrinite content Vi is significantly higher than other indicators. The other three indicators in the organic aspect C/H, FC/V and vitrinite average reflectance Rave almost have the same effect weight. In the inorganic aspect, the weight of G is greater than B/A and DT, which indicated that G is the best indicator of the inorganic indicators to reflect the agglomeration characteristics of coal. Based on the weight values of the seven indices, the expression of the comprehensive coal quality index of coal agglomeration characteristics was obtained by using a linear fitting:

ash composition in Table 5, the slagging conditions of the five coal samples were evaluated. The results are shown in Table 7. There is a certain difference between the four discriminant indices. Specifically, the discriminant results of B/A and G are more consistent, and the comprehensive reliability is relatively higher. In addition, the paper also uses the ash melting point instrument and experimentally measured the ash melting temperature of five kinds of coal samples. The 385

Fuel 231 (2018) 379–386

L. Guo et al.

Acknowledgment

Rx = 0.11C(C / H ) + 0.1C(FC / V ) + 0.26C(V ) + 0.12C(Rave) + 0.09C(DT ) + 0.14C(B / A) + 0.18C(G)

(8)

This work was supported by Postdoctoral Scientific Research Developmental Fund of Heilongjiang Province (Grant No.: LBHQ16088).

where C(C/H), C(FC/V), C(V), C(Rave), C(DT), C(B/A), C(G) were the normative value of seven indicators. The relationships between the normative value and the original value were as follows:

References

C(C / H ) = 0.41C / H −5.53

(9)

C(FC / V ) = 1.10FC / V −1.25

(10)

C(V ) = 0.03V −1.90

(11)

C(Rf ) = 2.13Rf −1.90

(12)

C(DT ) = −0.0067DT + 10

(13)

C(B / A) = 9.10B / A−0.64

(14)

C(G) = −0.08G + 7.96

(15)

[1] Datta S, Sarkar P, Chavan PD, et al. Agglomeration behaviour of high ash Indian coals in fluidized bed gasification pilot plant. Appl Therm Eng 2015;86:222–8. [2] Ma Z, Iman F, Lu P, et al. A comprehensive slagging and fouling prediction tool for coal-fired boilers and its validation/application. Fuel Process Technol 2007;88(11–12):1035–43. [3] Meng F, Gupta S, French D, et al. Characterization of microstructure and strength of coke particles and their dependence on coal properties. Powder Technol 2017;320:249–56. [4] Dai BQ, Low F, De Girolamo A, et al. Characteristics of ash deposits in a pulverized lignite coal-fired boiler and the mass flow of major ash-forming inorganic elements. Energy Fuels 2013;27(10):6198–211. [5] Loison R, Foch P, Boyer A. Coke: quality and production. Elsevier; 2014. [6] Dıez MA, Alvarez R, Barriocanal C. Coal for metallurgical coke production: predictions of coke quality and future requirements for cokemaking. Int J Coal Geol 2002;50(1):389–412. [7] Zhang L, Liu W, Men D. Preparation and coking properties of coal maceral concentrates. Int J Min Sci Technol 2014;24(1):93–8. [8] Zhang Q, Wu X, Feng A, et al. Prediction of coke quality at Baosteel. Fuel Process Technol 2004;86(1):1–11. [9] Speight JG. Handbook of coal analysis. John Wiley & Sons; 2015. [10] Ghosh A, Chatterjee A. Iron making and steelmaking: theory and practice. PHI Learning Pvt. Ltd.; 2008. [11] Hower JC, Wild GD, Eble CF. Coal quality considerations in the future production of metallurgical coal in eastern Kentucky. CQ-J Coal Qual 1994;13(1):1–6. [12] Miller BG. Coal energy systems. Academic Press; 2004. [13] Liu B, He Q, Jiang Z, et al. Relationship between coal ash composition and ash fusion temperatures. Fuel 2013;105:293–300. [14] Shao J, Lee DH, Yan R, et al. Agglomeration characteristics of sludge combustion in a bench-scale fluidized bed combustor. Energy Fuels 2007;21(5):2608–14. [15] Wang L, Skjevrak G, Hustad JE, et al. Sintering characteristics of sewage sludge ashes at elevated temperatures. Fuel Process Technol 2012;96:88–97. [16] Pronobis M. Evaluation of the influence of biomass co-combustion on boiler furnace slagging by means of fusibility correlations. Biomass Bioenergy 2005;28(4):375–83. [17] Lindström E, Sandström M, Boström D, et al. Slagging characteristics during combustion of cereal grains rich in phosphorus. Energy Fuels 2007;21(2):710–7. [18] Zou ZH, Yi Y, Sun JN. Entropy method for determination of weight of evaluating indicators in fuzzy synthetic evaluation for water quality assessment. J Environ Sci 2006;18(5):1020–3. [19] Shannon CE. A mathematical theory of communication. ACM SIGMOBILE Mobile Comput Commun Rev 2001;5(1):3–55. [20] Tian QH, Du YX. Study of performance evaluation for mechanical products based on entropy fuzzy comprehensive review. China Manuf Inf 2004;33(3):97–9. [21] Fang DC, Liu GL, Zhang LP. The application of information entropy in investment decision. Value Eng 2004;2:115–7. [22] Xu SQ, Hu ZG, Liu Q, et al. Multi-objective decision analysis of diversion standards based on entropy. China Rural Water Hydropower 2004;8:45–7. [23] Zheng F, zhi Wang Q. Application of matlab and entropy in comprehensive evaluation[C]//2010 National Vocational Education of Communications and Information Technology Conference (2010 NVCIC). 2010. [24] Gray RM. Entropy and information theory. Springer Science & Business Media; 2011. [25] Qiu W. Management decision and applied entropy. Beijing: China Machine Press; 2002. [26] Nishioka K, Yoshida S. Strength estimation of coke as porous material. Trans Iron Steel Inst Jpn 1983;23(5):387–92. [27] Chelgani SC, Hower JC, Hart B. Estimation of free-swelling index based on coal analysis using multivariable regression and artificial neural network. Fuel Process Technol 2011;92(3):349–55. [28] Ryan B, Leeder R, Price JT, et al. The effect of coal preparation on the quality of clean coal and coke. British Columbia Geological Survey Branch: Geological Fieldwork; 1998. p. 247–75. [29] Mochizuki Y, Ono Y, Uebo K, et al. The fate of sulfur in coal during carbonization and its effect on coal fluidity. Int J Coal Geol 2013;120:50–6. [30] Jing A. Investigation on assessing slagging index of coal ash both at home and abroad. Power Syst Eng 1994;10(2):1–18. [31] Contreras ML, Arostegui JM, Armesto L. Arsenic interactions during co-combustion processes based on thermodynamic equilibrium calculations. Fuel 2009;88(3):539–46. [32] Zevenhoven M, Yrjas P, Hupa M. Ash-forming matter and ash-related problems. Handbook of Combustion; 2010. [33] Wu X, Zhang X, Yan K, et al. Ash deposition and slagging behavior of Chinese Xinjiang high-alkali coal in 3 MW th pilot-scale combustion test. Fuel 2016;181:1191–202. [34] Vassilev SV, Vassileva CG, Baxter D, Andersen LK. Relationships between chemical and mineral composition of coal and their potential applications as genetic indicators. Part 1. Chemical characteristics. Geol Balcanica 2010;39(3):21–41. [35] Lu T, Li KZ, Zhang R, et al. Addition of ash to prevent agglomeration during catalytic coal gasification in a pressurized fluidized bed. Fuel Process Technol 2015;134:414–23.

Substituting (9)–(15) into (8), the final expression of Rx was obtained:

Rx = 0.0451C / H + 0.11FC / V + 0.0078V + 0.2556Rf −0.000603DT + 1.274B / A−0.0144 G+ 0.7879

(16)

Using the formula (16), the agglomeration evaluation values of the five coal samples were calculated. Comparing the degrees of agglomeration of the five coal samples in Table 9 with the evaluation values in Table 10, the limits of the discriminant index can be roughly obtained as shown in Table 11. The entropy method is a statistical weight analysis method. The number of samples does affect the accuracy of the weight evaluation result, but it may not influence the principle correctness of the method. This paper is only laboratory-scale research whose primary aim is to consider a new method for the prediction of coal agglomeration. The evaluation results will be more accurate with more coal samples. 4. Conclusions This study investigated the effects of various coal quality indices on the agglomeration characteristics during the coal combustion process. There was a positive correlation between the C/H, FC/V and the CRC value of coal. The vitrinite was the most important coal component that caused coal agglomeration, such that the higher the content was, the stronger the coal agglomeration was. However, besides the content of vitrinite, the vitrinite average reflectivity also had a significant effect on the coal agglomeration. The agglomeration ability of the vitrinite with a lower average reflectivity will be reduced. Among the various indices of the minerals, B/A and G are more comprehensive, and the predicted result is relatively accurate. Moreover, seven important indices were selected to represent the coal characteristics, which include C/H, FC/V, vitrinite content Vi, vitrinite average reflectance Rave, B/A and G. Their influence weights on the coal agglomeration are creatively analyzed by the entropy method. It was found that in terms of organic components, the content of vitrinite has the greatest influence on the agglomeration. In terms of inorganic components, G is a more reliable index. According to the calculated weight of the seven coal quality indices, combined with the crucible combustion test results, a comprehensive coal quality index to evaluate the characteristics of coal agglomeration was proposed, and the corresponding empirical formula and discriminant limits were obtained. Compared with the previous research, the index is more comprehensive and reliable and has strong engineering application value. Due to the limited amount of data in this paper, some conclusions may have some deviations, but the conclusion is still useful for industrial applications. 386