- Email: [email protected]
Application of chemometrics to study the kinetics of coal pyrolysis: A novel approach
Fuel 90 (2011) 3299–3305
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Application of chemometrics to study the kinetics of coal pyrolysis: A novel approach Puja Khare ⇑, B.P. Baruah, P.G. Rao Coal Chemistry Division, North-East Institute of Science and Technology (CSIR-NEIST), Jorhat, 785 006 Assam, India
a r t i c l e
i n f o
Article history: Received 26 October 2010 Received in revised form 17 March 2011 Accepted 19 May 2011 Available online 1 June 2011 Keywords: Chemometric Principal component analysis Hierarchical clustering analysis Coals
a b s t r a c t In present investigation, chemometric tools, principal component analysis (PCA) and Hierarchical clustering analysis (HCA) are used to get the linkage between the coal properties and kinetics of pyrolysis. Thermo gravimetric analysis (TGA) of 10 perhydrous Indian coals was done. Devolatilization of these coals showed five independent reactions. Kinetic parameters were calculated for individual reaction. Activation energy and weight loss of each reaction has been analyzed as a function of coal properties (moisture, volatile matter, ash, fixed carbon, carbon, hydrogen, nitrogen and sulfur). By applying chemometric, was extracted information about the linkage between activation energies of each reaction and coal properties. The mathematical treatment of data has provided conclusions on properties of coal and kinetic parameters. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Pyrolysis is an important intermediate stage in coal gasification, combustion, and liquefaction, and also considered as a simple and effective method for a clean conversion of coal [1]. The application of thermal analysis to study the behavior of fossil fuels during pyrolysis has gained much industrial importance [2]. Thermal analysis characterizes the physical and chemical properties of substances, depending on the temperature at defined heating rate (dynamics measurement) or on the time at a constant temperature (static measurement) [3]. In recent years, the application of thermal analysis techniques to study the combustion and pyrolysis behavior of fuels has gained a wide acceptance among research workers [4–6]. Kinetics of devolatilization is useful for accurate estimation of the reactivity of fossil fuels during their utilization and conversion processes (gasification, liquefaction). Devolatilization of solid fuel occurs by processes like dehydration, primary and secondary devolatilization and thermal degradation. Accurate knowledge of the devolatilization processes is necessary to develop predictive models for coal conversion processes. Application of chemometrics to TGA data at different steps of the devolatilization is of interest and useful. Chemometrics is a chemical discipline that uses mathematics, statistics and formal logic: (1) To design or to select optimal experimental procedures, (2) To provide maximum relevant chemical information by analyzing chemical data, and (3) To obtain knowledge about the chemical systems [7]. ⇑ Corresponding author. Fax: +91 (0) 376 2370011. E-mail address: [email protected] (P. Khare). 0016-2361/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2011.05.017
In present investigation, perhydrous Indian coals, which are suitable for conversion processes like liquefaction and coke making used for thermo gravimetric analysis. The kinetic parameters extracted from TGA are combined with differing chemometric tools such as correlation, principal component analysis (PCA) and cluster analyses. A general overview of the application of chemometrics, particularly in PCA techniques and soft modeling to TGA data is presented in the paper giving special attention to the more recent contribution on fuel conversion processes. The database can be interpreted in many ways, we have highlighted the usefulness of PCA in order to 1. Develop a relationship between the coal properties and kinetic parameters obtained from TGA. 2. Suitability of coal for particular conversion process i.e. by combining the physico-chemical properties of coal with that of the kinetic parameters obtained from TGA.
2. Materials and methods The freshly mined low rank coal samples from different coalfields of northeastern India were used in this study. The air-dried samples were ground to 0.211 mm before use. The proximate, ultimate and sulfur analyses were done by using a TGA 701 (Leco), True spec elemental analyzer (Leco) and S 144 DR sulfur determinator (Leco). Physico-chemical characteristics of these coals are summarized in Table 1. 2.1. TGA and DTG Differential thermal analysis was used to determine the pyrolysis behavior of perhydrous coals. Experiments were carried out in a
3300
P. Khare et al. / Fuel 90 (2011) 3299–3305
Table 1 Physico-chemical characteristics of sub-bituminous coals, and weight loss (%) for five reactions during pyrolysis.
M Ash VM FC C H N S H/C W1 W2 W3 W4 W5
C1
C2
C3
C4
C5
C6
C7
C8
C9
C10
2.9 20 35.6 41.5 74.6 7.5 1.2 2.9 1.21 0.22 2.82 19.36 8.63 16.24
3.1 11.5 41.4 44 64.1 5.72 1.4 2.9 1.07 0.39 7.33 18.97 10.85 18.98
1.5 11.5 40.5 46.5 64.3 6.07 1.1 4.46 1.13 0.15 3.4 20.27 9.87 18.91
2.7 12 35.5 49.8 70.1 6.1 1.3 3.98 1.04 0.21 2.67 19.2 11.2 20.6
2.9 7.1 42.7 47.3 68.6 5.5 1.1 3 0.96 0.25 2.88 19.67 10.3 20.1
3.4 11.9 37.2 47.5 70.8 4.7 1.1 4 0.80 0.32 6.4 18.2 11.4 19.2
2.8 17.6 36 45.6 71.4 2.7 1.1 4.5 0.45 0.27 2.67 19.01 8.6 19.8
2.1 14.4 38.1 45.4 81.8 5.8 1.4 2.2 0.85 0.25 2.1 18.1 8.2 18.7
3.2 3.9 40.4 52.5 75.9 5.3 1.1 2 0.84 0.26 2.65 18.3 9.99 17.2
2.8 16.3 36.1 44.8 82.3 5.9 1.4 2.4 0.86 0.31 1.92 17.2 10.2 18.2
VM: volatile matter, FC: fixed carbon; W1-W5: Weight loss during the reactions 1–5, respectively.
Leco TGA 701 thermal analysis system with 0.5 g each of coal samples in a stream of nitrogen with a flow rate of 40 ml/min and linear heating rate of 10 °C/min. Several preliminary thermo gravimetric runs were performed in order to investigate the sample mass influence on the devolatilization behavior. Variation in mass has negligible effect on the devolatilization of coal. 2.2. Calculation of kinetic parameters The kinetic parameters, activation energies and pre exponential factors of coal pyrolysis were determined by the integral method by applying Arrhenius equation. Arrhenius equation can be expressed as
dx=dt ¼ AexpðE=RTÞð1 xÞ
ð1Þ
where A is pre-exponential factor, E activation energy, T temperature, t time, X weight loss fraction or decomposition during pyrolysis and which can be calculated by
x¼
W0 Wt W0 Wf
ð2Þ
where W0 is the original mass of the test sample, Wt is the mass at time t or T and Wf is the final mass at the end of pyrolysis. The constant heating rate during pyrolysis is H = dT/dt for H being the heating rate. Rearranging the Eq. (1) and on integration gives
In½lnð1 xÞ=T2 ¼ In½AR=HEð1 2RT=EÞ E=RT
ð3Þ
The expression ln{AR[1-(2RT/E)]/HE in Eq. (3) is essentially constant for most of the values of E and temperature range of the pyrolysis. By plotting the left side of Eq. (3) against 1/T, a straight line is obtained indicating the process to be of first order reaction. From the slope, E/R, the activation energy E can be determined.
PCA technique extracts the Eigen values and Eigen vectors from the covariance matrix of original variables. It allows finding out association between variables, thus, reducing the dimensionality of the data set. The Eigen values of PCs are the measure of their associated variance, the participation of the original variables in the PCs is given by the loading, and the individual transformed observations are called scores [9,10]. PCA was performed on normalized (z-scale transformation) variables after sorting out the highly correlated variables from the data sets. The Bartlett’s sphericity test was applied to the correlation matrix of variables for assessing the adequacy of PCA [11]. PCs with Eigen value greater than 1 were considered. Here, PCA was performed with a view to establish linkage between the coal properties and kinetic parameters extracted from TGA. To validate the model result, correlation were also drawn to calculate theoretical activation energy. Cluster analysis is an exploratory multivariate method that can be used to describe the relationships among variables. Several mathematical criteria can be used to examine the similarity (or difference/distance) between variables and cases. For Hierarchical clustering analysis (HCA), the Ward’s method was used to get cleaner picture of clusters. This method is distinct from all other methods because it uses an analysis of variance approach to evaluate the distances between clusters. This method attempts to minimize the sum of square (SS) of any two clusters that can be formed each step. This method is regarded as a very efficient one. The joining or tree clustering method uses the dissimilarities (similarities) or distances between objects when using the clusters. Similarities (distance) are a set of rules that serves as criteria for grouping or separating items elucidean distance is chosen for the analysis.
3. Results 3.1. Thermal behavior and kinetics of coal pyrolysis
2.3. Data treatment and chemometric analysis Statistical treatment of data including correlation analysis, and principal component analysis (PCA) and Hierarchical clustering analysis (HCA) were performed using SPSS 15 statistical software. Principal component analysis (PCA) techniques have been widely applied in the treatment of datasets of such a high complexity [8]. PCA provides a new set of orthogonal variables, the principal components (PCs), generated so that each PC accounts for the maximum possible amount of variance contained in the original dataset. The PCs are obtained as linear combination of the original descriptors and are aligned along the directions of covariance of the data [9].
The curves of thermal analysis (TGA and DTG) of one of the coal samples are shown in Fig. 1. It is evident from the shape of the curves that pyrolysis proceeds through five different temperature regions. Hence devolatilization may occur via five reactions i.e. dehydration (reaction 1), primary devolatilization /thermal desorption of gases (reaction 2), thermal degradation (reaction 3), depolymerization/secondary devolatilization (reaction 4), and dehydrogenation/condensation (reaction 5) [12,13]. The activation energies are determined for each process reaction by applying first order Arrhenius equation on each independent reaction (calculation is given in the materials and methods). Activation energy (Ea) with standard deviation and r2 for each linear regression is shown for
3301
P. Khare et al. / Fuel 90 (2011) 3299–3305
Fig. 1. TGA (wt loss) and DTG (Rate of wt loss) curves for one of the perhydrous coals.
Activation energies of thermal degradation (reaction 3) vary from 141 to 157 kJ/mol, which are related to the plastic properties, swelling of coal by completion of the softening process, coal melts forming metaplast and re-solidifies as char. During thermal degradation, coal has undergone an intense degradation of its structure and greatest mass loss takes place due to degradation of carbonaceous matrix and the evolution of relatively high molecular weight species.
each independent reaction in Table 2. The activation energies and the weight loss of five reactions are reported as Ea1, Ea2, Ea3, Ea4 and Ea5 and W1, W2, W3, W4 and W5 respectively in the text. Analysis of variance (ANOVA) is applied on the activation energies of each set of reactions. Results of ANOVA show that activation energies are significantly different from each other (p < 0.01–0.001) which indicates that reactions are independent to each other. The activation energies of secondary devolatilization (reaction 4) range from 165 to 190 kJ/mol, which are higher than previous regions. In this region, formation of secondary char as the main pyrolysis products occurs due to re-solidification and secondary devolatilization processes, which are endothermic in nature [14]. Hence, the activation energies are higher in this region.
Activation energies for liberation of water of these coals varied between 92.0 and 104.2 kJ/mol. The possible reaction mechanism is shown in reaction I. The activation energies of primary devolatilization (reaction 2) range from 120.7 to 134.7 kJ/mol. During primary devolatilization, the coals undergo softening with the release of light species [14]. Thermal decomposition of the more labile structure of the coal matrix also takes place. In this stage, the volatiles (liquids and gases) are the main pyrolysis products.
The activation energies of dehydrogenation/condensation range between 113 and 124 kJ/moles. During reaction 5, formation of coke takes place with the evolution of hydrogen gas. H2 gas evolves
Table 2 Activation energies (kJ/mol) of sub-bituminous coals for five reactions during pyrolysis. Reaction-1
C1 C2 C3 C4 C5 C6 C7 C8 C9 C10
Reaction-2 2
Ea1
Sd
R
99.9 105 104 97 98 101 97 94 97 92
0.36 0.25 0.25 0.42 0.30 0.30 0.31 0.15 0.36 0.41
0.99 0.995 0.998 0.996 0.997 0.992 0.995 0.994 0.992 0.994
Reaction-3 2
Reaction-4 2
Ea2
Sd
R
Ea3
Sd
R
129 135 140 131 133 121 130 123 125 120
0.15 0.10 0.15 0.16 0.10 0.15 0.13 0.15 0.25 0.15
0.998 0.992 0.998 0.995 0.995 0.998 0.997 0.998 0.996 0.998
149.7 149.9 152 158 141 150 160 157 145 160
0.30 0.35 0.30 0.31 0.42 0.31 0.31 0.31 0.31 0.25
0.998 0.997 0.999 0.994 0.994 0.995 0.998 0.998 0.998 0.996
Sd; standard deviation (±), R2: regression coefficient.
Reaction-5 2
Ea4
Sd
R
Ea5
Sd
R2
165.2 183 170.5 169 185 186 190 183 184 183
0.20 0.25 0.41 0.20 0.35 0.25 0.41 0.25 0.25 0.45
0.996 0.998 0.999 0.998 0.998 0.999 0.997 0.998 0.999 0.998
118 112 118 116 121 120 124 121 121 122
0.20 0.21 0.20 0.20 0.42 0.20 0.36 0.20 0.36 0.20
0.1 0.999 0.999 0.999 0.999 0.998 0.998 0.998 0.998 0.999
P. Khare et al. / Fuel 90 (2011) 3299–3305
1.00 1.00 0.32 1.00 0.40 0.30 1.00 0.08 0.57 0.09 1.00 0.12 0.45 0.47 0.18 1.00 0.33 0.65* 0.18 0.25 0.33
3.2. Chemometric analysis 3.2.1. Correlation Correlation analysis provide primary indicator of relation between the coal properties and kinetic parameters. Correlation analysis of all the data set show that the ash content of coals is negatively correlated with VM and carbon content (Table 3). Carbon is highly correlated with the sulfur content, while hydrogen content of coal was highly correlated with H/C ratio. Significant correlations (r = 0.63–0.92) are observed between coal properties and activation energies in different reactions. The properties of these coals moderately correlate with the weight loss rate in different reactions (r = 0.52–0.66). These correlation analysis provide preliminary idea about the relationship between coal properties and their kinetics during pyrolysis.
1.00 0.69* 0.43 0.27 0.64* 0.79* 0.24 0.28 0.21 0.20 1.00 0.74* 0.81* 0.28 0.07 0.32 0.22 0.05 0.16 0.15 0.66* 0.27 0.09 0.05 0.16 0.45** 0.55** 0.13
1.00 0.24 0.47 0.03 0.20 0.22 0.23 0.44 0.47** 0.74* 0.26 0.21 0.20 0.36 0.29 0.30 0.03
1.00 0.05 0.27 0.37 0.02 0.24 0.15 0.09 0.24 0.20 0.15 0.20 0.15 0.24 0.45** 0.29
1.00 0.10 0.38 0.63* 0.29 0.88* 0.83* 0.37 0.16 0.55** 0.13 0.67* 0.43 0.36 0.20
1.00 0.36 0.40 0.92* 0.11 0.11 0.23 0.78* 0.53* 0.24 0.07 0.07 0.23 0.64⁄
1.00 0.47** 0.21 0.31 0.25 0.49** 0.08 0.39 0.32 0.03 0.06 0.05 0.16
1.00 0.15 0.45 0.48** 0.26 0.15 0.03 0.26 0.23 0.74* 0.11 0.52**
1.00 0.45 0.43 0.36 0.81* 0.73* 0.18 0.20 0.09 0.08 0.54**
1.00 0.25 0.39 0.47 0.19 0.65* 0.28 0.10 0.10
1.00 0.03 0.17 0.12 0.30 0.49** 0.26 0.26
1.00 0.51 0.53** 0.14 0.47 0.38 0.44
W1 Ea1
P < 0.001. ** P < 0.01.
*
N H C FC VM Ash M
by breaking of C–H bonds with the formation of stronger bonds between the carbon atoms. This process is followed by condensation and contraction within the carbon hexagonal planes [15]. Activation energies of dehydrogenation/condensation are lower than previous stage, which may be due to exothermic contraction reaction [16].
1.00 0.16 0.08 0.15 0.06 0.21 0.08 0.24 0.24 0.06 0.46 0.31 0.39 0.03 0.66* 0.33 0.62* 0.53** 0.13
Table 3 Correlation among the coal properties, activation energies and wt loss.
M Ash VM FC C H N S H/C Ea1 Ea2 Ea3 Ea4 Ea5 W1 W2 W3 W4 W5
S
H/C
Ea2
Ea3
Ea4
Ea5
W2
W3
W4
W5
3302
3.2.2. Hierarchical clustering analysis HCA Kinetic energy of individual pyrolysis reactions is taken as major kinetic parameters for cluster analysis, which provides information about the rate of the individual reaction during pyrolysis. Kinetic energy of a reaction of devolatilization depends upon heating rate, atmosphere of the pyrolysis and the coal characteristics. Heating rate and pyrolysis atmosphere was same during thermal analysis of all the coal samples. Variations in the kinetic parameters of individual coal were due to their physicochemical characteristics. Hence, for HCA physicochemical parameters such as ash, moisture, volatile matter, fixed carbon, carbon, hydrogen, nitrogen, sulfur and H/C of coal were chosen. Hierarchical cluster analysis is applied on kinetics parameters and physico-chemical properties of coals (Fig. 2). Here, four situations were studied for the independent variables used in HCA process; physico-chemical parameters of coals, Kinetic energies, combination of both, clustering among the coal samples. Fig. 2a shows the grouping of physico chemical parameters of coals. First cluster has H, N, S, ash, W1, W2, W3, W4 and W5, while second cluster includes C, FC and VM. Dendrogram obtained for grouping of activation energies shows two clusters (Fig. 2b). First cluster has variables Ea1, Ea2 and Ea5. Ea3 and Ea4 showed their grouping in the second cluster. Fig. 2c shows the dendrogram obtained by clustering of all the variables. Two major clusters are obtained having six subsections. First cluster includes M, Ash, N, C, Ea4, Ea5, and W. Second cluster have VM, FC, S, H, H/C, W2, W3, W4, Ea1 and Ea2.This dendogram is very complex and discussed with the PCA. However, based on these parameters, coals can be categorized into three groups (2d). The first group includes C7, C8 and C10 coals. They have similarity in VM, ash and Ea1. The second group is characterized by a close link between C5, C6 and C9, which might be due to their similarity in C, Ea5 and Ea4. The third group is a link between C1, C2, C3,
3303
P. Khare et al. / Fuel 90 (2011) 3299–3305
Fig. 2. Hierarchical clustering analysis (HCA) of coal samples: (a) clustering among the coal properties, (b) clustering among activation energies of pyrolysis, (c) clustering among all variables, and (d) clustering among the coal samples.
and C4 due to similarity in their FC, VM, Ea5, W1 and W3. Results of HCA clearly differentiate these coals according to their properties.
Table 4 Results of principal component analysis. Components
3.2.3. Principal component analysis (PCA) Principal component analysis indicates that six Eigen vectors (Table 4), which can explain the majority of the variance (95%) of data. PCA suggests that relationship between coal properties and kinetic parameters can explain by six factors. Variables having loading more than 0.5 are only considered to explain each factor. Factor 1 is mostly associated with carbon (negatively), sulfur and Ea1, Ea2, Ea5 (negatively) and accounts for 24% of the total variance and represent the carbonaceous fraction of coal. It reflects that activation energies Ea1, Ea2, Ea5, W2 is significantly related to the carbon content of coal. Negative score of carbon in this factor indicates coal with high carbon content have low activation energies of dehydration, devolatilization and less weight loss during primary devolatilization (reaction 2) and thermal degradation (reaction 3). Grouping of all these variables can also be seen in the HCA analysis. Due to their negative score, C and Ea5 show close linkage between them in cluster analysis (Fig. 2c). While opposite is true for activation energy of dehydrogenation/condensation (Ea5). High score of sulfur in this factor is due to transformation of sulfur into nascent sulfur above 350 °C and fixation on coal matrix. Factor 2 accounts for 20% of the total variance (Table 4). This factor has high score of Ea5, hydrogen content, H/C ratio, Ea4, W5. It reflects that H and H/C content of coal is responsible for the activation energy of char re-solidification (Ea4) and dehydrogenation (Ea5) and weight loss of dehydrogenation. HCA analysis also showed closed linkage between the H, H/C (Fig. 2c). Factor 3 explains 17% of the total variance. Moisture, weight loss during dehydration (W1) and thermal degradation (W3) have high scores in this factor, while weight loss during primary devol-
C Ea1 Ea2 W2 S Ea5 H H/C Ea4 W5 M W1 W3 VM Ea3 FC Ash W4 N % of variance
1
2
3
4
5
6
0.96 0.90 0.79 0.74 0.68 0.61 0.09 0.29 0.19 0.17 0.12 0.20 0.52 0.33 0.19 0.14 0.07 0.29 0.23 24
0.01 0.23 0.13 0.02 0.31 0.60 0.95 0.91 0.84 0.72 0.07 0.20 0.05 0.04 0.28 0.14 0.02 0.17 0.21 20
0.02 0.13 0.48 0.59 0.26 0.20 0.12 0.13 0.38 0.29 0.90 0.83 0.66 0.03 0.26 0.02 0.09 0.57 0.07 17
0.14 0.19 0.23 0.14 0.52 0.02 0.09 0.16 0.32 0.17 0.04 0.19 0.48 0.92 0.74 0.17 0.66 0.07 0.13 14
0.15 0.11 0.04 0.03 0.09 0.08 0.08 0.02 0.05 0.41 0.15 0.20 0.17 0.18 0.17 0.88 0.72 0.66 0.17 11
0.14 0.16 0.05 0.19 0.29 0.43 0.16 0.12 0.09 0.10 0.17 0.37 0.03 0.03 0.44 0.22 0.11 0.16 0.92 9
atilization (W2) and secondary devolatilization (W4) have moderate scores. It indicates that moisture content of coal controls the weight loss due to dehydration (reaction 1) and some extent to primary devolatilization (reaction 2). However, it has no effect on the activation energies of dehydration. M and W1 also show closed linkage with each other in HCA (Fig. 2c). Factor 4 accounts for 14% with high loadings of VM (negatively) and activation energy of thermal degradation (Ea3) and moderate loadings of sulfur and ash. This fraction reflects that activation
3304
P. Khare et al. / Fuel 90 (2011) 3299–3305
energy of thermal degradation (Ea3) (plastic properties) is related to volatile matter and ash content. Increase in volatile matter ease the activation of metaplast formation. Formation of metaplast also decreases with an increase in ash and sulfur content. This grouping is slightly different with the HCA, which show clustering of ash with Ea3, but distance with volatile matter. This might be due to negative loading of VM in this factor. Factor 5 explains 11% of the total variance with high scores of FC, ash and weight loss of secondary devolatilization is high. Fifth factor indicates that fixed carbon (FC) and ash content (negatively) affect the weight loss during re-solidification of char. However, activation energy in this process is independent to these parameters. HCA analysis also shows the linkage between FC and W4, while distance with ash content. Factor 6 accounts for 9%of total variance with high loadings of nitrogen and low loadings of activation energy of thermal degradation (Ea3) and activation energy of dehydrogenation (Ea5). Six factor loaded with nitrogen and suggest that devolatilization of coal is independent to the nitrogen concentration. Linkage of N with Ea3 is also shown in the HCA analysis (Fig. 2c). 4. Discussions It can be seen from the shape of TGA–DTG analysis that devolatilization of these coals occurs via five major processes. Depending upon the temperature range, it is hypothesized that dehydration, primary devolatilization, thermal decomposition, secondary devolatilization and dehydrogenation are the major processes occurring during devolatilization (reactions 1–5) [12]. Correlations among all the parameters indicate the linking between the physico-chemical parameter with activation energies. HCA of physico-chemical properties indicate that the physicochemical properties of these coals can be classified as inorganic fraction (cluster 1) and volatile matter/carbonaceous faction (cluster 2) (Fig. 2a). While clustering of activation energies suggests that coals can be classified into two groups. First group has similarities in their activation energies of dehydration, primary devolatilization and dehydrogenation (cluster 1), while second group has similarities in activation energies of thermal degradation and secondary devolatilization. HCA and PCA applied on the physico-chemical and kinetic parameters are complimentary to each other. Score plot of first three factors is shown in the Fig. 3. It indicates the similarity between the clustering of variables obtained in dendrogram (Fig. 2c). Results indicate that the activation energy for dehydration of coal is independent to the moisture content and dependent on carbon content of the coal. In lower rank coal, large volume of water is retaining in coal structure due to hydrogen bonds. Miura
et al., [15] reported that enthalpy of formation of water depend upon the hydrogen bonding during pyrolysis. The coals with high carbon content have less number of hydrogen bonding due to the presence of more hydrocarbon. Less availability of oxygen for hydrogen bonding ease the dehydration during pyrolysis. Coals with high carbon content have low activation energy and weight loss of primary devolatilization. In this reaction, loss of light hydrocarbon (C–C bond breaking), alcohol and acid (C–O bond breaking) and H2S (C–S bond breaking) may occur. High VM in the coals is attributed to the release of more hydrocarbons [17]. Lowering in activation energy with an increase in carbon content is due to the breaking of C–C and C–O bond energies. The bond energy of C–C bonds (bond energy = 348 kJ/mol) is lower than bond energies of C–O (bond energy = 360 kJ/mol). The activation energy of thermal degradation of coals (Ea3) depends upon their volatile matter and ash content of coal. At low temperatures, the volatile matters have been released from the coal matrix rather than being formed by the thermal breakdown of coal [18]. Hence, it is obvious that high VM and low ash content ease the release of volatile compounds. Association of sulfur with thermal degradation of coals (Ea3) is due to release of H2S from alkyl and aryl sulfides between the 440 and 630 °C temperature ranges [19]. The coals with high hydrogen content have low activation energy of secondary devolatilization. Aliphatic compounds have higher H/C ratios than aromatic ones, which ease the process of secondary devolatilization. However, weight loss in secondary devolatilization also depends upon the FC and ash content. Negative score of ash in factor five (PC 5) indicates that it might catalyze the depolymerization of coal. Coals with high carbon content have low activation energy of condensation and dehydrogenation. High H/C atomic ratio in coal is indicative of easy dehydrogenation and coalescence of higher molecular weight substances [19,20]. HCA and PCA can be used successfully to correlate the properties of coal with activation energies. This model can be refined by taking more parameters and gases released during the pyrolysis and could be useful for other solid fuels as well. 5. Conclusion Chemometrics analysis clearly identified that the effect of moisture, volatile matter, ash, fixed carbon, carbon, hydrogen, nitrogen and sulfur on kinetics of various reactions of devolatilization (dehydrogenation, primary devolatilization, thermal degradation, secondary devolatilization/depolymerization, dehydrogenation/ condensation). Carbon content controls the dehydrogenation, primary devolatilization and dehydrogenation, while H and H/C ratio control the activation energies of secondary devolatilization and dehydrogenation. Ash catalyzed the dehydrogenation reaction. VM has major effect on the metaplast formation during coal pyrolysis. Acknowledgement Authors are thankful to Ministry of Coal (Project No EE-42), Govt of India, New Delhi for financial assistance. References
Fig. 3. Plot of scores of first three principal components.
[1] Wagner N, Hlatshwayo B. The occurrence of potentially hazardous trace elements in five Highveld coals, South Africa. Int J Coal Geol 2005;63:228–36. [2] Vejahati F, Xu Z, Gupta R. Trace elements in coal: association with coal and minerals and their behavior during coal utilization – a review. Fuel 2010;89:904–11. [3] Kok MV. Recent developments in the application of thermal analysis techniques in fossil fuels. J Therm Anal Cal 2008;91:763–73. [4] Kok MV. Thermal investigation of Seyitomer oil shale. J Therm Anal Cal 2001;369:149–55.
P. Khare et al. / Fuel 90 (2011) 3299–3305 [5] Iordanidis A, Georgakopoulos A, Markova K. Application of TG–DTA to the study of Amynteon lignites, northern Greece. Thermochim Acta 2001;371:137–41. [6] Alonso MJG, Borrego A G, Alvarez D, Kalkreuth W, Menéndez R. Physicochemical transformations of coal particles during pyrolysis and combustion. Fuel 2001;80:1857–70. [7] Esteban M, Ariño C, Díaz-Cruz JM. Chemometric for the analysis of voltametric data. Trends Anal Chem 2006;25:86–92. [8] Marengo E, Gennaro MC, Robotti E, Rossanigo P, Rinaudo C, Roz-Gastaldi M. Investigation of anthropic effects connected with metal ions concentration, organic matter and grain size in Bormida river sediments. Anal Chim Acta 2006;560:172–83. [9] Singh KP, Malik A, Singh VK, Mohan D, Sinha S. Chemometric analysis of groundwater quality data of alluvial aquifer of Gangetic plain, North India. Anal Chim Acta 2005;550:82–91. [10] Khare P, Baruah BP. Chemometric analysis of trace and heavy metal distribution in raw and thermally treated high sulphur coals. Fuel Proc Technol 2010;91:1691–701. [11] Farmer VC. The infrared spectra of minerals. Monograph No 4. London: Mineralogical Society; 1974. [12] Baruah BP, Khare P. Pyrolysis of high sulphur Indian coals. Energy Fuels 2007;21:3346–52.
3305
[13] Khare P, Baruah BP. Thermogravimetric analysis of perhydrous Indian coals. Energy Sources (accepted); 2011. [14] Coats AW, Edfern JP. Kinetic parameters from thermo gravimetric data. J Polym Sci Part B: Polym Lett 1965;3:917–20. [15] Strezov I, Lucas JA, Strezov L. Experimental and modeling of the thermal regions of activity during pyrolysis of bituminous coals. J Anal Appl Pyrol 2004;71:375–92. [16] Miura K, Mae K, Li W, Kusakawa T, Morozumi F, Kumano A. Estimation of Hydrogen bond distribution in coal through the analysis of OH stretching bands in diffuse reflectance infrared spectrum measured by in-situ technique. Energy Fuels 2001;15:599–610. [17] Cai J, Wang Y, Zhou L, Huang Q. Thermogravimetric analysis and kinetics of coal/plastic blends during co-pyrolysis in nitrogen atmosphere. Fuel Proc Technol 2008;89:21–7. [18] Khare P, Baruah BP. Structural parameters of perhydrous Indian coals. Int J Coal Prep Util 2010;30:44–67. [19] Radenovic A. Pyrolysis of coal. Kemija u Industriji 2006;55:311–9. [20] Davis KA, Hurt RH, Yang NYC, Headley TJ. Evolution of char chemistry, crystallinity, and ultrafine structure during pulverized-coal combustion. Combust Flame 1995;100:31–40.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Application of chemometrics to study the kinetics of coal pyrolysis: A novel approach Puja Khare ⇑, B.P. Baruah, P.G. Rao Coal Chemistry Division, North-East Institute of Science and Technology (CSIR-NEIST), Jorhat, 785 006 Assam, India
a r t i c l e
i n f o
Article history: Received 26 October 2010 Received in revised form 17 March 2011 Accepted 19 May 2011 Available online 1 June 2011 Keywords: Chemometric Principal component analysis Hierarchical clustering analysis Coals
a b s t r a c t In present investigation, chemometric tools, principal component analysis (PCA) and Hierarchical clustering analysis (HCA) are used to get the linkage between the coal properties and kinetics of pyrolysis. Thermo gravimetric analysis (TGA) of 10 perhydrous Indian coals was done. Devolatilization of these coals showed five independent reactions. Kinetic parameters were calculated for individual reaction. Activation energy and weight loss of each reaction has been analyzed as a function of coal properties (moisture, volatile matter, ash, fixed carbon, carbon, hydrogen, nitrogen and sulfur). By applying chemometric, was extracted information about the linkage between activation energies of each reaction and coal properties. The mathematical treatment of data has provided conclusions on properties of coal and kinetic parameters. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Pyrolysis is an important intermediate stage in coal gasification, combustion, and liquefaction, and also considered as a simple and effective method for a clean conversion of coal [1]. The application of thermal analysis to study the behavior of fossil fuels during pyrolysis has gained much industrial importance [2]. Thermal analysis characterizes the physical and chemical properties of substances, depending on the temperature at defined heating rate (dynamics measurement) or on the time at a constant temperature (static measurement) [3]. In recent years, the application of thermal analysis techniques to study the combustion and pyrolysis behavior of fuels has gained a wide acceptance among research workers [4–6]. Kinetics of devolatilization is useful for accurate estimation of the reactivity of fossil fuels during their utilization and conversion processes (gasification, liquefaction). Devolatilization of solid fuel occurs by processes like dehydration, primary and secondary devolatilization and thermal degradation. Accurate knowledge of the devolatilization processes is necessary to develop predictive models for coal conversion processes. Application of chemometrics to TGA data at different steps of the devolatilization is of interest and useful. Chemometrics is a chemical discipline that uses mathematics, statistics and formal logic: (1) To design or to select optimal experimental procedures, (2) To provide maximum relevant chemical information by analyzing chemical data, and (3) To obtain knowledge about the chemical systems [7]. ⇑ Corresponding author. Fax: +91 (0) 376 2370011. E-mail address: [email protected] (P. Khare). 0016-2361/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2011.05.017
In present investigation, perhydrous Indian coals, which are suitable for conversion processes like liquefaction and coke making used for thermo gravimetric analysis. The kinetic parameters extracted from TGA are combined with differing chemometric tools such as correlation, principal component analysis (PCA) and cluster analyses. A general overview of the application of chemometrics, particularly in PCA techniques and soft modeling to TGA data is presented in the paper giving special attention to the more recent contribution on fuel conversion processes. The database can be interpreted in many ways, we have highlighted the usefulness of PCA in order to 1. Develop a relationship between the coal properties and kinetic parameters obtained from TGA. 2. Suitability of coal for particular conversion process i.e. by combining the physico-chemical properties of coal with that of the kinetic parameters obtained from TGA.
2. Materials and methods The freshly mined low rank coal samples from different coalfields of northeastern India were used in this study. The air-dried samples were ground to 0.211 mm before use. The proximate, ultimate and sulfur analyses were done by using a TGA 701 (Leco), True spec elemental analyzer (Leco) and S 144 DR sulfur determinator (Leco). Physico-chemical characteristics of these coals are summarized in Table 1. 2.1. TGA and DTG Differential thermal analysis was used to determine the pyrolysis behavior of perhydrous coals. Experiments were carried out in a
3300
P. Khare et al. / Fuel 90 (2011) 3299–3305
Table 1 Physico-chemical characteristics of sub-bituminous coals, and weight loss (%) for five reactions during pyrolysis.
M Ash VM FC C H N S H/C W1 W2 W3 W4 W5
C1
C2
C3
C4
C5
C6
C7
C8
C9
C10
2.9 20 35.6 41.5 74.6 7.5 1.2 2.9 1.21 0.22 2.82 19.36 8.63 16.24
3.1 11.5 41.4 44 64.1 5.72 1.4 2.9 1.07 0.39 7.33 18.97 10.85 18.98
1.5 11.5 40.5 46.5 64.3 6.07 1.1 4.46 1.13 0.15 3.4 20.27 9.87 18.91
2.7 12 35.5 49.8 70.1 6.1 1.3 3.98 1.04 0.21 2.67 19.2 11.2 20.6
2.9 7.1 42.7 47.3 68.6 5.5 1.1 3 0.96 0.25 2.88 19.67 10.3 20.1
3.4 11.9 37.2 47.5 70.8 4.7 1.1 4 0.80 0.32 6.4 18.2 11.4 19.2
2.8 17.6 36 45.6 71.4 2.7 1.1 4.5 0.45 0.27 2.67 19.01 8.6 19.8
2.1 14.4 38.1 45.4 81.8 5.8 1.4 2.2 0.85 0.25 2.1 18.1 8.2 18.7
3.2 3.9 40.4 52.5 75.9 5.3 1.1 2 0.84 0.26 2.65 18.3 9.99 17.2
2.8 16.3 36.1 44.8 82.3 5.9 1.4 2.4 0.86 0.31 1.92 17.2 10.2 18.2
VM: volatile matter, FC: fixed carbon; W1-W5: Weight loss during the reactions 1–5, respectively.
Leco TGA 701 thermal analysis system with 0.5 g each of coal samples in a stream of nitrogen with a flow rate of 40 ml/min and linear heating rate of 10 °C/min. Several preliminary thermo gravimetric runs were performed in order to investigate the sample mass influence on the devolatilization behavior. Variation in mass has negligible effect on the devolatilization of coal. 2.2. Calculation of kinetic parameters The kinetic parameters, activation energies and pre exponential factors of coal pyrolysis were determined by the integral method by applying Arrhenius equation. Arrhenius equation can be expressed as
dx=dt ¼ AexpðE=RTÞð1 xÞ
ð1Þ
where A is pre-exponential factor, E activation energy, T temperature, t time, X weight loss fraction or decomposition during pyrolysis and which can be calculated by
x¼
W0 Wt W0 Wf
ð2Þ
where W0 is the original mass of the test sample, Wt is the mass at time t or T and Wf is the final mass at the end of pyrolysis. The constant heating rate during pyrolysis is H = dT/dt for H being the heating rate. Rearranging the Eq. (1) and on integration gives
In½lnð1 xÞ=T2 ¼ In½AR=HEð1 2RT=EÞ E=RT
ð3Þ
The expression ln{AR[1-(2RT/E)]/HE in Eq. (3) is essentially constant for most of the values of E and temperature range of the pyrolysis. By plotting the left side of Eq. (3) against 1/T, a straight line is obtained indicating the process to be of first order reaction. From the slope, E/R, the activation energy E can be determined.
PCA technique extracts the Eigen values and Eigen vectors from the covariance matrix of original variables. It allows finding out association between variables, thus, reducing the dimensionality of the data set. The Eigen values of PCs are the measure of their associated variance, the participation of the original variables in the PCs is given by the loading, and the individual transformed observations are called scores [9,10]. PCA was performed on normalized (z-scale transformation) variables after sorting out the highly correlated variables from the data sets. The Bartlett’s sphericity test was applied to the correlation matrix of variables for assessing the adequacy of PCA [11]. PCs with Eigen value greater than 1 were considered. Here, PCA was performed with a view to establish linkage between the coal properties and kinetic parameters extracted from TGA. To validate the model result, correlation were also drawn to calculate theoretical activation energy. Cluster analysis is an exploratory multivariate method that can be used to describe the relationships among variables. Several mathematical criteria can be used to examine the similarity (or difference/distance) between variables and cases. For Hierarchical clustering analysis (HCA), the Ward’s method was used to get cleaner picture of clusters. This method is distinct from all other methods because it uses an analysis of variance approach to evaluate the distances between clusters. This method attempts to minimize the sum of square (SS) of any two clusters that can be formed each step. This method is regarded as a very efficient one. The joining or tree clustering method uses the dissimilarities (similarities) or distances between objects when using the clusters. Similarities (distance) are a set of rules that serves as criteria for grouping or separating items elucidean distance is chosen for the analysis.
3. Results 3.1. Thermal behavior and kinetics of coal pyrolysis
2.3. Data treatment and chemometric analysis Statistical treatment of data including correlation analysis, and principal component analysis (PCA) and Hierarchical clustering analysis (HCA) were performed using SPSS 15 statistical software. Principal component analysis (PCA) techniques have been widely applied in the treatment of datasets of such a high complexity [8]. PCA provides a new set of orthogonal variables, the principal components (PCs), generated so that each PC accounts for the maximum possible amount of variance contained in the original dataset. The PCs are obtained as linear combination of the original descriptors and are aligned along the directions of covariance of the data [9].
The curves of thermal analysis (TGA and DTG) of one of the coal samples are shown in Fig. 1. It is evident from the shape of the curves that pyrolysis proceeds through five different temperature regions. Hence devolatilization may occur via five reactions i.e. dehydration (reaction 1), primary devolatilization /thermal desorption of gases (reaction 2), thermal degradation (reaction 3), depolymerization/secondary devolatilization (reaction 4), and dehydrogenation/condensation (reaction 5) [12,13]. The activation energies are determined for each process reaction by applying first order Arrhenius equation on each independent reaction (calculation is given in the materials and methods). Activation energy (Ea) with standard deviation and r2 for each linear regression is shown for
3301
P. Khare et al. / Fuel 90 (2011) 3299–3305
Fig. 1. TGA (wt loss) and DTG (Rate of wt loss) curves for one of the perhydrous coals.
Activation energies of thermal degradation (reaction 3) vary from 141 to 157 kJ/mol, which are related to the plastic properties, swelling of coal by completion of the softening process, coal melts forming metaplast and re-solidifies as char. During thermal degradation, coal has undergone an intense degradation of its structure and greatest mass loss takes place due to degradation of carbonaceous matrix and the evolution of relatively high molecular weight species.
each independent reaction in Table 2. The activation energies and the weight loss of five reactions are reported as Ea1, Ea2, Ea3, Ea4 and Ea5 and W1, W2, W3, W4 and W5 respectively in the text. Analysis of variance (ANOVA) is applied on the activation energies of each set of reactions. Results of ANOVA show that activation energies are significantly different from each other (p < 0.01–0.001) which indicates that reactions are independent to each other. The activation energies of secondary devolatilization (reaction 4) range from 165 to 190 kJ/mol, which are higher than previous regions. In this region, formation of secondary char as the main pyrolysis products occurs due to re-solidification and secondary devolatilization processes, which are endothermic in nature [14]. Hence, the activation energies are higher in this region.
Activation energies for liberation of water of these coals varied between 92.0 and 104.2 kJ/mol. The possible reaction mechanism is shown in reaction I. The activation energies of primary devolatilization (reaction 2) range from 120.7 to 134.7 kJ/mol. During primary devolatilization, the coals undergo softening with the release of light species [14]. Thermal decomposition of the more labile structure of the coal matrix also takes place. In this stage, the volatiles (liquids and gases) are the main pyrolysis products.
The activation energies of dehydrogenation/condensation range between 113 and 124 kJ/moles. During reaction 5, formation of coke takes place with the evolution of hydrogen gas. H2 gas evolves
Table 2 Activation energies (kJ/mol) of sub-bituminous coals for five reactions during pyrolysis. Reaction-1
C1 C2 C3 C4 C5 C6 C7 C8 C9 C10
Reaction-2 2
Ea1
Sd
R
99.9 105 104 97 98 101 97 94 97 92
0.36 0.25 0.25 0.42 0.30 0.30 0.31 0.15 0.36 0.41
0.99 0.995 0.998 0.996 0.997 0.992 0.995 0.994 0.992 0.994
Reaction-3 2
Reaction-4 2
Ea2
Sd
R
Ea3
Sd
R
129 135 140 131 133 121 130 123 125 120
0.15 0.10 0.15 0.16 0.10 0.15 0.13 0.15 0.25 0.15
0.998 0.992 0.998 0.995 0.995 0.998 0.997 0.998 0.996 0.998
149.7 149.9 152 158 141 150 160 157 145 160
0.30 0.35 0.30 0.31 0.42 0.31 0.31 0.31 0.31 0.25
0.998 0.997 0.999 0.994 0.994 0.995 0.998 0.998 0.998 0.996
Sd; standard deviation (±), R2: regression coefficient.
Reaction-5 2
Ea4
Sd
R
Ea5
Sd
R2
165.2 183 170.5 169 185 186 190 183 184 183
0.20 0.25 0.41 0.20 0.35 0.25 0.41 0.25 0.25 0.45
0.996 0.998 0.999 0.998 0.998 0.999 0.997 0.998 0.999 0.998
118 112 118 116 121 120 124 121 121 122
0.20 0.21 0.20 0.20 0.42 0.20 0.36 0.20 0.36 0.20
0.1 0.999 0.999 0.999 0.999 0.998 0.998 0.998 0.998 0.999
P. Khare et al. / Fuel 90 (2011) 3299–3305
1.00 1.00 0.32 1.00 0.40 0.30 1.00 0.08 0.57 0.09 1.00 0.12 0.45 0.47 0.18 1.00 0.33 0.65* 0.18 0.25 0.33
3.2. Chemometric analysis 3.2.1. Correlation Correlation analysis provide primary indicator of relation between the coal properties and kinetic parameters. Correlation analysis of all the data set show that the ash content of coals is negatively correlated with VM and carbon content (Table 3). Carbon is highly correlated with the sulfur content, while hydrogen content of coal was highly correlated with H/C ratio. Significant correlations (r = 0.63–0.92) are observed between coal properties and activation energies in different reactions. The properties of these coals moderately correlate with the weight loss rate in different reactions (r = 0.52–0.66). These correlation analysis provide preliminary idea about the relationship between coal properties and their kinetics during pyrolysis.
1.00 0.69* 0.43 0.27 0.64* 0.79* 0.24 0.28 0.21 0.20 1.00 0.74* 0.81* 0.28 0.07 0.32 0.22 0.05 0.16 0.15 0.66* 0.27 0.09 0.05 0.16 0.45** 0.55** 0.13
1.00 0.24 0.47 0.03 0.20 0.22 0.23 0.44 0.47** 0.74* 0.26 0.21 0.20 0.36 0.29 0.30 0.03
1.00 0.05 0.27 0.37 0.02 0.24 0.15 0.09 0.24 0.20 0.15 0.20 0.15 0.24 0.45** 0.29
1.00 0.10 0.38 0.63* 0.29 0.88* 0.83* 0.37 0.16 0.55** 0.13 0.67* 0.43 0.36 0.20
1.00 0.36 0.40 0.92* 0.11 0.11 0.23 0.78* 0.53* 0.24 0.07 0.07 0.23 0.64⁄
1.00 0.47** 0.21 0.31 0.25 0.49** 0.08 0.39 0.32 0.03 0.06 0.05 0.16
1.00 0.15 0.45 0.48** 0.26 0.15 0.03 0.26 0.23 0.74* 0.11 0.52**
1.00 0.45 0.43 0.36 0.81* 0.73* 0.18 0.20 0.09 0.08 0.54**
1.00 0.25 0.39 0.47 0.19 0.65* 0.28 0.10 0.10
1.00 0.03 0.17 0.12 0.30 0.49** 0.26 0.26
1.00 0.51 0.53** 0.14 0.47 0.38 0.44
W1 Ea1
P < 0.001. ** P < 0.01.
*
N H C FC VM Ash M
by breaking of C–H bonds with the formation of stronger bonds between the carbon atoms. This process is followed by condensation and contraction within the carbon hexagonal planes [15]. Activation energies of dehydrogenation/condensation are lower than previous stage, which may be due to exothermic contraction reaction [16].
1.00 0.16 0.08 0.15 0.06 0.21 0.08 0.24 0.24 0.06 0.46 0.31 0.39 0.03 0.66* 0.33 0.62* 0.53** 0.13
Table 3 Correlation among the coal properties, activation energies and wt loss.
M Ash VM FC C H N S H/C Ea1 Ea2 Ea3 Ea4 Ea5 W1 W2 W3 W4 W5
S
H/C
Ea2
Ea3
Ea4
Ea5
W2
W3
W4
W5
3302
3.2.2. Hierarchical clustering analysis HCA Kinetic energy of individual pyrolysis reactions is taken as major kinetic parameters for cluster analysis, which provides information about the rate of the individual reaction during pyrolysis. Kinetic energy of a reaction of devolatilization depends upon heating rate, atmosphere of the pyrolysis and the coal characteristics. Heating rate and pyrolysis atmosphere was same during thermal analysis of all the coal samples. Variations in the kinetic parameters of individual coal were due to their physicochemical characteristics. Hence, for HCA physicochemical parameters such as ash, moisture, volatile matter, fixed carbon, carbon, hydrogen, nitrogen, sulfur and H/C of coal were chosen. Hierarchical cluster analysis is applied on kinetics parameters and physico-chemical properties of coals (Fig. 2). Here, four situations were studied for the independent variables used in HCA process; physico-chemical parameters of coals, Kinetic energies, combination of both, clustering among the coal samples. Fig. 2a shows the grouping of physico chemical parameters of coals. First cluster has H, N, S, ash, W1, W2, W3, W4 and W5, while second cluster includes C, FC and VM. Dendrogram obtained for grouping of activation energies shows two clusters (Fig. 2b). First cluster has variables Ea1, Ea2 and Ea5. Ea3 and Ea4 showed their grouping in the second cluster. Fig. 2c shows the dendrogram obtained by clustering of all the variables. Two major clusters are obtained having six subsections. First cluster includes M, Ash, N, C, Ea4, Ea5, and W. Second cluster have VM, FC, S, H, H/C, W2, W3, W4, Ea1 and Ea2.This dendogram is very complex and discussed with the PCA. However, based on these parameters, coals can be categorized into three groups (2d). The first group includes C7, C8 and C10 coals. They have similarity in VM, ash and Ea1. The second group is characterized by a close link between C5, C6 and C9, which might be due to their similarity in C, Ea5 and Ea4. The third group is a link between C1, C2, C3,
3303
P. Khare et al. / Fuel 90 (2011) 3299–3305
Fig. 2. Hierarchical clustering analysis (HCA) of coal samples: (a) clustering among the coal properties, (b) clustering among activation energies of pyrolysis, (c) clustering among all variables, and (d) clustering among the coal samples.
and C4 due to similarity in their FC, VM, Ea5, W1 and W3. Results of HCA clearly differentiate these coals according to their properties.
Table 4 Results of principal component analysis. Components
3.2.3. Principal component analysis (PCA) Principal component analysis indicates that six Eigen vectors (Table 4), which can explain the majority of the variance (95%) of data. PCA suggests that relationship between coal properties and kinetic parameters can explain by six factors. Variables having loading more than 0.5 are only considered to explain each factor. Factor 1 is mostly associated with carbon (negatively), sulfur and Ea1, Ea2, Ea5 (negatively) and accounts for 24% of the total variance and represent the carbonaceous fraction of coal. It reflects that activation energies Ea1, Ea2, Ea5, W2 is significantly related to the carbon content of coal. Negative score of carbon in this factor indicates coal with high carbon content have low activation energies of dehydration, devolatilization and less weight loss during primary devolatilization (reaction 2) and thermal degradation (reaction 3). Grouping of all these variables can also be seen in the HCA analysis. Due to their negative score, C and Ea5 show close linkage between them in cluster analysis (Fig. 2c). While opposite is true for activation energy of dehydrogenation/condensation (Ea5). High score of sulfur in this factor is due to transformation of sulfur into nascent sulfur above 350 °C and fixation on coal matrix. Factor 2 accounts for 20% of the total variance (Table 4). This factor has high score of Ea5, hydrogen content, H/C ratio, Ea4, W5. It reflects that H and H/C content of coal is responsible for the activation energy of char re-solidification (Ea4) and dehydrogenation (Ea5) and weight loss of dehydrogenation. HCA analysis also showed closed linkage between the H, H/C (Fig. 2c). Factor 3 explains 17% of the total variance. Moisture, weight loss during dehydration (W1) and thermal degradation (W3) have high scores in this factor, while weight loss during primary devol-
C Ea1 Ea2 W2 S Ea5 H H/C Ea4 W5 M W1 W3 VM Ea3 FC Ash W4 N % of variance
1
2
3
4
5
6
0.96 0.90 0.79 0.74 0.68 0.61 0.09 0.29 0.19 0.17 0.12 0.20 0.52 0.33 0.19 0.14 0.07 0.29 0.23 24
0.01 0.23 0.13 0.02 0.31 0.60 0.95 0.91 0.84 0.72 0.07 0.20 0.05 0.04 0.28 0.14 0.02 0.17 0.21 20
0.02 0.13 0.48 0.59 0.26 0.20 0.12 0.13 0.38 0.29 0.90 0.83 0.66 0.03 0.26 0.02 0.09 0.57 0.07 17
0.14 0.19 0.23 0.14 0.52 0.02 0.09 0.16 0.32 0.17 0.04 0.19 0.48 0.92 0.74 0.17 0.66 0.07 0.13 14
0.15 0.11 0.04 0.03 0.09 0.08 0.08 0.02 0.05 0.41 0.15 0.20 0.17 0.18 0.17 0.88 0.72 0.66 0.17 11
0.14 0.16 0.05 0.19 0.29 0.43 0.16 0.12 0.09 0.10 0.17 0.37 0.03 0.03 0.44 0.22 0.11 0.16 0.92 9
atilization (W2) and secondary devolatilization (W4) have moderate scores. It indicates that moisture content of coal controls the weight loss due to dehydration (reaction 1) and some extent to primary devolatilization (reaction 2). However, it has no effect on the activation energies of dehydration. M and W1 also show closed linkage with each other in HCA (Fig. 2c). Factor 4 accounts for 14% with high loadings of VM (negatively) and activation energy of thermal degradation (Ea3) and moderate loadings of sulfur and ash. This fraction reflects that activation
3304
P. Khare et al. / Fuel 90 (2011) 3299–3305
energy of thermal degradation (Ea3) (plastic properties) is related to volatile matter and ash content. Increase in volatile matter ease the activation of metaplast formation. Formation of metaplast also decreases with an increase in ash and sulfur content. This grouping is slightly different with the HCA, which show clustering of ash with Ea3, but distance with volatile matter. This might be due to negative loading of VM in this factor. Factor 5 explains 11% of the total variance with high scores of FC, ash and weight loss of secondary devolatilization is high. Fifth factor indicates that fixed carbon (FC) and ash content (negatively) affect the weight loss during re-solidification of char. However, activation energy in this process is independent to these parameters. HCA analysis also shows the linkage between FC and W4, while distance with ash content. Factor 6 accounts for 9%of total variance with high loadings of nitrogen and low loadings of activation energy of thermal degradation (Ea3) and activation energy of dehydrogenation (Ea5). Six factor loaded with nitrogen and suggest that devolatilization of coal is independent to the nitrogen concentration. Linkage of N with Ea3 is also shown in the HCA analysis (Fig. 2c). 4. Discussions It can be seen from the shape of TGA–DTG analysis that devolatilization of these coals occurs via five major processes. Depending upon the temperature range, it is hypothesized that dehydration, primary devolatilization, thermal decomposition, secondary devolatilization and dehydrogenation are the major processes occurring during devolatilization (reactions 1–5) [12]. Correlations among all the parameters indicate the linking between the physico-chemical parameter with activation energies. HCA of physico-chemical properties indicate that the physicochemical properties of these coals can be classified as inorganic fraction (cluster 1) and volatile matter/carbonaceous faction (cluster 2) (Fig. 2a). While clustering of activation energies suggests that coals can be classified into two groups. First group has similarities in their activation energies of dehydration, primary devolatilization and dehydrogenation (cluster 1), while second group has similarities in activation energies of thermal degradation and secondary devolatilization. HCA and PCA applied on the physico-chemical and kinetic parameters are complimentary to each other. Score plot of first three factors is shown in the Fig. 3. It indicates the similarity between the clustering of variables obtained in dendrogram (Fig. 2c). Results indicate that the activation energy for dehydration of coal is independent to the moisture content and dependent on carbon content of the coal. In lower rank coal, large volume of water is retaining in coal structure due to hydrogen bonds. Miura
et al., [15] reported that enthalpy of formation of water depend upon the hydrogen bonding during pyrolysis. The coals with high carbon content have less number of hydrogen bonding due to the presence of more hydrocarbon. Less availability of oxygen for hydrogen bonding ease the dehydration during pyrolysis. Coals with high carbon content have low activation energy and weight loss of primary devolatilization. In this reaction, loss of light hydrocarbon (C–C bond breaking), alcohol and acid (C–O bond breaking) and H2S (C–S bond breaking) may occur. High VM in the coals is attributed to the release of more hydrocarbons [17]. Lowering in activation energy with an increase in carbon content is due to the breaking of C–C and C–O bond energies. The bond energy of C–C bonds (bond energy = 348 kJ/mol) is lower than bond energies of C–O (bond energy = 360 kJ/mol). The activation energy of thermal degradation of coals (Ea3) depends upon their volatile matter and ash content of coal. At low temperatures, the volatile matters have been released from the coal matrix rather than being formed by the thermal breakdown of coal [18]. Hence, it is obvious that high VM and low ash content ease the release of volatile compounds. Association of sulfur with thermal degradation of coals (Ea3) is due to release of H2S from alkyl and aryl sulfides between the 440 and 630 °C temperature ranges [19]. The coals with high hydrogen content have low activation energy of secondary devolatilization. Aliphatic compounds have higher H/C ratios than aromatic ones, which ease the process of secondary devolatilization. However, weight loss in secondary devolatilization also depends upon the FC and ash content. Negative score of ash in factor five (PC 5) indicates that it might catalyze the depolymerization of coal. Coals with high carbon content have low activation energy of condensation and dehydrogenation. High H/C atomic ratio in coal is indicative of easy dehydrogenation and coalescence of higher molecular weight substances [19,20]. HCA and PCA can be used successfully to correlate the properties of coal with activation energies. This model can be refined by taking more parameters and gases released during the pyrolysis and could be useful for other solid fuels as well. 5. Conclusion Chemometrics analysis clearly identified that the effect of moisture, volatile matter, ash, fixed carbon, carbon, hydrogen, nitrogen and sulfur on kinetics of various reactions of devolatilization (dehydrogenation, primary devolatilization, thermal degradation, secondary devolatilization/depolymerization, dehydrogenation/ condensation). Carbon content controls the dehydrogenation, primary devolatilization and dehydrogenation, while H and H/C ratio control the activation energies of secondary devolatilization and dehydrogenation. Ash catalyzed the dehydrogenation reaction. VM has major effect on the metaplast formation during coal pyrolysis. Acknowledgement Authors are thankful to Ministry of Coal (Project No EE-42), Govt of India, New Delhi for financial assistance. References
Fig. 3. Plot of scores of first three principal components.
[1] Wagner N, Hlatshwayo B. The occurrence of potentially hazardous trace elements in five Highveld coals, South Africa. Int J Coal Geol 2005;63:228–36. [2] Vejahati F, Xu Z, Gupta R. Trace elements in coal: association with coal and minerals and their behavior during coal utilization – a review. Fuel 2010;89:904–11. [3] Kok MV. Recent developments in the application of thermal analysis techniques in fossil fuels. J Therm Anal Cal 2008;91:763–73. [4] Kok MV. Thermal investigation of Seyitomer oil shale. J Therm Anal Cal 2001;369:149–55.
P. Khare et al. / Fuel 90 (2011) 3299–3305 [5] Iordanidis A, Georgakopoulos A, Markova K. Application of TG–DTA to the study of Amynteon lignites, northern Greece. Thermochim Acta 2001;371:137–41. [6] Alonso MJG, Borrego A G, Alvarez D, Kalkreuth W, Menéndez R. Physicochemical transformations of coal particles during pyrolysis and combustion. Fuel 2001;80:1857–70. [7] Esteban M, Ariño C, Díaz-Cruz JM. Chemometric for the analysis of voltametric data. Trends Anal Chem 2006;25:86–92. [8] Marengo E, Gennaro MC, Robotti E, Rossanigo P, Rinaudo C, Roz-Gastaldi M. Investigation of anthropic effects connected with metal ions concentration, organic matter and grain size in Bormida river sediments. Anal Chim Acta 2006;560:172–83. [9] Singh KP, Malik A, Singh VK, Mohan D, Sinha S. Chemometric analysis of groundwater quality data of alluvial aquifer of Gangetic plain, North India. Anal Chim Acta 2005;550:82–91. [10] Khare P, Baruah BP. Chemometric analysis of trace and heavy metal distribution in raw and thermally treated high sulphur coals. Fuel Proc Technol 2010;91:1691–701. [11] Farmer VC. The infrared spectra of minerals. Monograph No 4. London: Mineralogical Society; 1974. [12] Baruah BP, Khare P. Pyrolysis of high sulphur Indian coals. Energy Fuels 2007;21:3346–52.
3305
[13] Khare P, Baruah BP. Thermogravimetric analysis of perhydrous Indian coals. Energy Sources (accepted); 2011. [14] Coats AW, Edfern JP. Kinetic parameters from thermo gravimetric data. J Polym Sci Part B: Polym Lett 1965;3:917–20. [15] Strezov I, Lucas JA, Strezov L. Experimental and modeling of the thermal regions of activity during pyrolysis of bituminous coals. J Anal Appl Pyrol 2004;71:375–92. [16] Miura K, Mae K, Li W, Kusakawa T, Morozumi F, Kumano A. Estimation of Hydrogen bond distribution in coal through the analysis of OH stretching bands in diffuse reflectance infrared spectrum measured by in-situ technique. Energy Fuels 2001;15:599–610. [17] Cai J, Wang Y, Zhou L, Huang Q. Thermogravimetric analysis and kinetics of coal/plastic blends during co-pyrolysis in nitrogen atmosphere. Fuel Proc Technol 2008;89:21–7. [18] Khare P, Baruah BP. Structural parameters of perhydrous Indian coals. Int J Coal Prep Util 2010;30:44–67. [19] Radenovic A. Pyrolysis of coal. Kemija u Industriji 2006;55:311–9. [20] Davis KA, Hurt RH, Yang NYC, Headley TJ. Evolution of char chemistry, crystallinity, and ultrafine structure during pulverized-coal combustion. Combust Flame 1995;100:31–40.