Higher heating value prediction of torrefaction char produced from non-woody biomass

Abstract

The higher heating value of five types of non-woody biomass and their torrefaction char was predicted and compared with the experimental data obtained in this paper. The correlation proposed in this paper and the ones suggested by previous researches were used for prediction. For prediction using proximate analysis data, the mass fraction of fixed carbon and volatile matter had a strong effect on the higher heating value prediction of torrefaction char of non-woody biomass. The high ash fraction found in torrefied char resulted in a decrease in prediction accuracy. However, the prediction could be improved by taking into account the effect of ash fraction. The correlation developed in this paper gave a better prediction than the ones suggested by previous researches, and had an absolute average error (AAE) of 2.74% and an absolute bias error (ABE) of 0.52%. For prediction using elemental analysis data, the mass fraction of carbon, hydrogen, and oxygen had a strong effect on the higher heating value, while no relationship between the higher heating value and mass fractions of nitrogen and sulfur was discovered. The best correlation gave an AAE of 2.28% and an ABE of 1.36%.

Keywords

higher heating value / correlation / biomass / proximate analysis / ultimate analysis

Introduction

The higher heating value is regarded as an important fuel property for combustion system design, and defined as the thermodynamic heat of combustion [ 1]. It is difficult to calculate the higher heating value of biomass by using the enthalpy of formation method because of the complexity of biomass structure and content [ 2]. However, the higher heating value can be determined by the bomb calorimeter and correlations in accordance with the proximate and elemental analysis [ 3]. For the bomb calorimeter, it is time consuming and involves a complex procedure [ 4]. The prediction of higher heating value by correlations using proximate and elemental analysis data, thus, is widely accepted [ 5– 9]. In addition, many researchers proposed the higher heating value correlation using the data of proximate analysis together with elemental analysis [ 10– 12].

For the higher heating value prediction by proximate analysis data, Cordero et al. [ 6] applied the correlation proposed by Jiménez and González [ 13] for the higher heating value prediction of lignocellulosic biomass and charcoals. It was found that the prediction result using this correlation had an error of 3.48% compared with the experimental data. Parikh et al. [ 14] predicted the higher heating value of solid fuels including coal, wood, shell, seed, stalks, hull, husk, dust, char, waste, industry waste, refuse, and municipal solid waste by using the correlation proposed by Cordero et al. [ 6], Demirbaş [ 15], Jiménez and González [ 13], Küçükbayrak et al. [ 16]. The absolute average errors (AAEs) were 13.81%, 10.81%, 17.91%, and 6.66%, respectively. Nhuchhen and Abdul Salam [ 4] proposed a linear and nonlinear correlation for higher heating value prediction and compared the predicted result obtained from the linear correlation with the correlation proposed by Sheng and Azevedo [ 1]. An AAE of 10.94% was found. A comparison of the predicted result obtained from the nonlinear correlation and the one proposed by Cordero et al. [ 6]; Demirbaş [ 15]; Parikh et al. [ 14] was also conducted. The AAEs were found to be 17.80%, 11.72%, and 12.63%, respectively.

For the higher heating value prediction by using elemental analysis data, Yin [ 17] proposed a prediction model using the data of 44 biomass samples from agricultural byproducts, (e.g. rice husk, sugarcane, bagasse, and corn straw) to wood (willow, bamboo, and oak). The predicted result was compared with the prediction using the correlation proposed by Channiwala and Parikh [ 18], Friedl et al. [ 2], Jenkins and Ebeling [ 8], Sheng and Azevedo [ 1], and Tillman [ 19]. The predicted result showed a mean absolute error (MAE) of 4.01 MJ/kg. Cordero et al. [ 6] predicted the higher heating value of lignocellulosic biomass and charcoals using the correlation proposed by Francis and Lloyd [ 20]. The predicted result showed a mean average error of 2% compared with experiments.

In the past, the higher heating value prediction was focused only on the biomass and coals. Due to development of mild pyrolysis process, i.e., torrefaction, attempt was made to use the correlations proposed by researchers to predict the higher heating value of torrefaction char [ 21– 23]. However, most investigations were conducted with woody biomass while the study of non-woody biomass was insufficient. The higher heating value prediction of five types of non-woody biomass including sugarcane leaves, oil palm frond, rice straw, corncob, and cassava rhizome using correlations of previous researches were conducted in this paper. The error evaluation of predicted result was also reported. In addition, the new correlation for higher heating value prediction was presented.

Materials and methods

Biomass samples

Five types of non-woody biomass were selected. The list of biomass types and their species was listed in Table 1.

Each biomass sample was ground into a particle size of 1 mm and dried in a hot air oven at 105°C for 24 hours. The moisture content of biomass samples after drying was (10±1)% on wet basis.

Torrefaction process

Each biomass sample of 40 g was torrefied in a fixed bed cylindrical reactor. An electrical heater was used as the heat source for the reactor. The heating rate was controlled at 20°C/min. Nitrogen at a flow rate of 0.1 dm³/min was used as purge gas for removing the oxygen and volatile inside the reactor. The torrefaction temperatures were varied at 220°C, 260°C, and 280°C. The torrefaction times for each temperature were set at 10 min, 15 min, 20 min, 35 min, 40 min and 60 min.

Torrefaction char analysis

The proximate analysis data of the torrefied char were obtained according to ASTM D 3175-07 (Volatile matter (VM)), ASTM D 3173-03 (Moisture), ASTM D 3174-04 (Ash), while elemental analysis data were obtained according to ASTM D 3178-84 (carbon and hydrogen), ASTM D 3179-02 (nitrogen), and ASTM D 3177-02 (sulfur). The fixed carbon (FC) value was determined by subtracting the sum of moisture, ash, and volatile matter from 100 percent and the oxygen can be calculated by subtracting the total percentages of nitrogen, hydrogen, sulfur, and carbon from 100. The higher heating value was measured using a bomb calorimeter according to ASTM D5865-07. In all experiments, the uncertainties were calculated at a 95% confidence interval. The higher heating value obtained by the bomb calorimeter (HHV_exp) and the proximate and elemental analysis data of various raw biomass and torrefaction char were tabulated in Table 2.

Correlations for higher heating value prediction

The correlations for prediction were gathered from the works of Sheng and Azevedo [ 1], Friedl et al. [ 2], Nhuchhen and Abdul Salam [ 4], Jenkins and Ebeling [ 8], Cordero et al. [ 6], Thipkhunthod et al. [ 12], Jiménez and González [ 13], Parikh et al. [ 14], Demirbaş [ 15], Yin [ 17], and Chaniwala and Parikh [ 18]. The details of each correlation were presented in Table 3.

Error evaluation

To evaluate an error of the prediction compared with the experimental result, the AAE was used as shown in Eq. (20). The over prediction or under prediction of correlation was indicated by the absolute bias error (ABE) as shown in Eq. (21).

(20) \mathrm{AAE=}{\displaystyle \frac{1}{n}}{\displaystyle \sum _{i=1}^n\left|{\displaystyle \frac{{\mathrm{HHV}}_{\mathrm{p}}-{\mathrm{HHV}}_{\exp }}{{\mathrm{HHV}}_{\exp }}}\right|}\times 100\%,

(21) \mathrm{ABE=}{\displaystyle \frac{1}{n}}{\displaystyle \sum _{i=1}^n\left[{\displaystyle \frac{{\mathrm{HHV}}_{\mathrm{p}}-{\mathrm{HHV}}_{\exp }}{{\mathrm{HHV}}_{\exp }}}\right]}\times 100\%,

where HHV_p and HHV_exp denote the higher heating value obtained from prediction and experiment, respectively.

Results and discussion

Predicted results using proximate analysis correlations

The predicted result of the higher heating value of five types of non-woody biomass using various correlations based on the proximate analysis data is illustrated in Figs. 1 and 2, and Table 4. The predicted result using Eqs. (1) to (3) corresponds well with the experimental data as seen in Fig. 1. The prediction using Eqs. (4) and (9) gives a fair result compared with the experimental data. However, the under prediction is clearly seen as depicted in Fig. 2. The prediction using Eqs. (5) to (11) results in a poor prediction as seen in Table 4. When the data of error evaluation are considered as shown in Table 4, it is found that Eq. (1) gives the best prediction result with a R² of 0.999, an AAE of 3.61%, and an ABE of −0.29%.

Figure 3 demonstrates the relationship between the higher heating value obtained from the experiment and the mass fraction, based on a dry basis, of FC, VM, and ash, respectively. There is a linear relationship between the higher heating value and the mass fraction of FC, VM, and Ash. Thus, the general form of prediction correlation for the torrefaction char of non-woody biomass can be written as,

(22) \mathrm{H}\mathrm{H}{\mathrm{V}}_{\mathrm{p}}=\alpha \mathrm{F}\mathrm{C}+\beta \mathrm{V}\mathrm{M}+\gamma \mathrm{A}\mathrm{s}\mathrm{h},

where α, β, and γ represent the arbitrary constant. The term HHV_p denotes the higher heating value obtained by prediction.

According to the equation of mass conservation, the relationship between the FC, VM and Ash mass fraction based on dry basis can be expressed as

(23) \mathrm{F}\mathrm{C}+\mathrm{V}\mathrm{M}+\mathrm{A}\mathrm{s}\mathrm{h}=\mathrm{1.}

Inserting Eq. (23) into Eq. (22), the result can be written as

(24) \mathrm{H}\mathrm{H}{\mathrm{V}}_{\mathrm{p}}=\alpha +\left(\gamma -\alpha \right)\mathrm{A}\mathrm{s}\mathrm{h}+\left(\beta -\alpha \right)\mathrm{V}\mathrm{M}.

When the plot is done by setting the vertical axis as HHV_p and the horizontal axis as VM, the term [α +(γ−α)Ash] is the intersection of Eq. (24) on the vertical axis. The change in ash fraction results in a shift of linear line while the slope is constant. According to Fig. 3(b), the lowest linear line shows the linear relationship of rice straw which has the highest ash fraction of all five types of non-woody biomass. In contrast, the highest linear line shows the linear relationship of oil palm frond which had the lowest ash fraction. It can be seen that the higher ash fraction result in a lower intersection on the vertical axis. Thus, the term (γ−α) must be less than zero or

(25)

In addition, Fig. 3(b) displays the negative slope of linear relationship between the higher heating value of torrefaction char and mass fraction of volatile matter, thus, the term (β−α) in Eq. (24) is less than zero, i.e.,

(26)

Equations (25) and (26) mean that Eq. (22) would be satisfied for predicting the higher heating value of torrefaction char of non-woody biomass when the value of α constant is higher than the value of γ and β constants. According to Table 4, Eqs. (1) to (4), Eqs. (7) to (9), and Eq. (11) are in the form of Eq. (22). However, only Eqs. (1) to (3) have the value of α and β constants that satisfy Eq. (26). This results in a good prediction when Eqs. (1) to (3) are used. Because of the low value of α and β constants in Eqs. (4) and (9), the prediction value is lower compared with the value obtained from the experiment. In other words, the under prediction will be obtained. For Eqs. (7), (8), and (11), the value of α and β constants are almost equal. Thus, the prediction result using these three equations are poor.

Although Eqs. (1) to (3) give a good predicted result from the view point of the overall data, the prediction in case of a high ash fraction found in biomass and torrefaction char, such as rice straw, does not support the experimental data as seen in Fig. 1. The reason for this is that the ash fraction does not show a clear effect when these three correlations are used. Three types of non-woody biomass used in this paper including rice straw, corncob, and cassava rhizome have a high variation in ash fraction when they are torrefied. Thus, it is necessary to develop a new correlation.

It can be concluded that the correlation suggested in literature which gives the best prediction for five types of non-woody biomass and their torrefaction char has an AAE of 3.61% and an ABE of −0.29%. The mass fraction of FC and VM has a strong effect on the higher heating value prediction of torrefaction char of non-woody biomass. The high ash fraction found in biomass and torrefaction char results in a decrease in prediction accuracy.

Developing of a new correlation and its predicted results

According to Eqs. (24) to (26), the value of α, β, and γ constants can be obtained by the multiple regression method. The experimental results obtained in this paper and the data reported by Cordero et al., 2001 [ 6] are used to calculate these three constants. The correlation developed in this paper can be written as

(27) \mathrm{H}\mathrm{H}{\mathrm{V}}_{\mathrm{p}}=\mathrm{35.4879}-\mathrm{0.3023}\mathrm{A}\mathrm{s}\mathrm{h}-\mathrm{0.1905}\mathrm{V}\mathrm{M}.

Table 5 shows the value of the constants α, β, and γ, and correlation coefficient (R²). The AAE and ABE from the prediction are also shown in Table 5. The AAE is 2.74% when the proposed correlation is used to predict only the experimental data conducted in this paper. Compared with the AAE obtained by using the correlation proposed by Cordero et al. [ 6] as shown in Table 4, the AAE of the proposed correlation is lower than that of Cordero et al.’s correlation.

Figure 4 shows the comparison of predicted result using the correlation proposed in this paper with the experimental data. The higher heating value prediction of rice straw using the correlation proposed in this paper is closer to the experimental data than the prediction using the correlation proposed by Cordero et al. [ 6] (see Fig.1(a)). This indicates that the effect of high variation of ash fraction in torrefaction char of non-woody biomass is taken into account in the correlation proposed in this paper.

For sugarcane leaves and oil palm frond, torrefaction results in a decrease in volatile mass fraction whereas it causes a slight change in ash fraction. Thus, the change in ash fraction does not affect the prediction accuracy when Eqs. (1), (2), (4), and (9) are used. However, for torrefaction of rice straw, cassava rhizome, and corncob, there is a significant change in ash fraction. This results in an increase in the error when Eqs. (1), (2), (4), and (9) are used. Although the ash fraction term is contained in Eq. (3), the multiplying factor of ash fraction is not large enough to cause a significant change of the higher heating value. The suitable multiplying factor of ash fraction is necessary to obtain an accurate prediction of torrefied char that has high ash fraction.

It can be concluded that the correlation proposed in this paper gives a better prediction than the correlation suggested by the previous work, and has an AAE of 2.74% and an ABE of 0.52%.

Predicted results using elemental analysis correlation

The predicted result of five types of non-woody biomass using various correlations based on elemental analysis data is shown in Table 4. Equations (12) to (14) lead to a good predicted result. Equation (15) and Eqs. (16) to (18) result in an over and under prediction, respectively. However, these equations give a fair result compared with the experimental data. Equation (19) results in a poor prediction. It is found that Eq. (12) gives the best prediction result with an AAE of 2.28%, and an ABE of 1.36%. Figure 5 shows the relationship between the higher heating value and the mass fraction based on a dry basis of carbon (C), hydrogen (H), oxygen (O), nitrogen (N), and sulfur (S). It is found that carbon, hydrogen, and oxygen have a strong effect on the higher heating value, while no relationship is found between the higher heating value and mass fraction of nitrogen and sulfur.

Conclusions

This paper predicted the higher heating value of five types of non-woody biomass and their torrefaction char using the correlations presented by the previous researches. The new correlation developed in this paper was also introduced. It was found that the correlation based on proximate analysis data suggested in literature which gives the best prediction for this work has an AAE of 3.61% and an ABE of −0.29%. The mass fraction of FC and VM had a strong effect on the higher heating value prediction of torrefaction char of non-woody biomass. The high ash fraction found in biomass and torrefaction char resulted in a decrease in prediction accuracy.

The prediction result of torrefaction char of non-woody biomass can be improved by taking into account the effect of ash fraction. The correlation developed in this paper gives a better prediction than the correlations suggested by the previous work, with an AAE of 2.74% and an ABE of 0.52%.

The best correlation based on elemental analysis data suggested by literature showed an AAE of 2.28% and an ABE of 1.36%. Carbon, hydrogen and oxygen had a strong effect on higher heating value, while no relationship is found between the higher heating value and the mass fraction of nitrogen and sulfur.

References

[1] Sheng C, Azevedo J L T. Estimating the higher heating value of biomass fuels from basic analysis data. Biomass and Bioenergy, 2005, 28(5): 499–507

[2] Friedl A, Padouvas E, Rotter H, Varmuza K. Prediction of heating values of biomass fuel from elemental composition. Analytica Chimica Acta, 2005, 544(1–2): 191–198

[3] Shen J, Zhu S, Liu X, Zhang H, Tan J. The prediction of elemental composition of biomass based on proximate analysis. Energy Conversion and Management, 2010, 51(5): 983–987

[4] Nhuchhen D R, Abdul Salam P. Estimation of higher heating value of biomass from proximate analysis: a new approach. Fuel, 2012, 99: 55–63

[5] Akkaya A V. Proximate analysis based multiple regression models for higher heating value estimation of low rank coals. Fuel Processing Technology, 2009, 90(2): 165–170

[6] Cordero T, Marquez F, Rodriguez-Mirasol J, Rodriguez J J. Predicting heating values of lignocellulosics and carbonaceous materials from proximate analysis. Fuel, 2001, 80(11): 1567–1571

[7] Erol M, Haykiri-Acma H, Küçükbayrak S. Calorific value estimation of biomass from their proximate analyses data. Renewable Energy, 2010, 35(1): 170–173

[8] Jenkins B M, Ebeling J M. Correlation of physical and chemical properties of terrestrial biomass with conversion. In: Proceedings of Energy from Biomass and Wastes IX, Chicago, 1985, 371–400

[9] Kathiravale S, Muhd Yunus M N, Sopian K, Samsuddin A H, Rahman R A. Modeling the heating value of Municipal Solid Waste. Fuel, 2003, 82(9): 1119–1125

[10] Callejón-Ferre A J, Velázquez-Marti B, López-Martínez J A, Manzano-Agugliaro F. Greenhouse crop residues: energy potential and models for the prediction of their higher heating value. Renewable  &  Sustainable  Energy  Reviews,  2011,  15(2):  948–955

[11] Graboski M, Bain R. Properties of biomass relevant to gasification [Fuel gas production]. Biomass gasification—principles and technology, 1981, 41–69

[12] Thipkhunthod P, Meeyoo V, Rangsunvigit P, Kitiyanan B, Siemanond K, Rirksomboon T. Predicting the heating value of sewage sludges in Thailand from proximate and ultimate analyses. Fuel, 2005, 84(7–8): 849–857

[13] Jiménez L, González F. Study of the physical and chemical properties of lignocellulosic residues with a view to the production of fuels. Fuel, 1991, 70(8): 947–950

[14] Parikh J, Channiwala S A, Ghosal G K. A correlation for calculating elemental composition from proximate analysis of biomass materials. Fuel, 2007, 86(12–13): 1710–1719

[15] Demirbaş A. Calculation of higher heating values of biomass fuels. Fuel, 1997, 76(5): 431–434

[16] Küçükbayrak S, Dürüs B, Meríçboyu A E, Kadioġlu E. Estimation of calorific values of Turkish lignites. Fuel, 1991, 70(8): 979–981

[17] Yin C Y. Prediction of higher heating values of biomass from proximate and ultimate analyses. Fuel, 2011, 90(3): 1128–1132

[18] Channiwala S A, Parikh P P. A unified correlation for estimating HHV of solid, liquid and gaseous fuels. Fuel, 2002, 81(8): 1051–1063

[19] Tillman D. Wood as an energy resource. New York: Academic, 1978

[20] Francis H E, Lloyd W G. Predicting heating value from elemental composition. Journal of Coal Quality, 1983, 2: 2

[21] Bridgeman T G, Jones J M, Williams A, Waldron D J. An investigation of the grindability of two torrefied energy crops. Fuel, 2010, 89(12): 3911–3918

[22] Chen W H, Cheng W Y, Lu K M, Huang Y P. An evaluation on improvement of pulverized biomass property for solid fuel through torrefaction. Applied Energy, 2011, 88(11): 3636–3644

[23] Medic D, Darr M, Shah A, Potter B, Zimmerman J. Effects of torrefaction process parameters on biomass feedstock upgrading. Fuel, 2012, 91(1): 147–154

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg