Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 107 (2017) 130 136

3rd International Conference on Energy and Environment Research, ICEER 2016, 7-11 September
2016, Barcelona, Spain

Prediction of Calorific Value of Biomass from Proximate Analysis

Aye zyuuran*, Serdar Yaman
Chemical Engineering Department, Istanbul Technical University, Istanbul, Turkey

Abstract

Biomass is one of the renewable and sustainable energy sources that does not lead greenhouse gas emissions. Efficient use of
biomass energy will help to solve problems resulting from fossil fuels. However, the main concern relevant to use of this energy
is mainly related to low calorific value of biomass. Therefore, calorific value is the key parameter to evaluate the fuel quality of a
special biomass material in energetic applications. In this context, twenty-seven different biomass species that represent very
wide range of biomass materials such as herbaceous and woody biomasses, nut shells, fruit stones, stem and husks, pulps, and
agricultural residues have been characterized by proximate analysis (moisture, volatile matter, fixed carbon, and ash contents).
Then, various empirical equations which contain linear and nonlinear terms have been tested in order to predict the higher
heating values (HHV) of full sample set from the proximate analysis results. It was concluded that since biomasses used in this
study have different structures and fuel characteristics, the predicted HHVs for the full sample set were a bit different from the
experimental HHVs and the r2 of these equations varied in the range of 0.812-0.837, while standard deviations were between
1.469 and 1.493 MJ/kg. Nevertheless, considering the number of the biomass species used in this study and their differences in
properties, these standard deviations may be regarded in the acceptable limits.
2016 The Authors. Published by Elsevier Ltd.
2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the scientific committee of the 3rd International Conference on Energy and Environment
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Research.
Peer-review under responsibility of the scientific committee of the 3rd International Conference on Energy and Environment Research.
Keywords: Biomass; calorific value; proximate analysis; prediction; correlation.

1. Introduction

Increasing energy need forces societies to find alternative energy sources that are cheap, abundant and have low

* Corresponding author. Tel.: +90 2122856638; fax: +90 2122853425.

E-mail address: kmayse@itu.edu.tr

1876-6102 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of the 3rd International Conference on Energy and Environment Research.
doi:10.1016/j.egypro.2016.12.149
 Aye zyuuran and Serdar Yaman / Energy Procedia 107 (2017) 130 136 131

impact on environment. Fossil fuels are highly far from meeting these requirements. From this point of view, biomass
is regarded as an energy source which is abundant, easy to find, renewable, and sustainable [1]. On the other hand, in
contrast to fossil fuels, biomass is allowed as CO2-neutral fuel since it does not contribute to the net increase in CO2
emissions in atmosphere. That is, the amount of CO2 formed upon burning of biomass is almost equivalent to the CO2
captured from atmosphere during photosynthesis of green biomass. For these reasons, biomass is of great interest in
renewable energy projects worldwide to mitigate greenhouse gases and to lower the share of fossil fuels in energy
consumption [2].
The main concern relevant to use of biomass energy resources is that the energy density of biomass is typically
lower than that of fossil fuels such as coals and the calorific value of biomass is closely affected from moisture
content that it may reach very high levels especially in case of green biomass and waste materials. Besides, the
experimental methods to estimate the heating value of a fuel are usually time consuming as well as expensive and
have higher possibilities of experimental errors. In addition, the heating value of biomass cannot be calculated
merely from the heats of formation of CO2, H2O and other products because biomass is a complex mixtures of
compounds, and the relevant bond energies cannot be estimated properly [3]. On the other hand, there still exist
extra unknown factors that influence the higher heating value (HHV) of biomass and such factors add more
difficulty in modeling of this property [3].
In addition, the volatile matter content of biomass reaches up to 90 wt.% in some cases and in turn fixed carbon
contents become too low [4]. The fixed carbon content of biomass can be easily connected with the calorific value
since it has a positive effect on the energy potential of biomass. On the other hand, this situation is much more
complicated for the volatile matter content. Because, high volatile matter content does not guarantee high calorific
value since some of the ingredients of volatile matter are formed from non-combustible gases such as CO2 and H2O.
Besides, ash has not only an inert effect on the calorific value of a fuel but also it shows some detrimental effects on
the apparent heat obtained from burning of biomass. That is, the energy need of ash forming inorganics for thermal
breakdown and phase transition is taken from the burning energy of biomass and it leads reduction in the calorific
value. These relations between the calorific value of biomass and the proximate analysis results (moisture, volatile
matter, fixed of carbon, and ash contents) motivates the researchers to develop various mathematical models to
predict the calorific value of biomass depending on the proximate analysis results [5-10]. Apart from these studies,
several alternative approaches have also been implemented based on ultimate analysis results (C, H, N, S, and O
contents), physical or chemical compositions of biomass, and even artificial neural network (ANN) was used for
modelling of HHVs [8]. Channiwala and Parikh [11] performed a detailed literature survey on these alternative
methods and the systematic development sequence of these procedures for proximate analysis as well as ultimate
analysis starting from 1880 can be seen in this paper. However, ultimate analysis and determination of the
macromolecular ingredients of biomass such as lignin, cellulose, and hemicellulose require sophisticated equipment
with good calibration and very sensitive analytical experiments. Likewise, the methods based on artificial neural
network necessitate very large number of experimental data to validate the predicted model and then test it using the
existing experimental data. In contrary to these methods, the proximate analysis of biomass only needs a simple oven
(for determination of moisture content) and a furnace (for determination of volatile matter and ash contents) which
can be easily found in an ordinary laboratory. Consequently, multiple linear regression (MLR) models that can be
used for modelling of calorific value of biomass have been developed basing on proximate analysis results of
miscellaneous biomass materials and the standard deviation of the predicted values are generally taken into
consideration to evaluate the suitability of the predicted models. In this context, the present paper aims to establish
various linear and non-linear empirical equations which can be used to predict the HHV values of a number of
different biomass species basing on the proximate analysis results, and to check their suitability considering standard
deviations and r2 values that basically measures the goodness-of-fit in the regression analysis.

2. Materials and methods

In this study, twenty-seven different biomass species have been used. All of these biomass samples were Turkish
origin and they were collected as residues or by-products from agricultural activities, forestry sector as well as food
industry throughout the country. Therefore, these samples represent very wide range of biomass sub-classes. Sample
names and codes of these biomasses are as follows: Elaeagnus (1), green bean stem and husk (2), red lentil hull (3),
132 Aye zyuuran and Serdar Yaman / Energy Procedia 107 (2017) 130 136

chickpea husk (4), cornstalk (5), tobacco waste (6), broad bean husk (7), apricot stone (8), apricot pulp (9), peach
pulp (10), damson plum stone (11), coconut shell (12), cornelian cherry stone (13), cacao husk (14), peanut husk
(15), broad bean husk (16), almond shell (17), thyme (18), walnut shell (19), locust bean (20), apple pulp (21),
sunflower stem (22), sour cherry stem (23), soybean residue (24), pea stem (25), pine cone (26), and peach stone
(27). As it can be seen from this long list, very wide range of samples such as herbaceous and woody biomasses, nut
shells, fruit stones, stem and husks, pulps, and agricultural residues were chosen for this study to investigate the
possibility of using proximate analysis results to predict HHV values of biomass materials that differ considerably in
terms of origin and fuel characteristics. However, this versatile nature of the biomass species used in this study also
indicate the difficulty of generating a general formula that represent all of the samples.
In sample preparation stage, these biomass species were kept in open trays for several days to get air-dried
samples to avoid biological decay of wet samples. The air-dried samples were milled with a special shredder
designed for leafy materials and then sieved to pass through a screen that has openings of 250 m. The milled and
sieved samples were stored in air-tight sample bottles to avoid further interaction with air. The proximate analysis
(moisture, volatile matter, ash, and fixed carbon contents) of the samples was carried out according to ASTM
standards. HHV (higher heating value) of the samples was determined by IKA C2000 model calorimeter. Proximate
analysis and HHV determinations were repeated several times to check the accuracy and the precision of the
experimental results, and data that deviate more than 5% of the average value was discarded to assure the
repeatability of the experimental data.

3. Results and discussion

The results of proximate analysis and HHV are given in Table 1. The abbreviations of VM and FC in this table
denote volatile matter and fixed carbon contents, respectively. Actually; moisture, volatile matter and ash contents
are determined experimentally in proximate analysis and then the fixed carbon content is calculated by subtraction
the sum of moisture, volatile matter, and ash contents from 100. On the other hand, VM, FC, and ash contents were
also normalized to get the moisture-free (dry) values since the comparison of the sample characteristics became
more meaningful in this way. For this reason, the moisture contents were excluded in Table 1.
The results in Table 1 revealed the fact that the proximate analysis results of samples changed in very wide
ranges. That is, the minimum value of volatile matter (70.29 %) belongs to locust bean (sample 20) while the
maximum value of volatile matter (86.42 %) belongs to peach stone (sample 27). Peach stone is a typical woody
ligno-cellulosic material for which the volatile matter content is seriously high. On the other hand, the lowest value
of fixed carbon (11.78 %) was determined in case of tobacco waste (sample 6), while the highest value of fixed
carbon (24.16 %) was observed for locust bean (sample 20). This also confirms that the fuel quality of tobacco
waste is very poor. Besides, the lower limit and the upper limit of fixed carbon indicate that the maximum value of
fixed carbon is more than twice of the minimum one. In addition to this, the minimum and the maximum values of
ash yields were determined as 0.15 % and 15.36 % for peanut husk (sample 15) and tobacco waste (sample 6),
respectively. In addition to low fixed carbon content, the high ash yield also confirms the poor fuel quality of
tobacco waste. On the other hand, these yields of ashes also revealed that the most important variations in results
took place in the ash yields since some of the biomass species are waste materials which are rich in ash forming
inorganics, while woody biomasses do not contain high amount of inorganics. Similarly, the HHVs also changed
considerably from sample to sample since they have very different fuel characteristics. Namely, the lowest HHV
value was determined as 14.51 MJ/kg for tobacco waste (sample 6) and the highest HHV value was determined as
21.23 MJ/kg for damson plum stone (sample 11). In this context, Nhuchhen and Abdul Salam [7] reported the
higher heating value results ranging from 5.63 to 23.46 MJ/kg whose lower limit (5.63 MJ/kg) is seriously lower
than the HHV results given in Table 1.
In order to establish mathematical equations between the HHV values of the biomass samples and their
proximate analysis results, some empirical equations have been developed. The coefficients in these equations were
calculated according to ordinary least squares (OLS) method. These equations given in Table 2 have both linear
terms and some nonlinear terms of volatile matter (VM), fixed carbon (FC), and ash (A). Namely, equations 1-4
contain only linear terms, while equation 5-8 are comprised of combinations of both linear and nonlinear terms.
Thus, influences of proximate analysis results on HHV were tested considering various situations.
 Aye zyuuran and Serdar Yaman / Energy Procedia 107 (2017) 130 136 133

Equation 1 considers effects of all of variables of proximate analysis. Therefore, FC, VM, and A were included to
this equation. Alternatively, effect of ash was excluded in Equation 2, and FC and VM which are the main sources
of the calorific value formed the equation. Equations 3 and 4 consider combined effects of FC and A as well as VM
and A, excluding effects of VM and FC, respectively. On the other hand, equations from 5 to 8 are comprised of
both linear and nonlinear parameters where inverse effects of either VM or FC were included to these equations.
Similarly, Nhuchhen and Abdul Salam [7] also used non-linear equations containing FC/VM, VM/FC, FC/ash,
Ash/FC, VM/ash, ash/VM, and combinations of these parameters to predict the HHVs. The equations tested to
predict the HHVs of full sample set based on the proximate analysis results are tabulated in Table 2.

Table 1. Analysis results

Moisture-free basis
Code VM (%) FC (%) Ash (%) HHV (MJ/kg)
1 76.43 22.67 0.90 19.80
2 77.88 12.52 9.60 16.86
3 74.73 20.03 5.24 18.27
4 79.70 16.32 3.98 18.26
5 76.79 16.02 7.19 16.55
6 72.85 11.78 15.36 14.51
7 74.97 17.33 7.70 16.24
8 82.03 16.26 1.61 19.79
9 78.80 20.12 1.08 18.52
10 71.21 23.34 5.44 18.23
11 82.33 16.67 1.00 21.23
12 83.01 15.97 1.02 20.24
13 79.27 18.67 2.06 19.84
14 75.03 20.56 4.41 17.85
15 79.32 20.53 0.15 19.16
16 74.88 19.40 5.73 16.80
17 81.70 14.82 3.48 19.53
18 75.04 16.00 8.96 18.16
19 82.98 16.12 0.90 20.03
20 70.29 24.16 5.54 18.04
21 82.56 15.13 2.31 19.85
22 75.09 13.81 11.10 16.18
23 77.48 17.16 5.36 18.27
24 80.01 13.54 6.45 19.26
25 74.67 12.27 13.06 16.31
26 80.92 17.48 1.60 20.07
27 86.42 12.61 0.97 20.31

All of linear equations in Table 2 (Eq. 1-4) gave very close results of coefficient of determination (r2) that this
coefficient varied in a very narrow interval of 0.826-0.827. Likewise, the standard deviations changed between
1.482-1.484 MJ/kg for these equations. On the other hand, r2 of the equations containing nonlinear parameters (Eq.
5-8) were between 0.812 and 0.837, while their standard deviations varied between 1.469 and 1.493 MJ/kg. As it
can be seen from these values, the proximate analysis results of biomass samples produced very near results
irrespective of the type of equations investigated in this study. Since, a number of different biomass species
accounted for the samples used in this study and their characteristics are highly diverse, it is not surprising to find r2
values which are not so close to 1. Nevertheless, the standard deviations were not awfully high and the maximum
value was only 1.493 MJ/kg. These numbers of standard deviations are comparable with the results in literature
found by Yin [6] who tested 44 different biomass samples and determined that HHV calculated from the proximate
analysis results has a standard deviation of 1.35 MJ/kg. Likewise, Gillespie et al. [4] also tried to predict HHV of
full sample set of many biomass materials basing on the analysis results and found that the value of r2 was
approximately 0.88.
134 Aye zyuuran and Serdar Yaman / Energy Procedia 107 (2017) 130 136

Fig. 1. Relations between experimental and predicted HHV value.

Fig. 1 shows the relations between the experimental and predicted HHVs for full set of the samples using all of
the eight equations. This figure also facilitates interpretation of the suitability of the predictions with respect to the
magnitude of HHV and evaluating the samples as subclasses in terms of their HHV levels. It is clear from the
distributions of the points on these figures that serious deviations took place between some of the experimental and
 Aye zyuuran and Serdar Yaman / Energy Procedia 107 (2017) 130 136 135

predicted values of HHV particularly in cases of the samples that have calorific values lower than 20 MJ/kg. Yin [6]
also reported the deviation between experimental and predicted values for low HHV samples. In contrast to this, the
samples that have HHVs around 20 MJ/kg fitted well and they situated almost on the regression lines. Besides, the
samples that have calorific value of approximately 18 MJ/kg showed that the predicted HHVs are usually lower than
the experimental HHVs. Actually its quite difficult to explain the reason of this fact properly. Basing on this, it can
be suggested that several sub-classes comprised of samples with relatively comparable calorific values may be
analyzed for a better prediction.

Table 2. Equations.
Eq. No Equation r2 St. dev. (MJ/kg)
1 HHV=167.2-1.449VM-1.562FC-1.846A 0.827 1.483

2 HHV=-17.507+0.3985VM+0.2875FC 0.826 1.484

3 HHV=22.3418-0.1136FC-0.3983A 0.827 1.482
4 HHV=10.982+0.1136VM-0.2848A 0.827 1.482

5 HHV=-18.37-0,8469FC-1,1251A+ 4420 0.837 1.493

VM
2394.7
6 HHV=44.336+0.286FC- 0.812 1.469
VM

656.2
7 HHV=28.296-0.2887A- 0.823 1.479
VM

35.8
8 HHV=18.297-0.4128A+ 0.832 1.487
FC

4. Conclusion

Heating value of biomass is necessary to analyze and design any bio-energy systems. For this reason, biomass
materials that represent very wide range of renewable energy sources have been characterized in this study by
applying proximate analysis to compare their fuel properties. The results of proximate analysis such as volatile
matter, fixed carbon, and ash have been used to predict the higher heating values (HHVs) of the biomass materials
through some linear and non-linear empirical equations. It can be concluded from the outcomes of this study that
this method is capable of predicting the HHVs with r2 values changing in the intervals of 0.826-0.827 and 0.812-
0.837 for linear and non-linear equations, respectively. However, in case of biomass species with relatively low
calorific value, the deviations between the experimental and predicted HHVs cannot be ignored and they may be
evaluated as a special group. Besides, satisfactory predictions could be obtained when the calorific value of biomass
is bigger than 20 MJ/kg. Therefore, it should be reasonable to make subclasses such as herbaceous samples, woody
samples, agricultural residues, etc. and each subclass may be evaluated separately.

References

[1] Basu P. Biomass gasification and pyrolysis-practical design and theory. Oxford: Elsevier; 2010. pp. 27-55.
[2] Klass DL. Biomass for renewable energy, fuels, and chemicals. London: Academic Press; 1998. pp. 35-37.
[3] Friedel A, Padouvas E, Rotter H, Varmuza K. Prediction of heating values of biomass fuel from elemental composition. Analytica Chimica
Acta 2005;544:191-198.
[4] Gillespie GD, Edward CD, Fagan CC, McDonnell KP, Bean CP. Prediction of quality parameters of biomass pellets from proximate and
ultimate analysis. Fuel 2013;111:771-777.
[5] Parikh J, Channiwala SA, Ghosal GK. A correlation for calculating elemental composition from proximate analysis of biomass materials.
Fuel 2007;86:1710-1719.
[6] Yin C. Y. Prediction of higher heating values of biomass from proximate and ultimate analyses. Fuel 2011;90:1128-1132.
[7] Nhuchhen DR, Abdul Salam P. Estimation of higher heating value of biomass from proximate analysis: A new approach. Fuel 2012;99:55-
63.
136 Aye zyuuran and Serdar Yaman / Energy Procedia 107 (2017) 130 136

[8] Shen JS, Zhu S, Liu X, Zhang H, Tan J. The prediction of elemental composition of biomass based on proximate analysis. Energy
Conversion and Management 2010;51:983-987.
[9] Erol M, Haykiri-Acma H, Kucukbayrak S. Calorific value estimation of biomass from their proximate analyses data. Renewable Energy
2010;35:170-173.
[10] M. Proximate analyses and predicting HHV of chars obtained from cocracking of petroleum vacuum residue with coal, plastics and biomass.
Bioresource Technology 2008;99:5043-5050.
[11] Channiwala SA, Parikh PP. A unified correlation for estimating HHV of solid, liquid and gaseous fuels. Fuel 2002;81:1051-1063.