You are on page 1of 10

Fuel 86 (2007) 1710–1719

www.fuelfirst.com

A correlation for calculating elemental composition from


proximate analysis of biomass materials
a,*
Jigisha Parikh , S.A. Channiwala b, G.K. Ghosal c

a
Chemical Engineering Department, Sarvajanik College of Engineering and Technology, Surat 395 001, India
b
Mechanical Engineering Department, Sardar Vallabhbhai National Institute of Technology, Surat 395 007, India
c
Petroleum Refining and Petrochemical Technology Department, Laxminarayan Institute of Technology, Nagpur, India

Received 8 December 2006; accepted 23 December 2006


Available online 23 February 2007

Abstract

Elemental composition of biomass is an important property, which defines the energy content and determines the clean and efficient
use of the biomass materials. However, the ultimate analysis requires very expensive equipments and highly trained analysts. The prox-
imate analysis on the other hand only requires standard laboratory equipments and can be run by any competent scientist or engineer.
This work introduces a general correlation, based on proximate analysis of biomass materials, to calculate elemental composition,
derived using 200 data points and validated further for additional 50 data points. The entire spectrum of solid lignocellulosic materials
have been considered in the derivation of the present correlation, which is given as: C = 0.637FC + 0.455VM, H = 0.052FC + 0.062VM,
O = 0.304FC + 0.476VM, where FC – 4.7–38.4% fixed carbon, VM – 57.2–90.6% volatile matter, C – 36.2–53.1% carbon, H – 4.36–8.3%
hydrogen and O – 31.37–49.5% oxygen in wt% on a dry basis. The average absolute error of these correlations are 3.21%, 4.79%, 3.4%
and bias error of 0.21%, 0.15% and 0.49% with respect to measured values C, H and O, respectively. The major advantage of these
correlations is their capability to compute elemental components of biomass materials from the simple proximate analysis and thereby
provides a useful tool for the modeling of combustion, gasification and pyrolysis processes.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Biomass materials; Proximate analysis; Ultimate analysis

1. Introduction average absolute error of 1.45% and average bias error of


0%. This generalized correlation, while being extremely
The field of thermo chemistry is one of the foundation useful in performance modeling of thermal systems, suffers
stones of our modern energy dependant society. The prox- from the drawback that it needs an elemental analysis as an
imate and ultimate analysis of biomass and coal are neces- input data, which needs expensive equipment and highly
sary for their efficient and clean utilization, while the HHV skilled analysts [2–4]. Considering this, another correlation
of these materials determine the quantitative energy con- was developed for predicting higher heating value (HHV)
tent of these fuels. There exists a variety of correlations of solid fuels from the proximate analysis, which is much
for predicting HHV from ultimate analysis of fuel [1]. easier to perform and needs only a balance, a furnace
One such correlation, presented in the earlier work by and moderately trained analysts [5]. This correlation has
Channiwala and Parikh [1], encompasses the complete been derived using 450 data points and additional 100 data
spectrum of fuel ranging from gaseous, liquid, solid and points used for its validation, and it offers predictions with
refuse derived fuels. This correlation offers predictions with average absolute error of 3.74% and average bias error of
0.12%.
*
Corresponding author. During this study, it was observed that despite the
E-mail address: kamal_parikh@yahoo.com (J. Parikh). rapidly developing biomass literature and many recent

0016-2361/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2006.12.029
J. Parikh et al. / Fuel 86 (2007) 1710–1719 1711

conferences on finding new ways of extracting energy out 2.2. Step 2: Selection of suitable data
of coal and biomass, relatively few efforts have been
devoted to finding correlations between feedstock composi- Through the process of collection and generation of
tion and properties. Another important fact revealed from data, information on about 250 biomass materials was col-
the literature review is that no correlation is available, to lected. Out of these, 200 data points have been used for the
predict elemental composition from proximate analysis of purpose of the derivation of the correlation, while 50 were
biomass materials except the correlation given by Vakkilai- used for validation of the correlation.
nen [6] specifically for black liquor only. The characteriza- In order to develop a generalized correlation to the
tion of biomass is vital for establishment of its potential as extent possible, care was taken to include data points of
well as the efficient operation of an energy conversion sys- the widest range of proximate analysis and ultimate analy-
tem. The analysis as above focuses towards the need for sis. Table 1 presents the summary of few such data points
development of correlations to predict elemental composi- used for derivation and validation of the correlation. In
tion from the proximate analysis of biomass materials. total, 250 data points comprising 28 data points of pits,
shells and seeds, 80 data points for bark, wood and energy
2. Derivation of the correlation of elemental composition crops, 45 data points for straws/stalks/cobs, 47 data points
with the proximate analysis for fibrous material/hull-husk/dust and 50 data points for
miscellaneous material like grass, leaves, waste etc. have
The steps involved in the development and derivation of been used for derivation and validation of the correlation.
correlations are listed below: The spectrum of samples was so selected that it approx-
imately represents the relative proportions of their occur-
2.1. Step 1: Collection and generation of data rence in nature and thus permits the derivation of a truly
useful correlation.
To obtain the correlation, proximate and ultimate anal- The range of data points considered for correlation
ysis data pertaining to different biomass-forest and agricul- between ultimate and proximate analysis includes the car-
tural wastes – have been used, so as to cover a wide range bon content between 36.2% and 53.1%, hydrogen content
of different values of fixed carbon, volatile matter, ash con- 4.36–8.3%, oxygen content 31.37–49.5%, volatile matter
tent as well as carbon, hydrogen and oxygen contents. content 57.2–90.6% and fixed carbon content 4.7–38.4%
(wt% on dry basis).
2.1.1. Collection of data
Data pertaining to a large number of biomass materials 2.3. Step 3: Selection of suitable forms of correlations
like pits, shells, seeds, cobs, energy crops, fuel wood, bark,
straws, stalks, hull-husk, fibrous material etc., were col- The proximate analysis on a dry basis gives volatile mat-
lected from the published literature. ter, ash and fixed carbon content whereas the ultimate
analysis shows carbon, hydrogen, oxygen etc., in the ele-
2.1.2. Generation of data mental composition. The fixed carbon indicates the extent
Since there are no standard sampling procedures speci- of non-volatile organic matter in the sample, which may
fied for biomass materials, the samples were collected with also contain oxygen along with hydrogen. The volatile
due care to get the most representative samples. The prep- matter content reflects extent of the volatile organic matter
aration of samples was carried out in accordance with in the sample. Considering both types of analysis i.e., ulti-
ASTM D 2013-86 [7]. The procedure requires samples to mate and proximate, which indicate the presence of
be in powder form of up to 250 lm grain size. The proxi- carbon, hydrogen, oxygen etc., in different forms, it is log-
mate analysis of samples were carried out in accordance ical to assume a relation between the results of these two
with ASTM D 3172-73(84) [8] standard. The moisture con- analyses.
tent in test samples was determined according to ASTM D In view of the above, it can be said that elemental car-
3173-87 [9] method in a Sartorious infrared moisture meter. bon, hydrogen and oxygen are proportional to the fixed
The volatile matter contents in the test samples were deter- carbon and volatile matter content of the material. Hence,
mined according to ASTM D 3175-89 [10] modified it is assumed that carbon, hydrogen and oxygen are inde-
method for sparking fuels. For coconut shell, fiber, pith pendently a function of fixed carbon (FC, wt%) and vola-
and bagasse, ash fusion was observed at 950 °C and hence tile matter (VM, wt%). Table 2 presents the different
these samples were analyzed for volatile matter contents forms of algebraic expressions with criteria for their selec-
only up to 750 °C. Determination of the ash content in tion. An exhaustive computer algorithm has been devel-
the test samples was carried out according to ASTM D oped based on a generalized method of least squares to
3174-89 [11] method in the electric muffle furnace. The evaluate the constants of these assumed algebraic expres-
fixed carbon content of the test samples was calculated sions. It is worth noting that the algorithm is capable of
by difference. The ultimate analysis of these samples was incorporating any number of variables and data points.
carried out according to ASTM Standard D 3176 to 79– Using this algorithm and 200 data points, the constants
84 [12–14]. of these algebraic expressions were evaluated.
1712
Table 1
Summary of few salient data used for derivation and validation of correlations
Serial no. Raw material Proximate analysis Ultimate analysis (wt% by dry basis) References
(wt% by dry basis) Carbon (%) Hydrogen (%) Oxygen (%)
FC VM ASH Measured Calculated Error Measured Calculated Error Measured Calculated Error
1 Olive pit 21.2 75.6 3.2 48.81 47.90 1.86% 6.23 5.79 7.07% 43.48 42.43 2.41% [15]
2 Peach pit 19.8 79.1 1.1 49.14 48.60 1.09% 6.34 5.93 6.41% 43.52 43.67 0.35% [16]
3 Coconut shell 22.1 77.19 0.71 50.22 49.20 2.03% 5.7 5.93 4.12% 42.56 43.46 2.12% [17]
4 Akhrot shell 18.78 79.98 1.2 49.81 48.35 2.92% 5.64 5.94 5.24% 42.94 43.78 1.96% [17]
5 Pistachio shell 16.84 82.03 1.13 48.79 48.05 1.52% 5.91 5.96 0.87% 43.41 44.17 1.74% [18]
6 Groundnut shell 19.48 73.72 6.8 45.72 45.95 0.51% 5.96 5.58 6.32% 41.02 41.01 0.02% PS
7 Brazil nut shell 22.2 76.1 1.7 49.15 48.77 0.78% 5.7 5.87 3.03% 42.8 42.97 0.40% [19]

J. Parikh et al. / Fuel 86 (2007) 1710–1719


8 Castor seed shell 20 72 8 44.25 45.50 2.82% 5.64 5.50 2.41% 41.94 40.35 3.79% [20]
9 Loblolly pine 33.9 65.7 0.4 56.3 51.49 8.55% 5.6 5.84 4.22% 37.7 41.58 10.29% [21]
10 Douglas fir bark 32.79 65.46 1.75 53.1 50.67 4.57% 6.1 5.76 5.51% 40.6 41.13 1.30% [22]
11 Wood bark 31.8 66.6 1.6 53.1 50.56 4.78% 6.1 5.78 5.20% 40.6 41.37 1.89% [23]
12 Wood chips 23.5 76.4 0.1 48.1 49.73 3.39% 5.99 5.96 0.52% 45.74 43.51 4.87% [15]
13 Canyon live oak 11.3 88.2 0.5 47.84 47.33 1.07% 5.8 6.06 4.41% 45.76 45.42 0.75% [16]
14 Madrone 15.1 84.5 0.3 48.56 48.07 1.02% 6.05 6.02 0.43% 45.08 44.81 0.59% [16]
15 Beech wood 24.6 74 0.4 49.5 49.34 0.32% 6.2 5.87 5.37% 41.2 42.70 3.65% [23]
16 Mango wood 11.36 85.64 2.98 46.24 46.20 0.08% 6.08 5.90 2.95% 44.42 44.22 0.45% [17]
17 Eucalyptus 16.15 82.62 1.23 47.1 47.88 1.66% 6.00 5.96 0.63% 45.43 44.24 2.63% PS
18 Subabul wood 18.52 81.02 1.2 48.15 48.66 1.06% 5.87 5.99 1.98% 44.75 44.20 1.24% [17]
19 Casuarina 19.58 78.58 1.83 48.5 48.23 0.56% 6.04 5.89 2.48% 43.32 43.36 0.08% [17]
20 Sudan grass 18.6 72.75 8.65 44.58 44.95 0.83% 5.35 5.48 2.39% 39.18 40.28 2.82% [18]
21 Babul 19.00 79.28 1.72 45.39 48.18 6.14% 6.12 5.90 3.54% 46.44 43.51 6.30% PS
22 Red wood 19.92 79.72 0.36 50.64 48.96 3.31% 5.98 5.98 0.03% 42.88 44.00 2.62% [18]
23 Spruce wood 28.3 70.2 1.5 51.9 49.97 3.72% 6.1 5.82 4.52% 40.9 42.02 2.73% [23]
24 Peltophorum 15.82 82.45 1.73 45.99 47.59 3.48% 6.09 5.93 2.55% 45.88 44.06 3.98% PS
25 Orchard 14.6 83.3 2.1 49.15 47.20 3.96% 5.95 5.92 0.44% 43.24 44.09 1.96% [16]
26 Poplar 16.35 82.32 1.33 48.45 47.87 1.20% 5.85 5.95 1.78% 43.69 44.15 1.06% [18]
27 Blockwood 14.59 83.32 2.09 46.9 47.20 0.65% 6.07 5.92 2.40% 43.99 44.10 0.24% [17]
28 Corncob 18.54 80.1 1.36 46.58 48.26 3.60% 5.87 5.93 1.03% 45.46 43.76 3.73% [18]
29 Straw 18.3 77.3 3.4 47.1 46.83 0.58% 6 5.74 4.26% 43.3 42.36 2.18% [24]
30 Alfalfa seed straw 20.15 72.6 7.25 46.76 45.87 1.91% 5.4 5.55 2.76% 40.72 40.68 0.09% [18]
31 Wheat straw 23.5 63 13.5 45.5 43.63 4.10% 5.1 5.13 0.55% 34.1 37.13 8.89% [23]
32 Barley straw  1 13.29 82.41 4.3 45.47 45.96 1.08% 5.61 5.80 3.40% 44.57 43.27 2.92% [25]
33 Rye straw-kustro 15.01 83.02 1.97 46.63 47.34 1.51% 5.62 5.93 5.48% 45.85 44.08 3.86% [26]
34 Paddy straw 11.8 72.7 15.5 35.97 40.60 12.86% 5.28 5.12 3.01% 43.08 38.19 11.35% [20]
35 Corn stover 14.5 78.1 7.4 46.5 44.77 3.72% 5.81 5.60 3.68% 39.67 41.58 4.82% [27]
36 Cotton stalk 19.9 62.9 17.2 39.47 41.30 4.63% 5.07 4.93 2.67% 38.09 35.99 5.51% [15]
37 Sugarcane bagasse 14.95 73.78 11.27 44.8 43.09 3.81% 5.35 5.35 0.03% 39.55 39.66 0.29% [18]
38 Coconut coir 29.7 66.58 3.72 50.29 49.21 2.14% 5.05 5.67 12.32% 39.63 40.72 2.75% [18]
39 Coconut fibre 26.6 70.6 2.8 46.43 49.07 5.68% 5.49 5.76 4.93% 43.78 41.69 4.77% [20]
40 Pine needles 26.12 72.38 1.5 48.21 49.57 2.82% 6.57 5.85 11.02% 43.72 42.39 3.03% [28]
J. Parikh et al. / Fuel 86 (2007) 1710–1719 1713

2.4. Step 4: Selection of the best correlation


References

To facilitate the selection of correlation, the average


[18]
[29]
[30]

[31]

[20]
[32]
[20]

[18]

[16]
PS

absolute and bias errors for each of these correlations as


derived above were computed as follows:
4.07%
7.61%
0.99%

0.10%
2.22%

0.28%
0.27%
1.38%

0.67%

2.07%
Error

Average Absolute Error


n  
1X  Calculated Value  Measured Value
¼  100% ð1Þ
n Measured Value
Calculated

i¼1

Average Bias Error


41.17
33.76
37.52

43.93
42.21

43.04
43.76
40.47

36.63

45.92

n  
1X Calculated Value  Measured Value
¼ 100%
n i¼1 Measured Value
Oxygen (%)
Measured

ð2Þ
39.56
31.37
37.89

43.88

43.16
43.88
41.04

36.38

44.99
41.3

The correlation giving minimum error levels over 200 data


points was selected as the best one.
3.00%
3.17%
0.56%

2.68%
1.23%

4.13%
3.53%
5.44%

5.75%

1.09%

The following are the correlations which turned out to


Error

be the best in this manner with an average absolute error


of 3.21%, 4.79%, 3.4% and bias error of 0.21%, 0.15%
and 0.49% with respect to measured values of C, H and
Calculated

O, respectively:
4.55
5.07

5.89
5.62

5.91
5.77

5.88

5.57

4.96

6.10

C ¼ 0:637FC þ 0:455VM ðwt%Þ ð3Þ


H ¼ 0:052FC þ 0:062VM ðwt%Þ ð4Þ
Hydrogen (%)

O ¼ 0:304FC þ 0:476VM ðwt%Þ ð5Þ


Measured
5.46

6.07

5.69
6.13

5.89

5.26

6.03

Covering the range of various contents as shown below:


4.7
5.1

5.7

57:2% 6 VM 6 90:6%
3.96%
1.94%
5.52%
5.40%

0.67%
3.15%

2.15%

0.67%

1.64%

3.26%

4:7% 6 FC 6 38:4%
Ultimate analysis (wt% by dry basis)

Error

36:2% 6 C 6 53:1%
4:36% 6 H 6 8:3%
Calculated

31:37% 6 O 6 49:5%
47.17
46.66
36.56
41.02

47.18
47.94

48.69

47.00

40.24

47.08

where VM, FC, C, H and O represents volatile matter,


fixed carbon, elemental carbon, hydrogen and oxygen con-
Carbon (%)

tent of biomass material, respectively, expressed in mass


Measured

percentages on dry basis.


45.37
45.77

38.92

47.50

47.67

46.69

39.59

48.67

From the data in Table 1, there is a good relation


38.7

49.5

between the determined and the calculated elemental com-


position and Eqs. (3)–(5) have given acceptable approxima-
5.89

2.30
ASH

tions to the measured values.


24.6
15.8

2.5

17.6
3.4

1.7

6.3

0.2
(wt% by dry basis)
Proximate analysis

3. Validation of the correlation


73.02

82.70

82.1
69.3

74.7

76.5

69.7

67.3

90.6
VM

63

To confirm the validity of these equations, a variety of


various samples were examined. Table 1 shows the results
21.09

15.00

15.4
12.4
14.9

21.9

21.8

15.1

9.2

obtained, which demonstrate that the equations provide a


FC

24

simple and rapid way to obtain good estimates of ele-


Raw material

Alabama oak

mental composition of various varieties of lignocellulosic


Peanut hulls

wood waste
RH  CP 4
Rice husk-

Cottongin
W-pallets
Tea bush
Saw dust

materials.
Coconut

Tan oak

PS, present study.


waste

Validation of the correlation developed under the pres-


patni

trash

ent work has been carried out by comparison of computed


and measured values of elemental analysis over 50 data
Serial

points. The measured and computed values of elemen-


no.

41
42
43

44
45

46
47
48

49

50

tal analysis of data have been presented graphically in


1714
Table 2
List of algebraic equations tested for the development of correlation
Serial Proposed equation Criteria for selection
no.
1 C = aFC + b(FC/VM) + cVM H = aFC + b(FC/VM) +cVM O = aFC + b(FC/VM) + cVM Assuming elemental carbon C,
H, O is distributed in fixed
carbon and volatile matter
linearly and is also a function
of its ratio
2 C = aFC + bFC2 + c(FC * VM) + dVM H = aFC + bFC2 + c(FC * VM) + dVM O = aFC + bFC2 + c(FC * VM) + dVM Assuming elemental carbon C,
+eVM2 + eVM2 + eVM2 H, O to be polynomial
function of its fixed carbon

J. Parikh et al. / Fuel 86 (2007) 1710–1719


and volatile matter content
3 C = a1FC + b1VM H = a1FC + b1VM O = a1FC + b1VM Assuming elemental carbon C,
H, O to be linear function of
its fixed carbon and volatile
matter content
4 C = a1FC + b2VMC2 H = a1FC + b2VMC2 O = a1FC + b2VMC2 Assuming elemental carbon C,
H, O to be linear function of
its fixed carbon and nonlinear
function of its volatile matter
content
5 C = a3FCd3 + b2VMC2 H = a3FCd3 + b2VMC2 O = a3FCd3 + b2VMC2 Assuming nonlinear effect of
volatile matter and nonvolatile
matter on elemental C,
H and O
6 C = a4FCd4 + b1VM H = a4FCd4 + b1VM O = a4FCd4 + b1VM Assuming linear effect of
volatile matter and nonlinear
effect of nonvolatile matter on
elemental C, H and O
7 C = FC + bVM H = FC + bVM O = FC + bVM Assuming linear effect of
volatile matter and
considering nonvolatile matter
as constant
C, wt% dry basis; H, wt% dry basis; O, wt% dry basis; FC, wt% dry basis; VM, wt% dry basis.
J. Parikh et al. / Fuel 86 (2007) 1710–1719 1715

Pit / Shell / Seeds Pit / Shell / Seeds


55 50
-- - - - - +/-7 % Line -- - - - - +/-7 % Line
AV. ABS ERROR = 3.23 % AV. ABS ERROR = 2.38 %
AV. BIAS ERROR = -1.37 % AV. BIAS ERROR = - 0.24 %
NO. OF DATA = 18 NO. OF DATA = 18

50 45

Predicted Value
Predicted Value

(O%)
(C%)

40
45

35
40 35 40 45 50
40 45 50 55
Measured Value (O%)
Measured Value (C%)
Fig. 1c. Comparison between measured and calculated elemental oxygen
Fig. 1a. Comparison between measured and calculated elemental carbon (pit/shell/seed).
(pit/shell/seed).

Pit / Shell /Seeds


7
-- - - - - +/-7 % Line
Wood / Bark / Energy Crops / Pruning
55
AV. ABS ERROR = 6.83 % -- - - - - +/- 7 % Line
AV. BIAS ERROR = 3.05 % AV. ABS ERROR = 2.55 %
NO. OF DATA = 18 AV. BIAS ERROR = - 0.85 %
NO. OF DATA = 61
6
Predicted Value

50
Predicted Value

(C%)
(H%)

5 45

40
4 40 45 50 55
4 5 6 7 Measured Value (C%)
Measured Value (H%)
Fig. 2a. Comparison between measured and calculated elemental carbon
Fig. 1b. Comparison between measured and calculated elemental hydro- (wood/bark/energy crops/pruning).
gen (pit/shell/seed).

Figs. 1a–1c, 2a–2c, 3a–3c, 4a–4c, 5a–5c. Figs. 6a–6c present Wood / Bark / Energy Crops / Pruning
this comparison for a whole range of fuels. The error band 7
-- - - - - +/-7 % Line
of ±7% has been also shown in these figures. The study of AV. ABS ERROR = 2.66 %
this comparison indicates that the average absolute errors AV. BIAS ERROR = 0.28 %
for pit/shell/seeds, wood/bark/energy crops/pruning, NO. OF DATA = 61
Predicted Value

straw/stalk, hull/husk/dust, miscellaneous material and


(H%)

whole range of biomass materials are 3.23%, 2.55%,


5.5
3.71%, 3.42%, 4.06% and 3.2%, respectively, for carbon
content, 6.83%, 2.66%, 5.67%, 6.49%, 5.64% and 4.78%,
respectively, for hydrogen content and 2.38%, 2.39%,
4.47%, 4.06%, 4.89% and 3.4%, respectively, for oxygen
content while the average bias errors are 1.37%,
0.85%, 2.83%, 0.48%, 2.26% and 0.21%, respectively, 4
for carbon content, 3.05%, 0.28%, 0.43%, 1.78%, 4 5 6 7
2.26% and 0.17%, respectively, for hydrogen content Measured Value (H%)
and 0.24%, 0.92%, 2.62%, 1.86%, 1.35% and 0.51%, Fig. 2b. Comparison between measured and calculated elemental hydro-
respectively, for oxygen content. gen (wood/bark/energy crops/pruning).
1716 J. Parikh et al. / Fuel 86 (2007) 1710–1719

Wood / Bark
Hull/ /Energy
Husk / Crops
Dust / Pruning Straw / Stalk
507 50
-- --
- ------ - +/-7+/-7
% % Line
Line -- - - - - +/-7 % Line
AV.
AV. ABSABS ERROR
ERROR= =2.39 % %
6.49 AV. ABS ERROR = 4.47 %
AV.
AV. BIAS
BIAS ERROR
ERROR = =0.92
-1.78
% % AV. BIAS ERROR = -2.62 %
NO.
NO. OFOF DATA = =61 29
DATA 45 NO. OF POINTS = 23
Value

456
Value

Predicted Value
(O%)
Predicted

(O%)
(H%)

40
Predicted

405
35

354 30
4 5 6 7 30 35 40 45 50
35 40 45 50
Measured Value (H%) Measured Value (O%)
Measured Value (O%)
Fig. 3c. Comparison between measured and calculated values oxygen
Fig. 2c. Comparison between measured and calculated elemental oxygen
(straw/stalk).
(wood/bark/energy crops/pruning).

Straw / Stalk Hull / Husk / Dust


55 55
-- - - - - +/-7 % Line
-- - - - - +/-7 % Line
AV. ABS ERROR = 3.71 %
AV. ABS ERROR = 3.42 %
AV. BIAS ERROR = 2.83 % AV. BIAS ERROR = -0.48 %
50 NO. OF POINTS = 23 50 NO. OF DATA = 29
Predicted Value

Predicted Value
(C%)
(C%)

45 45

40 40

35 35
35 40 45 50 55 35 40 45 50 55
Measured Value (C%) Measured Value (C%)

Fig. 4a. Comparison between measured and calculated elemental carbon


Fig. 3a. Comparison between measured and calculated elemental carbon
(hull/husk/dust).
(straw/stalk).

Straw / Stalk
7 Hull / Husk / Dust
-- - - - - +/-7 % Line
AV. ABS ERROR = 5.67 %
7
-- - - - - +/-7 % Line
AV. BIAS ERROR = 0.43 % AV. ABS ERROR = 6.49 %
NO. OF POINTS = 23 AV. BIAS ERROR = -1.78 %
NO. OF DATA = 29
6
Predicted Value

6
Predicted Value
(H%)

(H%)

5 5

4 4
4 5 6 7 4 5 6 7
Measured Value (H%) Measured Value (H%)

Fig. 3b. Comparison between measured and calculated elemental hydro- Fig. 4b. Comparison between measured and calculated elemental hydro-
gen (straw/stalk). gen (hull/husk/dust).
J. Parikh et al. / Fuel 86 (2007) 1710–1719 1717

Hull / Husk / Dust Miscellaneous


50 50
-- - - - - +/-7 % Line -- - - - - +/-7 % Line
AV. ABS ERROR = 4.06 % AV. ABS ERROR = 4.89 %
AV. BIAS ERROR = 1.86 % AV. BIAS ERROR = 1.35 %
NO. OF DATA = 29. NO. OF DATA = 25
45 45

Predicted Value
Predicted Value

(O%)
40
(O%)

40

35 35

30 30
30 35 40 45 50 30 35 40 45 50
Measured Value (O%) Measured Value (O%)
Fig. 4c. Comparison between measured and calculated elemental oxygen Fig. 5c. Comparison between measured and calculated elemental oxygen
(hull/husk/dust). (miscellaneous).

Miscellaneous
55 Whole Range
-- - - - - +/- 7 % Line 60
AV. ABS ERROR = 4.06 % -- - - - - +/-7 % Line
AV. BIAS ERROR = 2.26 %
AV. ABS ERROR = 3.20 %
NO. OF DATA = 25
55 AV. BIAS ERROR = 0.20 %
50 NO. OF DATA = 156
Predicted Value
Predicted Value

50
(C%)
(C%)

45 45

40
40
35

35 30
35 40 45 50 55 30 40 50 60
Measured Value (C%) Measured Value (C%)

Fig. 5a. Comparison between measured and calculated elemental carbon Fig. 6a. Comparison between measured and calculated elemental carbon
(miscellaneous). (whole range of biomass materials).

Miscellaneous Whole Range


7 8
-- - - - - +/- 7 % Line -- - - - - +/-7 % Line
AV. ABS ERROR = 5.64 % AV. ABS ERROR = 4.78 %
AV. BIAS ERROR = -2.26 % AV. BIAS ERROR = -0.17 %
NO. OF DATA = 25 NO. OF DATA = 156
7
Predicted Value

6
Predicted Value

(H%)
(H%)

5
5

4 4
4 4.5 5 5.5 6 6.5 7 4 4.5 5 5.5 6 6.5 7 7.5 8
Measured Value (H%) Measured Value (H%)

Fig. 5b. Comparison between measured and calculated elemental hydro- Fig. 6b. Comparison between measured and calculated elemental hydro-
gen (miscellaneous). gen (whole range of biomass materials).
1718 J. Parikh et al. / Fuel 86 (2007) 1710–1719

Whole Range 6. Limitations of the present correlations


55
-- - - - - +/-7 % Line
AV. ABS ERROR = 3.4 % The correlations for estimation of C, H and O from
50 AV. BIAS ERROR = 0.51 %
NO. OF DATA = 156 proximate analysis are derived based on available data in
the literature wherein measurement of oxygen is not explic-
Predicted Value

45
itly defined. Further, material may contain Nitrogen and
(O%)

40
Sulfur too. As a result, the use of these correlations needs
a prior knowledge of nitrogen and sulfur to make more
35 correct predictions.

30 References

25 [1] Channiwala SA, Parikh PP. A unified correlation for estimating HHV
25 35 45 55 of solid, liquid and gaseous fuels. Fuel 2002;81:1051–63.
Measured Value (O%) [2] Kucukbayrak S, Durus B, Mericboyu AE, Kadioglu E. Estimation of
calorific values of Turkish lignites. Fuel 1991;70:979–81.
Fig. 6c. Comparison between measured and calculated elemental oxygen [3] Cordero T, Marquez F, Rodriquez – Mirasol J, Rodriguez JJ.
(whole range of biomass materials). Predicting heating values of lignocellulosic and carbonaceous mate-
rials from proximate analysis. Fuel 2001;80:1567–71.
[4] Fernandez P, Diaz RM, Xiberta J. Correlations of properties of
Spanish coals with their natural radionuclides contents. Fuel
4. Utility of the correlations 1997;76(10):951–5.
[5] Parikh J, Channiwala SA, Ghosal GK. A correlation for calculating
Having established the validity and merits of the pro- HHV from proximate analysis of solid fuels. Fuel 2005;84:487–94.
posed correlations, a few of its applications have been sum- [6] Vakkilainen EK. Estimation of elemental composition from proxi-
marized below: mate analysis of black liquor. PAPERI JA PUU – Paper Timber
2000;82(7):450–4.
[7] ASTM D 2013-86, Standard method of preparing coal samples for
(i) The correlations can be used for computation of ele- analysis, in gaseous fuels; coal and coke section 5, Annual book of
mental composition (C, H and O) and HHV of any standards, vol. 05-05; 1989. p. 226.
biomass material, from its proximate analysis. [8] ASTM D 3172 – 73(84), Standards method of proximate analysis of
coal and coke, in gaseous fuels; coal and coke section 5, Annual book
(ii) They are useful in performance modeling, combus-
of ASTM standards, vol. 05-05; 1989. p. 299.
tion, gasification and pyrolysis processes. [9] ASTM D 3173-87, Standards test method for moisture in the analysis
(iii) The correlation provides the facility of using elemen- sample of coal and coke, in gaseous fuels; coal and coke section 5,
tal component as an algebraic expression in terms of Annual book of ASTM standards, vol. 05-05; 1989. p. 300.
fuel constituents, which in turn is useful in studying [10] ASTM D 3175-89, Standards test method for volatile matter in the
the energy balance and influence of proximate analy- analysis sample of coal and coke, in gaseous fuels; coal and coke
section 5, Annual book of ASTM standards, vol. 05-05; 1989. p. 305.
sis of a fuel on process performance. [11] ASTM D 3174-89, Standards test method for ash in the analysis
sample of coal and coke, in gaseous fuels; coal and coke section 5,
Annual book of ASTM standards, vol. 05-05; 1989. p. 302.
5. Conclusions [12] ASTM D 3176-84, Standards method for ultimate analysis of coal and
coke, in gaseous fuels; coal and coke, section 5, vol. 05-05; 1989. p. 308.
[13] ASTM D 3178-84, Standards test method for carbon and hydrogen in
The present correlations have been derived based on a the analysis sample of coal and coke, in gaseous fuels; coal and coke,
large number of data points having widely varying proxi- section 5, vol. 05-05; 1989. p. 315.
mate and elemental composition and encompassing all cat- [14] ASTM D 3179-84, Standards test method for nitrogen in the analysis
egories of solid lignocellulosic materials. This means that, sample of coal and coke, in gaseous fuels; coal and coke, section 5,
within the specified range of FC, VM, ash, carbon, hydro- vol. 05-05; 1989. p. 519.
[15] Jenkins BM. Downdraft gasification characteristics of major Califor-
gen and oxygen as wt%, and within the stated average nia residue derived fuels. Ph.D thesis, University of California, Davis,
absolute error limit of 3.21%, 4.79% and 3.4% with respect 1980.
to measured values of C, H and O, respectively, the present [16] Rossi A. Fuel characteristics of wood and non-wood biomass fuels.
correlations may be accepted as ‘general correlations’ for In: Tillman DA, Jahn EC, editors. Progress in biomass conversion,
estimating elemental composition of biomass materials on vol. 5. New York: Academic Press; 1984. p. 69.
[17] Channiwala SA. On biomass gasification process and technology
dry basis. development – some analytical and experimental investigations. Ph.D
The main advantage of these correlations is, based on thesis, Mechanical Engineering Department, IIT, Mumbai; 1992.
using only proximate analysis data, a rapid, easy and [18] Jenkins BM, Ebeling JM. Correlation of physical and chemical
economical estimation of the elemental components. This properties of terrestrial biomass with conversion: symposium energy
from biomass and waste IX IGT; 1985. p. 371.
may be of particular interest in the contexts where more
[19] Bonelli PR, Della Rocca PA, Cerrella EG, Cukierman AL. Effect of
sophisticated and expensive equipments for experimental pyrolysis temperature on composition, surface properties and thermal
elemental components measurement are not always degradation rates of Brazil nut shells. Bioresour Technol
available. 2001;76:15–22.
J. Parikh et al. / Fuel 86 (2007) 1710–1719 1719

[20] Grover PD, Iyer PVR, Rao TR. Biomass – thermochemical charac- [27] Gregory MK, Rose GR, Nandi SP, Onischak M, Zabransky RF,
terization. third ed. IIT Delhi: MNES; 2002. Babu SR. Development of biomass gasification to produce substitute
[21] Risser PG. Agricultural and forestry residues. In: Soffer SS, Zaborsky fuel. In: Proceedings 14th biomass thermochemical conversion
OR, editors. Biomass conversion processes for energy and fuels. New contracot’s meeting; 1982. p. 71–147.
York: Plenum Press; 1981. p. 25–56. [28] Grover PD, Anuradha G. Thermochemical characterization of
[22] Tran DQ, Rai C. A kinetic model for pyrolysis of Douglas fir bark. biomass for gasification, Report on physico-chemical parameters of
Fuel 1978;57:293–8. biomass residues, IIT – Delhi; 1988.
[23] Demirbas A. Calculation of higher heating values of biomass fuels. [29] Mansaray KG, Ghaly AE. Determination of kinetic parameters of
Fuel 1997;76(5):431–4. rice husks in oxygen using TGA. Biomass Bioenerg 1999;17:19–31.
[24] Sainz-Diaz CI, Griffiths AJ. Activated carbon from solid wastes using [30] Maheshwari RC. Utilization of rice husk as fuel. Ph.D thesis,
a pilot scale batch flaming pyrolyzer. Fuel 2000;79:1863–71. Agricultural Engineering Department, IIT, Kharagpur, India; 1975.
[25] Ghaly AE, Ergudenler A. A comparative study on the thermal [31] Boley EC, Landers WS. Entrainment drying and carbonization of
decomposition of four cereal straws in an oxidizing atmosphere. wood waste, Report of investigation 7282, Bureau of Mines,
Bioresour Technol 1994;50:201–8. Washington; 1969.
[26] Ghaly AE, Ergudenler A. Reaction kinetics of rye straw for [32] Reina J, Velo E, Puigjaner L. Kinetic study of the pyrolysis of waste
thermochemical conversion. Can J Chem Eng 1994;72:651–6. wood. Ind Eng Chem Res 1998;37:4290–5.

You might also like