Thermogravimetric analysis and the optimisation of bio-oil yield from fixed-bed pyrolysis of rice husk using response surface methodology (RSM)

Page 1

Tp

KSa

b

c

a

ARRAA

KRFTD

1

ocswbgflwt

m(otcM

0d

Industrial Crops and Products 33 (2011) 481–487

Contents lists available at ScienceDirect

Industrial Crops and Products

journa l homepage: www.e lsev ier .com/ locate / indcrop

hermogravimetric analysis and the optimisation of bio-oil yield from fixed-bedyrolysis of rice husk using response surface methodology (RSM)

hairuddin Md Isaa,∗, Suhardy Daudb, Nasrul Hamidina, Khudzir Ismail c,aiful Azhar Saada, Farizul Hafiz Kasima

School of Environmental Engineering, Universiti Malaysia Perlis, P.O Box 77, d/a Pejabat Pos Besar 01007, Kangar, Perlis, MalaysiaSchool of Material Engineering, Universiti Malaysia Perlis, P.O Box 77, d/a Pejabat Pos Besar 01007, Kangar, Perlis, MalaysiaFuel Combustion Research Laboratory, School of Applied Science, Universiti Teknologi Mara 02600, Arau, Perlis, Malaysia

r t i c l e i n f o

rticle history:eceived 12 February 2010eceived in revised form 12 October 2010ccepted 28 October 2010vailable online 26 November 2010

eywords:

a b s t r a c t

The effects of pyrolysis temperature, heating rate, particle size, holding time, and gas flow rate were inves-tigated to optimize bio-oil yield from rice husk pyrolysis. Thermogravimetric analysis showed thermaldegradation of hemicellulose, cellulose and lignin, indicating faster decomposition of cellulose com-pared to lignin. The optimisation process was analysed by employing central composite design (CCD)in response surface methodology (RSM) using Design Expert Version 7.5.1 (StatEase, USA). A two-levelfractional factorial was initially carried out and followed by RSM. The statistical analysis showed thatpyrolysis temperature, heating rate, particle size and holding time significantly affected the bio-oil yield.

ice huskixed-bed pyrolysishermogravimetricesign-expert

By utilising response surface method, these four factors were investigated, analysed and optimal con-ditions were obtained at pyrolysis temperature of 473.37 ◦C, heating rate of 100 ◦C/min, particle sizeof 0.6 mm and holding time of 1 min. Confirmation runs gave 48.30% and 47.80% of bio-oil yield com-pared to 48.10% of predicted value. Furthermore, the pyrolytic bio-oils obtained from fixed-bed pyrolysis

chromzer,

were examined using gasmethods, elemental analy

. Introduction

Biomass resources have attracted many researchers to studyn how these could be utilised into renewable energy. This is aommendable effort to find alternatives in order to reduce fos-il fuel consumption. Petroleum continues to supply most of theorld demand for transportation fuels and commodity chemicals,

ut there are signs this precious resource is depleting. As reported,lobal impact of greenhouse warming due to carbon emission hasuelled the need to utilise biomass resources to overcome the prob-em. Due to the lower contents of sulphur and nitrogen in biomass

aste, its energy utilization also creates less environmental pollu-ion and health risk than fossil fuel combustion (Tsai et al., 2007).

Biomass is composed mainly of carbohydrate compounds whoseain constituents are the elements of carbon, hydrogen and oxygen

Ozcimen and Karaosmanoglu, 2004). Current energy consumption

f biomass accounted for about 14% of total world energy consump-ion – fourth only to coal, oil, and natural gas – and developingountries accounted for 75% of the biomass utilization (Peter, 2002;atti, 2004; Li et al., 2008). Biomass has certain advantage to fix car-

∗ Corresponding author. Tel.: +60 4 9798991; fax: +60 4 9798636.E-mail addresses: [email protected], dr [email protected] (K.M. Isa).

926-6690/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.indcrop.2010.10.024

atographic/mass spectroscopy (GC/MS), Fourier transform infrared (FTIR)pH probe and bomb calorimeter.

© 2010 Elsevier B.V. All rights reserved.

bon dioxide balance in the atmosphere by photosynthesis process(Beis et al., 2002). The use of biomass can reduce the dependencyon the limited fossil fuels and there is the advantage of reduced netcarbon dioxide emissions (Park et al., 2009).

The pyrolysis process involves a cracking process on polymericstructure to convert the biomass into char and volatile matter.Numerous studies have been carried out to pyrolyse biomass forbio-oil (Onay et al., 2001; Onay and Kockar, 2004; Acikgoz andKockar, 2007; Demiral and Sensoz, 2008; Putun et al., 2008). Theidea of this research is to maximise the conversion of condensablegas into bio-oil and to minimise the char and gas emission throughfix-bed pyrolysis. A good understanding of the pyrolysis process isrequired to elucidate on bio-oil optimisation. Factors such as par-ticle size, heating rate, pyrolysis temperature, nitrogen flow-rate,holding time, condensation temperature and chemical treatmentsare important in the thermochemical conversions.

Biomass is composed of cellulose, hemicellulose and lignin(relatively more thermally stable compared to cellulose and hemi-cellulose) (Isa et al., 2009). Fast-pyrolysis liquid is a mixture of

many compounds, can be classified as acids, aldehydes, sugars,and is derived from the carbohydrate fraction; and phenolic com-pounds, aromatic acids, derived from the lignin fraction. The liquidis highly unstable due to highly oxygenated. The bio-oils are highlycorrosive due to the presence of organics acids primarily from the

Page 2

482 K.M. Isa et al. / Industrial Crops and

Table 1Main characteristics of rice husk.

Characteristics Values

Ultimate analysisa

Carbon 47.28%Hydrogen 5.25%Nitrogen 0.58%Sulfur 0.23%

hba

m3rotpot

123

4

2

2

frwtrccawoo

range in two-level fractional factorial experimental study. The

Heating value analysis 3988 kcal/kg

a As received.

emicellulosic contents of biomass. Cellulose produces such com-ustible volatiles acetaldehyde, propenal, methanol, butanedione,nd acetic acid.

In Malaysia, the land area for paddy planting is approxi-ately 6.8 × 105 ha (Khush, 1997). With the huge area, a total of

.1 tonnes/ha were produced and approximately 8.4 × 105 tonnesice husk were generated annually. Rice husk, which is a by-productf the rice hulling industry, is among the few agricultural residueshat can be readily obtained in huge amount in one location. Thisaper presents a study on the thermal decomposition of rice husk inrder to exploit its biomass into value-added products. The objec-ives of the study are:

. To obtain detailed data of rice husk decomposition.

. To obtain proximate analysis.

. To provide sufficient data for the thermochemical conversion inmicro-reactor.

. To optimise bio-oil yield by employing design-expert software.

. Materials and methods

.1. Materials

The rice husk used in this study was collected from a rice millactory in Kedah, Malaysia. The rice husk was air-dried for 8 h toemove external moisture. Before the experiments, the samplesere ground in a high-speed rotary cutting mill and sieved to obtain

he desired particle size. Elemental analyses were carried out on theice husk using a CHNS/O (Perkin Elmer), elemental analyzer. Thealorific values of the samples were determined by using a bombalorimeter (Model: c2000 basic, IKA co, Germany). The main char-

cteristics of the rice husk are given in Table 1. Standard celluloseas purchased from Asas Kimia Sdn Bhd, and standard lignin was

btained from the Fuel Combustion Research Laboratory, Facultyf Applied Science, UiTM.

Fig. 1. Schematic diagram o

Products 33 (2011) 481–487

2.2. Proximate analysis

Proximate analysis was performed according to ASTM D2974using the thermogravimetric analysis (TGA) DTA/DSC TA Model SDTQ600 for determination of moisture, volatile matter, fixed carbonand ash in biomass. The pyrolysis and combustion of rice husk wasperformed under inert nitrogen gas and purified air with a constantflow rate of 100 ml/min and heating rate of 20 ◦C/min. The sample(20 mg) was weighed directly into the alumina crucible and thetemperature was kept isothermal for 0.5 min until a steady con-dition was obtained before ramping to the desired temperature(Ismail et al., 2005). The experiments were replicated at least twiceto obtain reproducibility.

2.3. Thermogravimetric analysis

Fundamental tests of the pyrolysis for the rice husk were con-ducted using thermo-gravimetric (TG) analyzer. The experimentswere carried out by using a TGA/DSC 1 mettler Toledo. The sampleweight loss and the rate of weight loss were recorded continuouslyunder dynamic conditions, as functions of time or temperature, inthe range of 25–800 ◦C. During the test the sample was uniformlyspread into crucible and automatically weighed. The nitrogen gaswas used for the pyrolysis with a flow rate of 150 ml/min. The heat-ing rate of 20 ◦C/min was selected for this test. The temperature wasincreased to the setting value of 800 ◦C at the pre-selected heatingrate and it was kept constant at this value until steady conditionswere obtained. This study was to observe the decomposition ofhemicellulose, cellulose and lignin. Based on results of the TG anal-ysis, thermal stability data of the rice husk was analysed. The TGanalysis was also conducted on standard lignin and cellulose forcomparison.

2.4. Pyrolysis experimental procedure

Fast pyrolysis was conducted in a fixed-bed reactor as shownin Fig. 1. The experimental device consisted of a tube reactorwith a volume of 250 cm3, equipped with a K-type thermocou-ple, an electric heater, a nitrogen cylinder, a cold trap, waterbath and product effluent. The water bath was connected tothe cold trap to measure the condensation temperature. In thisstudy, the condensation temperature was set at 0 ◦C for all exper-iments. Sweeping gas was allowed to flow in according to set

particle sizes of the rice husk used in this experiment were pre-pared at 0.35, 0.60, 0.90, 1.20 and 1.50 mm. Pyrolysis experimentswere run according to fractional factorial and central compositedesign. For all experiments, whenever the pyrolysis temperature

f the fix-bed reactor.

Page 3

K.M. Isa et al. / Industrial Crops and Products 33 (2011) 481–487 483

Fig. 2. The proximate analysis.

Fig. 3. TGA and DTG for rice husk.

Page 4

4 s and Products 33 (2011) 481–487

w2

2

2

ecdvsotrtityrwt

2

cumtmewpt

y

wcxegomps

2

em(wIpgHsts

84 K.M. Isa et al. / Industrial Crop

as reached, it was immediately cooled down to 100 ◦C withinmin.

.5. Experimental design

.5.1. Two-level fractional factorial designIt is possible to investigate a number of variables using fewer

xperiments, by using a fractional or screening design. They arealled fractional as they use a fraction of the corresponding fullesign, and particularly important to screen or sieve out the mainariables. The assumption of certain high order interactions in theystem is negligible, thus information on the main effects and lowrder interaction may be obtained by running only a fraction ofhe complete factorial experiment. It is always best to use highestesolution (resolution V) for fractional factorial designs to screenhe factors to be explored. In resolution V designs have two-factornteractions aliased with three-factor interactions. The five factorsested in fractional factorial design (25–1) were holding time, pyrol-sis temperature, heating rate, particle size and gas flow rate. Theesponse measured was the bio-oil yield. The significant factorshich affected the response were identified, and investigated in

he RSM experiment for optimisation.

.5.2. Response surface methodology (RSM)The optimisation process was analysed by employing central

omposite design (CCD) in response surface methodology (RSM)sing Design Expert Version 7.5.1 (StatEase, USA). In doing the opti-ization, a number of factors that can be explored are 2–5. Usually,

he effects estimated from RSM are main effects, interactions andodel curvature. In this study, central composite design (CCD) was

mployed for the prediction and verification of model equation asell as the optimization of the response as the function of the inde-endent parameters. The quadratic equation model for predictinghe optimal point was expressed according to Eq. (1):

= ˇ0 +k∑

i=1

ˇixi +k∑

i=1

ˇiix2i +

∑

i=1

∑

j=i+1

ˇijxixj + ε (1)

here ˇ0, ˇi, ˇii and ˇij are regression coefficients for the inter-ept, linear, quadratic and interaction coefficients respectively andi and xj are the coded independent variables. The variability wasxplained by the multiple coefficient of determination, R2 whichave the overall predictive capabilities of the model obtained. Goalsf RSM are to determine the conditions of the inputs which opti-ize the output and to determine empirical models which allow

redictions of process performance in some region of the optimi-ation.

.6. Characterization

The bio-oils characterized in this study were obtained fromxperimental conditions that gave the maximum bio-oil yields. Ele-ental analyses were carried out on the bio-oil using a CHNS/O

Perkin Elmer), elemental analyzer. The calorific values of the oilere determined by using a bomb calorimeter (Model: c2000 basic,

KA co, Germany). The acidity of the product was measured by aH probe with digital meter (3305 Jenway). The gas chromato-

raphic/mass spectroscopy (GC/MS) was performed by using aewlett-Packard HP 5890-series II gas chromatograph with mass

elective detector, a Hewlett-Packard HP 5927A type. The Fourierransforms infra-red (Model: Spectrum 400 (Perkin Elmer)) analy-is was conducted to determine the functional group.

Fig. 4. TG curves of standard cellulose and lignin.

3. Results and discussion

3.1. Proximate analysis and pyrolysis behaviour of rice husk

The proximate analysis for rice husk is shown in Fig. 2. The con-tent of volatile matter was 57.89%, 10.48% of moisture, 15.47% offixed carbon and 15.92% of ash. At temperature of 100–105 ◦C, thecurves indicated 10.48% of moisture loss. Abdullah and Gerhauser(2008) measured 83.86% of volatile matter for empty fruit bunch.Luo et al. (2004) reported volatile matter content for rice straw as67.40%.

The TG analysis usually is used for thermal characterization, cal-culation of the pyrolysis conversion and evaluation of reactivity andkinetic parameters. In this research, the purpose is to view the ther-mal degradation curves of hemicellulose, cellulose and lignin. Basedon TG data, the data of thermal stability of rice husk was analysed.Fig. 3 shows the typical thermal decomposition for cellulose andlignin portion. It was observed that initial curve indicated moistureloss of up to 135.86 ◦C followed by the degradation of holocelluloseover 250 ◦C. Up to 400.65 ◦C, the remaining residue was 44.22% andabout 55.80% was decomposed. It was observed that the rice huskdecomposition occurred at the two stages during pyrolysis. Thedestructive reaction of cellulose started between 250 ◦C and 380 ◦Cand most of the volatile materials decomposed within these tem-peratures. Therefore, the remaining residue above this temperaturerange was considered char. The mass rapidly decreased due to cel-lulose volatilization and followed by slow mass decrease of lignin.In DTG, it can be seen that fast decomposition occurred at 347.30 ◦Cwhich indicates the high reactivity of cellulose compared to lignin.A small shoulder at nearly of 280 ◦C indicates the decompositionof hemicellulose. Generally the lignin is harder to decompose thanthe cellulose since part of lignin consists of benzene rings (Sharmaet al., 2004; Gani and Naruse, 2007).

Typical TG curves and DTG evolution profiles for standard cel-lulose and lignin are illustrated in Figs. 4 and 5. Apparently, thestandard cellulose started to degrade faster in comparison to thestandard lignin as observed by the huge percent of weight loss in theTG curves of the former. This proves that cellulose is more reactivethan lignin. The percentage of weight loss can be shown clearly byobserving the DTG evolution profiles in Fig. 5. From the DTG evolu-tion profiles, the standard cellulose material started to decompose

at temperature ranging from 250 ◦C to about 390 ◦C, with Tmax atca. 370 ◦C. The lignin, however, decomposed at a wider temperatureranging from 250 ◦C to ca. 500 ◦C with a broader evolution profile.The differences in the DTG evolution profiles showed the differ-

Page 5

K.M. Isa et al. / Industrial Crops and Products 33 (2011) 481–487 485

eathasd

3

wuwatysvbya

Studentized Residuals

Nor

mal

% P

roba

bilit

y

Normal Plot of Residuals

-2.09 -1.05 0.00 1.05 2.09

1

510

2030

50

7080

9095

99

Fig. 5. DTG curves of standard cellulose and lignin.

nce in the complexity of the organic components in the cellulosend lignin. This information is very important in pyrolysis processo maximise the liquid contents. Li et al. (2004) reported that theigher lignin composition of the apricot stone led to higher chars compared with that of legume straw. The pyrolysis temperaturehould be correctly set in order to minimise char caused by ligninegradation.

.2. Factor affecting bio-oil yield

All factors were screened by using fractional factorial designith 25–1 (resolution V) and 8 replicates at centre points to makep 24 runs. It can be observed that the factors of A, B, C, and Eere further away from a straight line which indicated less noise

s shown in Fig. 6. Analysis of variance (ANOVA) was performed andhe ‘prob > F’ less than 0.05 indicates factors of heating rate, pyrol-sis temperature, holding time and particle size are statistically

ignificantly. Gas flow rate was found insignificant with ‘prob > F’alue of 0.48. Holding time shows strong influences on producingio-oil and it is presented by F-value of 91.04. The F-value for pyrol-sis temperature, particle size and heating rate were 28.10, 28.10nd 7.02, respectively. Curvature test is significant and shows that

Half-Normal Plot

Hal

f-N

orm

al %

Pro

babi

lity

|Effect|

0.00 1.16 2.31 3.47 4.62

0102030

50

70

80

90

95

99

A

B

C

D

E

AE

Fig. 6. Half normal plot of bio-oil yield.

Fig. 7. Normal plot of residual for bio-oil yield.

optimisation can be investigated. Lack of fit test is not significantindicating that the model is fit.

3.3. Optimisation

The significant factors screened in a fractional factorial designwere investigated by using CCD for optimisation. In this exper-iment, 30 runs were required to be investigated. The tabulatedexperiments and yields are given in Table 2. Nitrogen flow rate andcondensation temperature were kept constant at 100 cm3/min and0 ◦C, respectively. Quadratic model was suggested with R2 value of0.9643 and indicated that 96.43% of the total variation in the bio-oilyield was attributed to the experimental variables studied. ANOVAtest was conducted and shows that A, B, C and D are significantwith p-value of <0.05. The coded values of A, B, C and D representpyrolysis temperature, heating rate, particle size and holding time.The data are normally distributed and shown in the normal plot ofresidues in Fig. 7. Residuals vs predicted and residual vs run werechecked and show an equal variance and stable. Lack of fit tests sug-gests a quadratic model with F-value of 4.13. The model was alsoexamined for any transformation that could have been employedbut the Box–Cox plot did not suggest any transformation for theresponse. A regression analysis was conducted and made by usingcoded values from the estimation of data and expressed by Eq. (2):

Bio-oil yield = 38.00 + 1.75A + 2.08B − 1.08C − 2.58D + 0.13AB

− 0.37AC − 0.25AD − 0.38BC − 0.10BD − 1.75A2

− 0.75B2 + 0.25C2 + 0.38D2 (2)

The result of regression analysis seems to suggest that bio-oilyield was only affected by the main factors since the interactions ofAB, AC, AD, BC, BD and CD are insignificant with p-value > 0.05. Byusing RSM, every factor was analysed and high bio-oil yield could beobtained at temperature nearly of 500 ◦C, high heating rate, smallparticle size, and low holding time. At temperature of 600 ◦C, pyrol-ysis process favors gas generation and liquid yield was decreasing.

As reported by Luo et al. (2004), the high pyrolysis temperaturewould lead to a high proportion of CO and CH4, with low proportionof CO2. Char generation was high at temperature of about 400 ◦C;as such, liquid yield was low.

Page 6

486 K.M. Isa et al. / Industrial Crops and Products 33 (2011) 481–487

Table 2Design matrix using CCD.

Run Pyrolysis temperature (◦C) Heating rate (◦C/min) Particle size (mm) Holding time (min) Bio-oil yield (%)

1 400 50 0.6 2 362 600 50 0.6 2 403 400 100 0.6 2 404 600 100 0.6 2 455 400 50 1.2 2 336 600 50 1.2 2 357 400 100 1.2 2 408 600 100 1.2 2 429 400 50 0.6 4 33

10 600 50 0.6 4 3511 400 100 0.6 4 3412 600 100 0.6 4 3713 400 50 1.2 4 3014 600 50 1.2 4 3215 400 100 1.2 4 3216 600 100 1.2 4 3417 300 75 0.9 3 2618 700 75 0.9 3 3619 500 25 0.9 3 3020 500 125 0.9 3 4021 500 75 0.3 3 4022 500 75 1.5 3 3823 500 75 0.9 1 4424 500 75 0.9 5 3525 500 75 0.9 3 3826 500 75 0.9 3 3827 500 75 0.9 3 3928 500 75 0.9 3 3729 500 75 0.9 3 38

0.9 3 38

o3idctt

cTcopcowwveaa5

ttasabllod

Table 3Main characteristics of bio-oil.

Characteristics Values

Ultimate analysisC 50.6 wt.%H 40.8 wt.%O 7.6 wt.%N 0.4 wt.%S 0.3 wt.%

GC/MS dataAcetic acid 14.5 wt.%Phenol 3.3 wt.%Toluene 5.3 wt.%

30 500 75

The particle size is very important and could influence the yieldf bio-oil. Based on Onay et al. (2001) at a higher heating rate of00 ◦C/min the overall conversion of the pyrolysis was only approx-

mately 2% higher than that of the lower heating rate of 100 ◦C/minue to the particle range. This study demonstrates that it is suffi-ient that a particle size of 0.60 mm be heated up by 100 ◦C/mino yield the highest amount of bio-oil of 45% at 600 ◦C and holdingime set at 2 min.

Process optimization was conducted to determine the optimalonditions of the process to obtain high percentage of bio-oil yield.he results from the software suggested the optimal conditionsan be attained at pyrolysis temperature of 473.37 ◦C, heating ratef 100 ◦C/min, particle size of 0.60 mm, holding time at 1 min andredicted bio-oil yield was 48.10% (Fig. 8). Confirmation runs wereonducted with the above conditions and gave 48.30% and 47.80%f bio-oil. It was indicated that the experimental values obtainedere in good agreement with the values predicted from the models,hich recorded small errors between the predicted and the actual

alues, which was only 0.42% and 0.62%. Beis et al. (2002) discov-red that the maximum pyrolysis oil yield of 44% was obtainedt a final temperature, particle size range, sweeping gas flow ratend heating rate equal to 500 ◦C, 0.425–1.25 mm, 100 cm3/min and◦C/min, respectively.

The bio-oil yield could be improved by certain factors such ashe increase of heating rate, suitable particle size, condensationemperature, lower ash content and lignin influences. Ozcimennd Karaosmanoglu (2004) revealed that sweeping gas usage withlow pyrolysis did not have any effect on the product yields. Ganind Naruse (2007) suggested that the volatilization behaviour of

iomass depends on its own component such as the cellulose and

ignin content. Abdullah and Gerhauser (2008) concluded that byowering the ash content would result in a substantial increase inrganics yield, while reaction water, char and gas yields are alleclined.

Benzofuran,2,3-dyhydro- 6.6 wt.%Phenol,2-methoxy 2.2 wt.%1,2-Benzencarboxylic acid 0.8 wt.%

3.4. Chemical characterization

The FTIR analysis shows that the presence of carbonyl groupwas in peaks between 1620 and 1700 cm−1, stretching vibrationindicative of the carboxylic acid, phenols and ketones. Mono andpolycyclic and substituted aromatic groups were indicated by theabsorption peaks between 1420 and 1610 cm−1.The bending vibra-tion between 1350 and 1475 cm−1 represented of C–H, indicativeof alkane groups in the bio-oil. Previous works have shown a sim-ilar trend (Onay and Kockar, 2003; Acikgoz et al., 2004; Yorgunand Simsek, 2008). The C C stretching vibrations was absorbed at1520 cm−1, indicative of alkenes and aromatics. The data from FTIRcan be used as a quick technique to elucidate on oxygen content orcarbonyl group.

Table 3 shows the elements and the composition of the producedbio-oil. It demonstrated a high concentration of O with 50.60 wt.%.An amount of 40.80 wt.% of C and small concentrations of S and N aredetected in the bio-oil. In the GC/MS analysis, the presence of 5.30%

Page 7

K.M. Isa et al. / Industrial Crops and Products 33 (2011) 481–487 487

imisat

taipprAtfiwam

4

dldmdprrtt4homaap

A

iptc

Fig. 8. An opt

oluene shows the significance of combustible materials, and it is ingreement with the presence of aromatic hydrocarbon as indicatedn FTIR. Phenolics are derived from lignin by cracking the phenyl-ropane units of the macromolecule lattice and were shown by theresence of 3.30 wt.% of phenol. The pyrolysis of hemicellulose hadeleased galacturonic acids and contributed to acid compositions.n acetic acid was found at approximately14.50 wt.%. As a result,

he bio-oil pH was tested and gave the value in mean of 2.96 fromve samples indicated corrosive characteristic. Low heating valueas tested and recorded 18.02 MJ/kg. The low calorific values usu-

lly because the presence of water content. However, we have noteasured the water content in this experiment.

. Conclusion

Thermogravimetric analysis was carried out and shows fasterecomposition of cellulose compared to the slower reaction rate of

ignin. It can be concluded that cellulose is more reactive than ligninue to its complexities. The useful information from thermogravi-etric analysis was used in pyrolysis experiments. In this study, the

esign-expert software was successfully employed to optimise theroduction of bio-oil in the fix-bed reactor. Factors such as heatingate, pyrolysis temperature, particle size, holding time and gas flowate were tested which indicate significantly affected the produc-ion of bio-oil. By utilising RSM, these factors were investigated, andhe optimal conditions were obtained at pyrolysis temperature of73.37 ◦C, heating rate of 100 ◦C/min, particle size of 0.60 mm andolding time at 1 min. Confirmation runs gave 48.30% and 47.80%f bio-oil yield compared to 48.10% of predicted value. It is esti-ated that via optimisation of rice husk pyrolysis, we can produce

pproximately 4 × 105 tonne of bio-oil per year in Malaysia. The cat-lytic pyrolysis is recommended to overcome the high oxygenatedroblem in the oil produced.

cknowledgments

The authors would like to thank MOSTI (Malaysia) for fully fund-ng the work published in this paper. All experimental works wereerformed at University Malaysia of Perlis and UiTM Arau. Specialhanks are due to Asc. Prof. Ahmad Nazri Abdullah (Unimap) for hisountless assistance.

ion of bio-oil.

References

Abdullah, N., Gerhauser, H., 2008. Bio-oil derived from empty fruit bunches. Fuel 87,2606–2613.

Acikgoz, C., Kockar, O.M., 2007. Flash pyrolysis of linseed (Linum usitatissimum L.)for production of liquid fuels. Journal of Analytical and Applied Pyrolysis 78,406–412.

Acikgoz, C., Onay, O., Kockar, O.M., 2004. Fast pyrolysis of linseed: product yieldsand compositions. Journal of Analytical and Applied Pyrolysis 71, 417–429.

D2974: Annual Book of ASTM Standards, vol. 5.05. The American Society for Testingand Materials (ASTM), Philadelphia, PA.

Beis, S.H., Onay, O., Kockar, O.M., 2002. Fixed-bed pyrolysis of safflower seed: influ-ence of pyrolysis parameters on product yields and compositions. RenewableEnergy 26, 21–32.

Demiral, I., Sensoz, S., 2008. The effects of different catalysts on the pyrolysis ofindustrial wastes (olive and hazelnut bagasse). Bioresource Technology 99,8002–8007.

Gani, A., Naruse, I., 2007. Effect of cellulose and lignin content on pyrolysis andcombustion characteristics for several types of biomass. Renewable Energy 32,649–661.

Isa, Md.K., Soe, M.M., Ab Rahim, M.A., Abdullah, M.F., Ismail, K., Ahmad, R, Saad, S.A.,2009. Characterization and thermal degradation of rice husk via TGA. Interna-tional Review of Chemical Engineering Rapid Communications 4, 334–338.

Ismail, K., Zakaria, Z., Ishak, M.A.M., 2005. Proceeding of 22nd, International Pitts-burgh Coal Conference, Pittsburgh, USA.

Khush, G.S., 1997. Origin, dispersal, cultivation and variation of rice. Plant MolecularTechnology 35, 25–34.

Li, S., Xu, S., Liu, S., Yang, C., Lu, Q., 2004. Fast pyrolysis of biomass in free-fall reactorfor hydrogen-rich gas. Fuel Processing Technology 85, 1201–1211.

Li, Z., Zhao, W., Meng, B., Liu, C., Zhu, Q., Zhao, G., 2008. Kinetic study of corn strawpyrolysis: comparison of two different three-pseudocomponent models. Biore-source Technology 99, 7616–7622.

Luo, Z., Wang, S., Liao, Y., Zhou, J., Gu, Y., Cen, K., 2004. Research on biomass fastpyrolysis for liquid fuel. Biomass Bioenergy 26, 455–462.

Matti, P., 2004. Global biomass fuel resources. Biomass Bioenergy 27, 613–620.Onay, O., Kockar, O.M., 2003. Slow, fast and flash pyrolysis of rapeseed. Renewable

Energy 28, 2417–2433.Onay, O., Kockar, O.M., 2004. Fixed-bed pyrolysis of rapeseed (Brassica napus L.).

Biomass Bioenergy 26, 289–299.Onay, O., Beis, S.H., Kockar, O.M., 2001. Fast pyrolysis of rape seed in a well-swept

fixed-bed reactor. Journal of Analytical and Applied Pyrolysis 58/59, 995–1007.Ozcimen, D., Karaosmanoglu, F., 2004. Production and characterization of bio-oil

and biochar from rapeseed cake. Renewable Energy 29, 779–787.Park, H.J., Park, Y.K., Dong, J.I., Kim, J.S., Jeon, J.K., Kim, S.S., Kim, J., Song, B., Park, J., Lee,

K.J., 2009. Pyrolysis characteristics of Oriental white oak: kinetic study and fastpyrolysis in a fluidized bed with an improved reaction system. Fuel ProcessingTechnology 90, 186–195.

Peter, M.K., 2002. Energy production from biomass (part 1): overview of biomass.Bioresource Technology 83, 37–46.

Putun, E., Ates, F., Putun, A.E., 2008. Catalytic pyrolysis of biomass in inert and steam.Fuel 87, 815–824.

Sharma, R.K., Wooten, J.B., Baliga, V.L., Lin, X., Chan, W.G., Hajaligol, M.R., 2004.Characterization of chars from pyrolysis of lignin. Fuel 83, 1469–1482.

Tsai, W.T., Lee, M.K., Chang, Y.M., 2007. Fast pyrolysis of rice husk: product yieldsand compositions. Bioresource Technology 98, 22–28.

Yorgun, S., Simsek, Y.E., 2008. Catalytic pyrolysis of Miscanthus x giganteus overactivated alumina. Bioresource Technology 99, 8095–8100.