The yield of CNSL increased with increase in pressure, temperature and mass flow rate of SC-CO2.. Keywords: Cashew nut shell; Supercritical fluid extraction; Carbon dioxide; Mass transfer
Trang 1Extraction of cashew (Anacardium occidentale) nut shell liquid
using supercritical carbon dioxide
Energy Systems Engineering, Indian Institute of Technology, Bombay, Powai, Mumbai 400 076, India Received 7 August 2004; received in revised form 4 April 2005; accepted 6 April 2005
Available online 6 June 2005
Abstract
This work investigated the extraction of cashew nut shell liquid (CNSL) using supercritical carbon dioxide (SC-CO2) Effects of process parameters such as extraction pressure, temperature and flow rate of SC-CO2 were investigated The yield of CNSL increased with increase in pressure, temperature and mass flow rate of SC-CO2 However, under different operating conditions, the composition of CNSL varied The study of physical properties and chemical composition of the oil obtained through super crit-ical fluid extraction (SCFE) showed better quality as compared to the CNSL obtained through thermal route Experimental results were compared with diffusion based mass transfer model Based on this simple model, extraction time was optimized
2005 Elsevier Ltd All rights reserved
Keywords: Cashew nut shell; Supercritical fluid extraction; Carbon dioxide; Mass transfer model; Optimum extraction time
1 Introduction
India is the largest producer and processor of cashews
(Anacardium occidentale) in the world (Das and Ganesh,
2003) In India, cashew cultivation covers a total area of
about 0.77 million hectares of land, with an annual
pro-duction of over 0.5 million metric tonnes of raw cashew
nuts The average productivity per 100,000 m2is around
760 kg The world production of cashew nut kernel was
907,000 metric tonnes in 1998 (Smith et al., 2003) The
cashew nut shell liquid (CNSL) is reported to be
15–20% by weight of the unshelled nut in Africa and
25–30% by weight in India (Das and Ganesh, 2003)
Considering the shell weight is about 50% of the weight
of the nut-in-shell (NIS), the potential of CNSL is about
450,000 metric tonnes per year In India, processed
cash-ew dominates more than half the world cashcash-ew market
The residue after extraction of CNSL is shell cake,
which is a very useful fuel and a substitute for fire wood The innumerable applications of CNSL are based on the fact that it lends itself to polymerization by various means
Various methods have been reported in literature for the extraction of CNSL from CNS, which include, open pan roasting, drum roasting, hot oil roasting, cold extru-sion, solvent extraction, etc The extraction through vac-uum pyrolysis has been reported recently byDas et al (2004) and Tsamba (2004) The extraction of CNSL using supercritical carbon dioxide (SC-CO2) has also been reported by Shobha and Ravindranath (1991)
andSmith et al (2003) Conventionally, both the quantity and quality (com-position of CNSL) vary with the method of extraction of CNSL Various authors have reported varied composi-tion of CNSL extracted CNSL extracted by cold extru-sion method is reported to contain approximately 70% anacardic acid, 18% cardol and 5% cardanol and the balance consisting of substituted phenols and less polar substances (http://www.epa.gov/chemrtk/cnsltliq/c13793 tp.pdf) Das (2004) has also reported CNSL extracted
doi:10.1016/j.biortech.2005.04.009
25726875.
Trang 2by cold extrusion method with 90% anacardic acid and
nearly 10% cardol According toTyman (1979), natural
CNSL contains nearly 64% anacardic acid, 11% cardol,
traces of cardanol, 2–3% of 2-methyl cardol and rest
polymeric material About 52% cardanol, 10% cardol
and 30% polymeric material (Das, 2004) constitutes
the Technical Grade CNSL A composition 64.8%
card-anol, 20.5% cardol, 2.8% 2-methyl cardol, and rest as
non-volatile polymeric material have also been reported
(Tyman, 1975) for the Technical Grade CNSL The
CNSL obtained through vacuum pyrolysis is cardanol
rich It is reported to have cardol, substituted phenols,
di-n-octyl phthalate, bis(2-ethyl hexyl) phthalate, etc
The use of this CNSL as a potential fuel in internal
com-bustion engine has also been suggested (Das, 2004)
However, the composition of CNSL obtained by SCFE
has not been reported in the literature
SCFE, as mentioned by the authorsShobha and
Rav-indranath (1991) and Smith et al (2003), has inherent
advantages over other extraction methods such as no
polymerization of CNSL, requirement of less amount
of solvent, and no extraction of undesirable coloured
compounds In view of this, the present work is an
at-tempt to study the effect of operating parameters on
the yield and quality of CNSL extracted through SCFE
method A simple mathematical model is also developed
for optimization of profit and energy/yield The
objec-tive of the study is also to demonstrate the feasibility
of the component separation of CNSL using SCFE,
par-ticularly the higher molecular substances like cardanol
2 Experimental procedure
Cashew nut shell (CNS) obtained from Pondicherry
was used for the present study The shells were ground
to small particles (to pass through 8 mesh screen) and
weighed and then placed in the extractor Carbon
di-oxide (99.9%) supplied by M/s Sicgil Corporation,
Bombay, was used as the supercritical fluid (SCF) for
extraction of oil from CNS
SCFE unit supplied by M/s Deven Supercritical was
used for the present study Carbon dioxide from the
cyl-inder passed through a pre-cooler, a positive
displace-ment pump, and a pre-heater before it entered the
bottom of the extraction vessel (The extraction vessel
was maintained at a predefined temperature.) The flow
of carbon dioxide was controlled by a needle valve
and was measured by a gas flow meter with an accuracy
of ±0.01 kg/h A variable frequency drive pump
con-trolled the pressure in the vessel to an accuracy of
±0.1 bar Extracted oil was recovered by expansion of
the loaded solvent stream to ambient pressure in a glass
separator Extract was collected and weighed at a fixed
time interval of 30 min (cumulatively) by closing the
needle valve The needle valve was then opened and
extraction process continued for the next interval Runs were carried out for 6 h at the pressures ranging from
200 to 300 bar at 25 bar intervals The extract of each run was analysed by Gas Chromatograph Mass Spec-troscopy (GC-MS) and Fourier Transform Infra-Red Spectroscopy (FTIR)
3 Results and discussion 3.1 Effect of pressure on yield of CNSL The total yield at various pressures from 200 bar to
300 bar, keeping other parameters constant, is shown
in Fig 1 Temperature and mass flow rate of carbon dioxide were kept constant at 333 K and 1.0 kg/h respec-tively Evidently total yield increased with increase in pressure from 200 bar to 300 bar—the yield being four
to five times higher at 300 bar than at 200 bar for the same consumption of SC-CO2 (at the same flow rate and temperature) This could be explained by the fact that the extraction capacity of solvent at the supercriti-cal state was density dependent It was also observed that the rate of extraction was high during initial phase
of extraction as the material is loaded with oil The rate
of extraction decreased at later stages as shown inFig 2 The FTIR analysis was used to identify the components, particularly ÔcardanolÕ
3.2 Effect of pressure on the yield of cardanol The samples were analysed by FTIR, GC-MS and Ultraviolet (UV) spectroscopy The FTIR and GC-MS aided in identifying the functional groups present and the components, respectively The UV spectroscopy, cal-ibrated for a commercial grade refined cardanol sample, was used to determine the approximate percentage of cardanol in the samples The results of FTIR, GC-MS and UV spectroscopy are summarized in Table 1 It was interesting to note that ÔacidÕ group was traced by
Fig 1 Variation in yield of CNSL and cardanol with extraction pressure (extraction time 270 min, extraction temperature 333 ± 0.5 K and mass flow rate of SCF 1 ± 0.01 kg/h).
Trang 3FTIR only for CNSL obtained above 225 bar However,
GC-MS did not identify Ôanacardic acidÕ as a major
group and therefore, was assumed to be in traces
GC-MS and FTIR analysis showed that at lower
pressure CNSL mainly consisted of cardanol Amount
of cardanol in CNSL decreased with increase in pressure
from 86% at 200 bar and 333 K to 63% at 300 bar and
333 K.Fig 1shows the CNSL yield, percentage
carda-nol in CNSL extracted and percentage cardacarda-nol
ex-tracted based on original CNS used This percentage
cardanol was the product of the percentage yield of
CNSL with percentage cardanol in CNSL The cardanol
yield, therefore, was higher at higher pressure
3.3 Effect of temperature on yield of CNSL
It is known that the yield of extract depends on the
change in density and volatility of SCF With increase
in temperature, the density of SCF decreased while
vol-atility increased (Mukhopadhyay, 2000) Hence
experi-ments were carried out at isochoric density by
modifying pressure The effect of temperature on total
yield of CNSL is shown inFig 3 It could be seen that
with increase in temperature, total yield of CNSL
in-creased at a given mass flow rate and density
3.4 Effect of mass flow rate of SCF on the
yield of CNSL
It is well understood that with increase in solvent to
solid ratio, the rate of extraction is enhanced, and hence
extraction time is reduced The effect of mass flow rate
of SCF on total yield is shown inFig 4 It was observed
that with increase in flow rate of SCF, total CNSL yield
increased However, due to lower retention time,
load-ing of SCF was lower, thereby reducload-ing the capacity
Fig 2 Cumulative yield of CNSL at different extraction pressure.
temperature 333 ± 0.5 K.
Trang 43.5 Properties of oil extracted by SC-CO2
3.5.1 Physical properties of CNSL
The physical properties of the oil extracted at various
operating conditions were studied using standard test
procedures It was observed that the calorific value of
the oil was almost same for all extraction conditions
while the density was in the narrow range of 0.92–
0.934 kg/m3
Table 2gives the comparison of physical properties of
oil extracted at 300 bar and 333 K—using SCFE with
IS:840 (1964)and oil obtained through vacuum
pyroly-sis It could be observed that the moisture content, den-sity and viscoden-sity of oil obtained through SCFE were very close to IS:840 (1969) specifications, whereas these properties were better as compared to CNSL obtained through vacuum pyrolysis Other properties of CNSL extracted using SCFE were very close to that obtained
by vacuum pyrolysis The comparison of this oil with oil obtained through vacuum pyrolysis was relevant in terms of CNSL as a potential bio fuel
3.5.2 Chemical composition of CNSL The oil obtained at various operating parameters was analysed for chemical compositions Table 3shows the main components present in CNSL obtained at various operating parameters It was noted that the main com-ponent in CNSL was cardanol with side chain having 13–17 carbon atoms; however, their concentration was different at different operating parameters This could
be attributed to the enhanced decarboxylation at higher pressure (Hazen et al., 2002)
3.6 Residue analysis The residue, after extracting oil from CNS at 300 bar and 333 K in supercritical fluid extractor was pyrolysed
at 773 K under vacuum of 700 mm of Hg The vapours were condensed to find the condensates The weight of the condensed oil was hardly found 2% suggesting al-most complete extraction of CNSL
4 Mathematical model and optimization
In extraction, the solute from the cell matrix dis-solved into the bulk fluid Extraction of the solute be-comes simple when it is free on the surface of solids
On the other hand, as the solute interacts with cell ma-trix its extraction becomes difficult For natural material with a high initial content of extractable, the rate of extraction remains constant at the initial period As the outer surface of the solid is depleted of the extract-able solute, solute from the core of the solid requires more time to reach the fluid–solid interface This results
in a drop in the rate of extraction with time ( Ganga-dhara Rao, 1990)
For extraction, several models have been proposed: unsteady-state packed bed mass transfer model (Mukhopadhyay, 2000), shrinking core leaching model (Mukhopadhyay, 2000; Goto et al., 1996), empirical models (Subra et al., 1998;Chrastil, 1982), etc The un-steady-state packed bed mass transfer model represents the concentration profile of the SCF solvent phase in the extractor with respect to time and length of bed
In this model all constituents are clubbed together as a solute, as it is believed that they would have similar mass transfer characteristics (Mukhopadhyay, 2000) The
Fig 4 Cumulative yield of CNSL at different extraction flow rates of
333 ± 0.5 K.
Fig 3 Cumulative yield of CNSL at different extraction temperatures.
Trang 5shrinking–core leaching model accounts for
intra-parti-cle diffusion, external fluid film mass transfer and axial
dispersion Several empirical models have been
pro-posed: Christil model, Kinetic model, etc The Christil
model (Chrastil, 1982) provides yield as a function of
density of SCF and extraction temperature It gives
con-stant rate of extraction and hence, only applicable for
the initial period of extraction In kinetic model (Subra
et al., 1998) the rate of extraction is assumed to decrease
exponentially and the rate constant is determined by
regressing experimental data This model is independent
of matrix parameters
With the following assumptions, the kinetic model
was used in this study Essential oil is assumed to be
uni-formly distributed over cell matrix Axial dispersion is neglected Solid ground particles have uniform struc-ture Flow rate of SCF, system temperature and pres-sure are constants and velocity of SCF through extractor is negligible If C is mass fraction of solute
in SCF over given period of time, Cinf is total amount
of solute present in solid, the kinetic model This model
is expressed as:
K is rate constant Through experimental studies, it has been observed that the rate constant, K, depends
on pressure, temperature and mass flow rate of the solvent
Table 2
Comparison of oil obtained by SCFE and thermal method
( Das and Ganesh, 2003 )
Absolute viscosity (cSt) at
(ASTM D445-88)
O (by difference)
Table 3
Identification of components in CNSL obtained at different operating parameters using GC-MS
Operating parameters
(P bar/T K)
Retention time (min)
peak in GC spectra
Trang 6K¼ K0þ KP P þ KF F þ KT T ð2Þ
where, P, F and T are pressure (bar), mass flow rate
of solvent (kg/h) and temperature (K) respectively By
regressing the experimental results, following values
are obtained: K0=0.004091, KP= 2.34· 10(5), KF=
0.0030923, KT=8.70563 · 10(6)
Positive values of KP and KF indicate that with
in-crease in pressure and mass flow rate of solvent, the rate
constant and hence the yield increases This is in
accor-dance with variation of CNSL yield with pressure and
mass flow rate of CO2as shown in Figs 2 and 4 The
negative value of temperature co-efficient in Eq.(2)
sug-gests that with increase in temperature, the yield of
CNSL decreases However, its sensitivity on CNSL yield
is very low
4.1 Optimum extraction time
The CNSL yield can be predicted using the kinetic
model for given operating parameters It is well known
that the yield depends on extraction pressure,
tempera-ture, mass flow rate of the solvent and time of
extrac-tion The operating parameters may be determined to
find maximum profit
The pump, pre-cooler and pre-heater of SCFE pilot
plant consume the energy The pump increased the
pres-sure of liquid CO2from cylinder pressure to extraction
pressure The energy required (WP) for driving a pump
depends on the density of CO2(q), mass flow rate (F),
differential pressure (pressure difference between
deliv-ery pressure and cylinder pressure) and the mechanical
efficiency (gm,P) of the pump
WP¼FðP PCÞ
It is to be ensured that CO2must be in liquid phase
before it enters the pump Carbon dioxide from cylinder
enters the pre-cooler and gets converted into liquid
phase The energy required (WC) for driving the
pre-cooler is expressed as:
3600
CPfðTsup TsatÞ þ hfgþ CPlðTsat TsubÞ
COP
ð4Þ where CPf and CPlare specific heats of gaseous and
li-quid CO2at cylinder pressure; hfg is the latent heat of
vaporization at cylinder pressure; Tsup is temperature
of CO2at exit of cylinder; Tsatis saturated temperature
at cylinder pressure; Tsubis temperature of liquid CO2at
the exit of pre-cooler and COP is the coefficient of
per-formance of a pre-cooler
Pre-heater increases the temperature of CO2 to
extraction temperature before it enters the extraction
vessel If h1 and h2 are specific enthalpies of CO2 at
the entry and exit of pre-heater, the energy required
for pre-heater (W ) is expressed as:
WH¼Fðh2 h1Þ
The specific energy (kW h/g m) required running SCFE pilot plant for extraction time, t, is:
E¼ðWPþ WCþ WHÞ t
Extraction of CNSL from CNS using SCFE is a semi-continuous process At the end of each batch, the extrac-tor is depressurized, fed with fresh raw material and again pressurized for extraction period There is a non-productive down time between two consecutive extraction batches Moreover, the rate of extraction is higher in the initial phase of extraction due to easily available oil on the outer surface of the matrix The rate
of extraction diminishes with time Hence, it is also pos-sible to determine the time of extraction of CNSL to get maximum profit per day Daily energy cost, cost for recycling the CO2, revenues from selling of CNSL and labor cost are considered for the formulation of profit function:
Daily profit ¼ revenues from CNSL
daily energy cost
cost of recycling of CO2 labor cost The objective function (daily profit) is optimized with following constraints:
190 bar < P < 300 bar 310 K < T < 333 K 0.8 kg=h < F < 1.3 kg=h
Optimizing above objective function, the values of operating parameters i.e extraction pressure, tempera-ture, mass flow rate of CO2and time of extraction were decided The results obtained by model for daily profit optimization suggested that the daily profit was mum for maximum possible pressure (300 bar), maxi-mum flow rate (1.3 kg/h) and minimaxi-mum temperature (310 K) The optimum extraction time was 145 min cor-responding to maximum daily profit The oil content in raw CNS was 29% (by weight) Though, it was possible
to extract total oil by SCFE method, yet, based on the parameters obtained by optimizing the objective func-tion for daily profit using model, it was found that only 13% of oil extraction (corresponding to optimized time
of extraction) maximizes the profit This was due to the fact that the rate of extraction of oil at the later stage
of extraction was low, which resulted into increase in oil extraction cost
5 Conclusions
In this work, the effect of various operating parame-ters on the yield of CNSL was studied It was observed that extraction pressure was the key parameter, which
Trang 7regulated the total yield It was also noticed that the
sol-ubility of supercritical fluid varied with extraction
pres-sure, temperature and mass flow rate of SCF The yield
obtained was clear and with light yellow in colour The
properties of CNSL obtained through SCFE were
com-pared with oil specifications mentioned inIS:840 (1964)
for CNSL The moisture content, density and viscosity
of the oil obtained through SCFE were very close to
CNSL specifications The chemical analysis of CNSL
ob-tained in present study showed that it mainly conob-tained
cardanol (70–90%) It hardly contained anacardic acid,
while traces of cardol were found only at high pressures
This suggested that selective separation of components
was possible by SCFE The study of effect on percentage
yield of cardanol in CNSL at various pressures showed
that with increase in pressure concentration of cardanol
decreased due to extraction of components with higher
molecular weight along with cardanol at higher pressure
However, total yield of the cardanol from CNS could be
predicted using the product of yield of CNSL from CNS
and yield of cardanol from CNSL in turn
The study showed that the model developed could be
used to predict the yield of CNSL in the pressure range
of 190–300 bar Considering the down time for loading
and unloading of feed in each batch, the daily profit
optimization gave the optimum values for extraction
pressure, temperature, flow rate of solvent and also time
of extraction The results of model suggested that
in-stead of complete extraction of CNSL from CNS,
par-tial extraction of CNSL would give more daily profit
References
Chrastil, J., 1982 Solubility of solids and liquids in supercritical gases.
Journal of Physical Chemistry 86, 3016.
Das, P., Ganesh, A., 2003 Bio-oil from pyrolysis of cashew nut shell—
A near fuel Biomass and Bioenergy 25, 113–117.
Das, P., 2004 Studies on pyrolysis of sugarcane baggase and cashewnut shell for liquid fuels Ph.D dissertation, Indian Institute
of Technology, Bombay, India.
Das, P., Sreelatha, T., Ganesh, A., 2004 Bio-oil from pyrolysis of cashew nut shell-characterization and related properties Biomass and Bioenergy 27, 265–275.
Gangadhara Rao, V.S., 1990 Studies on supercritical fluid extraction Ph.D dissertation, Indian Institute of Technology, Bombay, India Goto, M., Roy, B.C., Hirose, T., 1996 Shrinking–core leaching model for supercritical fluid extraction Journal of Supercritical Fluids 9, 128–133.
Hazen, R.M., Boctor, N., Brandes, J.A., Cody, G.D., Hemley, R.J., Sharma, A., Yoder Jr., H.S., 2002 Journal of Physics: Condensed Matter 14, 11489–11494.
http://www.epa.gov/chemrtk/cnsltliq/c13793tp.pdf , date visited 30th March, 2005.
IS:840, 1964 Indian standards: Specifications for cashew nut shell liquid (CNSL) Revised.
Mukhopadhyay, M., 2000 Natural Extract Using Supercritical Carbondioxide CRC Press, Boca Raton.
Shobha, S.V., Ravindranath, B., 1991 Supercritical carbon dioxide and solvent extraction of the phenolic lipids of cashew nut (Anacardium occidentale) shells Journal of Agricultural Food Chemistry 39, 2214–2217.
Smith, R.L., Malaluan, R.M., Setianto, W.B., Inomata, H., Arai, K.,
2003 Separation of cashew (Anacardium occidentale) nut shell liquid with supercritical carbon dioxide Journal of Bioresource Technology 88, 1–7.
Subra, P., Castellani, S., Jestin, P., Aoufi, A., 1998 Extraction of
b-carotene with supercritical fluids—Experiments and modeling Journal of Supercritical Fluids 12, 261–269.
listserv.repp.org/pipermail/gasification/2004-September/007105.html > date visited 30th March, 2005.
Tyman, J.H.P., 1975 Long-chain phenols IV quantitative determina-tion of the olefinic composidetermina-tion of the component phenols in cashew nut shell liquid Journal of Chromatography 111, 277– 284.
Tyman, J.H.P., 1979 Non-isoprenoid long chain phenols Chemical Society Review 8 (4), 522.