School of Chemical Engineering, Universiti Sains Malaysia, Engineering Campus, 14300 Nibong Tebal, SPS Penang, Malaysia
Received 29 May 2003; received in revised form 18 March 2004; accepted 30 March 2004
Abstract
A substantial amount of fatty acid mixture residue is generated in the palm oil-based oleochemical industries in Malaysia.
The recovery of fatty acids residue is dicult and not economical; therefore, its alternative usage is explored. The fatty
acid mixture can be converted to liquid hydrocarbon fuel and chemicals using catalytic technology. In the present study, the
catalytic cracking of fatty acid mixture residue to liquid hydrocarbon fuel was studied over HZSM-5 catalyst in a xed-bed
microreactor at atmospheric pressure. The eect of reaction temperature (400450
C to 230
C
and with a fatty acid mixture residue feed rate (weight
hourly space velocity, WHSV) of 2.54:5 h
1
while
the fatty acid mixture to catalyst ratio was varied
between 6 and 10. HZSM-5 (1:0 g) was loaded over
0:2 g of quartz wool supported with a stainless steel
mesh in the microreactor placed in the vertical tube
furnace (model no. MTF 10/25/130, Carbolite) and
the temperature was monitored by a thermocouple
positioned in the center of the catalyst bed. Nitrogen
gas was passed through the reactor for 1 h before
the fatty acid mixture was fed using a syringe pump
(Cole Parmer model no. E-74900-05). Once steady
state had reached in the reactor, the liquid product (or-
ganic and aqueous fractions) was collected in a liquid
sampler while the gaseous product was collected in a
gas-sampling bulb. The process run time of each ex-
periment varied from 2 to 4 h depending on the feed
rate and fatty acid mixture/catalyst ratio. The uncon-
verted fatty acid mixture was separated from the liquid
product in a microdistillation unit (Buchi B850, GKR)
operated at 200
C) (g/g cat) (h
1
)
A B C
400 6 2.5 1 1 1 95.1 61.2 38.3 22.2
425 6 2.5 0 1 1 95.0 58.2 40.3 27.4
450 6 2.5 1 1 1 97.8 57.4 41.6 32.5
400 8 2.5 1 0 1 97.4 62.2 39.8 24.0
425 8 2.5 0 0 1 96.0 59.6 40.0 26.6
450 8 2.5 1 0 1 98.3 55.8 40.9 33.4
400 10 2.5 1 1 1 91.4 61.9 38.4 19.3
425 10 2.5 0 1 1 95.4 59.3 42.1 27.0
450 10 2.5 1 1 1 97.9 56.4 36.1 33.1
400 6 3.5 1 1 0 73.8 49.2 34.9 7.9
425 6 3.5 0 1 0 87.9 59.0 42.5 16.7
450 6 3.5 1 1 0 95.3 59.7 41.2 26.7
400 8 3.5 1 0 0 73.8 54.4 39.6 8.5
425 8 3.5 0 0 0 94.9 60.1 42.1 26.0
450 8 3.5 1 0 0 95.7 59.0 42.9 27.2
400 10 3.5 1 1 0 67.3 46.8 34.0 11.0
425 10 3.5 0 1 0 86.7 56.9 41.7 18.0
450 10 3.5 1 1 0 95.1 61.4 44.5 24.4
400 6 4.5 1 1 1 67.6 42.3 31.7 8.0
425 6 4.5 0 1 1 79.8 49.9 34.6 21.5
450 6 4.5 1 1 1 96.0 62.8 42.4 26.2
400 8 4.5 1 0 1 67.3 45.1 28.3 9.1
425 8 4.5 0 0 1 91.6 59.6 44.9 20.5
450 8 4.5 1 0 1 96.2 59.0 42.1 27.1
400 10 4.5 1 1 1 68.2 45.7 33.1 12.0
425 10 4.5 0 1 1 74.3 50.7 35.0 13.8
450 10 4.5 1 1 1 92.4 52.3 40.8 30.9
Repeated trial
425 8 3.5 0 0 0 87.9 57.7 37.6 20.3
425 8 3.5 0 0 0 93.6 61.4 45.3 17.7
425 8 3.5 0 0 0 93.5 64.0 47.6 19.0
425 8 3.5 0 0 0 91.1 64.6 41.8 15.0
425 8 3.5 0 0 0 92.2 61.6 47.1 21.7
Y.-S. Ooi et al. / Biomass and Bioenergy 27 (2004) 477484 481
dened as
Conversion (wt%)
=
[gas (g) + OLP (g) + water (g) + coke (g)]
fatty acid mixture feed(g)
100%; (1)
Yield (wt%) =
desired product (g)
fatty acid mixture feed (g)
100%; (2)
where the OLP represents organic liquid product
and the total products are the sum of products gas,
OLP, water and coke. The major components of the
gaseous products were propane, propylene, butane
and butylenes. The OLP contains liquid hydrocar-
bons, which belong to gasoline fraction, kerosene
fraction and diesel fraction, respectively. The gaso-
line fraction contains hydrocarbons rich in aromatics.
Since it is complex mixture of hydrocarbons therefore
the fraction is based on the boiling point cut rather
than the composition.
Table 4 shows that conversion of fatty acid mixture
was almost constant, which was above 90 wt% at the
feeding rate of 2:5 h
1
. These results indicated that
at low space velocity, the conversion of the fatty acid
mixture was almost independent of the fatty acid to
catalyst ratio and the reaction temperature. However,
the product distribution varied as the temperature and
fatty acid mixture to catalyst ratio were changed. The
yield of gaseous product increased with the tempera-
ture and independent of increase in fatty acid to cata-
lyst ratio, while the OLP yield decreased from 62.2 to
55:9 wt%. The increment of the gas yield was higher
than the increase in OLP yield. Once the feed rate was
increased from 2.5 to 4:5 h
1
, the conversion of the
fatty acid mixture was in the range of 6780 wt% at
lower reaction temperature (400 and 425
C). How-
ever, the yield of organic liquid product as well as the
gasoline fraction increased at the expense the yield of
gaseous product. This was possibly due to the sec-
ondary cracking reaction, which was not active at
lower reaction temperature. However, no signicant
trend in the conversion was observed when the space
velocity was increased from 2.5 to 4:5 h
1
for the re-
action temperature higher than 425
C. The eect of
space velocity was found signicant at lower reaction
temperature (400
C).
Since several variables were studied, thus the con-
ditions which signicantly aected the product distri-
bution were to be identied statistically. Therefore,
the analysis of variance (ANOVA), was performed at
95% level of condence for the designed experiments
using the Design-Expert software (version 6.0.6). The
signicance of the individual and iteration of the fac-
tors were determined by the F-values. Table 5 shows
the summary of the results obtained. It was found that
the combined eects of the three factors were negligi-
ble for all the responses considered. The reaction tem-
perature and space velocity were signicant variables
for conversion as well as the yield of all products and
should be included in the model developed. The fatty
acid mixture to catalyst ratio did not play important
role in the conversion and product distribution except
coke formation (not shown). The coke formation can
be reduced by varying the temperature and WHSV at
constant fatty acid mixture to catalyst ratio.
In order to quantify the curvature eects, the data
from the experimental results were tted to a quadratic
equation. The test of the model adequacy was done
through lack-of-t F-tests [11]. Eqs. (3)(6) were
chosen to t the data after elimination of insignicant
variables and their interactions.
Statistical model (built with the codied factors):
P
C
= 90:68 + 9:04A 1:32B 6:11C
2:59A
2
4:20B
2
+ 2:60C
2
+0:48AB + 5:96AC 0:43BC
1:76B
2
C + 0:35BC
2
; (3)
Y
OLP
= 60:57 + 3:14A 0:45B 3:66C
2:64A
2
2:76B
2
1:50C
2
0:97AB + 4:82AC 0:58BC; (4)
Y
G
= 43:39 + 3:02A 0:081B 1:36C
2:28A
2
1:84B
2
2:29C
2
0:37AB + 2:51AC + 0:32BC; (5)
Y
Gas
= 19:25 + 7:75A + 0:023B 4:25C
0:37A
2
1:17B
2
+ 4:82C
2
0:10AB + 1:81AC + 0:32BC; (6)
482 Y.-S. Ooi et al. / Biomass and Bioenergy 27 (2004) 477484
Table 5
The analysis of variance (ANOVA) for the conversion, OLP yield, gasoline fraction yield and gas yield
Source of variance Sum of squares Degree of freedom Mean square F-value P-value
Conversion, P
C
Mean 2:496 10
5
1 2:496 10
5
Linear 2446 3 815 23.8 0.0001
Quadratic 219 3 73 5.2 0.0071
Cubic 96 7 14 1.0 0.4860
Model 2974 12 248 19.4
Residual 295 20 13
Lack of t 264 15 18 2.9 0.1237
Pure error 31 5 6
Signicant model term: A; C; B
2
; AC
OLP yield, Y
OLP
Mean 1:028 10
5
1 1:028 10
5
Linear 422 3 141 5.7 0.0036
Quadratic 190 3 63 6.6 0.0024
Cubic 76 7 11 1.2 0.3542
Model 906 9 101 10.5 0.0001
Residual 211 22 10
Lack of t 178 17 10 1.6 0.3156
Pure error 33 5 7
Signicant model term: A; C; A
2
; B
2
and AC
Gasoline fraction yield,Y
G
Mean 50653 1 50,653
Linear 198 3 66 4.4 0.0121
Quadratic 160 3 53 6.4 0.0029
Cubic 6 7 1 0.07 0.9992
Model 436 9 48 5.8 0.0004
Residual 185 22 8
Lack of t 112 17 7 0.5 0.8959
Pure error 72 5 14
Signicant model term: A; AC
Gas yield, Y
Gas
Mean 14237 1 14,237
Linear 1405 3 468 35.4 0.0001
Quadratic 169 3 56 7.7 0.0011
Cubic 20 7 3 0.3 0.9401
Model 1614 9 179 24.5 0.0001
Residual 161 22 7
Lack of t 91 17 5 0.4 0.9371
Pure error 70 5 14
Signicant model term: A; C and C
2
where P
C
represents conversion of fatty acid, Y
represents product yield (Y
OLP
for OLP, Y
G
for gaso-
line fraction and Y
Gas
for gas) and the values of
A, B and C are in terms of coded factor, i.e. 1; 0
and 1. In order for a model to be reliable, the re-
sults should be predicted with reasonable accuracy
Y.-S. Ooi et al. / Biomass and Bioenergy 27 (2004) 477484 483
Actual Conversion
P
r
e
d
i
c
t
e
d
C
o
n
v
e
r
s
i
o
n
61.74
71.28
80.81
90.35
99.89
61.74 71.28 80.81 90.35 99.89
R
2
= 0.91
(a)
(b)
Actual OLP Yield
P
r
e
d
i
c
t
e
d
O
L
P
Y
i
e
l
d
42.09
47.73
53.37
59.01
64.64
42.09 47.73 53.37 59.01 64.64
R
2
= 0.81
Fig. 1. Experimental versus predicted values for (a) conversion of
fatty acid mixture; (b) organic liquid product (OLP) yield.
by the model and compared with the experimen-
tal results. A comparison between the experimental
values with the simulated values of conversion and
OLP yield from the model are presented in Fig. 1.
The values of correlation coecient for the con-
version and yield of OLP were 0.91 and 0.81, re-
spectively. Since these values are greater than 80%,
it shows a good agreement between experimental
data and predicted values [6]. Three-dimensional
plots (response surface) for the yield of gaso-
line fraction obtained from Eq. (5) are plotted in
Fig. 2. The gasoline yield was increased with the
31.9
35.1
38.2
41.3
44.4
G
a
s
o
l
i
n
e
Y
i
e
l
d
,
w
t
%
400
425
450
2.5
3.5
4.5
Temperature,
o
C WHSV, h
-1
Fig. 2. Response surface plot for gasoline fraction yield ob-
tained from the statistical model over HZSM-5 for fatty acid
mixture/catalyst ratio of 8.
decrease of WHSV at higher temperatures within
the experimental range. At WHSV between 3.5 and
4:5 h
1
, the maximum gasoline fraction yield of
44:4 wt% was achieved at temperature of 450
C.
The optimum values of reaction temperature of
440