You are on page 1of 8

23rd International Symposium on Transport Phenomena

Auckland, New Zealand


1922 November 2012

Modelling of the Gas-Liquid Transesterification Reaction to Produce Biodiesel

S.T. Hoh, M.M. Farid and J.J.J. Chen


Department of Chemical and Materials Engineering
University of Auckland, Auckland 1142, New Zealand

Abstract = density (kg/m3)


= dynamic viscosity (Pa.s)
A kinetic model is derived to provide fundamental understanding = surface tension of soybean oil = 0.025 kg/s2
of the gas-liquid transesterification reaction between triglycerides Re = Reynolds number =
and methanol catalysed by sodium methoxide to produce
biodiesel. The model is based on the fundamentals of the gas Sc = Schmidt number =
absorption process followed by a second-order liquid phase
chemical reaction. Important parameters for the model include We = Weber number =
the Hatta modulus (MH) and the enhancement factor (E). A
laboratory scale droplet reactor set up was used to carry out Sh = Sherwood number =
isothermal gas-liquid transesterification reactions between
soybean oil droplets and vapour methanol at 80, 90 and 100oC.
The gas-liquid transesterification reaction is an intermediate Subscripts
reaction which occurs both in the liquid film and bulk. The
overall process is predominated by mass transfer effects while the A, g = gas/continuous phase
liquid phase reaction rate constant still remains important. The B, l = liquid/dispersed phase
second-order reaction rate constants for the gas-liquid d = droplet
transesterification reaction at 80, 90 and 100oC were calculated to i = interfacial
be 0.0279, 0.237 and 1.796 m3/mol.s respectively. The activation j = jet
energy is rather high at 228 kJ/mol, indicating that the reaction is 0 = initial
very temperature sensitive.
Introduction
Nomenclature
Biodiesel consists of fatty acid methyl esters derived from the
a = interfacial area per unit volume of oil phase (m2/m3) transesterification of triglycerides, which are the building blocks
A = reactant A of all natural fats and oils, with alcohol. Methanol is most
Ar = Arrhenius Law Constant commonly used to produce methyl esters as biodiesel because of
= methanol vapour (gaseous/continuous phase) its abundance and low cost. Biodiesel is established as a potential
B = reactant B alternative to petroleum fuels after being the subject of numerous
= soybean oil (liquid/dispersed phase) researches in the past few decades. It is very attractive because of
C = instantaneous concentration (mol/m3) its non-toxicity and it may be readily used in pure form (B100) or
d = diameter (m) blended with petroleum fuels (B5 B20). In addition, no engine
dm = effective diameter of molecule (m) modifications are necessary to enable its usage. The major
D = diffusion coefficient of solute in a solvent (m2/s) advantages biodiesel have over conventional petroleum or fossil
DG = diglycerides fuels include [1-3]:
E, Ei = enhancement factor and infinite enhancement factor
respectively i. Renewability
Ea = Activation Energy (J/mol) ii. Biodegradability
HA = Henrys Law Constant of reactant A (Pa.m3/mol) iii. Reduced amount of harmful emissions (with exception to
k1 = second order reaction rate constant (m3/mol.s) nitrogen oxides)
kL = liquid film mass transfer coefficient (m/s) iv. Higher engine combustion efficiency
MG = monoglycerides v. Better lubricity and ignition properties which prolong
MH = Hatta modulus engine life
n = number of moles
R = universal ideal gas constant = 8.31434kg.m2/s2.mol.K Although biodiesel has a lot of benefits as a green fuel, it is also
T = temperature (K or oC) not short of weaknesses. One of the major problems that hinder
t = contact/residence time or reaction time (s) the feasibility of biodiesel as an economically competitive
TG = triglycerides alternative to fossil fuels is its higher selling price, contributed by
v = velocity (m/s) expensive feedstock prices and its lengthy production process [3,
VA , B = molar volume of reactant A or B (cm3/mol) 4].
Vb = molar volume at boiling point of reactant A
(cm3/mol) The commercial technology widely used today for biodiesel
X = conversion (%) manufacture is the base-catalysed transesterification method
z = stoichiometry constant (Figure 1). This is a homogeneous reaction between liquid oil and
= liquid holdup or the fraction of liquid volume to liquid methanol using a base catalyst. The base-catalysed process

reactor volume is currently most feasible because it only requires mild


processing conditions and within 30 - 60 minutes to obtain and the concentration difference at the interface (C Ai) and bulk
reasonable conversion of oil to biodiesel i.e. atmospheric liquid (CA0). As with any mass transfer operations, the gas-liquid
pressure and temperatures of approximately 60 - 70oC [2, 5]. interfacial area per unit reactor volume (a) is also a very
important parameter in the equation.
The process of gas absorption coupled with chemical reaction on
the other hand, must consider the kinetics of both mass transfer
and chemical reaction simultaneously [9]. In addition to the mass
transfer resistances described earlier, the chemical reaction terms
should also be included in the model. A general scheme
demonstrating this phenomenon is shown below:

k1 kG, kL
A (g) A (l ) z.B (l ) P (l ) (2)

In the case of gas-liquid transesterification, gaseous reactant (A)


would be methanol and liquid reactant (B) would be the
feedstock oil, while the products (P) are the methyl esters and
glycerol. The stoichiometric constant (z) is one third according to
the stoichiometric conditions of the transesterification reaction
between triglycerides and methanol:
Figure 1: The base-catalysed transesterification method for B+ 3A 3 Alkyl Esters + Glycerol (3)
biodiesel production [2]
The following mathematical description of the actual mechanism
taking place at the gas-liquid interface is based on the work of
However, this technology limits the operating temperature below Charpentier [9]. According to Equation 3, the following rate
the boiling point of methanol (~65oC) although it is generally equations may be written for the liquid phase transesterification
known that higher temperatures will increase the reaction rate reaction which is considered first-order with respect to the
(with accordance to Arrhenius Law). The immiscible oil and concentration of oil (B) and methanol (A), therefore second-order
methanol phases also pose as a mass transfer barrier during early overall [11].
stages of the reaction thus contributed to overall slower reaction
rates [2]. This may be improved by vigorous mixing [6] or the ( ) (4)
use of a common solvent [7].
Assuming steady state conditions, a material balance for both
reactants within a thin liquid film adjacent to the gas-liquid
Recognising the shortcomings of the current biodiesel processing interface may be written in terms of their concentrations (CA and
technology, Behzadi and Farid [8] developed a new technology CB) and their respective liquid diffusion coefficients (DA and DB).
to produce biodiesel in which reaction rates are drastically The distance from the gas-liquid interface within the liquid film
elevated via higher temperatures and improved mass transfer. is x.
This is as an effort to decrease processing costs through shorter
production times. A novel spray tower reactor was used to carry ( ) (5)
out the gas-liquid transesterification of different feedstock (beef
tallow or soybean oil) atomised into small droplets with methanol
( ) (6)
vapour and catalysed by sodium methoxide. Feedstock
conversion as high as 94 96% was achieved within 7 to 8 Equations 15 and 16 are subject to the following boundary
seconds and the biodiesel product was of standard quality. conditions at the gas-liquid interface (x = 0).

The objective of this work is to model the kinetics of the gas- (7)
liquid transesterification reaction between triglycerides and
methanol catalysed by sodium methoxide to produce biodiesel. While at the inner edge of the liquid film (x = ), the boundary
The model is based on the fundamentals of gas absorption condition for reactant B is:
followed by chemical reaction and provides fundamental
(8)
understanding to this process in addition to being suitable for
future design and scale-up purposes. The boundary condition for reactant A is determined by
considering that some amount reacts within the liquid film while
Theory and Background the rest is transferred to the liquid bulk. Taking the specific gas-
liquid interfacial area (interfacial area per unit reactor volume) as
Gas Absorption Followed by Chemical Reaction a and the liquid holdup (volume of liquid per total volume of
reactor) as , this liquid bulk may be written as:
In the absence of any chemical reaction, the absorption of a gas
into a liquid is purely physical, governed by mass transfer Liquid bulk = (9)
resistances and gas-liquid equilibrium (Henrys Law). At steady
state, the gas absorption rate ( ) can be expressed in terms of the With the (initial) concentration of reactants in the liquid bulk to
mass transfer coefficient and the drive for solute transfer partial be CA0 and CB0, the boundary condition for reactant A may be
pressure and/or concentration difference [9, 10]. written as:

= ( ) ( ) (1) ( ) ( ) (10)
Here, the gas side mass transfer is controlled by the gas phase Depending on how fast the liquid phase reaction is (k1 value),
mass transfer coefficient (kG) and the difference or partial this corresponds to different concentration profiles within the
pressure in the gas (PA) and at the gas-liquid interface (PAi). Once liquid film. Typical concentration profiles of the reactants (Cases
solute A enters the liquid phase, the mass transfer resistance A through H) are shown in Figure 2 [10].
now is governed by the liquid-film mass transfer coefficient (kL)

2
Enhancement Factor and Hatta Modulus
As discussed in the previous section, the enhancement factor
(Equation 11) is introduced as means to provide a simplified
approach to describe gas-liquid reactions. Instead of attempting
to solve Equations 5 through 10 simultaneously, the prediction of
E is now the key to model the process of gas absorption coupled
with chemical reaction the main objective of this work.
It was found that E may be expressed as a function of the Hatta
modulus (MH) and the infinite enhancement factor (Ei). This is
shown in Figure 3 based on the numerical solutions of van
Krevelens and Hoftjizer in 1954, and modified by Levenspiel
[10].

Figure 2: Typical concentration profiles of reactants near the gas-


liquid interface [10]
With reference to Figure 2, the different reaction schemes are
categorised accordingly [10]:
i. Case A: Instantaneous reaction at interface with low CB
ii. Case B: Instantaneous reaction at interface with high and
constant CB (pseudo-first order)
iii. Case C: Fast reaction in liquid film with low CB
iv. Case D: Fast reaction in liquid film with high and constant
CB (pseudo-first order)
Figure 3: E as a function of MH and Ei [10]
v. Case E and F: Intermediate reaction in both liquid film and
bulk The Hatta modulus (MH) is also known as the film conversion
parameter. It is the measure of the maximum possible chemical
vi. Slow reaction in liquid bulk with liquid film resistance conversion of a solute in the liquid film compared to the
vii. Slow reaction in liquid bulk with no mass transfer maximum transport through the film [10]. In the context of the
resistance gas-liquid transesterification reaction, this is written as:

To determine the most suitable reaction scheme to apply for


(13)
modelling purposes, the Hatta modulus (MH) must be calculated.
This will be discussed further in the next section. Therefore, it is an important parameter to consider when
At present, a complete analytical solution of Equations 5 through determining the concentration profiles of reactants as shown in
10 is not possible [9]. However, approximate solutions are Figure 2. As a general rule of thumb, the following applies to
realised by incorporating the enhancement factor E. The different reaction schemes [10]:
enhancement factor is a term that illustrates the effect of having a i. MH > 2.0: fast reactions occurring inside the liquid
liquid phase chemical reaction on the overall gas absorption rate film; Cases A, B, C and D
[10]:
ii. 0.02 < MH < 2.0: intermediate reactions occurring in
(11) both film and bulk; Cases E, F and G

The enhancement factor is dimensionless and takes a value of iii. MH < 0.02: extremely slow reactions occurring in liquid
1.00 or larger, therefore indicating that having a liquid phase bulk; Case H
chemical reaction will either augment the absorption rate (E > It may be observed in Figure 3 that for the region where MH >
1.00) or have no effect on it at all (E = 1.00). In other words, the 2.0, E increases with MH until it reaches an asymptotic value.
enhancement factor is a measure of the improvement of solute This value is known as the infinite enhancement factor (E i). It is
A absorption rate when there is a presence of a liquid phase the maximum achievable E if an infinitely fast reaction exists in
chemical reaction. the liquid film. In other words, it is the limiting value of E given
Consequently, for the heterogeneous gas-liquid reaction scheme the specific conditions of a particular gas-liquid reaction system.
presented in Figure 2, a general expression to represent the
( )( ) (14)
overall reaction rate with respect to the gaseous reactant A (-
RA) may be written [10]: If the values of MH and Ei are known, then the value of E may be
(12) predicted using Figure 3. The value of E may also be calculated
using the corresponding implicit mathematical representation of
Figure 3 developed by van Krevelens and Hoftjizer based on the
film model [9]:

3
( ) collected into T-03. Product samples were collected by closing

( ) V-02 and opening V-03.
(15)
( ) P
{ }
( ) P1
V-01

P2
It should be noted that many other researchers have also Compressed
Air Supply
T1
developed implicit and explicit mathematical solutions to E as a T-01 V-02

function of MH and Ei which fit Figure 3 very closely [9]. The Equipment & Instrument List T2
P-17

Annotation Description
most convenient correlation for gas absorption accompanied by a C-01 Droplet Reactor Column
H-02

second-order liquid phase reaction was developed by DeCoursey H-01


H-02
Methanol Heater
Methanol Condenser
C-01

[12] based on the Danckwerts (surface renewal) model: P-01 Methanol Pump
T-01 Soybean Oil Feed Tank
T-02 Methanol & Catalyst Feed Vessel
T-03 Used Methanol Vessel
[ ]
V-04
( )
(16) P1 Pressure Regulator H-01
( ) P2 Pressure Gauge
V-03
P-01
T1, T2 Thermocouple
V-01 Pressure Relief Valve
Derivation of Model V-02, -03, -04 Manual Valves Sample
collection

According to the conditions of the gas-liquid transesterification T-02 T-03


reaction conducted by Behzadi and Farid [8], the following
assumptions hold: Figure 4: Droplet reactor set up
i. The gas film resistance to mass transfer is negligible Samples collected were then analysed using Flame-Ionised
because pure methanol vapour was used. Detector Gas-Chromatography (GC-FID) to ascertain its
composition, mainly the mass fractions of mono-, di- and
ii. The liquid phase reaction is non-reversible because a large
triglycerides, and methyl esters.Eight isothermal gas-liquid
excess of methanol vapour was used (30:1 methanol to oil
transesterification reactions were performed and the results
molar ratio).
obtained. The conditions of the reactions are summarised in
In addition, the use of effective heating elements means the heat Table 1.
transfer resistance within the laboratory scale droplet reactor is
negligible because isothermal conditions were achieved within T-01
T v d32
the reactor. Run Pressure t (s)
(oC) (m/s) ( 106 m)
(kPa)
With accordance to the first assumption, the value of kG becomes 1 80 55.2 8.03 0.0567 492
infinitely large and thus Equation 12 reduces to:
2 80 82.7 11.35 0.0401 436
(17) 3 90 82.7 11.35 0.0401 428
4 90 55.2 8.03 0.0567 440
The kinetic model is derived by integrating Equation 17: 5 100 55.2 8.03 0.0567 408
( ) 6 100 82.7 11.35 0.0401 408
( ) (18) 7 100 110.3 14.22 0.0320 408
8 100 137.9 16.63 0.0274 408
Equation 18 is used as a model to describe the kinetics of gas-
liquid transesterification reaction to produce biodiesel. The first Table 1: Conditions of the isothermal gas-liquid
( ) transesterification experiments reaction temperature, T; driving
term on the left hand side [ ] corresponds to the mass pressure of the soybean oil jet; velocity of the soybean oil
transfer resistance while the second term [ ( )], the jet/droplets, v; residence time of the soybean oil droplets, t; and
the Sauter Mean Diameter of the soybean oil droplets, d32.
reaction kinetics resistance.
Modelling Section
Materials and Methods
The modelling of the gas-liquid transesterification reaction is
Experiment Set Up based on Equation 18. The main aim of the modelling work is to
Figure 4 shows a schematic representation of the laboratory set estimate the values of E and the true reaction rate constant (k1), in
( )
up used to perform gas-liquid transesterification reaction between addition to the contribution of the mass transfer term [ ]
soybean oil droplets with vapour methanol, catalysed by sodium
methoxide. This set up is an effort to model the behaviour of the and the reaction kinetics term [ ( )] respectively, towards
atomised edible grade beef tallow droplets produced in the pilot the overall gas-liquid process as represented by Equation 18.
scale spray tower reactor used by Behzadi and Farid [8]. Soybean
oil used in this work is of edible quality (cooking oil) and Firstly, the liquid phase mass transfer coefficient (kL) of soybean
purchased from a local supermarket. Analytical grade liquid oil droplets formed by the breakup of a laminar jet can be
methanol and pure sodium methoxide powder were purchased estimated using the following correlation [13, 14]:
from Sigma-Aldrich.
( )( )( ) (19)

Pressurised and heated soybean oil from T-01 flowed through a
0.34 mm diameter orifice to form a capillary jet which broke up The diffusion coefficients of methanol molecules in oil droplets
into droplets inside the electrically-heated reactor column (C-01) (DA) can be estimated using the Wilke-Chang (Equation 20) [15]
of 0.455 m length. At the same time, a 5 g/L solution of sodium and Scheibel (Equation 21) [16] correlations. An average of the
methoxide dissolved in pure methanol was pumped at 3.0 10 -4 two is taken for use in the model.
kg/s into a heating coil submerged under a hot water bath ( )
containing 50 vol% ethylene glycol (H-01), vaporised, and fed (20)
( )
into C-01 continuously. The vapour exits C-01 and was
condensed using a cooling water condenser (H-02), then (Where = 1.0 for unassociated solvents i.e. non-polar)

4
The dimensionless Henrys Law Constant of methanol vapour in
[ ( ) ] (21)
soybean oil is taken to be 0.15 at elevated temperatures (80
100oC) in this modelling work. As pure methanol vapour was
The self-diffusion coefficient of soybean oil molecules (DB) can
used in this work, the vapour phase concentration of methanol
be estimated using a correlation given by Dullien [17]:
(CAg) can be calculated using the Ideal Gas Law (Equation 24),
taking the partial pressure to be 1 atm.
(22)
(24)
The well-known oil-film experiment predicts the size of an oil
molecule to be 2.0 10-9 m [18]. Assuming that the molecules The gas-liquid interfacial area (a) is taken to be the surface area
are one molecular size apart in liquid state, thus the effective of individual soybean droplets per unit volume. Assuming
molecule diameter, dm = 4.0 10-9 m. spherical shaped drops, this is defined by the surface area per unit
volume of a sphere, and derived using the Sauter Mean diameter
The viscosity of soybean oil is given as a function of temperature of the droplets. Thus, the system boundary for consideration in
by Anand et. al. [19]: the model of Equation 18 is confined to one single soybean oil
droplet. Therefore, the liquid holdup = 1.0.
(23)
Calculated values of all important parameters which will enable
The interfacial concentration of methanol on the surface of the modelling of the gas-liquid transesterification reaction using
soybean oil droplets can be estimated using the Henrys Law Equation 18 to determine the enhancement factor (using Equation
Constant. Dimensionless Henrys Law constant of methanol 16) and the true reaction rate constant k1, are summarised in
vapour in paratherm oil was measured by Poddar and Sirkar [20] Table 3.
and presented in Table 2.
With reference to Equation 18, modelling of the soybean oil
conversion, XB (change of soybean oil concentration, CB) is not
T (oC) ( )
possible if k1 is unknown and vice versa. Therefore,
30.0 0.132 experimentally measured values of XB of each reaction run
40.0 0.144 (Table 1) was used to obtain CB, of which, together with the
49.8 0.145 calculated parameters in Table 3 were plugged into Equation 18
Table 2: Dimensionless Henrys Law Constant for methanol and iterated using Microsoft Excel to obtain k1, MH, and E values
vapour in paratherm oil [20] that satisfied Equation 18. Detailed results and discussion are
presented in the following section.

Run 103 DA 109 DB 109 d32 106 a Re We Sc Sh kL 103 CAi HA Ei


1 6.85 1.54 1.50 492 12195.1 352.7 732.7 4846.05 2592.62 1.22 230.08 440.4 13.83
2 6.85 1.54 1.50 436 13761.4 480.7 1349.3 4846.05 3270.37 1.59 230.08 440.4 13.83
3 4.52 2.40 2.34 428 14018.7 800.5 1584.3 2054.25 2060.70 1.42 223.74 452.87 14.19
4 4.52 2.40 2.34 440 13636.4 573.8 894.0 2054.25 1605.25 1.09 223.74 452.87 14.19
5 3.05 3.65 3.56 408 14705.9 819. 8 835.3 911.12 943.16 1.01 217.74 465.34 14.55
6 3.05 3.65 3.56 408 14705.9 1155.7 1669.6 911.12 1220.21 1.32 217.74 465.34 14.55
7 3.05 3.65 3.56 408 14705.9 1442.9 2504.8 911.12 1441.26 1.56 217.74 465.34 14.55
8 3.05 3.65 3.56 408 14705.9 1681.2 3340.0 911.12 1616.31 1.76 217.74 465.34 14.55
Table 3: Calculated reaction system parameters
Results and Discussion 100.00

90.00
The product samples collected from each experiment run 80.00

consisted of a single homogenous liquid phase indicating that the 70.00


Mass Fraction (%)

methanol vapour has been successfully dissolved in the oil phase. 60.00

Small orange-coloured blobs of glycerol can be observed at the 50.00

40.00
bottom of the sample containers. As of Equation 3, the formation
30.00
of glycerol indicated that the transesterification reaction took 20.00
place in the reactor. The composition of the product samples 10.00
were analysed using the Flame Ionised Detector Gas 0.00
Run 1 Run 2 Run 3 Run 4 Run 5 Run 6 Run 7 Run 8
Chromatography (GC-FID) and the results are summarised in Esters 0.81 0.74 3.98 6.63 5.33 4.44 3.36 3.63
Monoglycerides 0.65 0.73 2.02 3.82 3.77 2.69 2.92 1.67
Figure 5. Diglycerides 0.45 1.19 4.59 2.75 4.30 1.90 2.60 1.80
Triglycerides 97.10 97.34 89.41 86.81 86.03 90.98 91.12 92.90
Calculated soybean oil conversions as reported in Figure 5 are Oil Conversion 2.90 2.66 10.59 10.03 13.97 9.02 8.88 7.10

used to determine the second-order reaction rate constant (k1) for


the gas-liquid transesterification reaction. The values of k1, MH Figure 5: Composition of product samples as analysed by GC-
and E as determined by Equation 18 are summarised in Table 4. FID and calculated oil conversions.

5
Run T (oC) k1 (m3/mol.s) MH E The comparison of k1 values calculated from the present model
1 80 0.0123 0.11 1.01 (Equation 18) and the conventional second-order reaction
kinetics equation (Equation 26) is tabulated in Table 6. For the k1
2 80 0.0115 0.08 1.00
values calculated using Equation 26, the average of two replicas
3 90 2.07 1.57 1.82 corresponding to Runs 1 and 2, and Runs 3 and 4 was taken for T
4 90 0.896 1.35 1.65 = 80oC and T = 90oC respectively, while for T = 100oC, the k1
5 100 1.41 2.25 2.35 value is an average of four replicas corresponding to Runs 5 8.
6 100 0.894 1.38 1.68 The k1 values from Equation 18 are from Table 4.
7 100 1.39 1.45 1.73 T (oC) k1 (Equation 26) k1 (Equation 18)
8 100 0.960 1.07 1.45 80 0.0121 0.002 0.0279
Table 4: Reaction rate constants, MH and E determined from
experiments and Equation 18
90 0.0635 0.019 0.237
100 0.0766 0.012 1.796
The k1 values from Table 4 are used to plot ln k versus T-1 (K-1), Table 6: Comparison of k1 values calculated from Equation 18
a linear plot in accordance to the Arrhenius Law Equation and Equation 26
(Equation 25). The plot is presented as Figure 6.
From Table 6, the reaction rate constant calculated by both
( ) (25) equations increases with temperature in accordance with
Arrhenius Law. At 80oC, Equation 18 predicts a k1 value
1
approximately 2.3 times that of Equation 26. At 90oC, this
becomes 3.7 times while at 100oC, becomes 23.4 times.
0
0.00266 0.00268 0.0027 0.00272 0.00274 0.00276 0.00278 0.0028 0.00282 0.00284
At temperatures beyond the boiling point of methanol (60
65oC) usually used for liquid-liquid transesterification reactions,
-1 the reaction rate is higher and the mass transfer between reactants
becomes an important factor that contributes to the overall
Ln k

-2 process. This is illustrated by Table 6, where k1 values predicted


using Equation 18 (which takes into account the mass transfer
-3 effect), is consistently larger than the values determined by
Equation 26 (which solely considers reaction kinetics). As the
-4 reaction temperature increases, the reaction rate becomes higher
and the mass transfer effect becomes increasingly important.
-5
T-1 This is supported further by Table 7 which reports the
contribution of mass transfer (MT%) and reaction kinetics
Figure 6: Plot of ln k1 versus T-1 (K-1) (RXN%) to the overall gas-liquid transesterification reaction as
determined from Equation 18. For Runs 1 and 2 at 80oC, mass
From Figure 6, the activation energy Ea = 228 kJ/mol and Ar = transfer and reaction kinetics contributes approximately equally
1.48 1032. Using Equation 25, k1 values at 80, 90 and 100oC are to the overall process. At 90oC and higher as the reaction rate
calculated. Table 5 presents the calculated k1 values of the gas- expedites, the mass transfer contribution becomes dominant.
liquid transesterification reaction at different reaction
temperatures. Run MH E MT (%) RXN (%)
o 3 1 0.17 1.01 64.7 65.4
T ( C) k1 (m /mol.s)
2 0.13 1.01 55.5 44.5
80 0.0279
3 0.53 1.13 91.2 8.80
90 0.237
4 0.69 1.21 92.7 7.30
100 1.796
5 2.54 2.58 97.7 2.30
Table 5: Calculated k1 values
6 1.95 2.12 97.6 2.40
With reference to Table 5, the reaction rate constant increased 7 1.65 1.88 97.5 2.50
approximately 8.5 times and 7.6 times when the reaction 8 1.46 1.74 97.3 2.70
temperature increased from 80 to 90oC, and from 90 to 100oC
Table 7: Reaction parameters as predicted by Equation 18
respectively. As a rule of thumb, the reaction rate constant should
only double for every 10oC increase in temperature. However for Also referring to Table 7, MH < 2.0 with the exception of Runs 5
reactions with high activation energies, such as the gas-liquid and 6, of which the value of MH do not exceed 2.0 very
transesterification reaction (calculated to be Ea = 228 kJ/mol), it significantly. Thus, it may be concluded that the gas-liquid
may be an exception to the rule. Reactions with high activation transesterification reaction is a moderately fast reaction occurring
energies of about 280 kJ/mol and occurring at average both in the liquid film and bulk, corresponding most favourably
temperatures of between 0 400oC are known to be very to case E in Figure 2.
temperature sensitive, with the reaction rate doubling every 1.5 oC ( )
[10]. According to Equation 18, the mass transfer term [ ]
As an effort to decipher the effect of mass transfer to the gas- diminishes when the value of E is large. However, according to
liquid transesterification reaction, the conventional rate equation Table 7, the value of E is not high and it does not vary much.
for second-order reactions (Equation 26) [10], which do not take Other parameters do change much as well as shown in Table 3.
into account any mass transfer effects, is also used to calculate Therefore, the mass transfer term of Equation 18 predominates
the reaction rate constant k1, and compared to the values obtained and the value of k1 is important to the overall gas-liquid
in Table 4 which was obtained using Equation 18. transesterification reaction.
Figure 8 presents the plots of oil conversion versus time as
[ ] ( ) (26) simulated by Equation 18 to predict the completion time of the
(
gas-liquid transesterification reactions according to conditions
(Where ) outlined in Table 1. It should be noted that the x-axis is

6
magnified 10 times ( 101) with the exception of the plots for References
Runs 1 and 2 which were plotted on a 10 0 scale.
1. Knothe, G., J.H. Van Gerpen, and J. Krahl, The
100% biodiesel handbook. 2005, Champaign, Ill.: AOCS
90% Press. ix, 302 p.
80% 2. Mittelbach, M. and C. Remschmidt, Biodiesel : the
70% comprehensive handbook. 3rd ed. 2006, Austria: M.
Oil Conversion

60% Mittelbach. 332 p.


50% 3. Pahl, G., Biodiesel : growing a new energy economy.
40% 2nd ed. 2008, White River Junction, Vt.: Chelsea
30% Green Pub. Co. xxx, 368 p.
20% 4. Behzadi, S. and M.M. Farid, Review: Examining the
10% use of different feedstock for the production of
0% biodiesel. Asia-Pacific Journal of Chemical
0.0 2.0 4.0 6.0 8.0 10.0 12.0
Reaction time ( 10 s)
Engineering, 2007. 2(5): p. 480-486.
5. Zhang, Y., et al., Biodiesel production from waste
Run 1 Run 2 Run 3 Run 4
Run 5 Run 6 Run 7 Run 8
cooking oil: 1. Process design and technological
assessment. Bioresource Technology, 2003. 89(1): p.
Figure 8: Plots of oil conversion versus residence time by 1-16.
simulation 6. Noureddini, H. and D. Zhu, Kinetics of
transesterification of soybean oil. JAOCS, Journal of
The reaction completion times for Runs 1 8 (considered to be
when 99.9% oil conversion is achieved), are tabulated in Table 8. the American Oil Chemists' Society, 1997. 74(11): p.
1457-1463.
Run Reaction completion time (s) 7. Boocock, D.G.B., et al., Fast formation of high-
1 4.10 purity methyl esters from vegetable oils. JAOCS,
2 3.82 Journal of the American Oil Chemists' Society, 1998.
3 0.99 75(9): p. 1167-1172.
4 1.14 8. Behzadi, S. and M.M. Farid, Production of biodiesel
5 0.41
using a continuous gas-liquid reactor. Bioresource
6 0.39
Technology, 2009. 100(2): p. 683-689.
7 0.38
8 0.36
9. Charpentier, J.-C., Mass-Transfer Rates in Gas-
Table 8: Gas-liquid transesterification reaction completion times Liquid Absorbers and Reactors, in Advances in
Chemical Engineering. 1981, Academic Press. p. 1-
Equation 18 predicted that the gas-liquid transesterification 133.
reaction will reach completion in less than 5 seconds at 80oC. 10. Levenspiel, O., Chemical reaction engineering. 3rd
Behzadi and Farid [8, 21] obtained 97% conversion of beef
ed. 1999, New York: Wiley. xvi, 668 p.
tallow when atomised at 90oC, and 99.9% conversion when
atomised at 120oC, both in less than 7 seconds. However, it must 11. Freedman, B., R.O. Butterfield, and E.H. Pryde,
be noted that isothermal conditions was not achieved and there Transesterification kinetics of soybean oil 1. Journal
were temperature gradients within the spray reactor. The average of the American Oil Chemists' Society, 1986. 63(10):
reaction temperature when beef tallow was atomised at 90 oC was p. 1375-1380.
thus 80.5oC, and 88.0oC at 120oC atomisation. If isothermal 12. DeCoursey, W.J., Absorption with chemical reaction:
conditions were achieved, shorter reaction completion times development of a new relation for the Danckwerts
similar to those reported in Table 8 are expected. model. Chemical Engineering Science, 1974. 29(9):
Conclusions p. 1867-1872.
13. Srinivasan, V. and R.C. Aiken, Mass transfer to
The gas-liquid transesterification reaction can be described using
droplets formed by the controlled breakup of a
Equation 18. The gas-liquid transesterification reaction is an
intermediate reaction best described by Case E in Figure 2. The
cylindrical jet--physical absorption. Chemical
overall process is predominated by the mass transfer term Engineering Science, 1988. 43(12): p. 3141-3150.
( ) 14. Hoh, S.T., M.M. Farid, and J.J.J. Chen, Mass transfer
[ ] of Equation 18, but the reaction rate constant, k1 (i.e.
to droplets formed by the controlled breakup of a
reaction temperature) remains important. The gas-liquid cylindrical jet physical absorption. Chemical
transesterification reaches 99.9% conversion (completion) in less
Engineering Science, 2012. 73(0): p. 329-333.
than 5 seconds at 80oC which is comparable to the findings of
Behzadi and Farid [8]. The calculated k1 values are as reported in
15. Wilke, C.R. and P. Chang, Correlation of diffusion
Table 5 and it increases with reaction temperature in accordance coefficients in dilute solutions. AIChE Journal, 1955.
to Arrhenius Law. 1(2): p. 264-270.
16. Scheibel, E.G., Correspondence. Liquid Diffusivities.
Acknowledgements Viscosity of Gases. Industrial & Engineering
The authors would like to thank the University of Auckland for Chemistry, 1954. 46(9): p. 2007-2008.
providing technical and financial support towards this research 17. Dullien, F.A.L., Predictive equations for self-
work. The New Zealand International Doctoral Research diffusion in liquids: A different approach. AIChE
Scholarship awarded to S.T. Hoh is also gratefully Journal, 1972. 18(1): p. 62-70.
acknowledged. 18. Avison, J., The world of physics. 1984, Walton-on-
Thames: Nelson. 504 p.

7
19. Anand, K., A. Ranjan, and P.S. Mehta, Estimating 21. Behzadi, S., Production of biodiesel using a
the Viscosity of Vegetable Oil and Biodiesel Fuels. continuous gas-liquid spray reactor. 2009, Chemical
Energy & Fuels, 2009. 24(1): p. 664-672. and Materials Engineering)--University of Auckland,
20. Poddar, T.K. and K.K. Sirkar, Henry's Law Constant 2009. p. xiv, 214 leaves.
for Selected Volatile Organic Compounds in High-
Boiling Oils. Journal of Chemical & Engineering
Data, 1996. 41(6): p. 1329-1332.

You might also like