Modeling the methanolysis of triglyceride catalyzed by immobilized lipase in a continuous-flow packed-bed reactor

Applied Energy 126 (2014) 151–160

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Modeling the methanolysis of triglyceride catalyzed by immobilized lipase in a continuous-flow packed-bed reactor Dang-Thuan Tran a, Yi-Jan Lin b, Ching-Lung Chen a, Jo-Shu Chang a,c,d,⇑ a

Department of Chemical Engineering, National Cheng Kung University, Tainan 701, Taiwan Graduate Institute of Natural Products, Center of Excellence for Environmental Medicine, Kaohsiung Medical University, Kaohsiung 807, Taiwan c University Center for Bioscience and Biotechnology, National Cheng Kung University, Tainan, Taiwan d Research Center for Energy Technology and Strategy, National Cheng Kung University, Tainan, Taiwan b

h i g h l i g h t s  A Burkholderia lipase was immobilized on alkyl-grafted celite carriers.  Celite-alkyl-lipase catalyzed the methanolysis of triglyceride in packed-bed reactor.  The kinetics of the enzymatic transesterification follows Ping Pong Bi Bi mechanism.  Models were developed to discuss the mass transfer and enzyme kinetics in the PBR.

a r t i c l e

i n f o

Article history: Received 25 September 2013 Received in revised form 7 February 2014 Accepted 30 March 2014 Available online 8 May 2014 Keywords: Burkholderia lipase Immobilized enzyme Packed-bed reactor Ping Pong Bi Bi kinetics Transesterification Mathematical model

a b s t r a c t A Burkholderia lipase was immobilized on celite grafted with long alkyl groups. The immobilized lipasecatalyzed methanolysis of sunflower oil in a packed-bed reactor (PBR) follows the Ping Pong Bi Bi mechanism. The external mass transfer and enzymatic reaction that simultaneously occurred in the PBR were investigated via the mathematical models. The overall biodiesel production in the PBR was verified to work in an enzymatic reaction-limited regime. Triglyceride conversion and biodiesel yield were higher under a lower reactant feeding rate, while a larger amount of biocatalyst would be required to achieve the designated conversion rate if a higher reactant feeding rate was employed. The PBR can achieve nearly complete conversion of triglyceride at a biocatalyst bed height of 60 cm (ca. 29 g biocatalyst) and a flow rate of 0.1 ml min1, whereas the biodiesel yield was lower than 67%, probably due to the positional specificity of Burkholderia lipase and the accumulation of glycerol. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Fatty acid alkyl esters (biodiesel) produced from various sources of triglycerides, such as vegetable oils, algae [1,2] and animal fats, have received considerable attention as potential biofuels over the last decade [3,4]. Because it is renewable, biodegradable, and nontoxic, biodiesel has been used as a blending additive with fossil diesel or directive fuel in many diesel engines [3,4]. Biodiesel is conventionally produced via transesterification of triglyceride with alcohol (primarily methanol) catalyzed by various types of catalysts, including homogeneous catalysts (e.g., NaOH, KOH), heterogeneous catalysts (BaO, MgO/SiO2, SrO/SiO2, etc.), ⇑ Corresponding author at: Department of Chemical Engineering, National Cheng Kung University, Tainan 701, Taiwan. Tel.: + 886 6 2757575x62651; fax: +886 6 2357146. E-mail address: [email protected] (J.-S. Chang). http://dx.doi.org/10.1016/j.apenergy.2014.03.082 0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.

and biocatalysts (e.g., lipases) [5–9]. Among these, lipases can work well and achieve high conversion rates under mild conditions with less refined oils, which often contain a considerable amount of water and free fatty acids [5,6,10,8]. In contrast, chemical catalysts can be poisoned if the substrate source (either triglyceride or alcohol) contains an excess amount of free fatty acid or water, resulting in lower reaction rate [5,6]. It is thus beneficial to use lipase as the catalyst for the synthesis of biodiesel with less refined oils [11]. In addition, lipases can be immobilized on a variety of carrier materials to allow the reuse and recycling of the valuable enzymes, and to enable continuous biodiesel production with the aid of an appropriate bioreactor design, such as a fixed-bed reactor, for continuous biodiesel production [5,6,10,12]. To scale up the production of biodiesel, the enzymatic transesterification of triglyceride with alcohol is usually carried out in packed-bed reactor (PBR), in which the immobilized lipase is physically trapped inside the PBR column, while reactants are passed

152

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160

through the catalyst bed by pumping at a desired flow rate [13– 17]. The mass transfer resistance between the immobilized enzyme and substrate is the major factor affecting the global reaction rate of enzyme-catalyzed reaction [5,10], and this issue has been widely addressed in the literature, with, for example, research examining lactose hydrolysis by the b-galactosidase immobilized by cross-linking within the alginate and j-carrageenan matrix [18]. However, this has been rarely or unsystematically discussed for the case of the enzymatic transesterification of triglyceride and alcohol [15,16]. On the other hand, the kinetics of lipases in the catalytic transesterification of triglyceride and alcohol has recently been studied [15,19]. It was reported in such works that the alcoholysis triggered by lipase includes two substrates (alcohol and lipids), and that the alcohol may inhibit the activity of lipase during the enzymatic reactions. The Ping Pong Bi Bi mechanism is thus often used to describe the enzymatic transesterification of triglyceride with alcohol [10]. However, the validity of the related kinetic model of the effects of immobilized lipase on continuous biodiesel production in a fixed-bed reactor still needs to be verified, and this is one aim of the current work. Lipase produced from the Burkholderia sp. C20 strain isolated from food waste has been shown to successfully catalyze the alcoholysis of triglyceride with alcohol to form biodiesel [20–22]. In our previous work, the Burkholderia lipase was immobilized onto a cellulose nitrate (CN) membrane via filtration [21], or onto a celite carrier functionally modified with 3-amino-propyltriethoxysilane via covalent bonding [22]. While the immobilized lipases showed great thermostability, their specific activity was still low, thus making it impossible to achieve a high conversion rate of biodiesel in the transesterification of oil with methanol [21,22]. Recently, it was found that Burkholderia lipase has a high affinity with regard to hydrophobic surfaces, leading to the development of hybrid nanocomposite alkyl-grafted Fe3O4–SiO2 as new immobilization carrier for the lipase [12,23]. The resulting Fe3O4–SiO2alkyl-lipase had a specific activity of up to 1281 U g1 [23], which is significantly higher than that of celite-lipase (273.5 U g1) created by covalent bonding [22]. The immobilized lipase exhibits high tolerance to methanol and water, and has great potential in catalyzing the alcoholysis of various oil sources, such as olive oil [23] and microalgae oil [24,25], to produce biodiesel. However, due to the small particle size of the nanocomposite carriers, it is difficult to physically trap the immobilized lipase in the packedbed reactor for continuous biodiesel production [23–25]. In this work, a new immobilization system was developed by immobilizing the Burkholderia lipase on micro-size celite 545 grafted with the hydrophobic functional group. The resulting immobilized lipase has a suitable particle size (ca. 48.2 lm) for the use in continuous biodiesel production via a PBR. The kinetics of the celite-alkyl-lipase on catalyzing methanolysis of sunflower oil was examined in a PBR to validate the Ping Pong Bi Bi mechanism. In addition, the enzymatic transesterification in the PBR was mathematically modeled by simultaneously taking into account the substrate diffusion, external mass transfer of the substrate to the non-porous surface of the biocatalyst, and the enzymatic kinetics in order to evaluate the performance of the PBR. The mathematical model developed in this work is a useful tool to identify the required amount of biocatalyst by weight, the effective size of the PBR, as well as the flow rate of the reactant, in order to achieve the desired conversion of triglyceride and biodiesel yield. 2. Materials and methods 2.1. Chemicals n-Hexadecane (99%) was obtained from Alfa Aesar (Ward Hill, USA). [3-(Trimethoxysilyl)propyl] octadecyl dimethyl ammonium

chloride (72%), sunflower oil (99%), gum arabic (99%), methyl palmitate (99%), methyl oleate (99%), and methyl linoleate (99%) were purchased from Sigma-Aldrich (St. Louis, USA). Methanol (99.9%) was obtained from Mallinckrodt (St. Louis, USA). Bacto™ Yeast extract was obtained from Difco (Lawrence, USA). HEPES (99%) was obtained from Across (New Jersey, USA). Potassium chloride (99%) was obtained from Wako (Osaka, Japan). Ammonium sulfate (99%), magnesium (II) chloride hexahydrate (99%). Celite 545 was purchased from Showa Chemical Co., Ltd. (Tokyo, Japan). Ethanol (95%) was obtained from Taiwan Tobacco and Liquor Co. Inc. (Taipei, Taiwan).

2.2. Synthesis of alkyl-grafted celite One hundred grams of celite 545 was added into 1 l ethanol (95%) and 20 ml de-ionized water with vigorous mixing. The mixture was heated up to 40 °C. Ten milliliters of [3-(Trimethoxysilyl)propyl] octadecyl dimethyl ammonium chloride was added to the mixture with vigorous mixing, and the reaction was carried out at 40 °C and 400 rpm for 8 h. The particles were filtered and washed with water several times to completely remove unadsorbed materials, and then dried at 100 °C for 24 h.

2.3. Microorganism The lipase-producing bacterium was isolated from the food waste obtained from a food processing plant located in central Taiwan [20]. The strain was identified as Burkholderia sp. C20 by a 16S rDNA sequence comparison, with a NCBI accession number of AY845053 [14]. The strain was stored in 1 ml mixture of 0.5 ml glycerol (30%) and 0.5 ml LB (25 g l1) (Difco) medium under 80 °C as a stock source [20–22]. The stock microorganism was pre-cultured in 4 ml of LB medium at 30 °C, with a shaking speed of 200 rpm for 12 h. The pre-cultured bacterium was then transferred to a fermentor containing 2 l of optimal sterilized medium consisting of 2 g l1 yeast extract, 4.8 g l1 HEPES, 0.2 g l1 MgCl2.6H2O, 9.9 g l1 KCl, 6 g l1 (NH4)2SO4, 5.4 ml l1 sunflower oil, 5 ml l1 hexadecane [26]. Fermentative lipase production with Burkholderia sp. C20 was conducted in a 5 l fermentor at a controlled pH of 6.5, a temperature of 30 °C, an aeration rate of 1 vvm, and a stirring speed of 400 rpm during 30 h fermentation time [26]. After fermentation was complete, the fermentation broth was centrifuged at 9050g for 10 min to harvest the supernatant as crude lipase. After determining the lipase activity and protein concentration, the crude lipase was subsequently bound onto alkyl-grafted celite to prepare the immobilized lipase.

2.4. Lipase immobilization In brief, 100 g of the alkyl group grafted celite was added into a beaker containing 3 L harvested crude lipase as described in Section 2.3 under vigorous mechanical stirring at 600 rpm. The immobilization was conducted at a stable temperature of 25 °C, stirring rate of 600 rpm, pH of 6.67, and adsorption time of 24 h. The residual activity and adsorbed protein concentration on the microsize particles were regularly monitored at designated time intervals until the immobilization reached equilibrium state, which was defined as when the residual activity of lipase remained unchanged. The immobilized lipase was filtered and washed with de-ionized water several times until complete removal of the unbounded lipase. The resulting immobilized lipase (denoted as alkyl-celite-lipase) was then subjected to activity assay and used for further experiments.

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160

2.5. Experiments for kinetic analysis of the methanolysis of sunflower oil catalyzed by celite-alkyl-lipase A glass cylindrical tube (20 cm in length and 1 cm in internal diameter) packed with 4.3 g of celite-alkyl-lipase was used as the packed-bed reactor (PBR) for continuous biodiesel production (Fig. 1). The height of the biocatalyst bed was 18 cm. The reaction mixture had a total volume of 100 ml, in which the volume of triglyceride was varied between 10 and 90 ml at an interval of 20 ml, and the water content was set at 10% that of triglyceride on a weight basis, while methanol was added until the mixture reached the total volume of 100 ml. The initial concentration of triglyceride was set at 104–936 mol m3, with intervals of 208 mol m3. The volumetric flow rate of the reactant mixture was set at 0.2 ml min1, and the transesterification was carried out at 25 °C. The retention time of the reactants passing through the reactor was estimated as 60 min. Samples were regularly taken at the

Effluent

153

outlet of reactor for the analysis of fatty acid methyl esters content to estimate the initial reaction rate based on the weight basis of the catalyst used. The initial reaction rate was determined as mol of fatty acid methyl esters (FAMEs) m3 s1. The experiment was repeated three times. 2.6. Production of biodiesel in PBR Biodiesel was produced in a PBR via methanolysis of sunflower oil catalyzed by celite-alkyl-lipase, as schematically depicted in Fig. 1. A reservoir (R) containing an emulsified mixture of triglyceride (sunflower oil), methanol (molar ratio of methanol to triglyceride, 4:1), and water (10% of triglyceride weight) was mixed at 600 rpm using a magnetic stirrer (M) before the content was transferred by a peristaltic pump (P) to a cylindrical glass column (C) (H = 167 cm, I.D. = 1.5 cm) packed with 58 g celite-alkyl-lipase (ca. specific activity of 1154 U g1). The column was divided into eight sections, and each was packed with biocatalyst to a height of 20 cm. The sample collection points were connected to the column at intervals of 20 cm in the height of the biocatalyst bed in order to regularly collect sample for analysis. The flow rate of the reactants was controlled by the peristaltic pump, and samples were regularly taken to monitor the conversion of triglyceride to fatty acid methyl esters, and to see how it was affected by the height of the biocatalyst bed.

B 2.7. Analysis 2.7.1. Characterization of celite materials and celite-alkyl-lipase The BET surface area of celite materials and celite-alkyl-lipase was measured by nitrogen adsorption experiments, which were carried out at 77 K using a Micrometrics ASAP 2000 apparatus (Micromeritics Instrument Corporation, Norcross, USA). The physicochemical properties of the celite carriers and celite-alkyl-lipase are presented in Table 1.

B: Biocatalyst bed C: Column H = 167 cm I.D = 1.5 cm

2.7.2. Lipase activity assay Lipase activity, maximum adsorption capacity of protein on alkyl-grafted celite, and maximum protein binding efficiency (PBEmax) were determined following the procedures described in Tran et al. [23]. The results are presented in Table 2.

M: Magnetic stirrer P: Peristaltic pump R: Reservoir S: Sample collection point

S R

C

M P

Fig. 1. Schematic description of the packed bed reactor used for continuous production of biodiesel via methanolysis of sunflower oil catalyzed by immobilized Burkholderia lipase immobilized on alkyl group-grafted celite carrier. H, height of column; I.D., internal diameter of the column.

2.7.3. FAMEs analysis and intermediate compositions determination FAMEs were analyzed using a gas chromatograph (GC) equipped with Stabilwax column and flame ionization detector (FID). In brief, the GC was calibrated by methyl oleate, methyl linoleate, and methyl palmitate under various concentrations. During the analysis, the temperature of the injector and FID were controlled at 250 °C. The temperature of the column was raised programmatically. The temperature was initially maintained at 150 °C for 2 min, and was then raised to 250 °C at a rate of 10 °C min1. Finally, the temperature of the column was maintained at 250 °C for 5 min. In addition, the intermediate composition, which included 1,2diglyceride, 1,3-diglyceride, 2-monoglyceride, 1-monoglyceride, and free fatty acids, was analyzed with nuclear magnetic resonance (NMR) spectroscopy using the method described in our recent work [22]. 2.7.4. Kinetics analysis The kinetics of the enzymatic alcoholysis of triglyceride is known to follow the Ping Pong Bi Bi (PPBB) mechanism (Eq. (1)) [15,19]. In this study, the PPBB model was used to simulate the data of the initial rate of FAMEs formation, obtained as described in Section 2.5, via methanolysis of sunflower oil catalyzed by celite-alkyl-lipase.

154

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160

Table 1 Physicochemical properties of celite materials and celite-lipase.

a b c

Property

Celite

Alkyl-grafted celite

Celite-alkyl-lipase

Physical state Particle density (qparticle, kg m3)a Bulk density (qbulk, kg m3)a Particle size (dp, m)a Surface area (m2 g1)b Micropore volume (m3 g1)b Bed porosity (/)c External surface area ac (m2 m3)c

Solid 2650 352 48.2  106 23.2 0.00 0.87 1.62  104

Solid 2650 352 48.2  106 27.3 0.00 0.87 1.62  104

Solid 2650 352 48.2  106 55 0.00 0.87 1.62  104

Data obtained from the producer. Data obtained by analysis with Micrometrics ASAP 2000 apparatus (Micromeritics Instrument Corporation, Norcross, USA) at 77 K. q Data obtained by calculation via the following equations / ¼ 1  q bulk and ac ¼ 6ð1/Þ . dp particle

Table 2 Kinetic parameters of the Ping Pong Bi Bi mechanism estimated for celite-alkyl-lipase based on Eq. (1). Parameter

d ½TG 2

dz

U

d½TG V max ½TGi     ¼0 dz i þ ½TGi  K m;TG 1 þ K½M þ K m;M ½TG ½M

1

Vmax (mol m s ) Km,TG (mol m3) Ki,M (mol m3) Km,M (mol m3) R2 R2adjusted PBEmax (%)a Maximum adsorption capacity (mg protein g1 celite)b Specific activity of celite-alkyl-lipase (U g1)

0.49 163 11.64  104 105 0.997 0.996 87 21.64 1154

a Maximum protein binding efficiency (PBEmax) is calculated by PBEmax ¼ PPei  100%, where, Pe (mg protein) is the total amount of protein adsorbed at equilibrium state; Pi (mg protein) is the total amount of protein at initial time before immobilization. b Maximum adsorption capacity was evaluated using the adsorption isotherm of protein on alkyl-grafted celite, as described in Tran et al. [23].

V max ½TG   ¼ ½M K m;TG 1 þ K i;M þ K m;M ½TG þ ½TG ½M

ð1Þ

where vi is the initial reaction rate (mol m3 s1), Vmax is the maximum reaction rate (mol m3 s1), Km,TG and Km,M are the apparent Michaelis constants for triglyceride (TG) and methanol (M) (mol m3), and Ki,M is the inhibition constant for the methanol (mol m3); [TG] and [M] are concentrations of triglyceride and methanol (mol m3), respectively. The validation process was analyzed by using nonlinear regression tools in the Sigma plot software (version 10; Systat Software Inc., Richmond, CA, USA). The kinetic parameters were analyzed, and the results are presented in Table 2. 2.7.5. Modeling biodiesel production in a PBR using celite-alkyl-lipase It has been shown that, in an enzyme catalyzed-reaction, the global reaction rate can be presented by a simple rate equation using the apparent parameters and interfacial concentration (interfacial rate) [18,28,29]. The initial rate of the celite-alkyllipase-catalyzed methanolysis of sunflower oil, which follows the Ping Pong Bi Bi mechanism, was thus used to represent the global reaction rate.

r TG ¼ r 0TG ac ¼ v i ¼

V max ½TGi    ½M i K m;TG 1 þ K i;M þ K m;M ½TG þ ½TGi  ½M

ð3Þ

i;M

Values 3

vi

2

DAB

ð2Þ

The mathematical model for the substrate concentration profile in the liquid phase in the immobilized lipase-catalyzed reaction in an isothermal packed-bed reactor of a length z, operating under steady state, can be described as proposed in previous studies [18,28,29]. The mass balance based on the triglyceride substrate can be expressed as follows:

where DAB is the diffusivity of triglyceride to an emulsified mixture of triglyceride, methanol and water (m2 s1); U is the superficial velocity, U ¼ AQc (m s1); Q is the volumetric velocity (m3 s1); Ac is
2 cross sectional area of the column reactor, Ac ¼ p I:D: (m2); I.D. is 2 the internal diameter of column reactor (m); [TG] and [TGi] are the bulk and surface concentrations of triglyceride (mol m3), respectively. Danckwerts’ boundary conditions [30] for the methanolysis of sunflower oil catalyzed by celite-alkyl-lipase in the continuous-flow packed-bed reactor are as follows:

at z ¼ 0;

½TG ¼ ½TG0  þ

at z ¼ L;

d½TG ¼0 dz

DAB d½TG U dz

Moreover, at steady state, the molar flux caused by convective transport of triglyceride to the boundary layer created on the surface of the biocatalyst (WTG) is equal to the rate of disappearance of TG on the surface ðr0TG ; mol m2 s1), r 0TG ¼ W TG ¼ kc ð½TG  ½TGi . Therefore, the following equation is obtained

V max ½TGi    ¼ kc ac ð½TG  ½TGi Þ ½M i K m;TG 1 þ K þ ½TGi  þ K m;M ½TG ½M

ð4Þ

i;M

where R = kcac([TG]  [TGi]) is the mass transfer rate of triglyceride from the bulk liquid to a non-porous surface of celite-alkyl-lipase, at which the triglyceride is consumed by the transesterification reaction; kc is the mass transfer coefficient (m s1); ac is the external surface area to volume of the biocatalyst bed (m2 m3). The mass transfer rate R is considered to be equal to the global rate of reaction (mol m3 s1) [31]. In Eq. (4), the concentration of methanol [M] can be replaced by the difference between the initial and consumed concentration of methanol. The consumed methanol concentration in the transesterification can be estimated from the consumed concentration of triglyceride using a stoichiometric factor b. This factor would be 3 if one triglyceride molecule reacts with three methyl groups. However, since the Burkholderia lipase is 1,3-positionally specific, the triglycerides may mainly react with two methyl groups in methanol. In addition, the hydrolysis of triglyceride by water may also cause some effects. As a result, the b was reduced to 2.01 as indicated in Fig. 2. Eq. (3) can thus be re-arranged, as shown below.

V max ½TGi    ½M0 bð½TG0 ½TGÞ ½TGi  K m;TG 1 þ þ K m;M ½M0 bð½TG þ ½TGi  K i;M 0 ½TGÞ ¼ kc ac ð½TG  ½TGi Þ

ð5Þ

155

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160

where [TG0] and [M0] are the concentrations of triglyceride and methanol at the initial time (mol m3). Eqs. (3) and (5) can also be re-written in dimensionless form by defining the following variables,

(MathWorks, Massachusetts, USA). The samples were collected at eight points, as marked in Fig. 1 to correlate the experimental data with simulated results.One way to evaluate the performance of the reactor is to look at the substrate conversion, which is defined by:

z x¼ ; L

X TG ¼

S¼

½TG K m;TG

Based on this, Eqs. (3) and (5) can be transformed into the following equations: 2

1 d S dS   StðS  Si Þ ¼ 0 Pe dx2 dx

ð6Þ

Si

1 ¼ ðS  Si Þ Si Da a  bðS0  SÞ þ cdðS0 SÞ þ Si

ð7Þ

where

clet number; Pe ¼ Pe is the Pe

UL advective transport rate ¼ DAB convective transport rate

€ hler number; Da ¼ Da is the Damko ¼

V max K m;TG kc ac

reaction rate external mass transfer

St is the Stanton number; St ¼ ¼   ½M0  ; a¼ 1þ K i;M

b¼

bK m;TG ; K i;M

kc ac L kc ac kc ac ¼ ¼ U ðU=LÞ SV external mass transfer space velocity c¼

½M0  ; K m;M

d¼

bK m;TG K m;M

Danckwerts’ boundary conditions can also be expressed in dimensionless form as follows.

at x ¼ 0;

S ¼ S0 þ

at x ¼ 1;

dS ¼0 dx

1 dS Pe dx

The differential equation (6) can easily be integrated numerically by the Runge–Kutta method, and the required value of the interfacial concentration Si was calculated by solving the algebraic Eq. (7). This combination of equations was solved within x = [0, 1] using the ODE45 (Ordinary Differential Equations) solver in MATLAB 7.0

S0  S ; S0

0 6 X TG 6 1

Theoretically, the concentration of fatty acid methyl esters (FAMEs) produced via transesterification is equal to the concentration of methanol consumed. The conversion of biodiesel can be calculated based on the theoretical conversion by using the following equation:

X FAME ¼

    bðS0  SÞ 2:01 S0  S S ¼ 0:67 1  ¼ 3S0 3 S0 S0

ð9Þ

Therefore, it is possible to grasp a clear picture of the performance of a PBR by looking at the variation of the conversion of triglyceride and production of FAMEs, based on the variations of the Péclet (Pe), Damköhler (Da), and Stanton numbers (St). Prior to the determination of these dimensionless numbers, the diffusivity and external mass transfer coefficient should first be estimated, as described below. 2.7.5.1. Determination of the diffusivity of triglyceride in an emulsified mixture of sunflower oil, methanol and water (DAB). In the transesterification of triglyceride with methanol catalyzed by immobilized Burkholderia sp., the water involved in the reaction was added at 10% weight of triglyceride. Since methanol and water dissolve well with each other, they quickly became a homogeneous mixture. The diffusion of triglyceride in the homogeneous mixture of methanol and water was thus assessed to evaluate the diffusivity of triglyceride. Considering triglyceride as solute 1, while the mixture of methanol and water as a solvent 2 or Blend1, the diffusivity of solute 1 in solvent 2 is denoted as D12. Also, the molecular volume of triglyceride (V1; m3 mol1) and mixture of methanol and water (V2; m3 mol1) was estimated as shown in Table 3,

V 2 28:75 ¼ ¼ 0:03 < 1:5 958 V1 Therefore, the diffusivity of triglyceride in the methanol and water mixture was estimated using Eq. (10), as follows [32],

D12 ¼ 10  108

M1=2 Blend1 T

lBlend1 V 11=3 V 1=3 2

ð10Þ

where MBlend1 is the molecular weight of the methanol and water solution (kg mol1), T is the absolute temperature (K), and lBlend1 is the dynamic viscosity of the methanol and water mixture (kg(m s)1) (Table 3). Plugging all the values into Eq. (10) results in D12 = 7.1  1010 (m2 s1). 958 In contrast, since VV 21 ¼ 28:75 ¼ 33:3 > 1:5, the diffusivity of Blend1 in triglyceride can be estimated using Eq. (11) as follows [32].

D21 ¼ 8:5  108

Fig. 2. The correlation between the consumption of methanol and consumption of triglyceride.

ð8Þ

M 1=2 TG T 1=3 lTG V 1=3 1 V2

ð11Þ

where lTG is the dynamic viscosity of triglyceride (kg (m s)1) (Table 3). D21 = 5.1  1011 (m2 s1) was obtained according to Eq. (11). The mixture of triglyceride and Blend1 can be defined as Blend2. For concentrated solutions containing associating compounds, such as methanol and water, the correlation (Eq. (12)) proposed by Welty et al. [33] and Co et al. [34] can be used to estimate the diffusivity of triglyceride to the emulsified mixture of sunflower oil, methanol and water, as in Eq. (12), shown bellow.

156

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160

DAB lBlend2 ¼ ðD12 lBlend1 ÞxBlend1 ðD21 lTG ÞxTG DAB ¼

ðD12 lBlend1 Þ

xBlend1

ðD21 lTG Þ

or

xTG

ð12Þ

lBlend2

where DAB is the diffusivity of triglyceride in the emulsified mixture of triglyceride, methanol and water (m2 s1); x is the mole fraction of each composition; lBlend2 is the dynamic viscosity of the emulsified mixture of triglyceride, methanol, and water. From Eq. (12), DAB = 4.72  1011 (m2 s1) was obtained. This diffusivity can be used for further calculations if the model includes Eqs. (6) and (7). 2.7.5.2. Determination of external mass transfer coefficient (kc). With DAB = 4.72  1011 (m2 s1), the resulting Péclet numbers (Pe) are shown in Table 4. Since Pe in all cases is larger than 500, the following correlation can be applied for the low supercritical velocity region [35].

Sh ¼

kc d p ¼ 0:99ðReÞ1=3 ðScÞ1=3 DAB

kc ¼

0:99DAB ðReÞ1=3 ðScÞ1=3 dp

or ð13Þ qd U

where Re is the Reynolds number, Re ¼ lp ; Sh is the Sherwood k d number, Sh ¼ DcABp ; Sc is the Schmidt number, Sc ¼ DvAB ; dp is biocatalyst particle diameter, = 48.2  106 (m) (Table 1); q = qBlend2 = density of liquid fluid, = 898 (kg m3); l = lBlend2 is the dynamic viscosity of liquid fluid, = 12.33  103 (kg (m s)1); m = mBlend2 is the kinematic viscosity of liquid fluid, = 13.73  106 (m2 s1) (Table 3). 3. Results and discussions 3.1. Physical characterization of celite materials and celite-alkyl-lipase The celite 545 material is mainly composed of SiO2 (89%), Al2O3 (5.0%), Fe2O3 (1.5%), MgO (1.0%), and CaO (1.0%) with an average particle size of 48.2 lm. The celite particles and celite-alkyl-lipase are nonporous (Table 1). The BET surface areas measured on the celite, alkyl-grafted celite, and celite-alkyl-lipase are 23.2, 27.3, and 55.0 m2 g1, respectively. The increase in the surface area of alkyl-grafted celite and celite-alkyl-lipase is reasonable, because the grafted material and immobilized lipase contain an alkyl group and protein on the celite surface, thereby increasing the adsorption sites for nitrogen. 3.2. Determination of apparent kinetics parameters of the immobilized lipase As shown in Table 2, the methanolysis of sunflower oil catalyzed by celite-alkyl-lipase was well fitted with Eq. (1) (the Ping Pong Bi Bi model) with R2 = 0.997. The maximum reaction rate (Vmax) was determined as 0.49 mol m3 s1. The apparent Michaelis constants of triglyceride (Km,TG) and methanol (Km,M) to the immobilized lipase were estimated as 163 and 105 mol m3, respectively, while the inhibition constant of methanol (Ki,M) was estimated as 11.64  104 mol m3. The equation of the initial production rate of FAMEs is as follows:

vi ¼

0:49½TG   ½M 163 1 þ 11:64104 þ 105 ½TG þ ½TG ½M

The methanol inhibition constant (11.64  104 mol m3) of celite-alkyl-lipase is much higher than that of Mucor miehei lipase (3.3  104 mol m3) [19] and Candida antarctica (3.5  104 mol m3) [15]. With celite-alkyl-lipase, the Michaelis constant of triglyceride to celite-alkyl-lipase (163 mol m3) is almost 3-fold

and 2-fold less than that of lipase obtained from M. miehei (430 mol m3) and C. antarctica (250 mol m3), respectively. This indicates that the reaction rate of product formation with the immobilized Burkholderia lipase can reach maximum velocity (Vmax) more quickly. The maximum product formation rate for the methanolysis of triglyceride catalyzed by celite-alkyl-lipase is also far higher than that by M. miehei lipase (6.83  104 mol m3 s1) and C. antarctica lipase (3.27  102 mol m3 s1). This is reasonable, because the lipases from M. miehei and C. antarctica have been reported to have the optimal conversion of triglyceride to biodiesel at a molar ratio of methanol to triglyceride of 3:1 [36] and 1:1 [13,14,37], respectively, whereas the Burkholderia lipase works best at a molar ratio of methanol to triglyceride of 4:1 [23], indicating a higher methanol tolerance compared to both M. miehei lipase and C. antarctica lipase. The specific activity of celite-alkyl-lipase was determined as 1154 U g1 following the method described in our recent work [23]. This value is significantly higher than the specific activity (ca. 273.5 U g1) of celite-lipase created by using covalent bonding [22]. However, the specific activity of the celite-alkyl-lipase is lower than that of Fe3O4–SiO2-alkyl-lipase (ca. 1281 U g1) [23]. The maximum adsorption capacity of celite-alkyl-lipase (ca. 21.64 mg protein g1 alkyl-grafted celite) is also slightly lower than that of alkyl-grafted Fe3O4–SiO2 (29.45 mg protein g1), as estimated by Tran et al. [23]. These results indicate that the binding of Burkholderia lipase on carriers possessing hydrophobic surfaces is better than that obtained on supporters that have been functionally modified with chemically created covalent bonding. Moreover, the adsorption of lipase on alkyl-grafted Fe3O4–SiO2 is more efficient than on alkyl-grafted celite, as the dissociation constant of lipase immobilized on alkyl-grafted Fe3O4–SiO2 was nearly 5-fold less compared to that of lipase immobilized on alkyl-grafted celite [23]. The reason for this may be due to the smaller particle size (270–380 nm), larger BET surface area (128 m2 g1) and the porous structure of the alkyl-grafted Fe3O4–SiO2 support [23] when compared with alkyl-grafted celite, which has a particle size and BET surface area of 48.2 lm and 27.3 m2 g1, respectively, with a non-porous structure. The higher surface area and smaller pore size are beneficial to the adsorption of protein on the support. 3.3. Model simulation of PBR for continuous biodiesel production The diffusivity of triglyceride in an emulsified mixture of sunflower oil, methanol, and water was estimated at 25 °C as DAB = 4.72  1011 m2 s1, which is comparable to the molecular diffusion coefficient of sunflower oil through methanol (7.8  1010 m2 s1) estimated by Stamenkovic et al. [38] at 60 °C, but less than the diffusivity of coconut oil in methanol at 60 °C (1.1  109 m2 s1) [34]. It is also reasonable that the viscosity of triglyceride at 60 °C is lower than that at 25 °C, and thus the diffusion of triglyceride in methanol is easier at a higher temperature. The mass transfer coefficient of sunflower oil into a mixture of methanol and water was estimated as 8.22  1010 to 3.81  107 m s1 (Table 4), which is comparable to the 1.4  107 to 1.45  106 m s1 estimated by Stamenkovic et al. [39] at three typical temperatures of 10, 20, and 30 °C. These values are notably lower than the mass transfer coefficient of sunflower oil into methanol, which ranged from 1.2  104 to 5.5  104 m s1 at 60 °C [40], and 4.0  106 to 2.04  104 m s1, as estimated by Klofutar et al. [41] at 40 and 50 °C. This is reasonable, because at higher temperatures the viscosity of triglyceride decreased, thus increasing the diffusion rate of triglyceride into methanol. Table 4 shows the effect of flow rate on the estimated model parameters for continuous biodiesel production using a PBR. Based

157

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160 Table 3 Physicochemical properties of methanol, triglyceride, water, and emulsified mixture. Properties

Symbol

Methanol 1

Sunflower oil 3a

3a

Water

Blend1 3a

Viscosity (25 °C) Kinematic viscosity (25 °C)

l (kg (m s) ) m ¼ lq (m2 s1)

0.544  10 0.687b

48.98  10 53.83b

0.89  10 0.929b

Molecular weight Density (25 °C) Molecular volume Molar fraction Mass fraction

Mw (kg mol1) d (kg m3) V (m3 mol1) x y

0.032 792a 4.04  105j 0.46 0.4

0.872a 910a 9.58  104j 0.1 0.81

0.018 960a 18.75  106j 0.54 0.6

0.697  10 0.82c

Blend2 3e

0.025g 850h 28.75  106j 0.9 0.19

12.33  103f 13.73d nd 898i nd nd nd

Blend1: Mixture of methanol and water; Blend2: emulsified solution of triglyceride, methanol and water; nd: not determined; VBN: Viscosity Blending Number; x: molar fraction; y: mass fraction.



 VBNBlend1  10:975  0:8 14:534 VBNBlend1 ¼ ½yMeOH  VBNMeOH  þ ½yH2 O  VBNH2 O  VBNCH3 OH ¼ 14:534  ln½lnðmCH3 OH þ 0:8Þ þ 10:975

mBlend1 ¼ exp exp



VBNH2 O ¼ 14:534  ln½lnðmH2 O þ 0:8Þ þ 10:975    VBNBlend2  10:975  0:8 mBlend2 ¼ exp exp 14:534 VBNBlend2 ¼ ½yTG  VBNTG  þ ½yBlend1  VBNBlend1 

ð14Þ ð15Þ ð16Þ ð17Þ ð18Þ ð19Þ

VBNTG ¼ 14:534  ln½lnðmTG þ 0:8Þ þ 10:975

ð20Þ

VBNBlend1 ¼ 14:534  ln½lnðmBlend1 þ 0:8Þ þ 10:975

ð21Þ

lBlend1 ¼ mBlend1  qBlend1 lBlend2 ¼ mBlend2  qBlend2 q V CH OH þ qH2 O V H2 O qBlend1 ¼ CH3 OH 3

ð22Þ

V CH3 OHþH2 O

qBlend2 ¼ a b c d e f g h i j

Data Data Data Data Data Data Data Data Data Data

qTG V TG þ qBlend1 V Blend1 V TGþCH3 OHþH2 O

ð23Þ ð24Þ ð25Þ

obtained from producer. calculated via equation m ¼ lq. calculated via Eqs. (14)–(17). calculated via Eqs. (18)–(21). calculated via Eq. (22). calculated via Eq. (23). calculated via quation M Blend1 ¼ xCH3 OH M CH3 OH þ xH2 O MH2 O . calculated via Eq. (24). calculated via Eq. (25). calculated via equation V ¼ Mdw .

on the flow rates used, the PBR performance can be categorized into two different regimes. For the regime with Da > 1, the methanolysis of sunflower oil catalyzed by celite-alkyl-lipase is limited by external mass transfer. However, for the regime with Da < 1, the methanolysis process is limited by the transesterification reaction. The mixed regime is defined under the condition of Da = 1, at which both external mass transfer and reaction are competitive. As natural properties of celite-alkyl-lipase, the values of Vmax and Km,TG, determined as 0.49 mol m3 s1 and 163 mol m3, remained unchanged when it was used in PBR under the specified conditions. Therefore, the variation of Da is dependent on the variation of kc, which is estimated by the correlations among the Reynolds (Re), Sherwood (Sh), and Schmidt numbers (Sc), in which Re is a function of superficial velocity (U). As shown in Table 4, a higher U caused better advective transport (i.e., turbulence) between molecules, thus increasing the mass transfer rate coefficient (kc). As a consequence, the value of Da decreased. The region where the flow rate of the reactant increased from 0.1 to 1 ml min1 was controlled by a peristaltic pump. The resulting Da values were estimated as 4.73  102 to 2.19  102, indicating that the overall rate of the transesterification of sunflower oil with methanol catalyzed by celite-alkyl-lipase fell within the reactionlimited regime. To obtain Da equal to 1.00, the flow rate of the reactant needs to be decreased to 1.06  105 ml min1 to turn the overall rate of methanolysis into that seen with the mixed regime. In addition, the flow rate of the reactant should be further decreased to less than 107 ml min1 so that methanolysis occurs in the external mass transfer-limited regime. However, in practice

the flow rate of reactant cannot be set at lower than 103 ml min1, due to the limitations of the mechanical pump equipment. The data shown in Table 4 for the flow rates of less than 103 ml min1 were thus generated mathematically, based on the model equations. Therefore, in the experiments, the flow rate was set within the range of 0.1–1.0 ml min1, as these are practical parameters for PBR operation. As a result, the overall reaction for the methanolysis of sunflower oil catalyzed by celite-alkyl-lipase was characterized to fall within the reaction-limited regime. In contrast, the external mass transfer limitation regime cannot be achieved experimentally. An increase in the reactant flow rate would increase the advective transport rate rather than convective one, which is mainly caused by molecular diffusion of the reactant in the emulsified solution, and thus the value of Pe increased accordingly (Table 4). In contrast, an increase in the reactant flow rate would decrease the residence time or increase the space velocity of the reactant inside the PBR, leading to a decrease in the value of St (Table 4). The concentration of triglyceride, triglyceride conversion, and fatty acid methyl esters conversion profiles depended on the flow rate of reactant are presented in Figs. 3–5, respectively. The data show that increasing the reactant flow rate from 0.1 to 1 ml min1 resulted in a decrease of residence time (i.e., a decrease of the value of St number from 4.87  102 to 1.05  102). As a result, the profiles of the decrease of triglyceride concentration (Fig. 3), and the increase of conversion of triglyceride (Fig. 4) and FAME (Fig. 5) along with the biocatalyst bed height decreased accordingly. This is in agreement with the findings of Watanabe et al. [13] and Shimada et al. [14] with regard to the methanolysis of vegetable

158

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160

Table 4 The effect of flow rate on the parameters estimated from the model simulation of the experimental data obtained from the methanolysis of sunflower oil catalyzed by celite-alkyllipase in a packed-bed reactor. Q (m3 s1)

Q (ml min1) 8

U (m s1)

16

1.00  10 1.00  107 1.00  106 5.00  105 1.05  104 1.00  103 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

13

1.67  10 1.67  1015 1.67  1014 8.33  1013 1.75  1012 1.67  1011 1.67  109 3.33  109 5.00  109 6.67  109 8.33  109 1.00  108 1.17  108 1.33  108 1.50  108 1.67  108

9.42  10 9.42  1012 9.42  1011 4.71  109 9.89  109 9.42  108 9.42  106 1.88  105 2.83  105 3.77  105 4.71  105 5.65  105 6.59  105 7.53  105 8.47  105 9.42  105

kc (m s1)

Pe

Da

St

3.81  1010 8.22  1010 1.77  109 6.52  109 8.35  109 1.76  108 8.21  108 1.03  107 1.18  107 1.30  107 1.40  107 1.49  107 1.57  107 1.64  107 1.71  107 1.77  107

4.02  103 4.02  102 4.02  101 20.10 42.20 401.92 4.02  104 8.04  104 1.21  105 1.61  105 2.01  105 2.41  105 2.81  105 3.22  105 3.62  105 4.02  105

21.94 10.18 4.73 1.28 1.00 0.47 0.11 8.08  102 7.06  102 6.42  102 5.96  102 5.61  102 5.32  102 5.09  102 4.90  102 4.73  102

1.05  107 2.26  106 4.87  105 3.59  104 2.19  104 4.87  103 2.26  102 1.42  102 1.09  102 8.97  101 7.73  101 6.85   101 6.18  101 5.65  101 5.22  101 4.87  101

advective transport rate Pe ¼ DUL ¼ convective transport rate AB V max reaction rate Da ¼ K m;TG ¼ external mass transfer kc ac kc ac mass transfer c ac St ¼ kcUac L ¼ ðU=LÞ ¼ kSV ¼ external ; kc: External mass transfer coefficient that was determined in Section 2.7.5.2. space velocity

U ¼ AQc (m s1); Q is the volumetric velocity (m3 s1); Ac is the cross sectional area of the column reactor, Ac ¼ p reactor (m).


I:D: 2 2

(m2); I.D. is the internal diameter of the column

40

-3

Triglyceride concentration (mol m )

100

30

Biodiesel conversion (%)

80

20

10

0

60

40

20

0 0

20

40

60

80

100

120

0

20

40

60

80

100

120

Biocatalyst bed height (cm)

Biocatalyst bed height (cm) Fig. 3. Profile of triglyceride concentration as a function of the height of biocatalyst bed under different flow rate of the reactant (Q). (d), Q = 0.1 (ml min1); (s), Q = 0.2 (ml min1); (.), Q = 0.3 (ml min1); (4), Q = 0.4 (ml min1); (j), Q = 0.5 (ml min1); (h), Q = 0.6 (ml min1); (), Q = 0.7 (ml min1); (}), Q = 0.8 (ml min1); (N), Q = 0.9 (ml min1); (r), Q = 1 (ml min1); (—); simulation data.

Fig. 5. Profile of conversion of FAMEs as a function of the height of biocatalyst bed under different flow rate of the reactant (Q). (d), Q = 0.1 (ml min1); (s), Q = 0.2 (ml min1); (.), Q = 0.3 (ml min1); (4), Q = 0.4 (ml min1); (j), Q = 0.5 (ml min1); (h), Q = 0.6 (ml min1); (), Q = 0.7 (ml min1); (}), Q = 0.8 (ml min1); (N), Q = 0.9 (ml min1); (r), Q = 1 (ml min1); (—); simulation data.

Triglyceride concentration (%)

100

80

60

40

20

0 0

20

40

60

80

100

120

Biocatalyst bed height (cm) Fig. 4. Profile of conversion of triglyceride as a function of the height of biocatalyst bed under different flow rate of the reactant (Q). (d), Q = 0.1 (ml min1); (s), Q = 0.2 (ml min1); (.), Q = 0.3 (ml min1); (4), Q = 0.4 (ml min1); (j), Q = 0.5 (ml min1); (h), Q = 0.6 (ml min1); (), Q = 0.7 (ml min1); (}), Q = 0.8 (ml min1); (N), Q = 0.9 (ml min1); (r), Q = 1 (ml min1); (—); simulation data.

oil (mixture of soybean and rapeseed oils) catalyzed by immobilized C. antarctica lipase (Novozym 435) in a fixed-bed reactor. At three typical reactant flow rates of 0.1, 0.2, and 0.3 ml min1, the biocatalyst bed heights required to achieve nearly complete conversion of triglyceride should be greater than 48, 96, and 144 cm, respectively, corresponding to the required biocatalyst weights of 24, 47, 80 g, respectively (Figs. 3 and 4). Further increasing the flow rate of reactant to the PBR, the amount of biocatalyst required to achieve a high conversion of triglyceride becomes higher (Figs. 3 and 4). The conversion of FAMEs, however, only reached a maximum value of 67% (Fig. 5). This maximum FAME conversion could be due to the nature of the Burkholderia lipase used, which is known to be a 1,3-positional specific lipase [27]. Thus, at the point triglyceride was mostly converted to FAMEs and glycerol, large amounts of 1,2-diglyceride, 1,3-diglyceride, 2-monoglyceride, 1-monoglyceride, and free fatty acids also formed simultaneously as the intermediate products [27]. This state will remain for a long time, until the acyl groups migrate from

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160

the 2-position of 1,2-diglyceride and 2-monoglyceride to the 3and 1-positions of 1,3-diglyceride and 1-monoglyceride, which are preferably catalyzed by the celite-alkyl-lipase to form FAMEs [27]. The acyl migration is very slow at a low temperature, like 25 °C [27], and thus the maximum conversion of 67% was obtained. The other reason for the limited conversion of FAMEs could be the accumulation of a large amount of glycerol as the by-product of the methanolysis process. Glycerol has high viscosity, and thus disturbs the diffusion of substrates (1,2-diglyceride, 1,3-diglyceride, 2-monoglyceride, 1-monoglyceride) to lipase molecules, and prevent any further increase in FAME conversion [13,14]. Therefore, an increase of the height of the biocatalyst bed is not recommended, because this would significantly increase the accumulation of glycerol, which may also gradually increase the amount of un-reacted methanol [13,14], which then causes the deactivation of the enzyme at a greater height of the biocatalyst bed. One way to enhance FAME conversion is to connect a number of PBRs in series with simultaneous removal of glycerol in the effluent of each column, and this should be better than when using only one long packed-bed column [42,43]. By using this approach, the first column mainly converts triglyceride to FAMEs, glycerol, and intermediate products, while the latter columns could allow the acyl migration reaction to occur, and eventually convert the glycerolfree intermediate products to biodiesel. This approach has been demonstrated to successfully enhance the conversion of FAMEs during continuous PBR operations (data not shown). As shown in Figs. 3–5, the mathematical model established based on the integration of substrate diffusion, external mass transfer of the substrate, and kinetic reaction following the Ping Pong Bi Bi mechanism is able to provide good predictions with regard to the conversion of triglyceride and FAMEs in the methanolysis of sunflower oil catalyzed by celite-alkyl-lipase. The model developed in this work can be applied for the design of PBR and a series of PBRs used for the continuous conversion of biodiesel from triglyceride and methanol via enzymatic transesterification, using the immobilized lipase as biocatalyst. 4. Conclusions Mathematical models were successfully established in this work to describe the immobilized-lipase-catalyzed methanolysis of sunflower oil in a continuous PBR by considering both the external mass transfer of triglyceride to the surface of immobilized lipase and the enzymatic kinetics. The model predicts that the overall transesterification process is an enzymatic reaction that can be controlled based on the experimental conditions used. The conversion of triglyceride could reach nearly 100%, while the biodiesel yield only reached 67%, when the height of the biocatalyst bed was 60 cm (ca. 29 g biocatalyst) and the reactant flow rate was 0.1 ml min1. Acknowledgements This study was financially supported by the Research Grants (NSC99-2221-E-006-137-MY3 and NSC102-3113-P-006-016-) obtained from Taiwan’s National Science Council. The support of a Top University grant (also known as the ‘‘5-year-50-billion’’ grant), funded by the Ministry of Education, Taiwan, is also greatly appreciated. References [1] Daroch M, Geng S, Wang G. Recent advances in liquid biofuel production from algal feedstocks. Appl Energy 2013;102:1371–81. [2] Chen CY, Yeh KL, Aisyah R, Lee DJ, Chang JS. Cultivation, photobioreactor design and harvesting of microalgae for biodiesel production: a critical review. Bioresour Technol 2011;102:71–81.

159

[3] Demirbas A. Progress and recent trends in biodiesel fuels. Energy Convers Manage 2009;50:14–34. [4] Sharma YC, Singh B. Development of biodiesel: current scenario. Bioresour Technol 2009;13:1646–51. [5] Akoh CC, Chang SW, Lee GC, Shaw JF. Enzymatic approach to biodiesel production. J Agric Food Chem 2007;22:8995–9005. [6] Jothiramalingam R, Wang MK. Review of recent developments in solid acid, base, and enzyme catalysts (heterogeneous) for biodiesel production via transesterification. Ind Eng Chem Res 2009;48:6162–72. [7] Leung DYC, Wu X, Leung MKH. A review on biodiesel production using catalyzed transesterification. Appl Energy 2010;87:1083–95. [8] Talebian-Kiakalaieh A, Amin NAS, Mazaheri H. A review on novel processes of biodiesel production from waste cooking oil. Appl Energy 2013;104: 683–710. [9] Santori G, Nicola GD, Moglie M, Polonara F. A review analyzing the industrial biodiesel production practice starting from vegetable oil refining. Appl Energy 2012;92:109–32. [10] Fjerbaek L, Christensen KV, Norddahl B. A review of the current state of biodiesel production using enzymatic transesterification. Biotechnol Bioeng 2009;102:1298–315. [11] Yan Y, Li X, Wang G, Gui X, Li G, Su F, et al. Biotechnological preparation of biodiesel and its high-valued derivatives: a review. Appl Energy 2014;113:1614–31. [12] Liu CH, Huang CC, Wang YW, Lee DJ, Chang JS. Biodiesel production by enzymatic transesterification catalyzed by Burkholderia lipase immobilized on hydrophobic magnetic particles. Appl Energy 2012;100:41–6. [13] Watanabe Y, Shimada Y, Sugihara A, Noda H, Fukuda H, Tominaga Y. Continuous production of biodiesel fuel from vegetable oil using immobilized Candida antarctica lipase. J Am Oil Chem Soc 2000;77:355–60. [14] Shimada Y, Watanabe Y, Sugihara A, Tominaga Y. Enzymatic alcoholysis for biodiesel fuel production and application of the reaction to oil processing. J Mol Catal B: Enzym 2002;17:133–42. [15] Al-Zuhair S, Dowaidar A, Kamal H. Dynamic modeling of biodiesel production from simulated waste cooking oil using immobilized lipase. Biochem Eng J 2009;44:256–62. [16] Halim SFA, Kamaruddin AH, Fernando WJN. Continuous biosynthesis of biodiesel from waste cooking palm oil in a packed bed reactor: optimization using response surface methodology (RSM) and mass transfer studies. Bioresour Technol 2009;100:710–6. [17] Séverac E, Galy O, Turon F, Monsan P, Marty A. Continuous lipase-catalyzed production of esters from crude high-oleic sunflower oil. Bioresour Technol 2011;102:4954–61. [18] Mammarella EJ, Rubiolo AC. Predicting the packed-bed reactor performance with immobilized microbial lactase. Process Biochem 2006;41:1627–36. [19] Al-Zuhair S, Ling FW, Jun LS. Proposed kinetic mechanism of the production of biodiesel from palm oil using lipase. Process Biochem 2007;42:951–60. [20] Liu CH, Chen WM, Chang JS. Methods for rapid screening and isolation of bacteria producing acidic lipase: feasibility studies and novel activity assay protocols. World J Microbiol Biotechnol 2007;23:633–40. [21] Liu CH, Chang JS. Lipolytic activity of suspended and membrane immobilized lipase originating from indigenous Burkholderia sp. C20. Bioresour Technol 2008;99:1616–22. [22] Liu CH, Lin YH, Chen CY, Chang JS. Characterization of Burkholderia lipase immobilized on celite carriers. J Taiwan Inst Chem Eng 2009;40:359–63. [23] Tran DT, Chen CL, Chang JS. Immobilization of Burkholderia sp. lipase on a ferric silica nanocomposite for biodiesel production. J Biotechnol 2012;58: 112–9. [24] Tran DT, Yeh KL, Chen CL, Chang JS. Enzymatic transesterification of microalgal oil from Chlorella vulgaris ESP-31 for biodiesel synthesis using immobilized Burkholderia lipase. Bioresour Technol 2012;108:119–27. [25] Tran DT, Chen CL, Chang JS. Effect of solvents and oil content on direct transesterification of wet oil-bearing microalgal biomass of Chlorella vulgaris ESP-31 for biodiesel synthesis using immobilized lipase as the biocatalyst. Bioresour Technol 2013;135:213–21. [26] Liu CH, Huang CC, Wang YW, Chang JS. Optimizing lipase production from isolated Burkholderia sp. J Taiwan Inst Chem Eng 2012;43:511–6. [27] Tran DT, Lin YJ, Chen CL, Chang JS. Kinetics of transesterification of olive oil with methanol catalyzed by immobilized lipase derived from an isolated Burkholderia sp. strain. Bioresour Technol 2013;145:193–203. [28] Lortie R, Pelletier D. Comparison between dispersion and plug-flow models for fixed-bed enzyme reactors. AlChE J 1992;38:1477–80. [29] Lortie R. Evaluation of the performance of immobilized enzyme reactors with Michaelis–Menten kinetics. J Chem Technol Biotechnol 1994;60:189–93. [30] Danckwerts PV. Continuous flow systems: distribution of residence times. Chem Eng Sci 1953;2:1–13. [31] Radovich JM. Mass transfer effects in fermentations using immobilized whole cells. Enzyme Microb Technol 1985;7:2–10. [32] Reddy KA, Doraiswamy LK. Estimating liquid diffusivity. Poona, India: National Chemical Laboratory; 1967. [33] Welty JR, Wicks CE, Wilson RE, Rorrer G. Fundamentals of momentum, heat and mass transfer. 4th ed. New York: John Wiley & Sons; 2001. [34] Co CET, Tan MC, Diamante JAR, Yan LRC, Tan RR, Razon LF. Internal masstransfer limitations on the transesterification of coconut oil using an anionic ion exchange resin in a packed bed reactor. Catal Today 2011;174:54–8. [35] Prˇenosil JE, Kut ÖM, Dunn IJ, Heinzle E. Biocatalysis, 2. Immobilized biocatalysts. Ullmann’s Encyclopedia of Industrial Chemistry; 2009.

160

D.-T. Tran et al. / Applied Energy 126 (2014) 151–160

[36] Soumanou MM, Bornscheuer UT. Improvement in lipase-catalyzed synthesis of fatty acid methyl esters from sunflower oil. Enzyme Microb Technol 2003;33:97–103. [37] Shimada Y, Watanabe Y, Samukawa T, Sugihara A, Noda H, Fukuda H, et al. Conversion of vegetable oil to biodiesel using immobilized Candida antarctica lipase. J Am Oil Chem Soc 1999;76:789–93. [38] Stamenkovic OS, Veljkovic VB, Todorovic ZB, Lazic ML, Bankovic-Ilic IB, Skala DU. Modeling the kinetics of calcium hydroxide catalyzed methanolysis of sunflower oil. Bioresour Technol 2010;101:4423–30. [39] Stamenkovic OS, Todorovic ZB, Lazic ML, Veljkovic VB, Skala DU. Kinetics of sunflower oil methanolysis at low temperatures. Bioresour Technol 2008;99: 1131–40. [40] Klofutar B, Golob J, Likozar B, Klofutar C, Zˇagar E, Poljanšek I. The transesterification of rapeseed and waste sunflower oils: mass-transfer and

kinetics in a laboratory batch reactor and in an industrial-scale reactor/ separator setup. Bioresour Technol 2010;101:3333–44. [41] Frascari D, Zuccaro M, Paglianti A, Pinelli D. Optimization of mechanical agitation and evaluation of the mass-transfer resistance in the oil transesterification reaction for biodiesel production. Ind Eng Chem Res 2009;48:7540–9. [42] Hama S, Tamalampudi S, Yoshida A, Tamadani N, Kuratani N, Noda H, et al. Enzymatic packed-bed reactor integrated with glycerol-separating system for solvent-free production of biodiesel fuel. Biochem Eng J 2011;55:66–71. [43] Hama S, Tamalampudi S, Yoshida A, Tamadani N, Kuratani N, Noda H, et al. Process engineering and optimization of glycerol separation in a packed-bed reactor for enzymatic biodiesel production. Bioresour Technol 2011;102: 10419–24.