Integrated reactive absorption process for synthesis of fatty esters

Bioresource Technology 102 (2011) 490–498

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Integrated reactive absorption process for synthesis of fatty esters Anton Alexandru Kiss a,⇑, Costin Sorin Bildea b a b

AkzoNobel Research, Development & Innovation, Process Technology ECG, Velperweg 76, 6824 BM, Arnhem, The Netherlands University ‘‘Politehnica” of Bucharest, Centre for Technology Transfer in Process Industries, Polizu 1-7, 011061 Bucharest, Romania

a r t i c l e

i n f o

Article history: Received 6 May 2010 Received in revised form 13 August 2010 Accepted 22 August 2010 Available online 25 August 2010 Keywords: Reactive absorption Plantwide control Energy integration Green catalyst Biofuel

a b s t r a c t Reactive separations using green catalysts offer great opportunities for manufacturing fatty esters, involved in specialty chemicals and biodiesel production. Integrating reaction and separation into one unit provides key benefits such as: simplified operation, no waste, reduced capital investment and low operating costs. This work presents a novel heat-integrated reactive absorption process that eliminates all conventional catalyst related operations, efficiently uses the raw materials and equipment, and considerably reduces the energy requirements for biodiesel production – 85% lower as compared to the base case. Rigorous simulations based on experimental results were carried out using Aspen Plus and Dynamics. Despite the high degree of integration, the process is well controllable using an efficient control structure proposed in this work. The main results are provided for a plant producing 10 ktpy fatty acid methyl esters from methanol and waste vegetable oil with high free fatty acids content, using sulfated zirconia as solid acid catalyst. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction 1.1. Background Fatty esters are important chemicals used mainly in cosmetics, pharmaceuticals, cleaning products and in the food industry. However, in the last decade the main interest has shifted to the large scale production of biodiesel – hence the stronger market drive for more innovative and efficient processes. Biodiesel is a biodegradable and renewable alternative fuel with properties similar to petroleum diesel (Bowman et al., 2006; Balat and Balat, 2008). Typically, it is produced from green sources such as vegetable oils, animal fat or even waste cooking-oils from the food industry (Encinar et al., 2005; Kulkarni and Dalai, 2006). Nowadays, employing waste and non-edible raw materials is mandatory to comply with the economical, ecological and also ethical requirements for biofuels. Basically, the ‘food versus fuel’ competition can be easily avoided when the raw materials are waste vegetable oils, non-food crops such as Jatropha (Kumar and Sharma, 2005) and Mahua (Puhan et al., 2005) or castor oil (Canoira et al., 2010). However, waste raw materials can contain a substantial amount of free fatty acids (FFA), up to 100%. Consequently, an efficient continuous biodiesel process is required that uses a solid catalyst in order to suppress the costly downstream processing steps and waste treatment. ⇑ Corresponding author. Tel.: +31 26 366 1714; fax: +31 26 366 5871. E-mail addresses: [email protected], [email protected] (A.A. Kiss). 0960-8524/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2010.08.066

Unlike petroleum diesel that is mixture of hydrocarbons, biodiesel consists of fatty acid methyl esters (FAME). Fatty esters are currently produced by acid/base catalyzed trans-esterification, followed by several neutralization and product purification steps (Hanna et al., 2005; Meher et al., 2006). Nevertheless, all conventional methods suffer from problems related to the use of liquid catalysts, leading to severe economical and environmental penalties. Obviously, this has to change considering the impact of the large scale production. At present, the most common manufacturing technologies employ homogeneous catalysts, in batch or continuous processes where both reaction and separation steps can create bottlenecks. The literature overview reveals several processes currently in use at pilot or industrial scale: Batch processes allow better flexibility with respect to composition of the feedstock. The trans-esterification is performed using an acid or base catalyst, but the equipment productivity is low and operating costs are high (Hanna et al., 2005; Lotero et al., 2005). Continuous processes combine the esterification and trans-esterification steps, allowing higher productivity. Nonetheless, most of these processes are still plagued by the disadvantages of using homogeneous catalysts although solid catalysts emerged in the last decade (Dale, 2003; Kiss et al., 2006a). Several reactive-distillation processes were reported (He et al., 2006; Kiss et al., 2006b), along with the ESTERFIP-H process developed by the French Institute of Petroleum. The latter uses a solid catalyst based on Zn and Al oxides, requiring high temperature (210–250 °C) and pressure (30–50 bar).

A.A. Kiss, C.S. Bildea / Bioresource Technology 102 (2011) 490–498

Supercritical processes were developed to solve the problem of oil-alcohol miscibility that hinders the kinetics of trans-esterification, as well as to take advantage of not using a catalyst at all. Still, the operating conditions are quite severe (T > 240 °C, p > 80 bar) and therefore require special equipment (He et al., 2007). Enzymatic processes have low energy requirements, as the reaction is carried out at mild conditions – temperatures of 50–55 °C and ambient pressure. However, due to the lower yields and long reaction times the enzymatic processes cannot compete yet with other industrial processes (van Gerpen, 2005; Demirbas and Balat, 2006; Demirbas, 2008). Two-step processes are much simpler as the glycerides are hydrolyzed first to fatty acids that are then esterified in a second step to fatty esters (Kusdiana and Saka, 2004; Minami and Saka, 2006). These processes are now very attractive and gain market share due to obvious advantages: high purity glycerol is obtained as by-product of the hydrolysis step, and the esterification step can be performed using solid catalysts in an integrated reactiveseparation setup (Kiss et al., 2008; Kiss, 2009). The use of solid catalysts avoids the neutralization and washing steps, leading to a simpler and more efficient process. Besides, when the raw materials consist mainly of FFA, only the esterification step is required. The recent literature is quite abundant in studies on integrated processes, such as reactive distillation (Kiss et al., 2006b; Kiss et al., 2008) and reactive absorption (Noeres et al., 2003; Ghaemi et al., 2009). However, that vast majority of the reported studies on reactive separation processes for biodiesel production are solely based on conventional reactive distillation (He et al., 2006; Kiss et al., 2006b; Kiss et al., 2008), or alternatives such as entrainer-based reactive distillation (RD) and dual RD (Dimian et al., 2009). This work proposes a novel energy-efficient reactive absorption (RA) process for biodiesel production, which is very well controllable in spite of the high degree of integration. The integration of reaction and separation into one unit corroborated with the use of a heterogeneous catalyst offers major advantages such as: reduced capital investment, low operating costs, simplified downstream processing steps as well as no catalyst-related waste streams and no soap formation. The main results are given for a plant producing 10 ktpy biodiesel by esterification of methanol with free fatty acids, using sulfated zirconia as solid acid catalyst. 1.2. Problem statement Biodiesel is a very appealing but still expensive alternative fuel. The common problem of all conventional processes is the use of liquid catalysts that require neutralization and costly separation steps that generates salt waste streams (van Gerpen, 2005; Meher et al., 2006; Narasimharao et al., 2007). To solve these problems, in this work we make use of solid acids applied in an esterification process based on catalytic reactive separation. Such an integrated process is able to shift the chemical equilibrium to completion and preserve the catalytic activity of the solid acid by continuously removing the products (Kiss et al., 2006b). Moreover, compared to conventional processes the investment and operating costs are much reduced. However, the market pressure demands further decrease of the operating costs by replacing the raw materials with inexpensive waste oils (high FFA content), and reducing the energy requirements per ton of product (Janulis, 2004; Kiss, 2009). Several reactive separation processes based on fatty acids esterification were reported in the recent years, aiming at high performance and productivity, along with low energy requirements:  Reactive distillation: 191.2 kW h/ton biodiesel (Kiss et al., 2008).  Dual reactive distillation: 166.8 kW h/ton biodiesel (Dimian et al., 2009).

491

 Reactive absorption: 138.4 kW h/ton biodiesel (Kiss, 2009).  Heat-integrated reactive distillation: 108.8 kW h/ton biodiesel (Kiss, 2008). This work adds heat-integration to the reactive absorption process, thus allowing a major reduction of the energy requirements: from 138.4 down to 21.6 kW h/ton biodiesel, which is equivalent to only 15% of the reference RA process previously reported. Considering the high degree of integration, the controllability of the process is also carefully investigated. 2. Methods In this work, we use rigorous computer simulations to evaluate the integrated process for synthesis of fatty esters by reactive absorption. Starting from a base case design described in previous work, we apply pinch analysis in order to identify the possibilities of heat recovery by internal process/process exchange, as well as the optimal selection and placement of utilities. Then, steady state process simulation (Dimian, 2003) is employed to obtain a description of material and energy streams in the process with the scope of evaluating the plant and understanding its steady state behaviour. Finally, dynamic simulations are used to evaluate controllability of the plant and to demonstrate the effectiveness of a proposed plantwide control strategy. 2.1. Process integration The conceptual design of the process is based on a reactive absorption column (RAC) that integrates the reaction and separation steps into one operating unit. The chemical equilibrium is shifted towards products formation by continuous removal of the reaction products, instead of using an excess of a reactant – typically the alcohol. An additional flash vessel and a decanter are used to guarantee the high purity of both products. Since methanol and water are much more volatile than the fatty ester or fatty acid, these will separate without problems in the top of the column. Fig. 1a presents the flowsheet of this process based on conventional reactive absorption, as reported by Kiss (2009). The column has 15 theoretical stages with the reactive zone located between stages 3–12, the solid catalyst loading being 6.5 kg per stage. The fatty acid is pre-heated then fed as hot liquid in the top of the reactive column while a stoichiometric amount of alcohol is injected as vapor into the bottom of the column, thus creating a counter-current flow regime over the reactive zone. Water by-product is removed as top vapor, then condensed and separated in a decanter from which the fatty acids are recycled back to the column while water by-product is recovered at high purity. The fatty esters are delivered as high purity bottom product of the reactive column. The hot product is flashed first to remove the remaining methanol, and then it is cooled down and stored. The reference flowsheet presented in Fig. 1a is relatively simple, with just a few operating units, two cold streams that need to be pre-heated (fatty acid and alcohol) and two hot streams that have to be cooled down (top water and bottom fatty esters). Therefore the heat-integration was performed by applying previously reported heuristic rules (Hamed et al., 1996; Chen and Yu, 2003). Consequently, two feed-effluent heat exchangers (FEHE) should replace partially or totally each of the two heat exchangers HEX1 and HEX2. Fig. 1b illustrates the improved process including heat-integration around the reactive absorption column. The top vapor stream is used to pre-heat the fatty acid feed stream (FEHE1). An additional heat exchanger (HEX1) is required to finally heat the fatty acids to the desired reaction temperature. When changes of the production rate are desired, the duty of this heat exchanger

492

A.A. Kiss, C.S. Bildea / Bioresource Technology 102 (2011) 490–498

a

Fig. 2 shows the composite curves and the yearly cost as function of the minimum delta T. This confirms the heuristics findings, as there is just a minimal need for an external hot utility to preheat the fatty acid feed stream. 2.2. Process simulation

b

Process simulations embedding experimental results were performed using Aspen Plus (AspenTech, 2009). The RA column was simulated using the rigorous RADFRAC unit with RateSep (ratebased) model enabled, and explicitly considering three phase balances. The physical properties required for the simulation and the binary interaction parameters for the methanol–water and acid-ester pairs were available in the Aspen Plus database of pure components, while the other interaction parameters were estimated using the UNIFAC – Dortmund modified group contribution method (AspenTech, 2009). Using other state-of-the art estimation methods, such as UNIFAC and UNIFAC – Lyngby modified, leads to similar results. As phase splitting may occur (Kiss, 2009), the RA column is modeled using VLLE data. Phase splitting must be accounted for as the free water phase can deactivate the solid acid catalyst. Nevertheless, as revealed latter by the column composition profiles, the liquid molar fraction of water does not exceed 0.1 on any given stage. Therefore, the catalyst deactivation does not occur here, under the designed process conditions. The fatty components were conveniently lumped into one fatty acid and its fatty ester – according to the reaction: R  COOH þ CH3 OH $ R  COO  CH3 þ H2 O. Dodecanoic (lauric) acid/ester was selected as lumped component due to the availability of experimental results, kinetics and VLLE parameters for this

Fig. 1. Synthesis of fatty esters by reactive absorption base case flowsheet (a) and heat-integrated process (b).

can be changed to ensure the required temperature of the columninlet acid stream (F-ACID). The hot liquid product of the FLASH, a mixture of fatty esters, is used to pre-heat and vaporize the alcohol feed stream. If changes of the production are expected, the nominal design should include a bypass of the hot stream (dashed line in Fig. 1b) which can be used for control purposes.

Fig. 2. Composite curve (a) and yearly cost versus minimum delta T (b).

493

A.A. Kiss, C.S. Bildea / Bioresource Technology 102 (2011) 490–498

system (Kiss et al., 2006a; Dimian et al., 2009). Moreover, as lauric acid is among the lightest fatty acids, this represents the worst case scenario in terms of separation from alcohol. The assumption of lumping components is very reasonable since fatty acids and their corresponding fatty esters have similar properties. This approach was already reported to be successfully used to simulate other fatty esters production processes (Kiss et al., 2006b; Kiss et al., 2008; Dimian et al., 2009; Kiss, 2009). In this work, sulfated zirconia is considered as a solid acid catalyst, as kinetic data for the esterification with methanol is available from previous work (Kiss et al., 2008; Dimian et al., 2009; Kiss, 2009). The esterification reaction is reversible, hence the reaction rate accounts for both direct and reverse reactions:

r ¼ ðk1 W cat Þ C Acid C Alcohol  ðk2 W cat Þ C Ester C Water

Table 2 Design parameters for simulating the reactive absorption column. Parameter

Value

Units

Total number of theoretical stages Number of reactive stages Column diameter HETP Valid phases Volume liquid holdup per stage Mass catalyst per stage Catalyst bulk density Fatty acid conversion Fatty acid feed (liquid, at 160 °C) Methanol feed (vapor, at 65 °C) Production of biodiesel (FAME) RA column productivity

15 10 (from 3 to 12) 0.4 0.6 VLL 18 6.5 1050 >99.99 1167 188 1250 19.2

– – m m – L kg kg/m3 % kg/h kg/h kg/h kg FAME/kg cat/h

ð1Þ

where k1 and k2 are the kinetic constants for the direct (esterification) and reverse (hydrolysis) reactions, Wcat is the weight amount of catalyst, and CComponent is the molar concentration of the components present in the system. Note that by changing the weight amount of catalyst used (Wcat) the reaction rate can be similar for different catalysts. As water is continuously removed from the system, the reverse hydrolysis reaction is extremely slow hence the second term of the reaction rate can be neglected. Therefore, a simple kinetic expression can be used:

r ¼ ðk1 W cat Þ C Acid C Alcohol ¼ k C Acid C Alcohol

ð2Þ

k ¼ A exp ½Ea =ðRTÞ

ð3Þ

where CAcid and CAlcohol are mass concentrations, A is the Arrhenius factor (A = 250 kmol m3 kg2 s1) and Ea is the activation energy (Ea = 55 kJ/mol). 3. Results and discussion 3.1. Heat integrated design The integrated reactive absorption process was designed according to previously reported process synthesis methods for reactive separations (Noeres et al., 2003). The next simulation results are given for a plant producing 10 ktpy (1250 kg/h) fatty acid

methyl esters (FAME) from methanol and waste vegetable oil with highest free fatty acids (FFA) content, using solid acids as green catalysts. Table 1 presents the complete mass balance of the process, while Table 2 lists the main design parameters, such as column size, catalyst loading, and feed condition. High conversion of the reactants is achieved, with the productivity of the RA unit exceeding 19 kg fatty ester/kg catalyst/h. The purity specification is higher than 99.9%wt for the final biodiesel product (FAME stream). Note that the total amount of the optional recycle streams (REC-BTM) is not significant representing less than 0.9% of the biodiesel production rate. The liquid and vapor composition profiles along the reactive absorption column are plotted in Fig. 3a and b, respectively. The concentration of fatty ester and methanol increases from the top to bottom, while the concentration of the fatty acid and water increases from the bottom to the top of the column. Therefore, the top product of the reactive separation column is water vapors with small amounts of fatty acids, while the bottom liquid stream is the fatty esters product (biodiesel) with a limited amount of methanol. In particular cases, the concentration of methanol in the liquid phase could be further increased by working at higher pressure. Fig. 3c plots the temperature and reaction rate profiles in the reactive column. The RA column is operated in the temperature range of 135–165 °C, at ambient pressure. The reaction rate exhibits a maximum in the middle, where the reactive zone is placed.

Table 1 Mass balance of a 10 ktpy FAME process based on integrated reactive-absorption.

Temperature/(°C) Pressure/(bar) Vapor fraction Mole Flow/(kmol/h) Mass flow/(kg/h) Mass flow/(kg/h) Methanol Acid Water Fame Mass fraction Methanol Acid Water Fame Mole fraction Methanol Acid Water FAME

F-ACID

F-ALCO

BTM

REC-BTM

REC-TOP

TOP

WATER

FAME

160 1.05 0 5.824 1166.7

65.4 1.05 1 5.876 188.3

136.2 1.03 0 6.125 1261.3

146.2 1.216 1 0.252 11.3

51.8 1 0 0.059 9.369

162.1 1 1 5.886 114.4

51.8 1 0 5.828 105.06

30 0.203 0 5.873 1250

0 116.74 0 0

188.3 0 0 0

9.125 Trace Trace 1252.2

7.544 Trace Trace 3.764

0.002 9.218 0.24 0.846

0.103 9.233 105.2 0.846

0.101 0.016 104.93 Trace

1.581 Trace Trace 1248.4

0 1 0 0

1 0 0 0

0.007 Trace Trace 0.993

0.667 Trace 10 ppb 0.333

172 ppm 0.894 0.023 0.082

894 ppm 0.08 0.912 0.007

965 ppm 148 ppm 0.999 513 ppb

0.001 Trace Trace 0.999

0 1 0 0

1 0 0 0

0.046 Trace Trace 0.954

0.931 Trace 26 ppb 0.069

873 ppm 0.726 0.211 0.062

546 ppm 0.008 0.992 670 ppm

0.001 13 ppm 0.999 43 ppb

0.008 Trace Trace 0.992

A.A. Kiss, C.S. Bildea / Bioresource Technology 102 (2011) 490–498

a

Molar fraction (liq) 0

0

0.2

0.6

0.8

Water

3 Stage

0.4

a

100%

Purity / [wt%]

494

90%

1

Acid

6

Water

FAME

80% 70%

9 12

Methanol

60% 0.8

Ester

0.9

1

1.1

1.2

Molar ratio alcohol:acid / [-] 0.2

0.4

0.6

0.8

1

Molar fraction (liq) Molar fraction (vap) 0

0

0.4

0.6

0.8

Acid

3 Stage

0.2

1

Water

6

80

FEHE2

0.6

Tacid = 180 ºC

40

0.4 160 ºC

20

0.2

140 ºC

HEX1

0 0.8

0 0.9

1

1.1

1.2

Molar ratio alcohol:acid / [-]

c Ester

0

0.2

Methanol

0.4

0.6

0.8

1

Molar fraction (vap) Temperature / °C 130 0

140

150

160

170

3

8

Product rate / kmol/h]

12

c

1 0.8

140 ºC

60

9

15

Split 160 ºC

FEHE1 180 ºC

0.5

Recycle Bottom

140 ºC

0.4

7

Water 160 ºC

0.3

180 ºC

0.2

6 FAME

180 ºC

5

0.1

160 ºC

Recycle Top

140 ºC

4 0.8

0 0.9

1

1.1

Recycle rate / [kmol/h]

b

100

b

Split fraction

0

Heat duty / [kW]

15

1.2

Stage

Molar ratio alcohol:acid / [-] 6 9

Fig. 4. Purity of top (water) and bottom (FAME) products (a), duty of heat exchangers (b), product and recycle flow rates (c) versus molar reactants ratio (alcohol:acid).

Reaction rate

12 Temperature

15

0

0.5 1 1.5 2 2.5 Ester formation / kmol/hr

3

Fig. 3. Composition, temperature and reaction rate profiles along the column.

3.2. Sensitivity analysis In spite of the recent progress in understanding the feasibility, design and control of reactive separations, the conceptual design of RA may lead to several different process configurations and operating parameters. In this work, sensitivity analysis was used to evaluate the range of the operating parameters: reactants ratio, feed streams temperature, decanting temperature, flashing pressure, and recycle rates. The most significant results of the sensitivity analysis are presented in Fig. 4, for a constant acid flow rate of

1166.7 kg/h (5.824 kmol/h) and variable alcohol flow rate. The optimal molar ratio of the reactants (alcohol:acid) is very close to the stoichiometric value of one, as illustrated by Fig. 4a. If this ratio is higher than one (i.e. excess of alcohol) then the fatty acid is completely converted to fatty ester, but the excess of alcohol becomes a significant impurity in the top stream and thereafter in the water by-product. On the contrary, when the ratio is less than one (i.e. excess of fatty acids) then the purity of water by-product remains high, but the conversion of fatty acid is incomplete – hence the bottom product contains unreacted fatty acid that cannot be easily removed from the final product by simple flashing. Since the separation of fatty acids from fatty esters is more difficult than the separation of fatty acids from water, this situation should be avoided. In practice, using a very small excess of methanol (up to 1%) or an efficient control structure that can ensure the stoichiometric ratio of reactants is sufficient for the complete conversion of the fatty acids and prevention of the difficult separation previously mentioned. Note that integrated reactive

A.A. Kiss, C.S. Bildea / Bioresource Technology 102 (2011) 490–498 Table 3 Comparison between integrated reactive-absorption vs reactive-distillation processes (at a production rate of 1250 kg/h or 10 ktpy fatty esters). Equipment/parameter/units

RD

HIRD

RA

HIRA

Reactive column – reboiler duty (heater), kW HEX-1 heat duty (fatty acid heater), kW HEX-2 heat duty (methanol heater), kW Reactive column – condenser duty (cooler), kW HEX-3 water cooler/decanter, kW COOLER heat duty (biodiesel cooler), kW FLASH heat duty (methanol recovery), kW Compressor power (electricity), kW Reactive column, number of reactive stages Feed stage number, for acid/alcohol streams Reactive column diameter, m Reflux ratio (mass ratio R/D), kg/kg Boil-up ratio (mass ratio V/B), kg/kg Productivity, kg ester/kg catalyst/h Energy requirements per ton biodiesel, kWh/ton FAME Steam consumption, kg steam/ton FAME

136 95 8 72

136 0 0 72

n/a 108 65 n/a

n/a 27 0 n/a

6 141 0 0.6 10 3/10 0.4 0.10 0.12 20.4 191.2

6 38 0 0.6 10 3/10 0.4 0.10 0.12 20.4 108.8

77 78 0 0.6 10 1/15 0.4 n/a n/a 19.2 138.4

0 14 0 0.6 10 1/15 0.4 n/a n/a 19.2 21.6

295

168

214

34

separations can be easily controlled by PID controllers in a multiloop framework (Dimian et al., 2009; van Diggelen et al., 2010) or by using more advanced techniques such as model predictive control (Nagy et al., 2007). Fig. 4b shows the duty of heat exchangers versus the molar reactants ratio (alcohol:acid), for different temperatures of the acid feed. Vaporization of an increasing flow rate of alcohol requires more energy to be exchanged in FEHE2, which is realized by increasing the amount of hot liquid that passes through the heat exchanger. This can be accomplished if – in the base case design – only a certain fraction of the hot stream (80% in our base case design, variable ‘‘Split” – dashed line in Fig. 4b) is used for alcohol vaporization. The fraction of hot liquid that is necessary for alcohol vaporization increases to almost 100% when the alcohol flow rate

495

is increased to 120% of the base case value. The dependence of FEHE2 heat duty versus the alcohol:acid ratio is insensitive with respect to the temperature of the acid feed. Increasing the alcohol flow rate leads to a larger flow rate at the top outlet of the column. More energy is exchanged in FEHE1 with the result of lower duty for the heater HEX1. In general, higher acid temperatures require larger heat duties in both FEHE1 and HEX1. Fig. 4c illustrates the flow rates of products and recycle streams versus the molar reactants ratio (alcohol:acid), at different temperatures of the fatty acid feed stream. The flow rate of the FAME stream is unchanged when the reactants ratio changes. However, below the stoichiometric ratio of one, this stream contains large quantities of unreacted acid. The flow rate of the WATER product stream increases with the alcohol flow rate. Above the stoichiometric ratio of one, this stream contains water and the excess of methanol. Higher temperature of the acid stream leads to increasing the top recycle – because more acid is vaporized – and reduces the bottom recycle. Table 3 shows a head-to-head comparison of the novel heatintegrated RA process proposed in this study against the previously reported reference RD and RA processes (Kiss et al., 2008; Kiss, 2008, 2009). The heating and cooling requirements are figures that ultimately translate into equipment size and cost. Remarkably, the energy demand is less than 22 kW h/ton biodiesel – or 34 kg steam per ton of biodiesel produced. Also, compared to the reference reactive absorption base case (Kiss, 2009), the heating and cooling requirements are significantly reduced by –85% and –90%, respectively. It should be noted that other heat-integration alternatives are possible. For example, the top outlet stream could be used to vaporize the methanol feed in FEHE1, while the liquid product from the FLASH could be used to pre-heat the acid feed FEHE2. This alternative has slightly higher energy requirements compared to the flowsheet presented in Fig. 1b (29 kW instead of 27 kW), while the duty of the FAME cooler decreases to 3.3 kW. However, an additional heat exchanger (13 kW) is necessary to cool the top

Fig. 5. Plantwide control structure: concentration of acid in bottom stream is controlled by manipulating the alcohol flow rate. The temperature of the column-inlet acid stream is manipulated by a methanol concentration controller.

496

A.A. Kiss, C.S. Bildea / Bioresource Technology 102 (2011) 490–498

stream to the decanting temperature (50 °C), which is a drawback of this alternative. 3.3. Plantwide control Heat-integrated reactive absorption offers indeed significant advantages such as minimal capital investment and operating costs, as well as no catalyst-related waste streams and no soap formation. However, the controllability of the process is just as important as the capital and operating costs savings. In processes based on reactive distillation or absorption, feeding the reactants according to their stoichiometric ratio is essential to achieve high products purity (Dimian et al., 2009; Bildea and Kiss, in press). Thus, the fatty acid is completely converted to fatty esters when there is an excess of methanol, but the excess of methanol becomes an impurity in the top stream and thereafter in the water by-product. On the contrary, when there is an excess of fatty acids, the purity of water by-product remains high, but the conversion of fatty acids is incomplete. In the later case the bottom product contains unreacted fatty acids that can not be easily removed from the final product by simple flashing. Since the separation of fatty acids from fatty esters is more difficult than the separation of fatty acids from water, this situation should be avoided. It is important to remark that this constraint must be fulfilled not only during the normal operation, but also during the transitory regimes arising due to planned production rate changes or unexpected disturbances. However, the integrated biodiesel processes based on reactive absorption have fewer degrees of freedom compared to reactive distillation. This makes it challenging to correctly set the reactants feed ratio and consequently avoiding impurities in the products. In the following, we will prove that, in spite of the high degree of integration this reactive absorption process is very well controllable. A key result of this study is an efficient control structure that can ensure the stoichiometric ratio of reactants and fulfills the excess of methanol operating constraint, which is sufficient for the total conversion of the fatty acids and for prevention of the difficult separations (e.g. fatty acid – fatty ester). Fig. 5 presents the proposed plantwide control structure. The production rate is set by the flow rate of the acid stream (controller FC1). After mixing with the recycle and passing the feed-effluent heat exchanger FEHE1, the temperature of the acid fed to the column is controlled (by TC1) by manipulating the duty of the heat exchanger HEX1. Compared to Fig. 1b, more details concerning the practical implementation of the FEHE2 evaporator are given. The flow rate of alcohol evaporated and fed to the column is set by changing the split of the hot stream from the flash-outlet stream (controller FC2). The temperature of the alcohol stream

entering the column is therefore the boiling temperature. Because the hot FAME stream is not available during the startup of the plant, the evaporator should include the option of using another heat source. This is shown by the dashed drawing in Fig. 5. It should be noted that setting the flow rates such that the stoichiometric ratio is fulfilled is not possible using only the flow controllers FC1 and FC2 because of unavoidable measurement or control implementation errors. For this reason, an additional concentration controller is necessary. An excess of alcohol will have as result the complete consumption of acid and a drop of the acid concentration at the bottom of the column. On the other hand, large quantities of acid will be present when this reactant is in excess. We conclude that the imbalance in the alcohol:acid ratio can be detected by measuring the concentration of acid in the bottom outlet of the column. Therefore, the concentration controller CC2 is used to give, in a cascade manner, the correct setpoint to the alcohol flow rate controller FC2. In this way, when production rate changes are implemented by changing the acid flow rate, the correct ratio between reactants is achieved. However, for large production rate changes the overall reaction rate is not sufficiently large to guarantee the complete conversion of both reactants and high purity cannot be achieved. In particular, large amounts of methanol will be present in the top product of the absorption column. The concentration controller CC1 avoids this by increasing the setpoint of the temperature controller TC1. The result is higher temperatures inside the column, and therefore increased reaction rates. In addition, the control of the material inventory is achieved by conventional level and pressure control loops. Details of the controller action and controller tuning parameters are presented in Table 4. Fig. 6a and b depict the dynamic simulation results for the flowsheet presented in Fig. 5, when the recycle of methanol vapors is considered. The simulation starts from the steady state. At time t = 1 h, the acid flow rate is increased by 10%, from 1166.7 to 1282 kg/h (5.824 to 6.4 kmol/h). Then, at time t = 5 h, the acid flow rate is decreased to 1041.7 kg/h (5.2 kmol/h), representing 10% decrease with respect to the nominal value. The new production rate is achieved in about 1 h. The dynamic response of the water product stream is slower. The purity of FAME remains high throughout the dynamic regime, with the acid concentration below the 2000 ppm requirement of the ASTM D6751–08 standard (i.e. acid number less than 0.50 mg KOH/g biodiesel). Similarly, Fig. 6c and d show the dynamic simulation results for the same flowsheet presented in Fig. 5, but without the bottom recycle of methanol vapors. The same scenario is assumed for the dynamic simulation. Again, the production rate responds quickly to changes of acid flow rate (Fig. 6c). However, a degradation of dynamic performance is observed when the purities of the product

Table 4 Controller tuning parameters. Controller

Control action

Range of manipulated variable

Range of controlled variable

Setpoint

Bias

Gain [%/%]

Integral time [min]

Level LC1 (Buffer vessel) Level LC2 (Dec., org. lq.) Level LC3 (Dec., aq. lq.) Level LC4 (Alc. evap.) Level LC5 (Col. sump) Level LC6 (Flash) Pressure PC1 (Column) Pressure PC2 (Flash) Temp. TC1 (Acid inlet) Mole fraction CC1 (Methanol, column top) Flow rate FC2 (Methanol inlet) Mole fraction CC2 (Acid, col. btm.) Temperature TC2 (Flash)

Direct Direct Direct Reverse Direct Direct Direct Reverse Reverse Direct Reverse Direct Reverse

0–2350 kg/h 0–40 kg/h 0–210 kg/h 0–400 kg/h 0–2500 kg/h 0–2490 kg/h 0–12 kmol/h 0–1.22 kmol/h 0–0.2 GJ/h 140–180 °C 0.5–1 2–8 kmol/h 0–0.1 GJ/h

0–1 m 0–1 m 0–1 m 0–2 0–1.75 m 0–1 m 0.95–1.05 bar 0.18–0.22 bar 140–180 °C 1–5  103 4–8 kmol/h 1–2  103 130–140 °C

0.5 m 0.5 m 0.45 m 1 0.875 m 0.5 m 1 bar 0.2 bar Output of CC1 3  103 Output of CC2 1  103 137 °C

1180 kg/h 14.2 kg/h 105 kg/h 197 kg/h 1259 kg/h 1245 5.91 kmol/h 0.29 kmol/h 0.12 GJ/h 167.6 °C 0.9 6.15 kmol/h 0.01 GJ/h

1 1 1 1 10 1 1 1 1 0.1 1 0.2 1

– – – – 60 – 12 12 20 40 20 40 20

A.A. Kiss, C.S. Bildea / Bioresource Technology 102 (2011) 490–498

Flow rates / [kg/h]

a

1500

FAME

1250

Acid

1000

300

4. Conclusions

250

The novel heat-integrated reactive absorption process proposed in this work eliminates all conventional catalyst related operations, improves efficiency and considerably reduces the energy requirements for biodiesel production – 85% lower as compared to the base case. Another important result is an efficient control structure that ensures the required reactants ratio and fulfills the excess of methanol operating constraint that is sufficient for the total conversion of the fatty acids and for prevention of difficult separations. Remarkably, in spite of the high degree of integration this reactive absorption process is very well controllable as illustrated by the results of rigorous dynamic simulations.

200 Alcohol

750 500

150 100

Water

250

50

0 0

2

4

6

8

497

0 10

b

1

Mass fraction / [-]

Time / [h]

0.98

2000

0.96

1500

0.94

1000 500

0.92 Acid in FAME

0

2

4

6

8

Mass fraction / [ppm]

FAME

Water

0.9

Acknowledgements

2500

0 10

Time / [h]

Flow rates / [kg/h]

c

1500

FAME

300

1250

Acid

250

1000

200 Alcohol

750

150

500

100 Water

250

50

0

0 0

2

4

6

8

10

1 0.96

3500

FAME

3000 2500

0.92

2000

Water

1500

0.88

1000 0.84 Acid in FAME

0.8 0

2

4

6

8

500

Mass fraction / [ppm]

d Mass fraction / [-]

Time / [h]

0 10

Time / [h] Fig. 6. Dynamic simulation results for the flowsheet with (a, b) and without (c, d) methanol recycle: Acid flow rate disturbance of +10% at 1 h, and –10% at 5 h. Production rate changes are easily achieved and the products purity is maintained at high values.

streams are taken into account (Fig. 6d). Therefore, the alternative with methanol recycle is preferable.

C. S. Bildea acknowledges the support of CNCSIS – UEFISCSU, Project Number PNII – IDEI 1545/2008 – ‘‘Advanced modelling and simulation of catalytic distillation for biodiesel synthesis and glycerol transformation”. We also thank A. C. Dimian and G. Rothenberg (University of Amsterdam), as well as A. B. de Haan (Eindhoven University of Technology, The Netherlands) for the helpful discussions. References Technology, Aspen., 2009. Aspen Plus: User guide – volume 1 & 2. Burlington, MA. Balat, M., Balat, H., 2008. A critical review of bio-diesel as a vehicular fuel. Energy Convers. Manage. 49, 2727–2741. Bildea, C.S., Kiss, A.A., 2010. Dynamics and control of a biodiesel process by reactive absorption, Chem. Eng. Res. Des., doi: 10.1016/j.cherd.2010.05.007. Bowman, M., Hilligoss, D., Rasmussen, S., Thomas, R., 2006. Biodiesel: a renewable and biodegradable fuel. Hydrocarb. Process. 85, 103–106. Canoira, L., Galean, J.G., Alcantara, R., Lapuerta, M., Garcia-Contreras, R., 2010. Fatty acid methyl esters (FAMEs) from castor oil: production process assessment and synergistic effects in its properties. Renewable Energy 35, 208–217. Chen, Y.H., Yu, C.C., 2003. Design and control of heat-integrated reactors. Ind. Eng. Chem. Res. 42, 2791–2808. Dale, B., 2003. Greening the chemical industry: research and development priorities for bio-based industrial products. J. Chem. Technol. Biotechnol. 78, 1093–1103. Demirbas, M.F., Balat, M., 2006. Recent advances on the production and utilization trends of bio-fuels. Energy. Convers. Manage. 47, 2371–2381. Demirbas, A., 2008. Comparison of trans-esterification methods for production of biodiesel from vegetable oils and fats. Energy Convers. Manage. 49, 125–130. Dimian, A.C., 2003. Integrated design and simulation of chemical processes. Elsevier, Amsterdam. Dimian, A.C., Bildea, C.S., Omota, F., Kiss, A.A., 2009. Innovative process for fatty acid esters by dual reactive distillation. Comput. Chem. Eng. 33, 743–750. Encinar, J.M., Gonzalez, J.F., Rodriguez-Reinares, A., 2005. Biodiesel from used frying oil. Variables affecting the yields and characteristics of the biodiesel. Ind. Eng. Chem. Res. 44, 5491–5499. Ghaemi, A., Shahhosseini, S., Maragheh, M.G., 2009. Nonequilibrium modelling of reactive absorption processes. Chem. Eng. Commun. 196, 1076–1089. Hamed, O.A., Aly, S., AbuKhousa, E., 1996. Heuristic approach for heat exchanger networks. Int. J. Energy Res. 20, 797–810. Hanna, M.A., Isom, L., Campbell, J., 2005. Biodiesel: current perspectives and future. J. Sci. Ind. Res. 64, 854–857. He, B.B., Singh, A.P., Thompson, J.C., 2006. A novel continuous-flow reactor using reactive distillation for biodiesel production. Trans. ASAE 49, 107–112. He, H., Wang, T., Zhu, S., 2007. Continuous production of biodiesel fuel from vegetable oil using supercritical methanol process. Fuel 86, 442–447. Janulis, P., 2004. Reduction of energy consumption in biodiesel fuel life cycle. Renewable Energy 29, 861–871. Kiss, A.A., Dimian, A.C., Rothenberg, G., 2006a. Solid acid catalysts for biodiesel production – towards sustainable energy. Adv. Synth. Catal. 348, 75–81. Kiss, A.A., Rothenberg, G., Dimian, A.C., Omota, F., 2006b. The heterogeneous advantage: biodiesel by catalytic reactive distillation. Top. Catal. 40, 141–150. Kiss, A.A., Dimian, A.C., Rothenberg, G., 2008. Biodiesel by reactive distillation powered by metal oxides. Energy Fuels 22, 598–604. Kiss, A.A., 2009. Novel process for biodiesel by reactive absorption. Sep. Purif. Technol. 69, 280–287. Kiss, A.A., 2008. Biodiesel production by heat-integrated reactive distillation. Comp. Aided Chem. Eng. 25, 775–780. Kulkarni, M.G., Dalai, A.K., 2006. Waste cooking oil-an economical source for biodiesel: a review. Ind. Eng. Chem. Res. 45, 2901–2913. Kumar, N., Sharma, P.B., 2005. Jatropha Curcus – a sustainable source for production of biodiesel. J. Sci. Ind. Res. 64, 883–889.

498

A.A. Kiss, C.S. Bildea / Bioresource Technology 102 (2011) 490–498

Kusdiana, D., Saka, S., 2004. Two-step preparation for catalyst-free biodiesel fuel production: hydrolysis and methyl esterification. Appl. Biochem. Biotechnol. 115, 781–792. Lotero, E., Liu, Y.J., Lopez, D.E., Suwannakarn, K., Bruce, D.A., Goodwin, J.G., 2005. Synthesis of biodiesel via acid catalysis. Ind. Eng. Chem. Res. 44, 5353–5363. Meher, L.C., Vidya Sagar, D., Naik, S., 2006. Technical aspects of biodiesel production by trans-esterification: a review. Renewable Sustainable Energy Rev. 10, 248–268. Minami, E., Saka, S., 2006. Kinetics of hydrolysis and methyl esterification for biodiesel production in two-step supercritical methanol process. Fuel 85, 2479– 2483. Nagy, Z.K., Klein, R., Kiss, A.A., Findeisen, R., 2007. Advanced control of a reactive distillation column. Comp. Aided Chem. Eng. 24, 805–810.

Narasimharao, K., Lee, A., Wilson, K., 2007. Catalysts in production of biodiesel: a review. J. Biobased Mater. Bioenergy 1, 19–30. Noeres, C., Kenig, E.Y., Gorak, A., 2003. Modelling of reactive separation processes: reactive absorption and reactive distillation. Chem. Eng. Process 42, 157–178. Puhan, S., Vedaraman, N., Rambrahamam, B.V., Nagarajan, G., 2005. Mahua (Madhuca Indica) seed oil: a source of renewable energy in India. J. Sci. Ind. Res. 64, 890–896. Van Diggelen, R.C., Kiss, A.A., Heemink, A.W., 2010. Comparison of control strategies for dividing-wall columns. Ind. Eng. Chem. Res. 49, 288–307. Van Gerpen, J., 2005. Biodiesel processing and production. Fuel Process. Technol. 86, 1097–1107.