chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Dynamics and control of a biodiesel process by reactive absorption

Costin Sorin Bildea a , Anton A. Kiss b,
a b

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

a b s t r a c t
Integrated biodiesel processes based on reactive separations powered by solid acid/base catalysts are available nowadays, offering signicant 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 savings. In such processes the small number of degrees of freedom is a drawback which makes it difcult to correctly set the ratio of reactant feeds and consequently to avoid impurities in the products. This work considers the process control of biodiesel production by reactive absorption, the main result being an efcient control structure that ensures the excess of methanol that is necessary for the total conversion of the fatty acids and for prevention of the difcult separations, while maintaining high purity of the water by-product. Rigorous simulations were performed using Aspen Plus and Aspen Plus Dynamics as efcient computer-aided process engineering tools for a plant producing 10 ktpy biodiesel from waste vegetable oil with high free fatty acids content, using solid acids as green catalysts. This reactive absorption process eliminates all conventional catalystrelated operations, and efciently uses the raw materials and the reactor volume in an integrated setup that is well controllable in spite of the reduced number of degrees of freedom. 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. Keywords: Reactive absorption; Plantwide control; Dynamic modeling; Green catalysts; Biofuels

1.

Introduction

Biodiesel is an alternative sustainable fuel produced from green sources such as waste vegetable oils, animal fat or even frying-oils from the food industry (Encinar et al., 2005; Kulkarni and Dalai, 2006; Bowman et al., 2006; Balat and Balat, 2008). Such waste raw materials can contain a substantial amount of free fatty acids (FFAs), up to 100%. At present, employing waste and non-edible raw materials is mandatory to comply with the ecological and ethical requirements for biofuels (Demirbas and Balat, 2006). Moreover, the food versus fuel competition can be avoided when the raw materials used are waste oils or non-food crops such as Jatropha (Kumar and Sharma, 2005) or Mahua (Puhan et al., 2005). As a non-petroleum-based diesel fuel, biodiesel consists of fatty acid methyl esters (FAME), currently produced by acid/base-catalyzed (trans-)esterication, followed by several neutralization and purication steps (Ma and Hanna, 1999;

Meher et al., 2006). Nevertheless, all the conventional methods suffer from problems associated with the use of homogeneous acid or base catalysts, leading to serious economical and environmental consequences, especially considering the recent growth of the overall biodiesel production scale (Hanna et al., 2005). Nowadays, modern plants replaced the homogeneous catalysts with solid bases or acids, thus eliminating the salt waste streams and simplifying the downstream processing steps. Moreover, integrated processes based on reactive distillation (Kiss et al., 2006a,b, 2008; Dimian et al., 2009) or reactive absorption (Kiss, 2009) are now available, offering signicant advantages such as minimal capital investment and operating costs, as well as no catalyst-related waste streams and no soap formation. This work considers the process control of biodiesel production by reactive absorption. The results are given for a plant producing 10 ktpy biodiesel from waste vegetable oil with high free fatty acids content (up to 100%), using solid acids as green

Corresponding author. Tel.: +31 26 366 1714, fax: +31 26 366 5871. E-mail addresses: s bildea@upb.ro (C.S. Bildea), tony.kiss@akzonobel.com, tonykiss@gmail.com (A.A. Kiss). Received 16 February 2010; Received in revised form 12 May 2010; Accepted 28 May 2010 0263-8762/$ see front matter 2010 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.cherd.2010.05.007

188

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

Fig. 1 Flowsheet of a biodiesel production process by reactive absorption. catalysts. Compared to reactive distillation, the absence of a reboiler and a condenser makes reactive absorption a simpler process. However, the drawback is the small number of degrees of freedom that makes it difcult to set the reactants feed ratio correctly and consequently to avoid impurities in the products. Aspen Plus and Aspen Plus Dynamics were used as efcient computer-aided process engineering tool to perform rigorous simulations for testing various control structures. An efcient control structure is presented, that ensures the ratio of reactants that fullls an excess of methanol. This excess is sufcient for the total conversion of the fatty acids therefore preventing difcult separations, while keeping high the purity of the water by-product. was selected as lumped component due to the availability of experimental results, kinetics and VLLE parameters for this system (Kiss, 2009). The assumption of lumping components is very reasonable since fatty acids and their corresponding fatty esters have similar properties. This approach has been successfully used during the last decade to simulate other fatty esters production processes, based on reactive distillation for example (Kiss et al., 2008). The next simulations consider the UNIQUAC property model (Aspen Technology, 2009a,b,c), sulfated zirconia as solid acid catalyst (Kiss et al., 2006a,b, 2008; Dimian et al., 2009) and a simple but reliable kinetic expression (Kiss, 2009): r = kCAcid CAlcohol where CAcid and CAlcohol are the mass concentration of reactants, k = A exp(Ea /RT), A = 250 kmol m3 kg2 s1 and Ea = 55 kJ/mol. The production rate of the process considered in this work is 10 ktpy fatty acid methyl esters manufactured from 100% fatty acids and methanol. Fig. 1 presents the owsheet of a biodiesel process based on a reactive absorption (RA) column, while Table 1 presents the mass and energy balance.

2.

Process description

The fatty components were lumped into one fatty acid and its fatty ester, according to the reaction: RCOOH + CH3 OH RCOOCH3 + H2 O. Lauric acid/ester

Table 1 Mass balance of a 10 ktpy biodiesel process based on reactive absorption. F-ACID
Temperature [ C] Pressure [atm] Vapor fraction Mole ow [kmol/h] Mass ow [kg/h] Volume ow [l/min] Enthalpy [Gcal/h]] Mass ow [kg/h] Methanol Acid Water FAME Mass fraction Methanol Acid Water FAME Mole fraction Methanol Acid Water FAME 160 0.987 0 5.885 1176.5 25.1 0.948 1.51 104 1176.2 0.246 8.62 105 0 0.9998 2.0 104 0

F-ALCO
65.5 1.036 1 6.111 195.8 2730.6 0.291 195.8 0 0 0

BTM
135.5 1.017 0 6.111 1257.6 23.5 0.902 9.20 0.168 9.36 104 1248.2 7.32 103 1.34 104 0 0.9926

TOP
162.9 0.987 1 5.885 114.7 3555.7 0.337 8.83 103 9.52 105.2 8.6 105 7.7 105 0.083 0.917 0 4.7 105 0.008 0.992 0

REC-TOP
51.9 0.987 0 0.061 9.747 0.19 0.009 1.51 104 9.50 0.246 8.62 105 1.55 105 0.975 0.025 0 7.73 105 0.776 0.224 0

VAP
135.5 0.2 1 0.255 11.3 711.7 0.013 7.61 1.39 104 8.28 104 3.708

Water
51.9 0.987 0 5.824 104.9 1.81 0.395 8.68 103 16.8 103 104.91 0 8.27 105 1.6 104 0.9998 0 4.65 105 1.44 105 0.9999 0

FAME
30 0.2 0 5.857 1246.3 20.9 0.954 1.59 0.168 1.07 104 1244.5 1.28 103 1.35 104 0 0.9986 8.50 103 1.43 104 0 0.991

1 0 0 0

0.672 1.23 105 7.32 105 0.328

0 0.9976 2.32 103 0

1 0 0 0

0.047 1.37 104 0 0.953

0.932 0 1.81 104 0.068

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

189

Fig. 2 Control structure CS-1a: in addition to basic inventory control, a ratio controller xes the inlet ow rates. The column has 15 theoretical stages with a liquid holdup of 25 L. The catalyst loading is 6.5 kg per reactive stage. The RA column is operated in the temperature range of 135160 C, at ambient pressure, with a maximum reaction rate in the middle of the reactive zone. As phase splitting may occur when the molar fraction of water is higher than 0.25, the reactive-absorption column is modeled using VLLE data. Nevertheless, as revealed latter by the composition proles, the molar fraction of water in the liquid phase does not exceed 0.05 and therefore phase splitting does not occur here, under the designed process conditions. The fatty acid is pre-heated then fed as a hot liquid in the top of the reactive column while an amount of alcohol which ensures the complete conversion of the fatty acid is injected as vapor into the bottom of the column, thus creating a countercurrent ow 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 bottom product of the RA column. The hot product is ashed to remove the remaining methanol, and then it is cooled and stored. converted and the purity of water by-product is high, but the conversion of fatty acids is incomplete. In the later case the bottom product contains unreacted fatty acids that cannot be removed from the nal product by simple ashing. Since the separation of fatty acids from fatty esters is more difcult than the separation of fatty acids from water, this situation should be avoided. It is important to remark that this constraint must be fullled not only during the normal operation, but also during the transitory regimes arising due to planned production rate changes or unplanned disturbances.

4.

Results and discussion

Plantwide control concerns the strategy of solving the dynamic problems of an entire process, with the goals of safety and a required production (rate and quality), while minimizing the operating cost. In this section, two workable control structures are developed by addressing these tasks of plantwide control. To achieve this goal, dynamic simulation (performed in Aspen Plus Dynamics) will be used to understand the dynamic behavior of the process and to prove the effectiveness of the proposed control structures.

3.

Problem statement

4.1.

Basic control

Reactive absorption offers indeed signicant 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 savings. Therefore, it is important to note that reactive absorption has less degrees of freedom and therefore more difcult to control than reactive distillation (Nagy et al., 2007). In processes based on reactive distillation or absorption, feeding the reactants in a correct ratio is essential to achieve high products purity (Nagy et al., 2007; Dimian et al., 2009; Kiss et al., 2010). 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 methanol is completely

In general, the safety of a process is achieved by a combination of inherent safe design, safety relief valves and instrument protective functions. By controlling liquid levels and gas pressures, therefore ensuring that all the material is contained within the process boundaries, the control system supports safety. For the reactive-absorption column presented in Fig. 1, the levels of organic and aqueous phases in the decanter, the level in the sump of the column, and the level in the ash vessel should be controlled. These tasks can be achieved by employing, as manipulated variables, the ow rates of acid recycle, water by-product, column bottom and FAME product, respectively. The pressure at the top of the column can be controlled by the vapor product rate. Besides the safety of the process, controlling this variable is important because the pressure determines the temperature prole inside the column (boiling

190

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

Fig. 3 Dynamic simulation results for control structure CS-1a. Production rate changes are easily achieved. At t = 2 h, the fresh acid ow rate is increased by 10%, from 5.824 to 6.4 kmol/h. At t = 20 h, the fresh acid ow rate is decreased to 5.2 kmol/h. point of the reacting mixture) and the vapor ow rate. These control loops are presented in Fig. 2. The efuents (products) of the reactive distillation process are FAME, water and unreacted methanol. The amounts of methanol and acid in the FAME product must be kept low because this is the main product and its quality determines the performance of the process. The concentration of methanol is the result of the vapor-liquid equilibrium in the auxiliary ash vessel. The temperature of the separation is easily controlled by the ash duty. One way to achieve the required vacuum is by means of a positive displacement compressor. In this case, the pressure can be controlled by manipulating the ow rate of a small inlet stream of inert gas, as shown in Fig. 2. These control loops also ensure that a limited amount of FAME is lost with the methanol by-product stream. However, low acid concentration in the FAME product cannot be achieved and controlled through a simple vapor-liquid equilibrium, because of the very tight requirement. To accomplish this goal, the control system must ensure that the reaction proceeds fast enough and there is sufcient methanol such that the entire amount of acid fed in the process is consumed in the reaction. As a rst try, the ratio between the column-inlet ows of alcohol and acid is kept constant, in a feed-forward manner: the setpoint of the ow controller on alcohol stream is calculated as the measured column-inlet acid ow multiplied by the desired alcohol/acid ratio (Fig. 2). At the top of the column, the amount of acid found in the water by-product is the result of its solubility. The solubility is determined by temperature, which can be controlled with the cooling duty. However, methanol and water are completely miscible. As a result, the entire amount of methanol found in the vapor outlet of the column will end in the water byproduct. The simplest way to set the production rate is by changing the amount of fatty acid that is fed into the process. Because only tiny amounts of fatty acid should leave the process with FAME, water and excess methanol (stream VAP in Fig. 1), the ow rate of fresh acid (stream ACID in Fig. 1) directly determines the process throughput. It should be remarked that the production rate is not feedback-controlled, but it is being set as a consequence of changing the amount of fresh acid fed to the process. This strategy works well when the per-pass conversion of the limiting reactant is large (Kiss et al., 2002, 2003, 2007; Dimian and Bildea, 2008; Kiss, 2010), condition that is fullled here. Note that the buffer vessel where the streams of fresh and recycled acid are mixed (Fig. 2) is not necessary and it can be replaced by a simple mixer. However, it allows switching the ow and level control loops. Thus, the alternative control structure CS-2 shown in Fig. 13 and discussed later sets the

Fig. 4 Control structure CS-1a: composition, reaction rate, and temperature proles along the reactive-absorption column, for different production rate changes (10%).

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

191

Fig. 5 CS-1a: purity versus the alcohol/acid ratio at various temperatures of the column-inlet acid feed (left); Optimum ratio versus temperature of the column-inlet acid feed (right).

Fig. 6 Control structure CS-1b: the concentration of acid in bottom stream is controlled by manipulating the alcohol/acid ratio. production rate in an indirect manner, as a result of changing the amount of acid that is fed to the reactive-absorption column. Fig. 3 shows results of dynamic simulation when control structure CS-1a is applied. Starting from the steady state, at time t = 2 h, the fresh acid ow rate is increased by about 10%, from 5.824 to 6.4 kmol/h. The ratio control leads to a higher ow rate of the alcohol feed. The new FAME production rate is achieved in several hours and the new operating point is stable. At time t = 20 h, the fresh acid ow rate is decreased to 90% of the original value (5.2 kmol/h). Again, the production rate changes as expected. The second plot of Fig. 3 shows the purity of the FAME and water streams. Although the purity remains high, during the large-production operating regime the purity of the FAME product stream is below the 99.8% requirement. Moreover, there is too much methanol in the water stream during the low-production operation. Fig. 4 shows reaction rate, temperature and liquid phase composition proles along the reactive-absorption column, for different production rates. When the production rate is high, the reaction front moves to the bottom of the column. Because no reaction takes place in the upper part of the column, the acid is pushed downwards, with the result of large concentration of acid in the FAME product stream. Simultaneously, the temperature in the upper part of the column increases because of the higher fraction of acid. Note that the numbering of stages starts from the top, increasing to the bottom of the column, as illustrated by Fig. 1.

Fig. 7 Dynamic simulation results for control structure CS-1b. Production rate changes are easily achieved and the purity of FAME stream is maintained.

192

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

Fig. 8 Control structure CS-1b: composition, reaction rate, and temperature proles along the reactive-absorption column, for different production rate changes (10%).

4.2.

Control of FAME purity

The proles presented in Fig. 4 suggest that keeping the alcohol/acid ratio at the steady-state design value is not sufcient to achieve high purity of the FAME and water product streams when the production rate is modied. The inuence of the alcohol/acid ratio on the purity of the FAME and water streams is presented in Fig. 5, at steady state and for different values of the Acid inlet temperature. It can be seen that at low values of the alcohol/acid ratio, the FAME purity is low but the water purity is high. The purities are reversed when the alcohol/acid ratio is high, namely high purity of the FAME stream and low purity of the water stream. Fig. 5 also shows that there is a narrow range of alcohol/acid ratio values, depending on the acid feed temperature, for which both FAME and water products have a high purity. In addition to the drawback pointed out by dynamic simulation, the sensitivity analysis shows that control structure CS-1a lacks robustness with respect to measurement inaccuracies because any error of the acid ow rate measurement leads to deviation of the alcohol/acid

ratio from the required value and therefore to impure products. These results suggest that the manipulation of the alcohol/acid ratio could be used to limit the amount of acid that arrives in the bottom of the column during the highproduction operating regime. This change implies adding a concentration controller as shown in CS-1b presented in Fig. 6. Fig. 7 presents the dynamic simulation results for the control structure CS-1b when the same scenario as in Fig. 3 is applied. As with CS-1a, the operation is stable and production rate changes can be easily achieved. In addition, the quality of the FAME product is radically improved. However, the acid-concentration controller overreacts and large amounts of methanol can be found in the water stream during the high-throughput operating regime. Fig. 8 shows the reaction rate, temperature and liquid composition proles along the reactive-absorption column, for different production rates and control structure CS-1b, when production rate changes are applied. The variability with respect to the base case (design) is reduced and only ne

Fig. 9 Control structure CS-1c: concentration of acid in bottom stream is controlled by manipulating the alcohol/acid ratio. The temperature of the column-inlet acid stream is manipulated by a methanol concentration controller.

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

193

Fig. 10 Dynamic simulation results for control structure CS-1c. Production rate changes are easily achieved and the products purity is maintained at high values.

Fig. 11 Control structure CS-1c: composition, reaction rate, and temperature proles along the reactive-absorption column, for different production rate changes (10%). adjustments seem to be necessary. It should be remarked that the large amount of methanol in the water stream during the high-throughput regime is a steady-state effect, as shown by Fig. 7 (right). Therefore, it is not caused by a wrong tuning of the ratio controller. In order to keep the methanol concentration at the required value, an additional process variable must be used.

4.3.

Control of water purity

The right diagram of Fig. 5 shows that an increase of the acid temperature reduces the amount of alcohol that must be fed to the column in order to keep the purity of both products at high values. This effect can be explained by the additional contribution of higher temperature to higher reaction rate and

Fig. 12 Control structure CS-1c: composition, reaction rate, and temperature proles along the reactive-absorption column, for different catalytic activities.

194

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

Fig. 13 Control structure CS-2: the column-inlet ow rate of acid is xed. The fresh acid controls the level in the buffer tank.

Fig. 14 Dynamic simulation results for control structure CS-2. Production rate changes are easily achieved and the products purity is maintained at high values. At t = 2 h, the column-inlet acid ow rate is increased by 10%, from 5.887 to 6.5 kmol/h. At t = 20 h, the column-inlet acid ow rate is decreased to 5.24 kmol/h.

Table 2 Controller tuning parameters. Controller Control action

Direct Direct Direct Direct Direct Direct Reverse Reverse Direct

Range of manipulated variable

02350 kg/h 020 kg/h 0215 kg/h 02517 kg/h 02490 kg/h 012 kmol/h 01.22 kmol/h 018.7 104 kcal/h 140180 C

Range of controlled variable

01.5 m 00.75 m 00.75 m 01.5 m 01.5 m 0.951.05 bar 0.180.22 bar 140180 C 35 103

Setpoint

Bias

Gain [%/%]

Integral time [min]

60 min 60 min 60 min 60 min 60 min 12 min 12 min 20 min 20 min

Level (buffer vessel) Level (decanter, organic liquid) Level (decanter, aqua. phase) Level (column sump) Level (ash) Pressure (column) Pressure (ash) Temperature (acid inlet) Mole fraction (methanol, column top) Mole fraction (acid, column bottom) Temperature (methanol inlet)

0.75 m 0.37 m 0.35 m 0.75 m 0.75 m 1 bar 0.2 bar 160.6 C 4 103

1176.8 kg/h 10.12 kg/h 105.7 kg/h 1259 kg/h 1245 5.91 kmol/h 0.31 kmol/h 9.36 104 kcal/h 160.6 C

10 1 1 1 1 2 2 2 0.05

Direct

0.10.2

28 ppm

6 ppm

0.1688

20 min

Reverse

012 104 kcal/h

6080 C

65.4

5.9 104 kcal/h

20 min

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

195

therefore to complete acid conversion. Consequently, in the control structure CS-1c (Fig. 9), a second concentration controller adjusts in a cascade fashion the setpoint of the acid stream temperature controller in order to maintain the concentration of methanol in the vapor outlet below a certain value. During the high-throughput regime, increasing the temperature in the upper part of the column has the effect of increased reaction rate, and therefore reduces the need of an excess of methanol. Fig. 10 shows the dynamic simulation results for the control structure CS-1c. The same scenario was tested as for CS-1a and CS-1b. The production rates are changed as desired. The purity of the FAME product is practically unchanged throughout the dynamic regime, while the purity of water is signicantly better compared to the previous cases. The temperature, reaction rate and liquid composition proles are shown in Fig. 11. It can be seen that by making ne adjustments to the temperature prole by means of the acid feed temperature the control system keeps constant the position of the reaction front, while the reaction rate is modied according to the required production rate.

4.6.

Controller tuning

The control loops were tuned by a simple version of the direct synthesis method (Luyben and Luyben, 1997). According to this method, the desired closed-loop response for a given input is specied. Then, with the model of the process known, the required form and the tuning of the feedback controller are back calculated. For all controllers, the acceptable control error, max , and the maximum available control action, umax , were specied. Then the controller gain, expressed in engineering units, was calculated as Kc = umax / max and translated into percentage units. First order open-loop models were assumed, in order to calculate the integral time of the pressure, temperature and concentration control loops. As rough evaluations of the process time constants , 12, 20 and 20 min were used, respectively. It can be shown (Luyben and Luyben, 1997) that the direct synthesis method requires that the reset time of a PI controller is equal to the time constant of the process, i = . For the level controllers, a large reset time i = 60 min was chosen as no tight control is required. An overview of the controllers tuning parameters is given in Table 2.

4.4.

Inuence of the catalyst activity

Finally, the performance of the control system was tested for a deactivating catalyst. The investigated scenario assumed a uniform decrease of the pre-exponential factor from 250 kmol m3 kg2 s1 (at t < 2 h) to 200 kmol m3 kg2 s1 (at t = 40 h). The control system succeeded to keep the throughput and product quality unchanged. Fig. 12 shows proles obtained at different moments during the dynamic simulation. The control system compensates lower catalyst activity by higher temperature inside the column, with the result of a constant reaction rate. We conclude that control structure CS-1c achieves stable operation at different production rates and is robust with respect to uncertainty of the reaction rate.

5.

Conclusions

Integrated biodiesel processes based on reactive absorption have fewer degrees of freedom as compared to reactive distillation. This makes it difcult to correctly set the reactants feed ratio and consequently to avoid impurities in the products. A control structure that achieves this objective is the main result of this study. Basic quality control is achieved by keeping constant the ratio alcohol/acid at column inlet, in addition to inventory control loops (CS-1a, Fig. 2). When the process is disturbed or there are uncertainties in the measurements, measuring the acid concentration in the bottom stream of the column and adjusting the alcohol/acid ratio (CS-1b, Fig. 6) is necessary for maintaining a very small concentration of acid in the FAME product, an important quality feature of the biodiesel product. High purity of the water by-product requires control of the alcohol concentration in the column top stream by manipulating the column-inlet acid temperature (CS-1c, Fig. 9). In this way, the correct excess of alcohol is achieved resulting in high products purity. As an alternative, the column-inlet acid ow rate can be xed (CS-2, Fig. 13).

4.5.

Indirect setting of the production rate

Kiss et al. (2007) discuss another option for controlling the inventory of reactants in the process, namely xing the ow rate of the reactant at the inlet of the reactive section (stream F-ACID in Fig. 1) and bringing the reactant into the process by means of a level controller. This control structure, denoted by CS-2, is illustrated in Fig. 13. When the level and control loops around the buffer tank were switched, dynamic simulation showed that the gain of both controllers in the concentrationtemperature cascade at the top of the column had to be reduced in order to attain stable operation. Fig. 14 shows results of dynamic simulation for the following scenario: at t = 2 h, the fresh acid rate is increased by 10%, from 5.824 to 6.5 kmol/h. At t = 20 h, the fresh acid ow rate is decreased to 5.24 kmol/h. The production rate can be easily manipulated (although in an indirect manner) by changing the setpoint of the acid-ow controller. The purity of the product streams remains high, although, compared to Fig. 10, a slight degradation of the performance is observed. However, this could be as well the result of a wrong selection of the tuning parameters, and not a consequence of switching the level and ow control loops.

The performance of the proposed control structure was demonstrated (Figs. 7, 10 and 14) by rigorous dynamic simulations using Aspen Plus Dynamics as an efcient computer-aided process engineering tool.

Acknowledgements
C.S. Bildea acknowledges the support of CNCSIS UEFISCSU, project number PNIIIDEI 1543/2008 A nonlinear approach to conceptual design and safe operation of chemical processes. We also thank A. C. Dimian (University of Amsterdam) and A.B. de Haan (Eindhoven University of Technology, The Netherlands) for the helpful discussions.

196

chemical engineering research and design 8 9 ( 2 0 1 1 ) 187196

References
Aspen Technology., (2009a). Aspen Plus: User Guide, vols. 1 and 2. Aspen Technology., (2009b). Aspen Plus Dynamics: User Guide. Aspen Technology., (2009c). Aspen Physical Property SystemPhysical Property Models. Balat, M. and Balat, H., 2008, A critical review of bio-diesel as a vehicular fuel. J Energy Convers Manage, 49: 27272741. Bowman, M., Hilligoss, D., Rasmussen, S. and Thomas, R., 2006, Biodiesel: a renewable and biodegradable fuel. Hydrocarbon Process, 85: 103106. Demirbas, M.F. and Balat, M., 2006, Recent advances on the production and utilization trends of bio-fuels. Energy Convers Manage, 47: 23712381. Dimian, A.C. and Bildea, C.S., (2008). Chemical Process DesignComputer-aided Case Studies. (Wiley-VCH, Weinheim). Encinar, J.M., Gonzalez, J.F. and Rodriguez-Reinares, A., 2005, Biodiesel from used frying oil. Variables affecting the yields and characteristics of the biodiesel. Indus Eng Chem Res, 44: 54915499. Dimian, A.C., Bildea, C.S., Omota, F. and Kiss, A.A., 2009, Innovative process for fatty acid esters by dual reactive distillation. Comput Chem Eng, 33: 743750. Hanna, M.A., Isom, L. and Campbell, J., 2005, Biodiesel: current perspectives and future. J Sci Indus Res, 64: 854857. Kiss, A.A., Bildea, C.S., Dimian, A.C. and Iedema, P.D., 2002, State multiplicity in CSTRseparatorrecycle polymerization systems. Chem Eng Sci, 57: 535546. Kiss, A.A., Bildea, C.S., Dimian, A.C. and Iedema, P.D., 2003, State multiplicity in PFRseparatorrecycle polymerization systems. Chem Eng Sci, 58: 29732984. Kiss, A.A., Dimian, A.C. and Rothenberg, G., 2006, Solid acid catalysts for biodiesel productiontowards sustainable energy. Adv Syn Catal, 348: 7581.

Kiss, A.A., Rothenberg, G., Dimian, A.C. and Omota, F., 2006, The heterogeneous advantage: biodiesel by catalytic reactive distillation. Top Catal, 40: 141150. Kiss, A.A., Bildea, C.S. and Dimian, A.C., 2007, Design and control of recycle systems by non-linear analysis. Comput Chem Eng, 31: 601611. Kiss, A.A., Dimian, A.C. and Rothenberg, G., 2008, Biodiesel by reactive distillation powered by metal oxides. Energy Fuels, 22: 598604. Kiss, A.A., 2009, Novel process for biodiesel by reactive absorption. Sep Purif Technol, 69: 280287. Kiss, A.A., Bildea, C.S. and Grievink, J., 2010, Dynamic modeling and process optimization of an industrial sulfuric acid plant. Chem Eng J, 158: 241249. Kiss, A.A., (2010). Process Design and Control by Nonlinear Analysis: Applications to ReactorSeparatorRecycle Systems. (LAP Lambert Academic Publishing, Cologne, Germany). Kulkarni, M.G. and Dalai, A.K., 2006, Waste cooking oilan economical source for biodiesel: a review. Indus Eng Chem Res, 45: 29012913. Kumar, N. and Sharma, P.B., 2005, Jatropha curcusa sustainable source for production of biodiesel. J Sci Indus Res, 64: 883889. Luyben, W.L. and Luyben, M.L., (1997). Essentials of Process Control. (McGraw-Hill, New York). Ma, F. and Hanna, M.A., 1999, Biodiesel production: a review. Bioresour Technol, 70: 115. Meher, L.C., Vidya Sagar, D. and Naik, S., 2006, Technical aspects of biodiesel production by trans-esterication: a review. Renew Sust Energy Rev, 10: 248268. Nagy, Z.K., Klein, R., Kiss, A.A. and Findeisen, R., 2007, Advanced control of a reactive distillation column. Comput Aid Chem Eng, 24: 805810. Puhan, S., Vedaraman, N., Rambrahamam, B.V. and Nagarajan, G., 2005, Mahua (Madhuca indica) seed oil: a source of renewable energy in India. J Sci Indus Res, 64: 890896.