Applied Energy 99 (2012) 146–153

Contents lists available at SciVerse ScienceDirect

Applied Energy

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

Enhanced methanol recovery and glycerol separation in biodieselproduction – DWC makes it happen

Anton A. Kiss a,⇑, Radu M. Ignat b

a AkzoNobel Research, Development & Innovation, Process Technology ECG, Zutphenseweg 10, 7418 AJ Deventer, The Netherlandsb University ‘‘Politehnica’’ of Bucharest, Department of Chemical Engineering, Polizu 1-7, 011061 Bucharest, Romania

a r t i c l e i n f o a b s t r a c t

Article history:Received 31 January 2012Received in revised form 30 March 2012Accepted 12 April 2012Available online 4 June 2012

Keywords:Dividing-wall columnBiodiesel purificationDesign and controlProcess intensificationCapital and energy savings

0306-2619/$ - see front matter � 2012 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.apenergy.2012.04.019

⇑ Corresponding author. Tel.: +31 26 366 9420.E-mail addresses: Tony.Kiss@akzonobel.com, tonyk

Biodiesel is a renewable fuel that consists of fatty acids methyl esters – currently produced by trans-esterification of glycerides with methanol. After the biodiesel synthesis, the downstream processing stepsinvolve the purification of crude glycerol, as well as the separation of excess methanol (recyclable), glyc-erol by-product and water (from washing and pre-treatment steps). The separation of the ternary mix-ture methanol–water–glycerol is carried out in a conventional direct sequence that requires twodistillation columns and rather high amounts of energy.

This study proposes an efficient process intensification method for this ternary separation, namely theuse of a dividing-wall column (DWC) that is able to separate all products at high purity, in only oneequipment unit. AspenTech Aspen Plus and Aspen Dynamics were used as computer aided process engi-neering tools to perform the rigorous steady-state and dynamic simulations, as well as the optimizationof the new DWC separation alternative. In order to allow a fair comparison, all designs analyzed herewere optimized using the sequential quadratic programming (SQP) method. Remarkable, the resultsshow that the proposed DWC system requires 27% less energy and 12% lower investment costs, thus hav-ing a significant contribution towards inexpensive biodiesel production.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Renewable energy sources are developed to tackle the shortageof fossil fuels, the volatile oil prices and the increased greenhousegas emissions. At present, employing waste and non-edible rawmaterials is mandatory to comply with the ecological and ethicalrequirements for biofuels. Biodiesel is a renewable fuel consistingof fatty acid methyl esters (FAME), currently produced from greensources such as vegetable oils, animal fat or waste cooking-oils [1–4]. The most widespread manufacturing technologies use homoge-neous catalysts, in processes where both reaction and separationsteps can create bottlenecks. Although reactive separations wererecently proposed for the esterification of free fatty acids (FFA) toFAME [5–9], the vast majority of existing biodiesel plants are basedon trans-esterification of tri-glycerides leading to biodiesel andglycerol by-product [10–12]. The biodiesel production by trans-esterification (Fig. 1) requires several expensive downstream pro-cessing steps such as the purification of biodiesel, as well as theseparation of excess methanol, glycerol by-product and water –from the washing and FFA pre-treatment steps [13–17]. In practice,the separation of the ternary mixture methanol–water–glycerol is

ll rights reserved.

iss@gmail.com (A.A. Kiss).

carried out in a conventional direct sequence that requires two dis-tillation columns and plenty of energy (Fig. 2 top). Consequently,reducing the energy requirements is not only a challenge but alsoa great opportunity for improving the operation and economics ofbiodiesel processes [18,19].

Distillation is still the most important thermal separation tech-nology, but in spite of its widespread use and major benefits, a keydrawback is its high energy demand generating over 50% of plantoperating costs [20]. The last decades led to energy efficient distil-lation solutions based on process integration and intensificationtechniques, such as: cyclic distillation, heat-integrated distillationcolumn, reactive distillation, and thermally coupled columns[20–23,7,8,24,25].

Dividing-wall column (DWC) – used for ternary separations – isa practical implementation of the Petlyuk configuration, consistingof a pre-fractionator and a main distillation column. A vertical wallis inserted in a DWC at a position such that it splits the columnshell into two sections [23,24]. The feed stream is introduced intothe pre-fractionator (feed side of the column) while a side streamis removed from the main column (side stream section). The light-est component goes overhead as distillate and the heaviest compo-nent goes out as bottoms, while the side stream contains only theintermediate boiling component. At the top of the wall the refluxedliquid is split between the two sides of the column, while at thebottom the vapor flow is split according to the pressure drop on

Oil/fat

Alcohol/catalyst Separator

Wash column

Reactor

Dryer

Alcohol

Glycerol

Water

Biodiesel

Biodiesel

Glycerol recovery column

Glycerol + Water

Glycerol + Alcohol

+ Water

Glycerol + Alcohol

+ WaterAlcohol recovery column

Water

Fig. 1. Process flow diagram of a conventional biodiesel production process based on trans-esterification of tri-glycerides.

A.A. Kiss, R.M. Ignat / Applied Energy 99 (2012) 146–153 147

both sides of the wall (Fig. 2 btm). The DWC technology is veryappealing to industry as it confers major benefits over conven-tional distillation: up to 30% reduced investment costs and up to40% energy savings [24]. Remarkable, these DWC benefits are notlimited to ternary separations alone, but they can be present alsoin azeotropic and extractive distillation [26–28], as well as reactivedistillation [27–31].

Thermally coupled distillation columns and DWC were re-cently proposed for the biodiesel production, but mainly for thesynthesis of FAME by esterification of FFA in reactive distillationsystems [30,32,33] and not for the downstream processing steps.In order to reduce the number of distillation equipment units andthe energy requirements for the current biodiesel purification,this study proposes the use of a DWC for a single step methanolrecovery and glycerol separation. Aspen Plus and Aspen Dynamicswere used to perform rigorous steady-state and dynamic simula-tions, as well as the optimization of the new DWC separationalternative. In order to allow a fair comparison, all designs wereoptimized using the sequential quadratic programming (SQP)method.

Based on a solid theoretical and computational foundation, thesequential quadratic programming (SQP) method has become themost successful method for solving nonlinearly constrained opti-mization problems [34–39]. Nowadays, there are a large varietyof SQP techniques, from the initial approaches introduced in theearly 1960s, to the more state-of-the-art techniques, e.g. basedon the multiple shooting techniques. These SQP approaches havealready proved their advantages in many areas of applications inchemical engineering, such as model predictive control and dy-namic optimization, among others [40–46].

2. Problem statement

The biodiesel production by trans-esterification requires severalexpensive downstream processing steps such as the purification ofcrude glycerol, as well as the separation of excess methanol (recy-clable stream), glycerol by-product and water (from the washingand FFA pre-treatment steps). The problem is that the separationof the ternary mixture methanol–water–glycerol is still carried

out in a conventional direct sequence that requires two distillationcolumns (Fig. 2 top), a relatively large footprint and significantlyhigh amounts of energy. To solve this problem, this study proposesan efficient process intensification method for this ternary separa-tion, namely the use of a dividing-wall column (DWC) that is ableto separate all products at high purity, in only one equipment unit(Fig. 2 btm).

3. Results and discussion

In this study we consider the separation of the ternary mixturemethanol–water–glycerol as described in the literature [13–16].For the base case (direct distillation sequence) and the DWC alter-native the feed basis is 2900 kg/h – equivalent to a biodiesel pro-duction rate of about 100 ktpy. The target product purityconsidered for each product cut is min. 99.5 %wt. Note that thereactants and product amounts for trans-esterification are in themass ratio of 100:22:11 for biodiesel:methanol:glycerol, meaningthat 11 kg of glycerol is synthesized per each 100 kg of biodieselproduced [13]. Moreover, the amount of methanol introduced inthe reactor typically exceeds 50% of the required stoichiometric ra-tio, thus this must be recovered and recycled. It is also importantnoting that the composition of the feed mixture can vary, depend-ing on the type of the biodiesel production process – alkali or acidcatalyzed – and the amount of water produced in the FFA pre-treatment step and also used for the biodiesel washing [13–17].Typically, the weight amounts of methanol and glycerol are simi-lar, while water is present in lower quantities [13,16].

Steady-state simulations were carried out in AspenTech AspenPlus� using the rigorous RADFRAC unit enhanced with the RateSep(rate-based) model. The well-known MESH equations are govern-ing the process. Note that MESH is an acronym referring to the typeof equations: M – mass balance, E – equilibrium relationships, S –summation equations, and H – enthalpy balance. Due to the natureof the components involved in the separation, UNIQUAC was se-lected as the most accurate property model [30,9]. It is worth not-ing that the mixture does not form any azeotropes, and there is noliquid phase splitting possible at any given composition. Conse-quently, the residues curve maps and the ternary diagrams are

DC 1 DC 2

WaterMethanol

Glycerol

Feed

DC1-Btm

Methanol

Glycerol

Water

Feed

rL

rV

DWC

Fig. 2. Methanol–water–glycerol separation in a direct sequence (top) and adividing-wall column (btm).

Table 1Design parameters of an optimal direct sequence of two columns.

Design parameters Value Unit

Flowrate of feed stream 2900 kg/hFeed composition (mass fractions)Methanol:Water:Glycerol 0.473:0.054:0.473 –Temperature of feed stream 60 �CPressure of feed stream 1.2 barOperating pressure (DC1) 0.5 barOperating pressure (DC2) 0.5 barColumn diameter (DC1) 0.9 mColumn diameter (DC2) 0.4 mTotal number of stages (DC1) 14 –Total number of stages (DC2) 12 –Feed stage (DC1) 9 –Feed stage (DC2) 7 –Reflux ratio (DC1) 1.4 kg/kgReflux ratio (DC2) 0.2 kg/kgDistillate to feed ratio (DC1) 0.473 kg/kgDistillate to feed ratio (DC2) 0.1 kg/kgMethanol product purity 99.9/99.9 %wt/%molWater product purity 99.6/99.8 %wt/%molGlycerol product purity 99.9/99.9 %wt/%molReboiler duty DC1 1080 kWCondenser duty DC1 �1029 kWReboiler duty DC2 250 kWCondenser duty DC2 �120 kW

148 A.A. Kiss, R.M. Ignat / Applied Energy 99 (2012) 146–153

not included here since they do not provide any added valueinformation.

Both configurations (base case and DWC alternative) describedhereafter were optimized in terms of minimal energy demandusing the sensitivity analysis and sequential quadratic program-ming (SQP) tools from Aspen Plus [47,39]. Note that the SQP opti-mization method implemented in Aspen Plus is a state-of-the-art,quasi-Newton nonlinear programming algorithm – while the SQP-Biegler variant is the SQP implementation developed by Biegleret al. [48,49]. The default optimization procedure in Aspen Plus isto converge tear streams and the optimization problem simulta-neously, using the SQP method. When the SQP method is used toconverge tears and optimization problems simultaneously, thealgorithm is a hybrid of an infeasible path method where the tearsare not converged at each iteration but are converged at the opti-mum, and a feasible path method where the tears are converged ateach iteration of the optimization [47]. In practice, minimizing thetotal heat duty of the distillation sequence, constrained by the re-quired purity and recovery of the products, was carried out com-bining the sensitivity analysis and SQP tools from Aspen Plus.Several optimization variables were used: total number of stages,feed-stage location, side-stream location, location and length ofdividing wall, reflux ratio, liquid and vapor split [27,28].

3.1. Direct sequence

Fig. 2 (top) shows the direct distillation sequence as the conven-tional industrial practice for methanol recovery and glycerol sepa-ration [13–16]. Table 1 provides the optimal design parameters forthis base case, while Fig. 3 conveniently plots the temperature andliquid composition profiles. Methanol is collected as top distillatefrom the first distillation column (DC1), while the bottom product– consisting of water and glycerol – is fed to the second distillationcolumn (DC2). The recovered methanol is typically recycled back tothe trans-esterification reactor. High purity glycerol is obtained asbottom product, and water as distillate product of DC2, respec-tively. Both distillation columns are operated at 0.5 bar in orderto keep the temperature in the reboiler sufficiently low to preventthe degradation of glycerol.

3.2. DWC alternative

The DWC was designed according to the rules described in theliterature [50,23,51,24,52]. In order to design of the most energyefficient DWC we use the concept of minimum energy or minimumvapor flow rate (Vmin). Plotting the Vmin diagrams pioneered byHalvorsen and Skogestad [53,52] contributes to the understandingof the optimal operating point with the minimum energy require-ments. Sections with infinite number of stages are assumed, thusthe numbers given in the Vmin diagram are independent of the de-tailed stage design. In practice, the real number of stages is se-lected based on the required separation purity, while taking intoaccount the trade-off between the equipment cost and the energycost. Clearly, the real energy input is slightly higher than the min-imum requirements. Nevertheless, Vmin is a straightforward andcommon measure used to compare the energy consumption in dif-ferent distillation arrangements [53,52]. Fig. 4 presents the Vmindiagram for the ternary mixture methanol–water–glycerol – notedhere as A, B, C for convenience. The minimum amount of energy re-quired for the separation in a DWC is given by the highest peak inthe diagram (PA/B).

A sequence of two rigorous RADFRAC units was used to modelthe DWC as no off-the-shelf DWC unit is available in the currentcommercial process simulators. This configuration is thermallyequivalent to a DWC, as long as the temperature difference on both

DC2

DC1

Fig. 3. Temperature and composition profile in DC1 and DC2 units of the directdistillation sequence.

Fig. 4. Vmin diagram of the methanol–water–glycerol system.

Fig. 5. Temperature and composition profile in the dividing-wall column.

A.A. Kiss, R.M. Ignat / Applied Energy 99 (2012) 146–153 149

sides of the wall indicates that there is no heat transfer betweenthe two sides. The SQP optimization method and the efficient sen-sitivity analysis tool from Aspen Plus were also employed. Theobjective of the optimization is to minimize the total reboiler dutyrequired, as follows:

MinðQÞ ¼ f ðNT;NF;NSS;NDWS;NDWC;V ;RR; rV; rLÞSubject to ~ym P~xm

ð1Þ

where the optimization parameters used here are: total number ofstages (NT), feed location (NF), side-draw stage (NSS), wall size(NDWS) and location (NDWC), boilup rate (V), reflux ratio (RR), liquid

and vapor split (rL and rV), while ym and xm are the vectors of theobtained and required purities for the m products [27,28]. Note thatin order to determine the optimal ratio between the energy cost andthe number of stages, an additional objective function was used,Min NT (RR + 1), which approximates very well the minimum of to-tal annualized cost of a conventional distillation column [54].

Since both distillation columns of the direct sequence (Fig. 2top) are operated at the same pressure, DWC (Fig. 2 btm) can bean attractive alternative for the industrial implementation. Metha-nol is recovered as top distillate, glycerol as bottom product andwater is withdrawn as side stream of the main column. As startingpoint of the DWC simulation, the results obtained from the directsequence Aspen Plus model were used. In fact, these results

Table 2Design parameters of an optimal dividing-wall column.

Design parameters Value Unit

Flowrate of feed stream 2900 kg/hFeed composition (mass fractions)Methanol:Water:Glycerol 0.473:0.054:0.473 –Temperature of feed stream 60 �CPressure of feed stream 1.2 barOperating pressure 0.5 barColumn diameter 1.1 mNumber of stages pre-fractionator side 10 –Total number of stages DWC 30 –Feed stage pre-fractionator 7 –Side stream withdrawal stage 22 –Wall position (from/to stage) 15–25 –Distillate to feed ratio 0.473 kg/kgReflux ratio 0.83 kg/kgLiquid split ratio (rL) 0.42 kg/kgVapor split ratio (rV) 0.25 kg/kgMethanol product purity 99.9/99.8 %wt/%molWater product purity 99.6/99.8 %wt/%molGlycerol product purity 99.9/99.9 %wt/%molReboiler duty 975 kWCondenser duty �793 kW

Methanol

Glycerol

Water

Feed

rL

rV

DWC

LC

CC

CCCC

YG

CC

LC

CCCC

Fig. 6. Control structure of the DWC unit (4-point control: DB/LSV + rL).

150 A.A. Kiss, R.M. Ignat / Applied Energy 99 (2012) 146–153

provide the initial estimates for the total number of trays, the feedtray location, as well as the liquid and vapor split. The optimal de-sign of the DWC was obtained according to the method describedearlier. Fig. 5 plots the temperature and liquid composition profilesin the DWC unit, while the key parameters of the optimal designare presented in Table 2. Remarkable, the temperature differencebetween the two sides of the wall is very low – less than �25 �C– such conditions being easily achievable in the practical applica-tion with little heat transfer expected and negligible effect on thecolumn performance [23,24].

Table 3Tuning parameters of the PID controllers.

Controlled variable Manipulated variable Gain P (%/%)

xmethanol (in D) L 0.8xwater (in S) V 0.6xglycerol (in B) S 0.6yG (at top PF) rL 1Level reflux drum D 1Level reboiler B 1

3.3. Dynamics and control

DWC technology offers indeed key economic benefits, but thecontrollability of the process is just as important as the savingsin the capital and operating costs. Fig. 6 illustrates the proposedcontrol structure: two multi loops are needed to stabilize the col-umn and another three to maintain the set points specifying theproduct purities. The regulatory control focuses on maintainingthe specified product purities, while the inventory control aims tostabilize the column and to control the level in the reflux tankand the level in the reboiler. The level of the reflux drum and thereboiler can be controlled by the manipulated variables D (distil-late), and B (bottoms), respectively. The composition of the freeproduct streams is controlled by the remaining variables: L (liquidreflux), S (side-stream) and V (vapor boil-up). Note that the controlstructure DB/LSV was reported as the best option for a DWC in ourprevious studies [55,56,7,8].

An additional optimization loop is used to manipulate the liquidsplit, in order to control the heavy component composition in thetop of pre-fractionator (PF), and implicitly achieving minimizationof the energy requirements [53,52,57,56,58]. Note that any heavycomponent (glycerol) going out the top of the wall will end up alsoin the liquid flowing down in the main column and thus stronglyaffecting the purity of the side stream (S). Since the side streamis collected as a liquid product, it means that any small amountsof light impurity in the vapor phase will not significantly affectits composition. However, even tiny amounts of heavy impurityin the liquid phase will greatly affect the composition of the sidestream.

At the bottom of the dividing-wall section, the vapor flow issplit proportionally to the cross-sectional area of each side andthe hydrodynamic conditions (e.g. flow resistance). The cross-sec-tional area of each side is fixed by the physical location of the wall,and this is already set at the design stage hence it cannot be chan-ged later on, during operation [23]. Because the location of the wallfixes how the vapor flow splits between the two sides of the col-umn, the vapor split (rV) variable is not adjustable during operationfor control purposes [7,8]. Therefore, the vapor split ratio is notused as a manipulated variable in the dynamic simulations pre-sented hereafter.

Table 3 lists the tuning parameters of the PID controllers. Thecontrol loops were tuned by a simple version of the direct synthe-sis method [59]. Fig. 7 presents the results of the dynamic simula-tions for various industrially relevant disturbances in the set point(SP), the feed flow rate (F) and feed composition. The mass frac-tions of all components are returning to their set point within rea-sonable short times, thus proving that the system can successfullyreject the disturbances. Moreover, the dynamic response of theproposed control structure is clearly characterized by low over-shooting and short settling times.

3.4. Process comparison

In order to perform a fair comparison of the two process alter-natives, the total investment cost (TIC), total operating cost (TOC)

Int. time I (min) Drv. time D (min) Control direction

40 0 �20 0 �40 0 �20 0 +

100 0 +100 0 +

MethanolWaterGlycerol

Fig. 7. Results of the dynamic simulations at various disturbances in the set point(change from 0.995 to 0.999), feed flowrate (+10% F) and feed composition(±10% �water).

Table 4Head-to-head comparison of conventional vs. DWC alternative.

Key performance indicators Conventionalprocess

DWCalternative

Total investment cost (TIC) $563,429 $499,087Total operating costs (TOCs) $280,491 $208,402Total annual costs (TACs) $336,834 $258,310Specific energy requirements

(kW h/ton glycerol)967.5 709.3

CO2 emissions (kg CO2/h ton glycerol) 135.31 99.19

A.A. Kiss, R.M. Ignat / Applied Energy 99 (2012) 146–153 151

and total annual cost (TAC) were calculated. The equipment costsare estimated using correlations from the Douglas textbook tothe price level of 2010, as described by Dejanovic et al. [54]. TheMarshall and Swift equipment cost index (MSI) considered herehas a value of 1468.6. For a carbon steel column, the estimated costin US$ is given by the relation:

Cshell ¼ fpðMSI=280Þd1:066c h0:802

c ð2Þ

where fp is the cost factor (equal to 2981.68 in this case), dc is thecolumn diameter (calculated using the internals-sizing procedurefrom Aspen Plus) and hc its height (tangent-to-tangent) consideringa tray-spacing of 0.6 m. For heat exchangers (e.g. condensers andreboilers) the next expression was used to calculate the equipmentcost (US$):

Chex ¼ ðMSI=280ÞcxA0:65 ð3Þ

where cx = 1609.13 for condensers and 1775.26 for kettle reboilers,while A is the heat transfer area (m2). Also, a price of 600 US $/m2

was used for the sieve trays cost calculations.The following utility costs were considered: US $0.03/t cooling

water and US $13/t steam. For the TAC calculations, a plant lifetimeof 10 years was considered. While the accuracy of the correlationsis in the range of acceptable and realistic ±30%, this level of accu-racy is less important when comparing design alternatives sincethe error is consistent in all cases.

Table 4 provides a head-to-head comparison of the key perfor-mance economic indicators, while Fig. 8 conveniently illustratesthe costs and the CO2 emissions of the two processes considered.Remarkable, the DWC alternative is the most efficient in terms ofenergy requirements allowing energy savings of 27% while alsobeing the least expensive in terms of capital investment and oper-ating costs, leading to 25% lower total annual costs.

The energy requirements are closely linked to the CO2 emis-sions, but only when no heat integration is considered. When partof the process heat is reused instead of primary energy, then theCO2 emissions are lower as compared to the figure expected fromthe energy data. The CO2 emissions were calculated according tomethod described by Gadalla et al. [60]:

½CO2�emissions ¼ ðQ fuel=NHVÞðC%=100Þa ð4Þ

where a = 3.67 is the ratio of molar masses of CO2 and C, NHV is thenet heating value, and C% is the carbon content – dependent on thefuel. For natural gas, NHV is 48,900 kJ/kg and the carbon content is0.41 kg/kg. Hence the amount of fuel used can be calculated asfollows:

Q fuel ¼ Q proc=kproc � ðhproc � 419ÞðTFTB � T0Þ=ðTFTB � TstackÞ ð5Þ

where kproc (kJ/kg) and hproc (kJ/kg) are the latent heat and enthalpyof the steam, TFTB (K) and Tstack (K) are the flame and stack temper-ature, respectively.

The hourly rate of CO2 emissions for the conventional and DWC

Fig. 8. Comparison of conventional and DWC alternatives in terms of keyperformance indicators: CO2 emissions, total investment, operating and annualcosts.

152 A.A. Kiss, R.M. Ignat / Applied Energy 99 (2012) 146–153

alternatives are 183.19 kg/h and 136.34 kg/h, respectively. Table 4also lists the specific amount of CO2 emissions per ton of glycerol.As these emissions are closely linked to the amount of energy re-quired, it comes as no surprise that the DWC alternative exhibitsalso the lowest carbon footprint.

4. Conclusions

This study successfully demonstrated a novel application ofDWC technology for a single step methanol recovery and glycerolseparation in biodiesel production. Compared to an optimized con-ventional direct sequence of two distillation columns, the novelproposed DWC alternative reduces the energy requirements by27% and the equipment costs by 12%, thus leading to 25% savingsin the total annual costs. Notably, the novel separation scheme alsorequires less equipment units and reduced plant footprint. Basedon these results, the use of dividing-wall column in biodiesel pro-duction is especially interesting in case of building new large bio-diesel plants, but also in the case of revamping existing plants –where the equivalent Petlyuk configuration should be certainlyconsidered.

Acknowledgments

The financial support from the Sector Operational Program Hu-man Resources Development 2007-2013 of the Romanian Ministryof Labor, Family and Social Protection through the Financial Agree-ment POSDRU/88/1.5/S/61178 is gratefully acknowledged.

References

[1] Van Gerpen J. Biodiesel processing and production. Fuel Process Technol2005;86:1097–107.

[2] Bowman M, Hilligoss D, Rasmussen S, Thomas R. Biodiesel: a renewable andbiodegradable fuel. Hydrocarbon Process 2006;85:103–6.

[3] Lam MK, Lee MT, Mohamed AR. Homogeneous, heterogeneous and enzymaticcatalysis for transesterification of high free fatty acid oil (waste cooking oil) tobiodiesel: a review. Biotechnol Adv 2010;28:500–18.

[4] Leung DYC, Wu X, Leung MKH. A review on biodiesel production usingcatalyzed transesterification. Appl Energy 2010;87:1083–95.

[5] Kiss A. Separative reactors for integrated production of bioethanol andbiodiesel. Comput Chem Eng 2010;34:812–20.

[6] Kiss AA. Heat-integrated reactive distillation process for synthesis of fattyesters. Fuel Process Technol 2011;92:1288–96.

[7] Kiss AA, Bildea CS. Integrated reactive absorption process for synthesis of fattyesters. Bioresour Technol 2011;102:490–8.

[8] Kiss AA, Bildea CS. A control perspective on process intensification in dividing-wall columns. Chem Eng Process 2011;50:281–92.

[9] Kiss AA, Bildea CS. A review on biodiesel production by integrated reactiveseparation technologies. J Chem Technol Biotechnol 2012;87. http://dx.doi.org/10.1002/jctb.378.

[10] Meher LC, Vidya Sagar D, Naik S. Technical aspects of biodiesel production bytransesterification: a review. Renew Sustain Energy Rev 2006;10:248–68.

[11] Demirbas A. Comparison of transesterification methods for production ofbiodiesel from vegetable oils and fats. Energy Convers Manage2008;49:125–30.

[12] Qiu ZY, Zhao LN, Weather L. Process intensification technologies in continuousbiodiesel production. Chem Eng Process 2010;49:323–30.

[13] Dunford NT. Biodiesel production techniques, Robert M. Kerr Food &agricultural products center. Food technology fact sheet. FAPC-150; 2007. p.1–4.

[14] Fjerbaek L, Christensen KV, Norddahl B. A review of the current state ofbiodiesel production using enzymatic transesterification. Biotechnol Bioeng2009;102:1298–315.

[15] Helwani Z, Othman MR, Aziz N, Fernando WJN, Kim J. Technologies forproduction of biodiesel focusing on green catalytic techniques. Fuel ProcessTechnol 2009;90:1502–14.

[16] Chemstations Inc. Biodiesel in ChemCAD. White paper; 2010. p. 1–7.[17] Atadashi IM, Aroua MK, Abdul Aziz AR, Sulaiman NMN. Refining technologies

for the purification of crude biodiesel. Appl Energy 2011;88:4239–51.[18] Balat M, Balat H. Progress in biodiesel processing. Appl Energy

2010;87:1815–35.[19] Lin L, Zhou C, Saritporn V, Shen X, Dong M. Opportunities and challenges for

biodiesel fuel. Appl Energy 2011;88:1020–31.[20] Olujic Z, Jodecke M, Shilkin A, Schuch G, Kaibel B. Equipment improvement

trends in distillation. Chem Eng Process 2009;48:1089–104.[21] Asprion N, Kaibel G. Dividing wall columns: fundamentals and recent

advances. Chem Eng Process 2010;49:139–46.[22] Harmsen J. Process intensification in the petrochemicals industry: drivers

and hurdles for commercial implementation. Chem Eng Process 2010;49:70–3.[23] Dejanovic I, Matijaševic L, Olujic Z. Dividing wall column – a breakthrough

towards sustainable distilling. Chem Eng Process 2010;49:559–80.[24] Yildirim O, Kiss AA, Kenig EY. Dividing wall columns in chemical process

industry: a review on current activities. Sep Purif Technol 2011;80:403–17.[25] Maleta VN, Kiss AA, Taran VM, Maleta BV. Understanding process

intensification in cyclic distillation systems. Chem Eng Process2011;50:655–64.

[26] Bravo-Bravo C, Segovia-Hernández JG, Gutiérrez-Antonio C, Duran AL, Bonilla-Petriciolet A, Briones-Ramírez A. Extractive dividing wall column: design andoptimization. Ind Eng Chem Res 2010;49:3672–88.

[27] Kiss AA, Suszwalak D. J-P. C., Enhanced bioethanol dehydration by extractiveand azeotropic distillation in dividing-wall columns. Sep Purif Technol2012;86:70–8.

[28] Kiss AA, Suszwalak DJ-PC. Innovative dimethyl ether synthesis in a reactivedividing-wall column. Comput Chem Eng 2012;38:74–81.

[29] Kiss AA, Pragt H, van Strien C. Reactive dividing-wall columns – how to getmore with less resources? Chem Eng Commun 2009;196:1366–74.

[30] Kiss AA, Segovia-Hernandez JG, Bildea CS, Miranda-Galindo EY, Hernandez S.Reactive DWC leading the way to FAME and fortune. Fuel 2012;95:352–9.

[31] Hernandez S, Sandoval-Vergara R, Barroso-Munoz FO, Murrieta-Duenasa R,Hernandez-Escoto H, Segovia-Hernandez JG, et al. Reactive dividing walldistillation columns: simulation and implementation in a pilot plant. ChemEng Process 2009;48:250–8.

[32] Gomez-Castro FI, Rico-Ramirez V, Segovia-Hernandez JG, Hernandez S.Feasibility study of a thermally coupled reactive distillation process forbiodiesel production. Chem Eng Process 2010;49:262–9.

[33] Gomez-Castro FI, Rico-Ramirez V, Segovia-Hernandez JG, Hernandez-Castro S.Esterification of fatty acids in a thermally coupled reactive distillation columnby the two-step supercritical methanol method. Chem Eng Res Des2011;89:480–90.

[34] Wilson RB. A simplicial method for convex programming. PhD dissertation,Harvard University; 1963.

[35] Murray W. An algorithm for constrained minimization. In: Fletcher R, editor.Optimization. London and New York: Academic Press; 1969. p. 247–58.

A.A. Kiss, R.M. Ignat / Applied Energy 99 (2012) 146–153 153

[36] Gill PE, Murray W, Wright MH. Practical optimization. London: AcademicPress; 1981.

[37] Powell MJD. Variable metric methods for constrained optimization. In:Bachem A, Grotschel M, Korte B, editors. Mathematical programming: thestate of the art. Springer Verlag; 1983. p. 288–311.

[38] Boggs PT, Tolle JW. Sequential quadratic programming. Acta Numerica1995;4:1–51.

[39] Bartholomew-Biggs M. Nonlinear optimization with engineering applications.Springer Optimiz Appl 2008;19:1–14.

[40] Storen S, Hertzberg T. The sequential linear-quadratic programming algorithmfor solving dynamic optimization problems – a review. Comput Chem Eng1995;19:S495–500.

[41] Nagy ZK, Mahn B, Franke R, Allgower F. Efficient output feedback nonlinearmodel predictive control for temperature control of industrial batch reactors.Control Eng Pract 2007;15:839–59.

[42] Nagy ZK. Model based robust control approach for batch crystallizationproduct design. Comput Chem Eng 2009;33:1685–91.

[43] Simon LL, Nagy ZK, Hungerbuehler K. Model based control of a liquid swellingconstrained batch reactor subject to recipe uncertainties. Chem Eng J2009;153:151–8.

[44] Roman R, Nagy ZK, Cristea MV, Agachi SP. Dynamic modelling and nonlinearmodel predictive control of a fluid catalytic cracking unit. Comput Chem Eng2009;33:605–17.

[45] Mesbah A, Huesman AEM, Kramer HJM, Nagy ZK, van den Hof PMJ. Real-timecontrol of a semi-industrial fed-batch evaporative crystallizer using differentdirect optimization strategies. AIChE J 2011;57:1557–69.

[46] Alsumait JS, Sykulski JK, Al-Othman AK. A hybrid GA–PS–SQP method to solvepower system valve-point economic dispatch problems. Appl Energy2010;87:1773–81.

[47] Aspen Technology. Aspen Plus: user guide, vols. 1 & 2; 2010.

[48] Biegler LT, Cuthrell JE. Improved infeasible path optimization for sequentialmodular simulators. Part II: the optimization algorithm. Comput Chem Eng1985;9:257–67.

[49] Lang YD, Biegler LT. A unified algorithm for flowsheet optimization. ComputChem Eng 1987;11:143–58.

[50] Rangaiah GP, Ooi EL, Premkumar R. A simplified procedure for quick design ofdividing-wall columns for industrial applications. Chem Prod Process Model2009;4(1) [article 7].

[51] Rong BG. Synthesis of dividing-wall columns (DWC) for multicomponentdistillations – a systematic approach. Chem Eng Res Des 2011;89:1281–94.

[52] Halvorsen IJ, Skogestad S. Energy efficient distillation. J Nat Gas Sci Eng2011;3:571–80.

[53] Halvorsen IJ, Skogestad S. Optimizing control of Petlyuk distillation:understanding the steady-state behaviour. Comput Chem Eng1997;21:249–54.

[54] Dejanovic I, Matijaševic L, Olujic Z. An effective method for establishing thestage and reflux requirement of three-product dividing wall columns. ChemBiochem Eng Q 2011;25:147–57.

[55] Diggelen van RC, Kiss AA, Heemink AW. Comparison of control strategies fordividing-wall columns. Ind Eng Chem Res 2010;49:288–307.

[56] Kiss AA, Rewagad RR. Energy efficient control of a BTX dividing-wall column.Comput Chem Eng 2011;35:2896–904.

[57] Ling H, Luyben WL. New control structure for divided-wall columns. Ind EngChem Res 2009;48:6034–49.

[58] Rewagad RR, Kiss AA. Dynamic optimization of a dividing-wall column usingmodel predictive control. Chem Eng Sci 2012;68:132–42.

[59] Luyben WL, Luyben ML. Essentials of process control. New-York,USA: McGraw-Hill; 1997.

[60] Gadalla M, Olujic Z, de Rijke A, Jansens PJ. Reducing CO2 emissions of internallyheat-integrated distillation columns for separation of close boiling mixtures.Energy 2006;31:2409–17.