Assessment of four biodiesel production processes using HYSYS.Plant Alex H. West, Dusko Posarac, Naoko Ellis * Department of Chemical and Biological Engineering, The University of British Columbia, 2360 East Mall, Vancouver, BC, Canada V6T 1Z3 Received 24 February 2006; received in revised form 31 October 2007; accepted 3 November 2007 Available online 29 January 2008 Abstract Four continuous biodiesel processes were designed and simulated in HYSYS. The first two employed traditional homogeneous alkali and acid catalysts. The third and fourth processes used a heterogeneous acid catalyst and a supercritical method to convert a waste veg- etable oil feedstock into biodiesel. While all four processes were capable of producing biodiesel at high purity, the heterogeneous and supercritical processes were the least complex and had the smallest number of unit operations. Material and energy flows, as well as sized unit operation blocks, were used to conduct an economic assessment of each process. Total capital investment, total manufacturing cost and after tax rate-of-return were calculated for each process. The heterogeneous acid catalyst process had the lowest total capital invest- ment and manufacturing costs, and had the only positive after tax rate-of-return. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Biodiesel; Process design; Process simulation; Economic assessment 1. Introduction and background Recent concerns over diminishing fossil fuel supplies and rising oil prices, as well as adverse environmental and human health impacts from the use of petroleum fuel have prompted considerable interest in research and devel- opment of fuels from renewable resources, such as biodiesel and ethanol. Biodiesel is a very attractive alternative fuel, as it is derived from a renewable, domestic resource and can, therefore, reduce reliance on foreign petroleum imports. Biodiesel reduces net carbon dioxide emissions by 78% on a life-cycle basis when compared to conven- tional diesel fuel (Tyson, 2001). It has also been shown to have dramatic improvements on engine exhaust emissions. For instance, combustion of neat biodiesel decreases car- bon monoxide (CO) emissions by 46.7%, particulate matter emissions by 66.7% and unburned hydrocarbons by 45.2% (Schumacher et al., 2001). Additionally, biodiesel is biode- gradable and non-toxic, making it useful for transportation applications in highly sensitive environments, such as mar- ine ecosystems and mining enclosures. As shown in Eq. (1), biodiesel (alkyl ester) is usually produced by the transesterification of a lipid feedstock. Transesterification is the reversible reaction of a fat or oil (which is composed of tri-glycerides) with an alcohol to form fatty acid alkyl esters and glycerol. Stoichio- metrically, the reaction requires a 3:1 molar alcohol-to-oil ratio, but excess alcohol is usually added to drive the equi- librium toward the products side CH 2 –OOC–R 1 R 1 –COO–R’ CH 2 –OH | Catalyst | CH–OOC–R 2 + 3R’OH ⇔ R 2 –COO–R’ + CH–OH | | CH 2 –OOC–R 3 R 3 –COO–R’ CH 2 –OH ð1Þ Transesterification can be alkali-, acid- or enzyme-cata- lyzed; however, enzyme-catalysts are rarely used, as they are less effective (Ma and Hanna, 1999). The reaction can also take place without the use of a catalyst under 0960-8524/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2007.11.046 * Corresponding author. Tel.: +1 604 822 1243; fax: +1 604 822 6003. E-mail address: [email protected](N. Ellis). Available online at www.sciencedirect.com Bioresource Technology 99 (2008) 6587–6601
Transcript
Available online at www.sciencedirect.com
Bioresource Technology 99 (2008) 6587–6601
Assessment of four biodiesel production processes using HYSYS.Plant
Alex H. West, Dusko Posarac, Naoko Ellis *
Department of Chemical and Biological Engineering, The University of British Columbia, 2360 East Mall, Vancouver, BC, Canada V6T 1Z3
Received 24 February 2006; received in revised form 31 October 2007; accepted 3 November 2007Available online 29 January 2008
Abstract
Four continuous biodiesel processes were designed and simulated in HYSYS. The first two employed traditional homogeneous alkaliand acid catalysts. The third and fourth processes used a heterogeneous acid catalyst and a supercritical method to convert a waste veg-etable oil feedstock into biodiesel. While all four processes were capable of producing biodiesel at high purity, the heterogeneous andsupercritical processes were the least complex and had the smallest number of unit operations. Material and energy flows, as well as sizedunit operation blocks, were used to conduct an economic assessment of each process. Total capital investment, total manufacturing costand after tax rate-of-return were calculated for each process. The heterogeneous acid catalyst process had the lowest total capital invest-ment and manufacturing costs, and had the only positive after tax rate-of-return.� 2007 Elsevier Ltd. All rights reserved.
Keywords: Biodiesel; Process design; Process simulation; Economic assessment
1. Introduction and background
Recent concerns over diminishing fossil fuel suppliesand rising oil prices, as well as adverse environmentaland human health impacts from the use of petroleum fuelhave prompted considerable interest in research and devel-opment of fuels from renewable resources, such as biodieseland ethanol. Biodiesel is a very attractive alternative fuel,as it is derived from a renewable, domestic resource andcan, therefore, reduce reliance on foreign petroleumimports. Biodiesel reduces net carbon dioxide emissionsby 78% on a life-cycle basis when compared to conven-tional diesel fuel (Tyson, 2001). It has also been shown tohave dramatic improvements on engine exhaust emissions.For instance, combustion of neat biodiesel decreases car-bon monoxide (CO) emissions by 46.7%, particulate matteremissions by 66.7% and unburned hydrocarbons by 45.2%(Schumacher et al., 2001). Additionally, biodiesel is biode-gradable and non-toxic, making it useful for transportation
0960-8524/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.
applications in highly sensitive environments, such as mar-ine ecosystems and mining enclosures.
As shown in Eq. (1), biodiesel (alkyl ester) is usuallyproduced by the transesterification of a lipid feedstock.Transesterification is the reversible reaction of a fat or oil(which is composed of tri-glycerides) with an alcohol toform fatty acid alkyl esters and glycerol. Stoichio-metrically, the reaction requires a 3:1 molar alcohol-to-oilratio, but excess alcohol is usually added to drive the equi-librium toward the products side
CH2–OOC–R1 R1–COO–R’ CH2–OH
| Catalyst |
CH–OOC–R2 + 3R’OH ⇔ R2–COO–R’ + CH–OH
| |
CH2–OOC–R3 R3–COO–R’ CH2–OH
ð1Þ
Transesterification can be alkali-, acid- or enzyme-cata-lyzed; however, enzyme-catalysts are rarely used, as theyare less effective (Ma and Hanna, 1999). The reactioncan also take place without the use of a catalyst under
ATROR after tax rate-of-returnCAC auxiliary facility costCBM bare module capital costsFBM bare module factorB1, B2 bare module factor parametersA capacity parameterCCF contingency feeCFC fixed capital costFFA free fatty acidFM materials factor
NRTL non-random two liquidFP pressure factorCp purchase costK1, K2 purchase cost parametersTCI, CTCI total capital investmentsTMC total manufacturing costCTM total module costwt.% weight per centCWC working capital cost
6588 A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601
conditions in which the alcohol is in a supercritical state(Demirbas, 2002; Saka and Kusdiana, 2001).
Currently, the high cost of biodiesel production is themajor impediment to its large-scale commercialization(Canakci and Van Gerpen, 2001). The high cost is largelyattributed to the cost of virgin vegetable oil as feedstock.Exploring methods to reduce the production cost of biodie-sel has been the focus of much recent research. One methodinvolves replacing a virgin oil feedstock with a waste cook-ing oil feedstock. The costs of waste cooking oil are esti-mated to be less than half of the cost of virgin vegetableoils (Canakci and Van Gerpen, 2001). Furthermore, utiliz-ing waste cooking oil has the advantage of removing a sig-nificant amount of material from the waste stream – as of1990, it was estimated that at least 2 billion pounds ofwaste grease was produced annually in the United States(Canakci and Van Gerpen, 2001).
In the last few years, a number of new production meth-ods have emerged from laboratory/bench-scale researchaimed at reducing the cost of biodiesel (Canakci and VanGerpen, 2003; Delfort et al., 2003; Demirbas, 2002). Onesuch method uses alcohol in its supercritical state, andeliminates the need for a catalyst. Additionally, the super-critical process requires only a short residence time to reachhigh conversion (Kusdiana and Saka, 2004). Anotheroption is to use a solid catalyst (Abreu et al., 2005; Furutaet al., 2004; Suppes et al., 2004) which simplifies down-stream purification of the biodiesel. The catalyst can beseparated by physical methods such as a hydrocyclone inthe case where a multiphase reactor is used. Alternatively,a fixed bed reactor would eliminate the catalyst removalstep entirely.
Zhang et al. (2003a) developed a HYSYS-based processsimulation model to assess the technological feasibility offour biodiesel plant configurations – a homogeneousalkali-catalyzed pure vegetable oil process; a two step pro-cess to treat waste vegetable oil; a single step homogeneousacid-catalyzed process to treat waste vegetable oil; and ahomogeneous acid-catalyzed process using hexane extrac-tion to help purify the biodiesel. All four configurationswere deemed technologically feasible, but a subsequenteconomic analysis of the four designs revealed that the
one step acid-catalyzed process was the most economicallyattractive (Zhang et al., 2003b). Haas et al. (2006) devel-oped a versatile process simulation model to estimate bio-diesel production costs; however, the model was limited toa traditional alkali-catalyzed production method. In orderto determine whether the supercritical methanol or the het-erogeneous acid-catalyst process is a promising alternativeto the standard homogeneous catalytic routes, our aim wasto develop a process flowsheet and simulation, conduct aneconomic analysis of each process based on the materialand energy balance results reported by HYSYS, and carryout sensitivity analyses to optimize each process. Addition-ally, it was desired to automate the sizing and economiccalculations, whence they were incorporated into each sim-ulation by way of the spreadsheet tool available in HYSYS.The material and energy flows, as well as some unit param-eters were imported directly into the spreadsheet, therebyallowing the sizing and economic results to be updatedautomatically when any changes were made to the processflowsheet. Additional comparison is made to the simula-tion work by Zhang et al. (2003a) in order to ensure thatthe present simulations provide comparable results.
The homogeneous alkali-catalyzed system has been wellstudied, and optimum conditions at 1 atm pressure (60 �C,1 wt.% catalyst, six-to-one alcohol-to-oil molar ratio), areknown (Freedman et al., 1984). In order to prevent sapon-ification during the reaction, the free fatty acid (FFA) andwater content of the feed must be below 0.5 wt.% and0.05 wt.%, respectively (Freedman et al., 1984). Becauseof these limitations, only pure vegetable oil feeds areappropriate for alkali-catalyzed transesterification withoutextensive pre-treatment.
A homogeneous acid-catalyzed process can be employedto take advantage of cheaper feedstocks, such as wastecooking oil and animal-based tallow. The acid-catalyzedprocess can tolerate up to 5 wt.% FFA, but is sensitive towater content greater than 0.5 wt.%. The disadvantage ofthis method is that it is extremely slow at mild conditions:to achieve a 98% conversion, reaction for 48 h at 60 �C atan alcohol-to-oil ratio of 30:1 was required (Canakci andVan Gerpen, 1999). At higher temperatures and pressures(e.g., 100 �C and 3.5 bar) reaction times can be substan-
Table 1Catalysts and reaction parameters for heterogeneously catalyzed reactionsof soybean oil at 1 atm
Reaction parameters
Catalyst type Molarratio
Temperature(�C)
Conversion Time(h)
WO3/ZrO2 (Furuta et al.,2004)
40:1 >250 >90% 4
SO4/SnO2 (Furuta et al.,2004)
40:1 300 65% 4
SO4/ZrO2 (Furuta et al.,2004)
40:1 300 80% 4
SnO (Abreu et al., 2005) 4.15:1 60 94.7 3
A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601 6589
tially reduced (down to 8 h) to achieve similar conversion(99%) (Goff et al., 2004).
To utilize the advantages of the alkali- and acid-cata-lyzed processes, a process that combines both schemescan be created. FFA waste vegetable oil can be pre-treatedvia esterification with methanol in an acid-catalyzed envi-ronment to convert the FFA to methyl-ester. The acid-freeoil feedstock can then be treated by a base catalyst in themanner described above for the homogeneous alkali-cata-lyzed system. This process has the advantages that it canuse a cheaper feedstock such as waste vegetable oil, andemploy alkali-catalyzed transesterification which is fastand efficient (Canakci and Van Gerpen, 2001).
A process using a heterogeneous acid-catalyst is appeal-ing because of the ease of separation of a solid catalyst. Lote-ro et al. (2005) reports this advantage, coupled with theability of the acid functionality to process low cost, high freefatty acid feedstocks, will yield the most economical biodie-sel production method. As outlined in Table 1, a number ofsolid phase catalysts have been identified that hold potentialfor use. Research concerning heterogeneous catalysts is stillin the catalyst screening stage. Studies regarding reactionkinetics, as well as improving reaction parameters have yetto be conducted. In addition, studies to determine the effectsof free fatty acid concentration and water on the perfor-mance of the catalyst have been scarce.
Supercritical transesterification is also a potential alter-native to the standard homogenous catalytic routes. Super-critical transesterification using methanol has been shownto give nearly complete conversion in small amount of time(15 min) (Warabi et al., 2004). High temperatures (up to350 �C) and large alcohol-to-oil ratios (42:1) are requiredto achieve the high levels of conversion that have beenreported (Kusdiana and Saka, 2001). In addition to thehigh conversion and reaction rates, supercritical transeste-rification is appealing as it can tolerate feedstocks with veryhigh contents of FFAs and water, up to 36 wt.% and30 wt.%, respectively (Kusdiana and Saka, 2004).
2. Process simulation
To assess the technological feasibility and obtain mate-rial and energy balances for a preliminary economic analy-
sis, complete process simulations were performed. Despitesome expected differences between a process simulationand real-life operation, process simulators are commonlyused to provide reliable information on process operation,owing to their vast component libraries, comprehensivethermodynamic packages and advanced computationalmethods. HYSYS (HYprotech SYStem) Plant NetVer 3.2(ASPEN Tech, Cambridge MA) was used to conduct thesimulation. HYSYS was selected as a process simulatorfor both its simulation capabilities and its ability to incor-porate calculations using the spreadsheet tool. The firststeps in developing the process simulation were selectingthe chemical components for the process, as well as a ther-modynamic model. Additionally, unit operations and oper-ating conditions, plant capacity and input conditions mustall be selected and specified. The unit operations, plantcapacity and input conditions for the base cases, i.e.,homogeneous acid- and alkali-catalyzed processes, as wellas distillation column operating conditions, were selectedbased on the research done by Zhang et al. (2003a) toensure that each of the four processes simulated in thiswork could be compared in a consistent manner.
HYSYS library contained information for the followingcomponents used in the simulation: methanol, glycerol, sul-furic acid, sodium hydroxide, and water. Waste Canola oilwas represented by triolein and oleic acid. Accordingly,methyl-oleate, available in the HYSYS component library,was taken as the product of the transesterification reaction.Where a simulation required a feedstock with some amountof free fatty acids, oleic acid, also available in the HYSYSlibrary, was specified as the free fatty acid present.
Components not available in the HYSYS library werespecified using the ‘‘Hypo Manager” tool. Calcium oxide,calcium sulfate, phosphoric acid, sodium phosphate andtriolein were all specified in this manner. Specification ofa component requires input of a number of properties, suchas normal boiling point, density, molecular weight, as wellas the critical properties of the substance. Since triolein is acrucial component and is involved in operations requiringdata for liquid and vapour equilibria, great care was takenin specifying the values as accurately as possible. Values fordensity, boiling point, critical temperature, pressure andvolume were obtained from the ASPEN Plus componentlibrary and were input as 915 kg/m3, 846 �C, 1366 �C,470 kPa, 3.09 m3/kmol, respectively.
Owing to the presence of polar compounds such asmethanol and glycerol in the process, the non-randomtwo liquid (NRTL) thermodynamic/activity model wasselected for use as the property package for the simulation.Since some binary interaction parameters were not avail-able in the simulation databanks, they were estimated usingthe UNIFAC vapour–liquid equilibrium and UNIFACliquid–liquid equilibrium models where appropriate.
Plant capacity was specified at 8000 metric tonnes/yearbiodiesel (the same as in Zhang et al., 2003a). This trans-lated to vegetable oil feeds of roughly 1000 kg/h for eachprocess configuration.
6590 A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601
The process units common to all configurations includedreactors, distillation columns, pumps and heat exchangers.The homogeneous acid- and alkali-catalyzed processesincluded liquid–liquid extraction columns to separate thecatalyst and glycerol from the biodiesel. In contrast tothe base case scenarios, a gravity separation unit wasincluded in the supercritical methanol and heterogeneousacid-catalyst processes. In spite of the availability of kineticdata for the alkali-catalyzed, homogeneous acid-catalyzed,and supercritical processes (Freedman et al., 1986; Kusdi-ana and Saka, 2001), the reactors were modeled as conver-sion reactors since kinetic information for theheterogeneous acid-catalyzed process was unavailable.The reactors were assumed to operate continuously forall cases. Lab-scale reaction conditions and conversiondata available for all processes (Abreu et al., 2005; Warabiet al., 2004; Zhang et al., 2003a), were assumed to beappropriate for large-scale production, and set as the oper-ating conditions for each reactor. The following conver-sions were assumed for each process: 97%, 97%, 94% and98% for the alkali, acid, heterogeneous and supercriticalcases, respectively. The mono- and di-glyceride intermedi-ates were neglected during the reaction (Zhang et al.,2003a).
Multi-stage distillation was used to recover the excessmethanol, as well as in the final purification of biodiesel.Distillation columns were specified to meet or exceed theASTM standard for biodiesel purity, i.e., 99.65 wt.%.Reflux ratios for the heterogeneous acid-catalyzed andsupercritical cases were calculated by determining the min-imum reflux ratio using a shortcut distillation column, andthen multiplying by 1.5 to obtain the optimum reflux ratio(McCabe et al., 2001). The methanol recovery columnswere able to operate at ambient pressures (except in the
Fig. 1. (a) Pre-treated alkali-catalyzed process flowsheet. Pre-treatment compocomponent.
alkali-catalyzed case), while vacuum operation in themethyl-ester purification columns was necessary to keepthe temperatures of the distillate and bottoms streams atsuitably low levels, as biodiesel and glycerol are subjectto degradation at temperatures greater than 250 �C and150 �C, respectively (Goodrum, 2002; Newman, 1968).
3. Process design
Four continuous processes were simulated. The first wasbased on an alkali-catalyzed reaction to convert a wastevegetable oil feedstock containing 5 wt.% FFA that waspre-treated with an acid-catalyzed reaction (Fig. 1a). Thesecond was based on a homogeneous acid-catalyzed pro-cess using a waste cooking oil feedstock, containing5 wt.% free fatty acids. The third configuration employeda heterogeneous acid-catalyst, (tin(II) oxide), in a multi-phase reactor fed with waste vegetable oil, while the finalprocess used a supercritical methanol treatment of wastevegetable oil to produce biodiesel. Process flowsheets arepresented in Figs. 1–4.
The processes all followed the same general scheme. Thevegetable oil was transesterified in the first step, and thensent for downstream purification. Downstream purificationconsisted of the following steps: methanol recovery by dis-tillation; glycerol separation; catalyst neutralization andremoval (where appropriate) and methyl-ester purificationby distillation. Table 2 gives details for the unit operationsin each process. Tables 3–6 present the feed and productmaterial flow details for each process.
As illustrated in Table 2, there are also a number of keydifferences between the processes. The first difference iswith regards to the catalyst removal method. The solid cat-alyst in Process III is removed by a hydrocyclone before
nent. (b) Pre-treated alkali-catalyzed process flowsheet. Transesterification
Fig. 1 (continued)
Fig. 2. Homogeneous acid-catalyzed process flowsheet.
A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601 6591
methanol recovery; whereas the liquid phase catalyst inProcesses I and II is removed by washing the productstream with water in a liquid–liquid extraction column.
The acid-catalyst in Process II was removed as a solid pre-cipitate in separator X-100 after neutralization in reactorCRV-101. As in the homogeneous acid-catalyst process,
Fig. 3. Heterogeneous acid-catalyzed process flowsheet.
Fig. 4. Supercritical alcohol process flowsheet.
6592 A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601
the alkali-catalyst had to be neutralized before it could bedisposed of. The heterogeneous catalyst in Process IIIrequired no neutralization step and it was discarded as awaste product. However, it has the potential advantageof being recycled.
The second major difference is in the separation of glyc-erol from the biodiesel. In Processes I and II, glycerol isremoved by washing the product stream with water, andcollected in the bottoms product. In Processes III and IV,glycerol is separated from the biodiesel in a three-phase
separator by gravity settling. Krawczyk (1996) initially pro-posed gravity separation to remove glycerol; however,Zhang et al. (2003a) indicated from their simulation thatsatisfactory separation could not be achieved by gravityalone. In the present work, gravity separation was usedto separate the biodiesel from the glycerol, and a satisfac-tory separation was achieved. Note, however, that the cal-culations for this unit operation are based on parametersthat have been estimated by HYSYS and, therefore, maynot truly represent a real system. Additional experimental
Table 2Summary of unit operating conditions for each process
Pre-treated alkali-catalyzed Acid-catalyzed Heterogeneous acid-catalyzed Supercritical process
A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601 6593
data is needed to verify the applicability and results of thegravity separator, in order to use the unit block with con-fidence. In practice, a gravity separation unit has been usedon a pilot plant-scale to separate glycerol and biodiesel(Canakci and Van Gerpen, 2003). All processes producedbiodiesel at a higher purity than required by the ASTMstandard of 99.65 wt.%.
4. Equipment sizing
Process equipment was sized according to principlesoutlined in the literature (Seider et al., 2003; Turtonet al., 2003). The principal dimensions of each unit are pre-sented in Table 7. The equipment sizing calculations wereconducted using the Spreadsheet tool available within
Table 4Feed and product stream information for the alkali-catalyzed transesterification step of the pre-treated process
A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601 6595
ponent flowrates, changes to the desired recovery in the dis-tillations columns are automatically calculated and imple-mented, thus eliminating tedious recalculations steps.
4.1. Reactor vessels
Reactors were sized for continuous operation by divid-ing the residence time requirement by the feed flowratefor each process. Residence times were: 4 h, 4 h, 3 h and20 min for the pre-treated alkali-catalyzed, acid-catalyzed,heterogeneous acid-catalyzed and supercritical processes,respectively. The vessels were specified to have an aspectratio of three-to-one.
4.2. Columns
Distillation column diameters were sized by two meth-ods. An initial diameter was estimated from the F-FactorMethod (Luyben, 2002). If the column diameter was calcu-lated to be greater than 0.90 m (2.95 ft) it was specified as atray tower, and thus calculated from the flooding velocityusing the Fair correlation (Seider et al., 2003). Columnswith diameters calculated at less than 0.9 m were specifiedas a packed tower. The diameter of each packed columnwas calculated from the flooding velocity obtained fromthe Leva correlation (Seider et al., 2003).
Tray tower height was calculated by multiplying thenumber of actual stages by the tray spacing, and thenincreasing the result by 20% to provide height for the con-denser and reboiler.
Packed tower height was calculated by multiplying theheight equivalent of a theoretical plate (HETP) by the num-ber of stages calculated for the tower. HETP was assumedto equal the column diameter (Seider et al., 2003). As forthe height of a tray tower, the packed height was increasedby 20%. The liquid–liquid extraction columns were sizedaccording to the work of Zhang et al. (2003a).
4.3. Gravity separators
The gravity separators in the heterogeneous acid-cata-lyzed and supercritical processes were designed as vertical
process vessels with an aspect ratio of 2. They were sizedto allow for continuous operation, with a residence timeof 1 h.
4.4. Hydrocyclone
The initial dimensions of the hydrocyclone (used to sep-arate the solid catalyst from the product stream in ProcessIII) were calculated by the unit block in HYSYS. Thosedimensions were then manipulated slightly to obtain com-plete removal of the catalyst in the hydrocycloneunderflow.
5. Economic assessment
Since each process was capable of producing biodiesel atthe required purity, it was of interest to conduct an eco-nomic assessment to determine process viability, and deter-mine if any one process was advantageous over the others.As with the sizing calculations, all the economic calculationswere performed within the HYSYS spreadsheet. Addition-ally, the values presented for the economic analysis are thevalues obtained after performing sensitivity analyses andoptimization of each process. The details for the sensitivityanalyses and optimization studies are presented in Section 6of this paper. All parameters necessary to determine mate-rial and energy costs were imported to the spreadsheet fromthe flowsheet. Costing equations were incorporated directlyinto the spreadsheet as well. Individual unit costs were cal-culated, as well as figures for each process in its entirety.Incorporating the economic calculations into the simulationallowed for automatic recalculation of process economicsshould any process parameters, such as component flow-rates or unit operating conditions be changed. By integrat-ing sizing and economic calculations into each simulation,the potential to perform economic sensitivity analyses isautomatically built-in to each simulation.
5.1. Basis of calculations
Each process was based on a plant capacity of 8000 ton-nes/year biodiesel production. Operating hours were set at
6596 A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601
7920 h/year (assuming 330 operating days/year). The feed-stock was assumed free of water and solid impurities. Lowand high pressure steam were used as heating media, whilewater was used for cooling.
Each process was evaluated based on total capitalinvestment (TCI), total manufacturing cost (TMC), andafter tax rate-of-return (ATROR). The assessment per-formed in this work is classified as a ‘‘study estimate” (Tur-ton et al., 2003) with a range of expected accuracy from+30% to �20%. While the results of such a study will likelynot reflect the final cost of constructing a chemical plant,the technique is useful for providing a relative means tocompare competing processes.
5.2. Total capital investment
Table 8 gives a breakdown of the total capital invest-ment. It also presents the costs for the individual unit oper-ations in each process.
Bare module capital costs (CBM) consist of the purchasecost of a piece of equipment, multiplied by the bare modulefactor. Purchase costs were estimated for each piece ofequipment based on a capacity equation (Turton et al.,2003)
log10C�p ¼ K1 þ K2log10ðAÞ þ K3½log10ðAÞ�2 ð2Þ
where Ki is a constant specific to the unit type and A isthe capacity of the unit. Bare module cost was calculatedfrom
CBM ¼ C�pF BM ð3Þ
where FBM is given by
F BM ¼ B1 þ B2F MF P ð4Þwhere B1 and B2 are constants specific to the unit type, andFM and FP are the material and pressure factors, respec-tively. The constants, Ki and Bi, as well as the pressureand material factors were obtained from the literature(Turton et al., 2003). Eqs. (2)–(4) were encoded withinthe costing spreadsheet to allow for automatic cost updateswhen process parameters are changed.
Table 8Equipment sizes for various process units in all processes
Dimensions are diameter x height (m).a Indicates units that are pre-treatment process components.
5.3. Total manufacturing cost
Direct manufacturing expenses were calculated based onthe price and consumption of each chemical and utility.Chemical and utility prices are presented in Table 9 andmaterial flow information was obtained from HYSYS.Operator salary was estimated at $47,850/year, and itwas assumed that an operator worked 49 weeks/year, andthere were three 8-h shifts per day for the continuous plant(Zhang et al., 2003b). Table 10 presents a breakdown of thecomponents of the total manufacturing cost as well as theresults for each process. After tax rate-of-return is a generalcriterion for economic performance of a chemical plant. Itis defined as the percentage of net annual profit after taxes,relative to the total capital investment. Net annual profitafter taxes (ANNP) is half the net annual profit (ANP)assuming a corporate tax rate of 50%. The results for aftertax rate-of-return for each process are shown in Table 10.
As shown in Table 8, the transesterification reactorforms a large part of the capital cost, especially for Pro-cesses II and IV. The reactor in Process II was requiredto contain a large material flow at a long residence time.The presence of sulfuric acid as the catalyst required astainless steel reactor, resulting in a substantially higherreactor cost. Consequently the reactor in Process II wasmuch more expensive than in all other processes. Thesupercritical reactor was required to withstand a high pres-sure, and was constructed from stainless steel to preventoxidation and corrosion, hence its high cost. Distillationcolumns also contributed a significant part to the capitalcost of each process. Tower costs for the methyl-ester puri-fication tower were roughly equal between the processes, aseach tower was handling approximately the same materialflows and producing biodiesel at equal purities. The meth-anol recovery columns in Processes I and III were the leastexpensive, as they had the smallest material flow require-ments. In spite of Process IV having the smallest numberof unit operations, Process III had the smallest total capitalinvestment. This is due to the fact that Process IV requireda more expensive reactor in order to withstand the highpressures and corrosive conditions associated with the
id-catalyzed Heterogeneous acid-catalyzed Supercritical process
a Value from Turton et al., 2003.b Glycerol is used as a solvent in Process I, and its cost is assumed to be
750 $/tonne.
A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601 6597
supercritical state of the alcohol, as well as the larger meth-anol recovery tower. The total capital investment for Pro-cess I was less expensive than Process II even though ithad a greater number of required unit operations due tothe waste oil pre-treatment step. Again this is due todecreased material flows and reactor costs. The total capi-tal investment for Process I in the present work was calcu-lated to be $1.59 million, less than the value reported byZhang et al. (2003b) of $2.68 million. The difference lies
mostly in the lower costs calculated for the methanol recov-ery column and methyl-ester purification column, due tothe differences in sizing.
Results for the total manufacturing cost of each processare shown in Table 11. The direct manufacturing cost rep-resents between 71% and 77% of the total manufacturingcost in each process. The largest proportion of the directmanufacturing cost is due to the oil feedstock, namely39% for Process I, and around 43% for the other processes.Process III has the lowest total manufacturing cost. This isdue to both the ability of the process to use low cost wastevegetable oil, as well as the lower utility costs of the processresulting from the smaller material streams handled in theprocess. The total manufacturing cost of Process IV isslightly more than that of Process III, owing to the largeenergy requirements necessary to separate the methanolfrom the product stream. In spite of the use of an efficientreaction scheme in Process I, the total manufacturing costsare higher than in Process II because of the pre-treatmentstep: the additional catalyst and methanol consumed toconvert the FFA, and the large costs associated with usingglycerol as a washing agent in the first liquid washing col-umn add considerable expense to the process. This use ofglycerol for washing essentially negates the benefit of theglycerol production credits that the other processes are ableto realize.
Except for Process III, all processes had a negative aftertax rate-of-return. Process I had the lowest ATROR, at�22%, while Processes II and IV had ATRORs at �9%and �1%, respectively. The ATROR for Process III was58%, indicating that the process could earn a profit withoutany government subsidies. The value for ATROR reportedby Zhang et al. (2003b) for the pre-treated Process I was
Table 11Total manufacturing cost and after tax rate-of-return for each process
Maintenance & Repairs (M&R), 6% of CFC 0.08 0.11 0.03 0.11Operating supplies, 15% of M&R 0.01 0.02 0.00 0.02Lab charges, 15% of operating labour 0.09 0.09 0.09 0.09Patents and royalties, 3% TMC 0.16 0.15 0.12 0.14Subtotal ADMC 4.19 3.84 3.19 3.61
Indirect manufacturing cost
Overhead, packaging and storage, 60% of sum of operating labour, supervision and maintenance 0.45 0.46 0.42 0.47Local taxes, 1.5% of CFC 0.02 0.03 0.01 0.03Insurance, 0.5% of CFC 0.01 0.01 0.00 0.01Subtotal, AIMC 0.46 0.47 0.42 0.47
Depreciation 10% of CFC 0.14 0.18 0.05 0.19
General expenses
Administrative costs, 25% of overhead 0.11 0.12 0.10 0.12Distribution and selling, 10% of TMC 0.52 0.48 0.39 0.46Research & Development, 5% of TMC 0.26 0.24 0.19 0.23Subtotal 0.89 0.84 0.69 0.81
Total production cost 5.78 5.37 4.45 5.19Glycerine credit 0.58 0.61 0.57 0.60Total manufacturing cost, ATE 5.20 4.76 3.88 4.59Revenue from biodiesel 4.77 4.76 4.70 4.92Net annual profit �0.43 �0.005 0.82 0.33Income taxes, AIT 50% of ANP �0.21 �0.003 0.41 0.17Net annual after tax profit, ANNP �0.21 �0.003 0.41 0.17
After tax rate-of-return, I = [ANNP � ABD]/CTC (%) �22.2 �8.71 58.76 �0.90
Costs are reported as $millions.
6598 A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601
�51% was reasonably close to the value reported in thiswork. Additionally, our rate-of-return for Process II(�9%) was in close agreement with the value reported forthe acid-catalyzed case by Zhang et al. (2003b) of �15%.The relative economic order of Processes I and II (i.e., thatProcess II has an ATROR greater than that of Process I) aspresented in this work is also in agreement with that ofZhang et al. (2003b). As predicted by Lotero et al.(2005), the heterogeneous acid-catalyzed process was byfar the most economically attractive process.
6. Sensitivity analyses and optimization
Sensitivity analyses were conducted to determine theeffect on the process of variables that had some degree ofuncertainty; and to identify any operating specificationswithin an individual process that could be modified toimprove the process.
Since the conversion data for the heterogeneous acid-catalyzed and supercritical processes were taken frombench-scale research, the economics of scale may not beaccurately reflected. Thus the effect of reduced conversionon the overall process economics was examined for eachprocess. As shown in Fig. 5, conversion in the heteroge-neous acid-catalyzed process must drop to approximately85%, while conversion in the supercritical and homoge-neous acid-catalyzed processes must increase to almost100% before there is any overlap in the after tax rate-of-return. From this, it is clear that even at reduced reactorconversion, the heterogeneous process will still be advanta-geous over the supercritical and homogeneous acid-cata-lyzed processes.
Sensitivity analyses were performed for all processes todetermine the effect of changing methanol recovery in themethanol recovery distillation column on the ATROR. Inall cases except the pre-treatment case, increasing the meth-
-165
-135
-105
-75
-45
-15
15
45
75
80 85 90 95 100 105
Reaction Conversion (%)
ATR
OR
(%)
Base Catalyzed Homog. Acid Cat. Heterog. Acid Cat. Supercritical
Fig. 5. After tax rate-of-return vs. reaction conversion for all processes.
A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601 6599
anol recovery caused an increase in the ATROR, due todecreased methanol consumption in all cases. Methanolacts as a cosolvent (Chiu et al., 2005) increasing the solubil-ity of biodiesel in the glycerol phases. Therefore, reducingthe amount of methanol entering the three phase separator(HAC and SC processes) reduced the amount of biodiesellost in the glycerol stream, thereby boosting ATROR forboth processes. The effect of methanol recovery onATROR for the HAC process is illustrated in Fig. 6. Meth-anol recovery is limited to about 85%, as the bottomsstream temperature should not exceed 150 �C to preventglycerol decomposition. In order to increase the methanolrecovery, the distillation column was operated under vac-uum conditions. The effect of vacuum pressure (and there-fore cost of the vacuum system) on the ATROR wasinvestigated to determine if the cost of the vacuum system
50
51
52
53
54
55
56
57
58
0.75 0.8 0.85Methanol Re
ATR
OR
(%)
ATROR
Fig. 6. ATROR vs. methanol recovery in the
was offset by the increase in revenue that results fromhigher methanol recovery. As shown in Fig. 7, the additionof the vacuum system resulted in a decrease in ATROR.However, as the methanol recovery was increased undervacuum operation the ATROR increased, indicating thepotential for optimization of the column operating condi-tions to maximize the ATROR. Similar analyses wereconducted for the homogeneous acid-catalyzed and super-critical processes, but it was found that vacuum operationdid not provide any economic benefits, as the methanolrecovery was already greater than 99% and the bottomstemperature was within the allowable limit at ambient pres-sure operation. The HYSYS optimizer tool was used tovary the HAC methanol recovery in order to maximizethe ATROR, according to the following constraints: bot-toms temperature <150 �C; 1 kPa < column pressure
Fig. 7. ATROR vs. operating pressure in the methanol recovery column, HAC process.
6600 A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601
<100 kPa; and 85.0% < methanol recovery <99.9%. Anoptimum was found at a pressure equal to 40 kPa andmethanol recovery of 99.9%. Upon optimization the bot-toms temperature decreased from 149.9 �C to 145.5 �C,methanol recovery increased from 85% to 99.9% and theATROR increased slightly from 53.7% to 54.2%. In addi-tion to the financial incentive, including a vacuum systemreduces methanol consumption and eliminates 79,200 kg/year of methanol from the waste stream, greatly reducingthe environmental impact of the process.
Lastly, the effect of vacuum operation in the FAME dis-tillation columns was investigated for the heterogeneousacid-catalyzed and the supercritical processes, to determineif vacuum operation would result in a net savings due to adecrease in the heating and cooling duties on the column.Column heating and cooling loads did decrease; however,the savings in utilities cost were not enough to offset thecost of the vacuum system, and inclusion of a vacuum sys-tem therefore decreased the ATROR in both cases. Sincethe upper temperature limit of glycerol (150 �C) has notbeen exceeded at ambient operation, a vacuum systemwas deemed unnecessary for FAME distillation in bothprocesses. Vacuum operation for FAME distillation wasneeded in the homogeneous acid-catalyzed process to keepthe temperature of the distillate below 250 �C.
7. Conclusion
Four continuous processes to produce biodiesel at a rateof 8000 tonnes/year were designed and simulated inHYSYS: all were capable of producing ASTM gradebiodiesel. Subsequently, an economic assessment revealedthat the heterogeneous acid-catalyzed process has the low-est total capital investment, owing to the relatively small
sizes and carbon steel construction of most of the processequipment. Raw materials consumed in the processaccount for a major portion of the total manufacturingcost. Accordingly, Processes II, III and IV have muchlower manufacturing costs than Process I. The high solvent(glycerol for washing) and catalyst costs of Process I, thelarge excesses of methanol in Processes II and IV resultedin much higher manufacturing costs than in Process IIImaking it the only process to produce a net profit. Theafter tax rate-of-return for Process III was 59%, while Pro-cesses I, II and IV had rates-of-return of �22%, �9% and�1%, respectively.
Sensitivity analyses were conducted to identify any unitoperations where operating conditions could be modifiedto improve the process. Increasing methanol recovery ledto a greater ATROR. Accordingly, methanol recoverywas set as high as possible (>99%) before the glycerol deg-radation temperature (150 �C) was exceeded in the homo-geneous acid-catalyzed and supercritical processes. Use ofthe optimizer function indicated a vacuum system couldbe installed in the HAC process to increase methanolrecovery and consequently the ATROR, while keepingthe bottoms stream within the temperature limit.
An analysis of the effect of reaction conversion onATROR revealed that even at reduced reaction conversion(i.e., between 85% and 93%), the ATROR of the HAC pro-cess is greater than at 100% conversion of the homoge-neous acid and supercritical processes.
Therefore Process III, the heterogeneous acid-catalyzedprocess, is clearly advantageous over the other processes,as it had the highest rate-of-return, lowest capital invest-ment, and technically, was a relatively simple process.Further research in developing the heterogeneous acid-catalyzed process for biodiesel production is warranted.
A.H. West et al. / Bioresource Technology 99 (2008) 6587–6601 6601
Acknowledgement
The authors acknowledge the financial support of theNatural Sciences and Engineering Research Council ofCanada.
References
Abreu, F.R., Alves, M.B., Macedo, C.C.S., Zara, L.F., Suarez, P.A.Z.,2005. New multi-phase catalytic systems based on tin compoundsactive for vegetable oil transesterification reaction. Journal of Molec-ular Catalysis A: Chemical 227, 263–267.
Canakci, M., Van Gerpen, J., 1999. Biodiesel production via acidcatalysis. Transactions of the ASAE 42, 1203–1210.
Canakci, M., Van Gerpen, J., 2001. Biodiesel production from oils andfats with high free fatty acids. Transactions of the ASAE 44, 1429–1436.
Canakci, M., Van Gerpen, J., 2003. A pilot plant to produce biodieselfrom high free fatty acid feedstocks. Transactions of the ASAE 46,945–954.
Chiu, C.W., Goff, M.J., Suppes, G.J., 2005. Distribution of methanol andcatalysts between biodiesel and glycerin phases. AIChE Journal 51,1274–1278.
Delfort, B., Hillion, G., Le Pennec, D., Chodorge, J.A., Bournay, L.,2003. Biodiesel production by a continuous process using a heteroge-neous catalyst. Abstracts of Papers, 226th ACS National Meeting,New York, NY, United States, September 7–11, 2003, FUEL-066.
Demirbas, A., 2002. Biodiesel from vegetable oils via transesterification insupercritical methanol. Energy Conversion and Management 43,2349–2356.
Freedman, B., Pryde, E.H., Mounts, T.L., 1984. Variables affecting theyields of fatty esters from transesterified vegetable-oils. Journal of theAmerican Oil Chemists Society 61, 1638–1643.
Freedman, B., Butterfield, R.O., Pryde, E.H., 1986. Transesterificationkinetics of soybean oil. Journal of the American Oil Chemists Society63, 1375–1380.
Furuta, S., Matsuhashi, H., Arata, K., 2004. Biodiesel fuel productionwith solid superacid catalysis in fixed bed reactor under atmosphericpressure. Catalysis Communications 5, 721–723.
Goff, M.J., Bauer, N.S., Lopes, S., Sutterlin, W.R., Suppes, G.J., 2004.Acid-catalyzed alcoholysis of soybean oil. Journal of the American OilChemists Society 81, 415–420.
Goodrum, J.W., 2002. Volatility and boiling points of biodiesel fromvegetable oils and tallow. Biomass & Bioenergy 22, 205–211.
Haas, M.J., McAloon, A.J., Yee, W.C., Foglia, T.A., 2006. A processmodel to estimate biodiesel production costs. Bioresource Technology97, 671–678.
Krawczyk, T., 1996. Biodiesel. INFORM 7, 801–822.Kusdiana, D., Saka, S., 2001. Kinetics of transesterification in
rapeseed oil to biodiesel fuel as treated in supercritical methanol.Fuel 80, 693–698.
Kusdiana, D., Saka, S., 2004. Effects of water on biodiesel fuel productionby supercritical methanol treatment. Bioresource Technology 91, 289–295.
Lotero, E., Liu, Y.J., Lopez, D.E., Suwannakarn, K., Bruce, D.A.,Goodwin, J.G., 2005. Synthesis of biodiesel via acid catalysis.Industrial & Engineering Chemistry Research 44, 5353–5363.
Luyben, W.L., 2002. Plantwide Dynamic Simulators in Chemical Pro-cessing and Control. Marcel Dekker, New York.
McCabe, W.J., Smith, J.C., Harriott, P., 2001. Unit Operations ofChemical Engineering, sixth ed. McGraw Hill, Boston.
Newman, A.A., 1968. Glycerol. CRC Press, Cleveland.Saka, S., Kusdiana, D., 2001. Biodiesel fuel from rapeseed oil as prepared
in supercritical methanol. Fuel 80, 225–231.Schumacher, L.G., Marshall, W., Krahl, J., Wetherell, W.B., Grabowski,
M.S., 2001. Biodiesel emissions data from Series 60 DDC engines.Transactions of the ASAE 44, 1465–1468.
Seider, W.D., Seader, D., Lewin, D.R., 2003. Process Design Principles:Synthesis, Analysis and Evaluation, second ed. John Wiley & Sons,Chichester, UK.
Suppes, G.J., Dasari, M.A., Doskocil, E.J., Mankidy, P.J., Goff, M.J.,2004. Transesterification of soybean oil with zeolite and metalcatalysts. Applied Catalysis A: General 257, 213–223.
Turton, R., Bailie, R.C., Whiting, W.B., Shaeiwitz, J.A., 2003. Analysis,Synthesis, and Design of Chemical Processes, 2nd ed. Prentice Hall,Upper Saddle River, New Jersey.
Tyson, K.S., 2001. Biodiesel: Handling and Use Guidelines. NationalRenewable Energy Laboratory, Golden, CO.
Warabi, Y., Kusdiana, D., Saka, S., 2004. Reactivity of triglycerides andfatty acids of rapeseed oil in supercritical alcohols. BioresourceTechnology 91, 283–287.
Zhang, Y., Dube, M.A., McLean, D.D., Kates, M., 2003a. Biodieselproduction from waste cooking oil: 1. Process design and technologicalassessment. Bioresource Technology 89, 1–16.
