A Second Order Kinetics of Palm Oil Transesterification

Transcription

1 3-05 (O) The Joint Interntionl Conference on Sustinble nergy nd nvironment (S) 1-3 December 004, Hu Hin, Thilnd Second Order Kinetics of Plm Oil Trnsesterifiction Theeryut Leevijit 1,*, Worwut Wisutmethngoon, Gumpon Prteepchikul, Chrktir Tonguri nd Michel llen 1 The Joint Grdute School of nergy nd nvironment, King Mongkut s University of Technology Thonburi, Bngkok, Thilnd Fculty of ngineering, Prince of Songkl University, Songkhl, Thilnd bstrct: Trnsesterifiction of plm oil with methnol in the presence of NOH s ctlyst ws conducted in well-mixed btch rector t the reported optimum rection condition [1]. Smples were collected during the rection nd nlyzed by thin lyer chromtogrphy/flme ioniztion detector (TLC/FID) to determine the weight percentges of rection compositions on glycerol free bsis. The experimentl mole concentrtions were clibrted nd fitted to mthemticl model of second-order kinetics without shunt rection. The rection rte constnts were determined. The effect of mss trnsfer could be observed tht it ws eliminted; thus, the obtined rection rte ws true intrinsic rte of homogeneous rection. This kinetics coupled with the obtined rection rte constnts provided stisfctory mechnism with the experimentl result. The overll goodness of fit to predict the weight percentge of methyl esters in the product during the rection ws quntified through two stndrds: the correltion coefficient (R ) nd the men reltive devition (MRD). The R ws nd the MRD ws 1.835%. lthough, trnsesterifiction of plm oil consisted of three stepwise nd reversible rections, the rection rte constnts reveled tht the forwrd rections were the most importnt. In ddition, the mximum purity t the equilibrium ws 99.1 %wt. quilibrium ly towrds the products. Thus, glycerol seprtion ws not significnt fctor for producing sleble biodiesel by btch process. Keywords: Biodiesel, Plm Oil, Kinetics, Trnsesterifiction, Triglycerides, Methyl sters. 1. INTRODUCTION Presently, biodiesel hs become more ttrctive in mny countries, including Thilnd. It is renewble nd environmentl friendly energy resource; it cn be produced from vegetble oils nd tllow; nd it lso gives environmentl benefits, especilly crbon dioxide sving [,3]. In Thilnd, severl vegetble oils cn be used s feedstock; however, the highest potentil source is oil from plm fruits: in the yer 001, 535 kilotons of plm oil were produced [4]. Consequently, success of using biodiesel from plm oil commercilly is n importnt fctor for Thilnd s sustinble development; the positive economic effects nd n increse of energy security would be chieved. However, first of ll, the production technology must be developed. The most recent method for biodiesel production is btch trnsesterifiction processes; however, generlly, continuous processes give lower production cost nd more uniform product qulity thn btch processes. In order to design n efficient continuous rector for biodiesel production from plm oil, suitble kinetics mechnism nd rection rte constnts must be found. lthough there re number of kinetics studies in literture on trnseterifiction of esters with lcohol [5-7], only few reserches re concerned with trnsesterifiction of vegetble oils. Freedmn et. l. [8] studied the kinetics of the cid- nd bse-ctlyzed trnsesterifiction of soyben oil with 1-butnol nd methnol t 30: 1 nd 6:1 molr rtio of lcohol to oil. Noureddini nd Zhu [9] studied the kinetics of bse-ctlyzed trnsesterifiction of soyben oil with methnol t 6:1 molr rtio of lcohol to oil. Significntly, they found tht pseudo first-order kinetics provided stisfctory mechnism with experimentl results t lrge molr excess of lcohol nd second-order kinetics provided stisfctory mechnism with experimentl results t 6:1 molr rtio of lcohol to oil. For the ltter cse, n inclusion of shunt mechnism ws not necessry. Drnoko nd Cheryn [1] studied the kinetics of bse-ctlyzed trnsesterifiction of plm oil with methnol t 6:1 molr rtio of lcohol to oil. They reported tht the best kinetics mechnism ppered to be pseudo second-order. However, second-order mechnism is more convenient nd more suitble thn pseudo second-order mechnism for predicting conversion yield t ny rection time. Using Corresponding uthor: leetheeryut@yhoo.com second-order mechnism, the initil concentrtions of ll rectnts re needed; generlly, these concentrtions re known depend on the given condition. In contrst, using pseudo second-order mechnism, couple of n initil concentrtion of triglycerides (TG) nd initil highest concentrtions of diglycerides () nd monoglycerides () re needed. n initil concentrtion of TG is not problem, but the highest concentrtions of nd must be determined. The rection must be llowed to proceed for minute to know these concentrtions. Consequently, pseudo second-order mechnism cnnot predict conversion yield for ll rection times unlike second-order mechnism. However, there is no published deling with second-order kinetics of bse-ctlyzed trnsesterifiction of plm oil. The objectives of this work were to evlute second-order kinetics mechnism nd to find out the rection rte constnts for plm oil trnsesterifiction with methnol when NOH ws used s ctlyst. rection condition focused on the reported optimum condition for trnsesterifiction of plm oil t tmospheric pressure: 6:1 molr rtio of lcohol to oil, temperture 60 C, nd ctlyst concentrtion 1% wt of oil [1]. The obtined result cn be used to predict conversion yield of plm oil trnsesterifiction in continuous rectors efficiently.. MTRILS ND MTHODS.1 Mterils refined plm oil, commercil methnol of 95 % purity, nd commercil sodium hydroxide pellets were used in this study. The free cid content of the oil ws determined ccording to OCS officil method C 5-40 [10] to be 0.64 %. Chloroform, hexne, diethyl ether, formic cid, nd benzene used in the nlysis were ll of pro nlysis grde.. pprtus The experimentl set up is shown in Fig. 1. seled,000 ml glss rector equipped with mechnicl stirrer, digitl thermometer, nd smpling port ws used in this study. The rector ws immersed in constnt-temperture circulting wter bth, Model YL SB-4 & T-80 (Tokyo Rikkiki Co.,Ltd, Jpn); it ws cpble of controlling the temperture to within ±0.1 C. mechnicl stirrer, Type RR1 (Heidolph, Germnny), fitted with stinless steel 6-blde disk turbine 77

2 3-05 (O) The Joint Interntionl Conference on Sustinble nergy nd nvironment (S) 1-3 December 004, Hu Hin, Thilnd provided the sufficient mixing. The mixing intensity is presented by Reynolds number (N Re ) of impeller in q. (1) [11] nd SG (1) NRe = µ where N Re is Reynolds number; n is turbine speed (rpm); D is turbine dimeter (mm); SG is fluid specific grvity; nd µ is fluid dynmics viscosity (P.s). N Re ws referred to the strt of the rection; it ws round,000. During the rection, smples were withdrwn with ml pipettes through the smpling port. Smpling Mechnicl Stirrer Digitl Thermometer Constnt Temperture Wter Bth Chromrod type S-III qurtz rod (Mitshubishi Kgku Itron Inc., Jpn). The smples were diluted in chloroform nd 1 µl of the solution ws spotted on ech rod. The rods were developed in hexne/diethyl ether/formic cid (50:0:30 vol/vol/vol) for 8 cm nd hexne/benzene (1:1 vol/vol) for 10 cm. The rods were dried nd scnned under the following conditions: hydrogen flow rte 160 ml/min, ir flow rte.0 L/min, speed 30 s/scn. M, TG,, nd were effective seprted; pek res were clculted with chromtogrphy system ChromStr. The weight percentges (%wt) of the rection compositions on glycerol-free bsis were determined..6 Kinetics modeling Primrily, vegetble oils consist of TG round 90 to 98 % by weight of oil nd smll mounts of nd [14]. In this work, the initil composition of plm oil ws mesured. It consisted of TG 95.7%, 3.7%, nd 0.6%. Trnsesterifiction of vegetble oils with lcohol is multiple rections consisting of number of consecutive nd reversible rections. TG is converted stepwise to, nd finlly glycerol (GL) s following equtions [8]. TG k1 k k3 k4 k5 k6 GL () Fig. 1 xperimentl set up for btch trnsesterifiction..3 Procedures The rector ws initilly chrged with 650 g of refined plm oil. The rector ssembly ws then plced in the circultion wter bth nd heted to the designed temperture. mesured mount of methnol nd sodium hydroxide solution ws heted seprtely to the rection temperture nd dded to the rector. The mechnicl stirrer ws strted s soon s possible. The rection ws timed when the solution ws dded..4 Smpling Smples of bout ml were withdrwn t pre-specified time intervls. pproximtely 17 smples were collected during the rection (90 min). t the beginning of the rection, more frequent smpling ws required. ch smple ws collected in 0 ml test tube filled with 6 drops of 0.6 N sulfuric cid to neutrlize the ctlyst. The mixture ws then suddenly wshed twice with distilled wter: first 8 ml of distilled wter ws dded to the test tube nd the mixture ws wshed in 300 ml flsk filled with 00 ml of distilled wter. Upon wshing, glycerol, methnol nd remining sodium hydroxide or sulfuric cid were trnsferred to the wter phse. t ech time of neutrliztion nd wshing, vigorous mixing ws provided by the mixer. Seprte studies hd determined tht these procedures stopped the rection immeditely. Then the ester phse nd wter phse formed seprte liquid lyers. The ester phse ws then centrifuged to ensure thorough seprtion. The ester phse ws kept t 0 C until further nlysis..5 nlysis The smples were nlyzed by thin lyer chromtogrphy/flme ioniztion detector (TLC/FID) [1,13]. nlyses were performed using n Itronscn MK-6 with where nd denote the lcohol nd ester respectively. The experimentl mole concentrtions of TG,,, M, nd GL were clibrted by using the experimentl weight percentges of rection compositions, their moleculr weights, nd the mechnism of stepwise rections. This clibrtion ws bsed on the conservtion of mss of rection mixture nd the conservtion of mole of glycerol core. The moleculr weights of TG,, nd of plm oil nd moleculr weight of M were clculted from the known ftty cids of plm oil [15]; they were 849.5, 597.0, 344.5, nd 84.5 kg/kmol, respectively. moleculr weight of GL ws 9.1 kg/kmol [16]. The governing set of second-order rte equtions chrcterizing the stepwise rections for trnsesterifiction of TG, without the shunt rection, re s following [8,9]. d[ TG] = k1[ TG][ ] k[ ][ ] d[ ] = k1[ TG][ ] k[ ][ ] k3[ ][ ] k4[ ][ ] d[ ] = k3[ ][ ] k4[ ][ ] k5[ ][ ] k6[ GL][ ] (3) d[ GL] = k5[ ][ ] k6[ GL][ ] d[ ] = k1[ TG][ ] k[ ][ ] k3[ ][ ] k4[ ][ ] k5[ ][ ] k6[ GL][ ] d[ ] d[ ] = where k 1 to k 6 re rection rte constnts (L/mol.s); [TG], [] [], [GL], [], nd [] re mole concentrtions of TG,,, GL, lcohol, nd esters in rection mixture (mol/l). The differentition of mole concentrtions with respect to time on the left hnd side of q. (3) ws clculted from the experimentl mole concentrtions t vrious rection times by three-points numericl differentil formuls. These formuls cn be used both when the points re eqully time spced nd when the points hve chnge in time increment [17]. The coefficients of k 1 to k 6 on the right hnd side of q. (3) were 78

3 3-05 (O) The Joint Interntionl Conference on Sustinble nergy nd nvironment (S) 1-3 December 004, Hu Hin, Thilnd obtined from multipliction of experimentl mole concentrtions. Substitution of the differentition of concentrtions with respect to time nd the coefficients of k 1 to k 6 into q. (3) for ll mesured points nd rerrngement of the equtions gve the system of liner equtions of 6 unknowns in the following form n n M n n n n6 k k k k k k b1 b b3 = b4 M bn where n is equl to multipliction of number of sub-equtions in q. (3) with number of mesured points; 11 to n6 nd b 1 to b n re the known constnts obtined from experimentl. The bove system of liner equtions cn be solved by vrious numericl techniques [18]. However, lest-squre regression is the best method to find the best fit of k 1 to k 6 into the experimentl. Consequently, the lest-squres regression technique ws used. The bility of solver tool in Microsoft xcel 000 progrm cn be used to minimize vlue of sum of error squres ( ) s shown in following eqution. [ bi ( i1k1 i k i3k3 L i6k6 )] (5) i= 1 To correspond with the nture of reversible rections, the best fit of rection rte constnts in this study ws determined bsed on constrins tht k 1 to k 6 0. In fct, nture of numericl technique, nd initil guesses ffect the obtined results. Some initil guesses give divergent results; some initil guesses give convergent results. For the ltter cse, different initil guesses cn lso give different results. Hence, different initil guesses were used. The rnges of results for rection rte constnts were obtined. However, the rnges were very nrrow. = n.7 vlution To evlute second-order kinetics mechnism nd the obtined rection rte constnts, simultion results nd experimentl results must be compred. computer code for finite different method ws employed on progrm MTLB 6.1 to solve the governing set of second-order rte equtions, q. (3), for the whole of the experimentl rection time. The simultion results of concentrtions nd weight percentge of M in the product could be obtined t ny rection time. In fct, it is generl for numericl technique; size of time step ( t) used in simultion ffects the obtined results. The effect of size of time step ws tested. However, time step less thn 1 sec gve insignificnt different results compred to time step t 1 sec. s result, time step t 1 sec ws used for simultion. The overll goodness of fit to predict weight percentge of M in the product during the rection ws quntified through two stndrds: the correltion coefficient (R ) nd the men reltive devition (MRD). The MRD is n bsolute vlue tht cn give cler ide of the devition of the simultion result nd the experimentl result: 100 m WM,exp WM, sim (6) MRD (%) = W m i= 1 M, exp where W M is weight percentge of M in the product; subscript exp is denoted for the experimentl vlue; subscript sim is denoted for the simulted vlue; nd m is the number of points. (4) Concentrtion (mol/l) RSULTS ND DISCUSSIONS Bse-ctlyzed trnsesterifiction of refined plm oil t molr rtio of methnol to oil 6:1, temperture 60 C, NOH concentrtion 1 %wt of oil, mixing intensity N Re,000, nd tmospheric pressure ws studied. The experimentl results nd the simultion results for the concentrtion of the rection mixture during the rection in the significnt rnge (the first 10 min of the rection time) re shown in Fig.. Moreover, Figs. 3-4 clerly show the concentrtions of TG,, nd. The concentrtions of ll compositions for rection time from 10 min to 90 min were quite similr to the concentrtions t 10 min; however, they re not presented here Fig. xperimentl points nd simultion curves for the concentrtion of the rection mixture during the trnsesterifiction of plm oil t molr rtio 6:1, temperture 60 C, ctlyst concentrtion 1 %wt of oil, nd N Re =,000; ( ) methyl esters; ( ) triglycerides; ( ) diglycerides; ( ) monoglycerides; ( ) glycerol. The production rte of M in Fig. strts with sudden surge followed by lower production rte s rection pproches equilibrium. number of reserchers [1,9,1] observed sigmoidl pttern (S-shpe) for production of M. This pttern consists of slow rte t the beginning followed by sudden surge nd finlly slow rte gin. Noureddini nd Zhu [9] nd Boocock et. l. [19] explin the resons why the production rte of M occurs like this. Truly, rection mechnism for trnsesterifiction of vegetble oils consists of n initil mss trnsfer-controlled region followed by kineticlly controlled region. The lg time t the initil stge occurs becuse the effect of poor diffusion between lcohol nd oil; it is clled two-phse rection. Whenever methyl esters re formed, they ct s mutul solvent for the rectnts; single-phse system is formed nd the rection reches kineticlly controlled region. The beginning of kineticlly controlled region cn be observed t time corresponding to the beginning of the sudden surge in time-concentrtion digrm [9]. However, if sufficient mixing is supplied, lg time of mss trnsfer-controlled region cn be eliminted [9]. Fig. lso shows tht mss trnsfer-controlled region in this study disppers. Thus, the supplied mixing is sufficient to drive the rection to be homogeneous for ll rection times even in the initil stge of the rection. Furthermore, when n effect of mss trnsfer clled physicl effect is eliminted, the obtined rection rte is true intrinsic rection rte [0]. In summry, the obtined rection rte ws n intrinsic rection 79

4 3-05 (O) The Joint Interntionl Conference on Sustinble nergy nd nvironment (S) 1-3 December 004, Hu Hin, Thilnd rte of homogeneous rection. Concentrtion (mol/l) Fig. 3 xperimentl points nd simultion curve for triglycerides during the trnsesterifiction of plm oil t molr rtio 6:1, temperture 60 C, ctlyst concentrtion 1 %wt of oil, nd N Re =,000. Concentrtion (mol/l) Fig. 4 xperimentl points nd simultion curve for diglycerides nd monoglycerides during the trnsesterifiction of plm oil t molr rtio 6:1, temperture 60 C, ctlyst concentrtion 1 %wt of oil, nd N Re =,000; (, dshed line) diglycerides; (, full line) monoglycerides. Tble 1 The rection rte constnts (L/mol.s) k k k k k k In ddition, lthough trnsesterifiction of TG with lcohol consists of three stepwise nd reversible rections, the rection rte constnts in Tble 1 show tht the forwrd rections re much fster thn reverse rections. In the 1 st rection (TG ), the rection rte constnt is finite higher for the forwrd rection. For the nd nd the 3 rd rections ( nd GL), the rection rte constnts of the forwrd rections re 1.4 nd 65. times of the reverse rections. Indeed, the forwrd rections re the most importnt. purity of biodiesel is represented by the weight percentge of M in the product s shown by the following eqution W M Purity (%wt) = 100% W W W W M TG (7) where W M, W TG, W, W re weight percentges of M, TG,, nd in the product on glycerol-free bsis (%wt). Fig. 5 shows experimentl purity of the rection in the first 10 min of the rection time. The production rte of M ws rpid t the beginning of the rection followed by lower production rte s rection pproched equilibrium. The production rte in this study ws little fster thn the production rte tht Drnoko nd Cheryn [1] obtined, especilly t the beginning of the rection. It might be due to the effect of using NOH s ctlyst insted KOH. The weight percentge of M in the product surged from zero to 94.3 %wt within min. The production rte then decresed nd reched the equilibrium within round 0 min. The mximum purity t equilibrium ws 99.1 %wt. This result reveled tht the equilibrium ly towrds the products of the rection. Schwb et. l. [1] nd Freedmn et. l. [8] lso observed nd suggested tht the equilibrium of trnsesterifiction of vegetble oils ly towrds the production of ftty cid esters nd glycerol. ccording to the uropen Union stndrds for lterntive diesel fuels, the minimum cceptble purity of biodiesel is 96.5 %wt []. Thus, the effect of GL seprtion is not significnt fctor even in the lter stge of the rection for biodiesel production by btch process. Moreover, Fig. 5 lso shows tht simultion result fit well with experimentl result. The R nd the MRD between the simultion result nd the experimentl result for the purity during the rection were nd 1.835%, respectively. The highest concentrtions of nd were observed t 0.5 min. Their levels then decresed nd reched equilibrium. t this point, TG disppered, but little of nd still remined. Furthermore, Figs. -4 lso show tht simultion concentrtions of ll compositions fit well with experimentl concentrtions. Tht is, second-order kinetics mechnism coupled with the obtined rection rte constnts provides stisfctory mechnism for plm oil trnsesterifiction. The obtined rection rte constnts re shown in Tble 1. 80

5 3-05 (O) The Joint Interntionl Conference on Sustinble nergy nd nvironment (S) 1-3 December 004, Hu Hin, Thilnd Purity (%wt) Fig. 5 xperimentl points nd simultion curve for the purity during the trnsesterifiction of plm oil t molr rtio 6:1, temperture 60 C, ctlyst concentrtion 1 %wt of oil, nd N Re =, CONCLUSIONS The production rtes of M strted with sudden surge followed by lower production rte s the rection pproches equilibrium. n initil mss trnsfer-controlled region in this study ws very short. Hence, the mixing intensity (N Re =,000) could provide sufficient mixing to eliminte n effect of mss trnsfer for trnsesterifiction of plm oil. Moreover, the obtined rection rte could ensure tht it ws true intrinsic rection rte of homogeneous rection. second-order kinetics without shunt rection coupled with the obtined rection rte constnts provided stisfctory mechnism for plm oil trnsesterifiction. The R nd the MRD between the simultion result nd the experimentl result for the weight percentge of methyl esters in the product during the rection were nd 1.835%, respectively. The rection rte constnts lso reveled tht the forwrd rections were the most importnt. In ddition, the mximum purity t equilibrium ws 99.1 %wt. The equilibrium ly towrds the products of the rection. Thus, glycerol seprtion ws not significnt fctor for producing sleble biodiesel by btch process. CKNOWLMNTS The uthors cknowledge Ministry of University ffirs Thilnd nd The Joint Grdute School of nergy nd nvironment t King Mongkut s University of Technology Thonburi, Thilnd, for the scholrship nd reserch fund provided to T. Leevijit. [5] Frks, L., Schchter, O. nd Vromn, B.H. (1949) On the rte of cid-ctlyzed Trnsesterifiction of Butyl lcohol nd thyl cette, J. m. Chem. Soc., 71, pp [6] Dvies, B. nd Jeffreys, G.V. (1973) Continuous Trnsesterifiction of thyl lcohol nd Buthyl cette in Sieve Plte Column: II. Btch Rection Kinetics Studyies, Trns. Inst. Chem. Soc., 51, pp [7] Sridhrn, R. nd Mthi, I.M. (1974) Trnsesterifiction Rections, J. Sci. Ind. Res., 33, pp [8] Freedmn, B., Butterfield, R.O. nd Pryde,.H. (1986) Trnesterifiction Kinetics of Soyben Oil. JOCS, 63, pp [9] Noureddini, H. nd Zhu, D. (1997) Kinetics of Trnesterifiction of Soyben Oil, JOCS, 74, pp [10] Link, W.. (1989) Method C 5-40: Smpling nd nlysis of Commercil Fts nd Oils for Free Ftty cids, Officil Methods nd Recommended Prctices of the mericn Oil Chemists Society-4 th ed. [11] Oldshue, J.Y. (1983) Fluid Mixing Technology, Mc-Grw-Hill, United Sttes of meric. [1] Freedmn, B., Pryde,.H. nd Mounts, T.L. (1984) Vribles ffecting the Yields of Ftty sters from Trnesterified Vegetble Oils, JOCS, 61, pp [13] Diskou, M., Louloudi,. nd Ppynnkos, N. (1998) Kinetics of the Non-ctlytics Trnsesterifiction of Soyben Oil, Fuel, 77, pp [14] Srivstv,. nd Prsd, R. (000) Triglycerides-Bsed Diesel Fuels, Renewble & Sustinble nergy Reviews, 4, pp [15] Kincs, F.R. (1985) Met ft formultion, JOCS, 73, pp [16] Perry, R.H., Green, D.W. nd Mloner, J.O. (1997) Perry s Chemicl ngineers Hndbook-7 th ed., , Mc-Grw-Hill, ustrli. [17] Fogler, H.S. (1999) lements of Chemicl Rection ngineering 3 rd ed., , Prentice Hll, US. [18] Hoffmn, J.D. (1993) Numericl Methods for ngineers nd Scientists, McGrw-Hill, New York. [19] Boocock, G.B., Konr, S.K., Mo, V., Lee, C. nd iilign, S. (1998) Fst Formtion of High-Purity Methyl sters from Vegetble Oils, JOCS, 75, pp [0] Smith, J.M. (1981) Chemicl ngineering Kinetics-3 rd ed., Mc-Grw-Hill. [1] Schwb,.W., Bgby, M.O. nd Freedmn, B. (1987) Preprtion nd Properties of Diesel Fuels from Vegetble Oils, Fule, 66, pp [] Krosmnoglu, F., Cigizoglu, K.B., Tuer, M. nd rtekin, S. (1996) Investigtion of the Refining Step of Biodiesel Production, nergy Fuels, 10, pp RFRNCS [1] Drnoko, D. nd Cheryn, M. (000) Kinetics of Plm Oil Trnesterifiction in Btch Rector, JOCS, 77, (1), pp [] Krwczyk, T. (1996) Biodiesel-lterntive Fuel Mkes Inrods but Hurdles Remin, INFORM, 7, pp [3] Shy,.G. (1993) Diesel Fuel from Vegetble Oils: Sttus nd Opportunities, Biomss nd Bioenergy, 4, pp [4] Thi Prliment, 00. lterntive Fuels: thnol nd Biodiesel. 81