NON-ISOTHERMAL MODEL OF THE YEASTS GROWTH IN ALCOHOLIC FERMENTATIONS FOR HIGH QUALITY WINES

Transcription

1 NON-ISOTHERMAL MODEL OF THE YEASTS GROWTH IN ALCOHOLIC FERMENTATIONS FOR HIGH QUALITY WINES Pablo M. Aballay (a), Gustavo J.E. Scaglia (a), Martha D. Vallejo (b), Laura A. Roríguez (b), Oscar A. Ortiz (a) (a) Instituto e Ingeniería Química, Universia Nacional e San Juan, Av. Libertaor San Martín Oeste 1109, San Juan J5400ARL, Argentina (b) Instituto e Biotecnología, Universia Nacional e San Juan, Av. Libertaor San Martín Oeste 1109, San Juan J5400ARL, Argentina (a) paballay@unsj.eu.ar ABSTRACT Since, one of most critical variables influencing bioprocess in winemaking is temperature; a nonisormal phenomenological moel for yeast kinetics in winemaking fermentation was evelope. The propose moel, base on a previous publishe by authors, consiers a new expression for maximum specific growth rate an two more kinetic parameters epening on operation temperature. They are: specific eath rate an carbon ioxie at 85-95% of its maximum value. The evelope moel was valiate by accurately preicting growth of own lab-scale fermentations an, it also verifies to follow typical tren of literature experimental ata. For such purpose, moel performance between 10 to 40ºC was evaluate via simulations for constant an variable temperature preefine trajectories. Since obtaine results are satisfactory, this moel can be use to track complex temperature profiles to achieve high quality wines, as well as, in or control an optimization strategies. Keywors: non isormal operation, wine fermentation, phenomenological moelling, yeasts growth 1. INTRODUCTION Argentina is largest wine proucer in South America. A range of fine wines has raise ir incorporation in most important international markets in last years, an some of m are among top rate wines in worl. The customers increasing eman for high quality wines an its marke preferences for outstaning organoleptic properties of wine, presents new challenges for winemaking technology. The bioreactor bulk temperature is a well-known critical variable that etermine kinetics of fermentation (Coleman, Fish, an Block 2007). Temperature irectly influences on microbial ecology of grape must an biochemical reactions of yeasts (Fleet an Hear 1993). Moreover, it is known that Saccharomyces cerevisiae synsizes aroma compouns uring winemaking fermentations. It is also state that prouction, quality, quantity an rate of yeast-erive aroma compouns is affecte by temperature use. Typically, temperatures ranging between 15ºC (for white wines) an 30ºC (for re wines) are use. Furrmore, most winemaking fermentations are not carrie out at constant temperature. Experiments conucte at constant temperature, reveale that prouction of compouns relate to fresh an fruity aromas is favoure at temperatures near 15 C, while flowery relate aroma compouns are better prouce at 28 C (Molina, Swiegers, Varela, Pretorius, an Agosin 2007). In relation with some sensory-relevant flavour generation, it was suggeste that higher temperatures, near 28 ºC, are only beneficial at start of fermentation, an n lower temperatures will be avantageous ue to ecrease of volatility an removal of aroma compouns forme (Fischer 2007). It is evient that temperature strongly affects quality of wine (Torija, Rozès, Poblet, Guillamón, an Mas 2003), an new technologies must inclue variable temperature trajectories throughout fermentation. The evelopment of efficient control strategies for main operation variables in fermentations such as ph, temperature, issolve oxygen concentration; agitation spee, foam level, an ors nee accurate ynamic moels (Morari an Zafiriou 1997, Henson 2003). Also, wine fermentation moels are useful tools to assure wine quality an reproucibility among batches (Zenteno, Pérez-Correa, Gelmi, an Agosin 2010). In previous reports, authors have evelope isormal an non-isormal first-principles an hybri neural moels, an an improve isormal phenomenological moel with satisfactory capability to approximate wine fermentation profiles (Vallejo, Aballay, Toro, Vazquez, Suarez, an Ortiz 2005; Ortiz, Aballay, an Vallejo 2006; Aballay, Scaglia, Vallejo, an Ortiz 2008; Scaglia, Aballay, Mengual, Vallejo, an Ortiz 2009). Page 143

2 The objective of this work is to propose a nonisormal phenomenological moel for wine fermentation kinetics, able to preict with enough accuracy yeasts growth an track complex temperature profiles from 10 to 40ºC, to prouce wines with high quality. The propose moel couples mass an energy balances preicting behaviour of main state process variables: viable cells, substrate (total fermentable sugars) an ethanol concentrations, carbon ioxie, an bioreactor temperature. It is base on one evelope by Scaglia, Aballay, Mengual, Vallejo, an Ortiz (2009), that possesses a goo performance for isormal fermentations, an one presente by Aballay, Scaglia, Vallejo, an Ortiz (2008) for non-isormal fermentations. In latter case, operating temperature ranges from 20 to 30ºC. A set of orinary ifferential equations (ODE), incluing heat transferre between reactor an its cooling jacket, constitute present moel. Balances have been couple by means of Arrhenius equation which escribes temperature influence on cell growth (Aballay, Vallejo, an Ortiz 2006; Aballay, Scaglia, Vallejo, an Ortiz 2008) an eath rates (Phisalaphong, Srirattana, an Tanthapanichakoon 2006). Kinetic parameters of moel were ajuste using experimental ata obtaine from anaerobic labscale cultures of Saccharomyces cerevisiae (killer), an/or Cania cantarelli yeasts in Syrah must (regrape juice), see Toro an Vazquez (2002). In case of specific parameters in Arrhenius expression, y were ajuste by least-square metho. Since in practice, temperature in bioreactor must be maintaine constant at a certain level to avoi quality prouct ecrease, or varie tracking a preefine trajectory to achieve a wine with particular organoleptic properties (Ortiz, Vallejo, Scaglia, Mengual, an Aballay 2009), performance of moel was teste via simulation to valiate it. Results from moel simulations an valiation are shown. They state suitable agreement with own experimental an publishe ata, preicting fermentation evolution without significant retars. The latter allows moel application in avance control strategies for winemaking process. The work is organise as follows. First, labscale fermentation experiments carrie out with variable temperature to valiate moel to evelop are escribe. Secon, non-isormal kinetic moelling of bioprocess from formerly evelope isormal an non-isormal moels is presente. Thir, moel simulation results are compare to: literature ata to verify y well track growth trens an, own experimental ata for its valiation. Fourth, a iscussion on appliance of obtaine moel in complex control an optimization schemes in winemaking, an conclusions are expose. 2. MATERIALS AND METHODS Microorganism: Saccharomyces cerevisiae, (strain PM16, obtaine in our laboratory (b) ), maintaine in agar-yepd (yeast extract-peptone-extrose), an propagate in re-grape must. Culture meium: concentrate re-grape must, properly ilute to obtain 23ºBrix at 23ºC, initial ph was set to 3.5, an sterilize at 121ºC uring 20 minutes. Fermentations (FER3): 250 ml flasks containing 100 ml of sterile must was inoculate with 2x10 6 yeast, cappe with Muller s valves, an culture in anaerobic conitions, at temperature following sequence from 23ºC to 18ºC, presente in Fig. 4. Samples were taken each 6 hours uring first 7 ays an n each ay; yeasts were accounte by means a Neubauer chamber, fermente must was centrifugate an supernatant was maintaine for analytical eterminations. 3. MATHEMATICAL MODEL In winemaking conitions, main bio-reactions can be synsise by reuctive pathway S X + P +, this reaction means that substrates (S, glucose an fructose an sucrose, after ir hyrolysis as limiting substrate), in anaerobic conitions, are metabolise to prouce a yeast population (X), ethanol (P, mainly prouce by yeast through Emben- Meyerhof-Parnas metabolic pathway) an carbon ioxie ( ). The ethanol-formation reaction from glucose is: CH O 2CHCHOH+ 2CO (1) The metabolite accumulation in extra- meium has been moelle by a set of ODE base on mass balances on X, S, P an CO2 which change with time t [h] like in isormal moel of Scaglia, Aballay, Mengual, Vallejo, an Ortiz (2009), which can be seen for furr etails, an it is summarise as Eqs. 2 through 5: Viable cells: ( CO2 C 0295 ( )) X e S = Aμ t e + e S + Ks B X... X 1 + S Aμ m a ( S + KsB ) β ( CO2 C 0295 ( )) ( CO2 C 0295 ( )) m a ( CO2 C0 295 ( )) e S 1 CX K ( CO2 C 0295 ( )) ( CO2 C 0295 ( )) X + e + e t Substrate: S 1 S = X μm EX FX b t YX / S S + Ks B... (2) (3) Carbon ioxie : Page 144

3 CO2 S = Gμm X + c t S + Ks B 2 S + H μm X + IX e t ( S + KsB )( S + KsB ) Ethanol: (4) can be expresse in a general way as: { X t, S t, CO2 t an P t } = f ( X, S, CO2, μ m( T), K( T), CO2(95) ( T)). The mamatical expressions for three kinetic temperature-epenent parameters are given in Eqs. (7), (8) an (9): P 1 CO t Y t 2 = (5) CO2 / P Numerical values of preceing moel parameters an ir escription are shown in Table 1 in Appenix. Moel assumptions are: or mass balance parameters of moel, incluing ph, are constant. Fermentation is not nitrogen source-limite; this is viable, base on information about chemical composition of local re-grape musts. Moreover, local winemakers only a nitrogen supplementation, in excess, to correct white-grape musts. In energy balance (Eq. 6): heat losses ue to evolution, water evaporation an ethanol an flavour losses are neglecte; average grape juice-wine ensity an specific heat, an all physical properties are uniform in mass bulk. They are constant with (bioreactor) temperature T [K] an time. Convective heat transfer coefficient of fermentation mass, implicitly inclue in Eq. (6), is constant (Colombié, Malherbe, an Sablayrolles 2007). In cooling jacket: water properties variations an fouling factor in jacket sie are neglecte. Heat transfers by raiation an conuction are negligible. The non-isormal kinetic moel is constitute by mass balances of before-mentione moel an energy balance in reactor an its cooling water jacket. ( ρ ) r Vr Cpr T CO2 = YH / CO V 2 r Q (6) t t V r [m 3 ] is volume. Y H/CO2 [W h prouce/ ] is energy ue to carbon ioxie by bio-reaction. It was obtaine by stoichiometry (Eq. 1) from Y H / S, likely energetic yiel on substrate consume. Q [W] represents exchange heat between mass an cooling jacket (see etails in Aballay, Scaglia, Vallejo, an Ortiz 2008). ρ r [ m -3 ] an C pr [W h -1 K -1 ] are ensity an specific heat of mass. Mass an energy balances are couple by means of: Arrhenius equation for maximum specific growth an eath rates, μ m [h -1 ] an K [h -1 ] respectively, an polynomial regressions for imensionless coefficients L within μ m, an M within parameter for estimation of carbon ioxie at 85-95% of its maximum value (95). The above mentione bioprocess variables progress in time an, temperature influence on m an ir parameters T e μm = γ. L. 1+ e Ea R T ΔG R T (7) γ is maximum growth rate per Kelvin egree [h -1 K -1 ], L is a imensionless coefficient epening on temperature (Eq. 10), E a is activation energy for cell growth [kj kmol -1 ] an ΔG [kj kmol -1 ] is Gibbs free energy change of fermentation reaction. R is general gases constant [kj kmol -1 K -1 ]. K E RT = K,0 T e if T 304K (8) Orwise K replaces parameter D in moel of Scaglia Aballay, Mengual, Vallejo, an Ortiz (2009). K,0 is specific eath rate per Kelvin egree an E is activation energy for eath [kj kmol -1 ]. Moreover, parameters E a, ΔG, K,0, an E, were ajuste by least-square metho, using experimental ata obtaine from anaerobic lab-scale cultures of Saccharomyces cerevisiae (killer) an Cania cantarelli yeasts, with Syrah must in batch moe (Toro an Vazquez, 2002). CO = CO M (9) * 2(95) 2(95) CO * 2 (95) is a carbon ioxie value, chosen between 85% an 95% of total carbon ioxie at constant temperature (296K) an, M is a imensionless coefficient epening on temperature (Eq. 11). L M = f T - g T + h T - i T + j T - k (10) = lt - m T + n T - o T + p T - q (11) Where f, g, h, i, j, k, l, m, n, o, p, an q are own coefficients of moel, see Table 2 in Appenix. Initial conitions an anor parameters of moel use for simulating experimental fermentations from literature are escribe in Table 2 in Appenix. Those fermentations are mentione as: FERT, of Torija Rozès, Poblet, Guillamón, an Mas (2003), FER1 an FER2 (Toro an Vazquez 2002) an FER3 from own ata. The latter was carrie out to valiate present moel. In aition, maximum values of viable cells concentration achieve uring fermentations are inclue in Table 2 (Appenix) as well. Page 145

4 4. SIMULATIONS 4.1. Results The evelope moel was teste via simulations in similar conitions than experimental fermentations from literature. To carry out simulations, moel was coifie in Matlab TM software. Results are presente in this section. Results are expresse on normalise yeast concentrations in orer to allow comparisons between ifferent yeast strains, having ifferent masses. Figure 1, represents yeasts growth profiles attaine by simulations at ifferent constant temperatures ranging from 283K (10ºC) to 313K (40ºC) for same initial conitions of substrate an yeasts concentration. It can be seen that yeast growth an maximum cells concentration are favoure at temperatures between 290 an 300K. At lower temperatures than 290K, it is observe that yeast growth is elaye an maximum cells concentration achieve is lower than attaine between 290 an 300K. For initial temperatures higher than 300K, not only yeast growth is iminishe but also yeast eath is anticipate. This situation may be ue to ual effect of temperature over optimal growth conitions an ethanol-tolerance. same use to fit moel in Fig. 1. Thus, se yeasts may have similar performance but not exactly same, because each strain has a proper behaviour pattern. Although, comparison between Figs. 1 an 2 is promising, moel performance must be contraste with experiences carrie out at variable temperature profiles. Figure 2: Normalise Viable Cells Profiles: Experimental Fermentations at Different Temperatures ( K) (Torija, Rozès, Poblet, Guillamón, an Mas 2003); Max.= cfu ml Moel valiation The moel valiation was accomplishe by simulation as well, using initial conitions of ifferent own labscale experimental ata sets at ifferent constant an variable temperature profiles. As Fig. 3 shows that for fermentations FER1 an 2 (both at constant 296±1K), moel propose has an acceptable preiction with only up to 10 hours average in retar regaring experimental ata. Figure 1: Normalise Viable Cells Profiles: Moelle Results at Different Temperatures ( K); Max. = cfu ml -1 (Parameter Values from FER2) In orer to contrast simulation results obtaine with experimental ata, it was constructe a 3D-mesh plot (Fig. 2), taking ata from literature an reporting experiences of wine fermentations at ifferent constant initial temperatures (Torija, Rozès, Poblet, Guillamón, an Mas 2003). Comparing Figs. 1 an 2, it can be seen that behaviour of moel approximates appropriately to mentione experimental ata from literature. Even though, in last case (Fig. 2), at low temperatures, yeast growth seems to be more retare than in Fig. 1. It is necessary to point out that yeast use by Torija, Rozès, Poblet, Guillamón, an Mas (2003), was not X [Normalise yeasts concentration] Moelle Experimental Time [h] a Page 146

5 X [Normalise yeasts concentration] Moelle Experimental Time [h] Figure 3: Normalise Viable Cells Profiles: Moelle an Experimental Fermentations: (a) FER1 (Max.= cfu ml -1 ) an (b) FER2 (Max.= cfu ml -1 ) both of m at 296±1K Figure (4a), presents moel preictions an experimental results for fermentations performe at a preefine temperature profile, Fig. (4b), fixe from biochemical consierations on yeasts growth an yeastrelate aroma compouns. X [Normalise yeasts concentration] T [K] Moelle Experimental Time [h] Time [h] Figure 4: (a) Normalise Viable Cells Profiles: Moelle an Experimental Fermentation FER3 (Max.= cfu ml -1 ), at (b) a Specific fermentation temperature profile ( K) Furrmore, results in Fig. 4, constitute effective moel valiation since that it allows to preict viable yeasts population when an optimal temperature profile is state. b a b Table 1, illustrates a quantitative comparison of obtaine results in Figs. 3 an 4. Accoring to Scaglia, Aballay, Mengual, Vallejo, an Ortiz (2009): firstly, it is use mean absolute error (MAE, Eq. 12) that also has been use to preict biomass in this case, MAE = n 1 X mo n X exp (12) n is number of experimental ata, X mo preicte value of biomass (cells concentration) an, X exp experimental one. Subsequently, effectiveness of presente moel was assesse by means of percentage mean error (ME%, Eq. 13) with respect to experimental range of variable expresse by its maximum value (X exp,max ); this, also regars fermentation progress an its control (Malherbe, Fromion, Hilgert an Sablayrolles 2004). MAE ME% = 100 (13) X exp, max Lastly, in Table 1, it is expose that both errors are into a typical maximum limit in biotechnology an process engineering of 10% with respect to ata range of variable biomass, which is compensate with an experimental measurement error of about similar value. Funamentally, preicte profiles o not show appreciable time retars with respect to experimental ata an achieves an enhance precision by estimating biomass compare to own (Ortiz, Aballay, an Vallejo 2006), an or first-principles moels like ones of Coleman, Fish an Block (2007), an Phisalaphong, Srirattana, an Tanthapanichakoon (2006), respectively. This fact was attaine with an aitional critical variable as temperature an new parameters in propose moel. Hence, it woul be possible to apply it in control algorithms to track with proximity esire fermentation trajectories without significant elays in control actions. Such characteristic is particularly essential uring winemaking process, since a elaye control action on variables, such as temperature or ph, can generate a sluggish or stuck fermentation or egraation in organoleptic properties of wine. In aition, moel can be use at inustrial scale with some aaptation, given that, or nonisormal moels evelope from lab-scale alcoholic fermentations have been valiate or teste with goo performance, or highlighte ir possible aaptation, taking into account scale-up effects (Phisalaphong Srirattana, an Tanthapanichakoon 2006; Colombié, Malherbe, an Sablayrolles 2007, Malherbe, Fromion, Hilgert an Sablayrolles 2004; Coleman, Fish, an Block 2007). In work of Zenteno, Pérez-Correa, Gelmi, an Agosin (2010), moel was valiate for a 10 m 3 inustrial tank. Page 147

6 Table 1: Comparison between Simulate an Experimental Results for Viable Cells Concentration Fermentation MAE [10 6 cfu ml -1 ] ME% FER FER FER CONCLUSIONS A first-principles moel, for non-isormal alcoholic fermentations in wier conitions of winemaking temperature, has been presente in this work. Since bioprocess is strongly affecte by temperature in aroma an flavour prouction, final wine quality epens on monitoring an controlling on this variable. Therefore, moel obtaine consist of mass balances, preicting state variables (viable cells, substrate an ethanol concentrations, an ), couple with an energy balance of system. The latter is one by means of growth an eath parameters, an (95) parameter, all of m in function of temperature in an interval from 10 to 40 C. The evelope moel has been satisfactorily valiate via simulation with publishe an own experimental ata, showing a proper behaviour to preict growth kinetics at constant an variable preefine temperature profiles. This allows isposing of a reliable moel to: approximate state variables trajectories an propose avance control an optimization strategies. The moel valiation reaches to lab-scale winemaking fermentations. It is possible to use it at inustrial scale, in that case, it may be necessary inclue some aspects not consiere such as: mixing of fermentation mass an spatial concentration graients, heat transfer, etc. In aition, or topics will be inclue in next contributions, such as: to track or variables of bioprocess as, substrate an ethanol concentrations,, ensity an/or ph; to show an extensive sensitivity stuy for moel variables an parameters; to improve parameter estimation with artificial intelligence tools; to make efforts to reuce winery cooling requirements even though process emans specific cooling protocols to maintain low temperatures that protect wine quality. ACKNOWLEDGMENTS We gratefully acknowlege Universia Nacional e San Juan an National Council of Scientific an Technological Research (CONICET), Argentina, by financial support to carry out this work. APPENDIX Table 1: Coefficients an parameters values from isormal fermentation moel of Scaglia, Aballay, Mengual, Vallejo, an Ortiz (2009), use in present non-isormal moel for three fermentations. Description Unit Value* Fitting Coefficient FERT FER1 FER3 a b c e A B C E F G H I Ks Volume of mass per substrate mass Volume of mass per forme cells an time Specific rate of substrate consumption for maintenance per forme cells multiplie by time per forme cells Similar to G m m 3-1 hr hr hr Kinetic an Yiel Parameters Saturation coefficient in Mono s m equation Page 148

7 β Y X/S Y CO2/P Coefficient in Verlhurst s equation Forme cells per consume substrate Carbon ioxie yiel coefficient base on ethanol m 3-1 h *Values for fermentation FER2 are not shown here. Table 2: Initial conitions, coefficients an parameters use in propose non-isormal moel for three fermentations. Description Unit Value* Initial Conitions X(0) S(0) (0) P(0) T(0) Viable cells concentration (yeasts) Substrate concentration Carbon ioxie evolution Ethanol concentration Bioreactor temperature M-cfu m -3 ** FERT FER1 FER m m -3 0 m -3 0 K *** t(0) Time h 0 Maximum μ m (0) specific h growth rate K (0), (95)(0) Specific eath rate per Kelvin egree between 85-95% of maximum h m -3 (*) (*) (*) L(0) M(0) Q(0) Exchange heat between mass an cooling jacket W Maximum Value Achieve Viable cells X max concentration (yeasts) (**) M-cfu m Fitting Coefficients f g h i j k l m n o p q Physical-chemical an Kinetic Parameters ρ r C pr V r Y H / CO2 γ ΔG E a Density of mass Specific heat of mass Volume of mass Energy ue to carbon ioxie by bioreaction Maximum growth rate per Kelvin egree Gibbs free energy change of fermentatio n reaction Activation energy for cell growth m W h - 1 K m W h prouce/ of h -1 K kj kmol kj kmol Page 149

8 E K,0 * (95) R Activation energy for cell eath Specific eath rate per Kelvin egree between 85-95% of maximum at constant temperature General gases kj kmol h m kj kmol -1 K constant *Values for fermentation FER2 are not shown here. **Millions of Colony Forming Units per cubic meter. ***Different constant temperatures: 288, 293, 298, 303 an 308K. (*) Iem to * CO2(95) values. (**) Values accoring to temperature trajectory require. REFERENCES Aballay, P.M., Vallejo, M.D., Ortiz, O.A., Temperature control system for high quality wines using a hybri moel an a neural control system. Proceeings of XVI Congresso Brasileiro e Engenharia Química-COBEQ 2006, pp September, Santos (Brazil). Aballay, P.M., Scaglia, G.J.E., Vallejo, M.D., Ortiz, O.A., Non isormal phenomenological moel of an enological fermentation: moelling an performance analysis. Proceeings of 10th International Chemical an Biological Engineering Conference - CHEMPOR 2008, pp. 4-6 September, Braga (Portugal). Coleman, M.C., Fish, R., Block, D.E., Temperature-epenent kinetic moel for nitrogen-limite wine fermentations. Applie an Environmental Microbiology, 73 (18), Colombié, S., Malherbe, S., Sablayrolles, J.M., Moeling of heat transfer in tanks uring winemaking fermentation. Foo Control, 18, Fischer, U., Wine Aroma. In: Berger, R. G., e. Flavours an Fragrances, Chemistry, Bioprocessing an Sustainability. Berlin Heielberg: Springer, Fleet, G.H., Hear, G.M., Yeasts: growth uring fermentation. In: Fleet, G.M., e. Wine Microbiology an Biotechnology. Chur (Switzerlan): Harwoo Acaemic Publishers, Henson, M. A., Dynamic moeling an control of yeast cell populations in continuous biochemical reactors. Computers an Chemical Engineering, 27 (8-9), Malherbe, S., Fromion, V., Hilgert, N., Sablayrolles, J.M., Moeling effects of assimilable nitrogen an temperature on fermentation kinetics in enological conitions. Biotechnology an Bioengineering, 86 (3), Matlab TM, Version Release 14. User guie. Natick (Massachusetts, USA): The MathWorks, Inc. Molina, A.M., Swiegers, J.H., Varela, C., Pretorius, I.S., Agosin, E., Influence of wine fermentation temperature on synsis of yeast-erive volatile aroma compouns, Applie Microbiology & Biotechnology, 77, Morari, M., an Zafiriou, E., Robust Process Control. UK: Pearson Eucation. Ortiz, O.A., Aballay, P.M., Vallejo, M.D., Moelling of killer yeasts growth in an enological fermentation by means of a hybri moel. Proceeings of XXII Interamerican Congress of Chemical Engineering an V Argentinian Congress of Chemical Engineering. Innovation an Management for Sustainable Development, A. 13b-224, pp October, Buenos Aires (Argentina). Ortiz, O.A., Vallejo, M.D., Scaglia, G.J.E., Mengual, C.A., Aballay, P.M., Avance Temperature Tracking Control for High Quality Wines using a Phenomenological Moel. In: e Brito Alves, R.M., Oller o Nascimento, C.A., an Chalbau Biscaia Jr., E., es. 10th International Symposium on Process Systems Engineering - PSE2009, 27, Part A, Computer Aie Chemical Engineering - Series. Amsteram (The Nerlans): Elsevier B.V., Phisalaphong, M., Srirattana, N., Tanthapanichakoon, W., Mamatical moeling to investigate temperature effect on kinetic parameters of ethanol fermentation. Biochemical Engineering Journal, 28, Scaglia, G.J.E., Aballay, P.M., Mengual, C.A., Vallejo, M.D., Ortiz, O.A., Improve phenomenological moel for an isormal winemaking fermentation. Foo Control, 20, Torija, M.J., Rozès N., Poblet M., Guillamón J. M., Mas A., Effects of fermentation temperature on strain population of Saccharomyces cerevisiae. International Journal of Foo Microbiology, 80, Toro, M.E., Vazquez, F., Fermentation behavior of controlle mixe an sequential cultures of Cania cantarellii an Saccharomyces cerevisiae wine yeasts. Worl Journal of Microbiology & Biotechnology, 18, Vallejo, M. D, Aballay, P. M., Toro, M. E., Vazquez, F., Suarez, G. I., Ortiz, O. A., Hybri Page 150

9 Moeling an Neural Preiction of Wil Killer Yeast Fermentation Performance in a Winemaking Process. Proceeings of 2n Mercosur Congress on Chemical Engineering an 4th Mercosur Congress on Process Systems Engineering, Paper coe 230, pp August, Rio e Janeiro (Brazil). Zenteno, M.I, Pérez-Correa, J.R., Gelmi, C.A., Agosin, E., Moeling temperature graients in wine fermentation tanks. Journal of Foo Engineering, 99, Page 151