Bioprocess monitoring and chemometric modelling of wine fermentations

Date
2013-03
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: Wine fermentation is a continuously changing biological process whereby the raw product, grape juice is transformed into a high value product wine. In an ideal situation the fermentation kinetics of batch fermentations should follow the same trend over time. This however is not the case in industrial wine fermentations where significant batch-to-batch variation is present. The time trajectories of fermentation processes are therefore often unpredictable in absolute terms. The monitoring of substrate (sugar) and product concentrations (ethanol) as well as other quality parameters during a wine fermentation, is therefore of extreme importance to ensure effective control and management of wine fermentation processes. Conventional methods for fermentation monitoring are however costly, time consuming and often unreliable. For these reasons the modern wine industry requires rapid, reliable, non-destructive monitoring techniques which would meet the criteria of providing critical real-time process information that is displayed in easily interpretable graphical format, in order to ensure the highest quality and continuous consistency throughout all the stages of a process. This research study in particular, addressed the current need for alternative fermentation monitoring strategies that meet these criteria, by evaluating the potential use of spectroscopy as an analytical technique for fermentation monitoring. The overall objective of this study was to use chemometric modelling of information obtained by Fourier transform mid-infrared (FT-MIR) and near-infrared (FT-NIR) spectroscopy, to quantitatively and qualitatively monitor both alcoholic (AF) and malolactic fermentation (MLF) processes. Towards this objective 11 batch fermentations elaborated with Oenococcus oeni and Lactobacillus plantarum strains in respectively a co-inoculation and sequential inoculation scenario, were sampled and analysed at regular time intervals with FT-MIR and FT-NIR spectroscopy and enzymatic reference methods during 2011. Samples were also analysed by gas chromatography flame ionisation detection (GC-FID) and mass spectrometry (GC-MS) at two critical stages during fermentation, namely 50% completion of MLF and 100% completion of MLF, in order to obtain a profile of the evolution of the aroma compounds associated with each inoculation scenario. Three clearly defined research objectives were set for this project. The first objective involved the expansion of the existing quantitative platform for fermentation monitoring. Towards the outcomes of this objective, partial least squares (PLS) calibration models for prediction of malic acid and lactic acid in fermenting must and wines elaborated in our study, were established, based on the MIR and NIR spectra. The models showed excellent predictive abilities in independent test set validation. This outcome made a significant contribution to our existing PLS calibration capacity, particularly towards monitoring of MLF. Quantitative data obtained with the PLS models were also used to graphically project the rate of AF in the different batches, by non-linear fitted regression plots that easily visualised the overall patterns of sugar and ethanol metabolism in the different fermentations. The second research objective involved the qualitative monitoring of fermentations. This approach used FT-MIR and FT-NIR spectra together with chemometrics to identify trends between the different fermentation treatments. Principal component analysis (PCA) clearly projected the time trend from the onset of fermentation, through AF and MLF. No unique bacterial trend was however observed with spectroscopy. These results illustrate the potential of these techniques to be used for modelling of fermentations in industrial situations, through providing critical information about the evolution of the process. Furthermore, these techniques provide tools for identifying problematic and deviating fermentations. A spectral conformity test based on simple calculations of the standard deviation between the absorbance at each recorded wavenumber in the spectra, further confirmed identification of the critical fermentation stages. This technique by-passes the need for spectral interpretation and is a very useful addition, particularly from the industry perspective, to the portfolio of methods established in this study. The third research objective adressed the need to evaluate the possibility to discriminate between the different process stages and LAB treatments using univariate (ANOVA) and multivariate chemometric techniques such as PCA, PLS discriminant analysis (PLS-DA) and Soft Independent Modelling of Class Analogy (SIMCA) for possible future interpretative and classification purposes. The exploratory tool of PCA was used to investigate the similarities and differences between the chemical footprints of the different treatments. PCA showed clear differentiation between the two process stages using chemical quantified data. Differentiation between the LAB treatments was visible with PCA, showing a more prominent separation at 50% completion of MLF. Furthermore the ability of spectroscopy for potential classification was shown using PLS-DA and SIMCA. This profiling study can be seen as a preliminary study setting the ground work for further in-depth research into the profiling of different LAB treatments and inoculation strategies.
AFRIKAANSE OPSOMMING: Wyngisting is ‘n voortdurend veranderende biologiese proses waarvolgens die rou produk, druiwesap, in ‘n produk van hoë waarde, naamlik wyn, verander word. Onder ideale omstandighede sou die gistingskinetika van lotgistings dieselfde tendens oor tyd volg. Dit is egter nie die geval in industriële wyngistings nie, waar noemenswaardige wisseling van lot tot lot teenwoordig is. Die tydsbaan van gistingsprosesse is dus in baie gevalle in absolute terme onvoorspelbaar. Die monitering van substraat- (suiker) en produkkonsentrasies (etanol), sowel as ander kwaliteitsparameters tydens ‘n wyngisting, is van die uiterste belang om doeltreffende beheer en bestuur van wyngistingsprosesse te verseker. Konvensionele metodes vir die monitering van gisting is egter duur, tydrowend en ook soms onbetroubaar. Om hierdie redes vereis die moderne wynbedryf vinnige, betroubare en nie-destruktiewe moniteringstegnieke wat aan die kriteria sou voldoen vir die verskaffing van kritiese, intydse prosesinligting wat in ‘n maklik interpreteerbare grafiese formaat vertoon word om die hoogste kwaliteit en voortdurende konsekwentheid tydens al die stadia van ‘n proses te verseker. Hierdie navorsingstudie het in besonder die huidige behoefte aan alternatiewe gistingsmoniteringstrategieë wat aan hierdie kriteria voldoen, aangespreek deur die potensiële gebruik van spektroskopie as ‘n analitiese tegniek vir die monitering van gisting te evalueer. Die oorhoofse doelwit van hierdie studie was die chemometriese modellering van inligting wat met behulp van Fourier transform middel-infrarooi (FT-MIR) en naby-infrarooi (FT-NIR) spektroskopie verkry is om die prosesse van alkoholiese en appelmelksuurgisting (AMG) kwantitatief en kwalitatief te monitor. Ten einde hierdie doelwit te bereik, is 11 lotgistings met Oenococcus oeni en Lactobacillus plantarum rasse uitgevoer in scenario’s van ‘n gesamentlike inokulasie en opeenvolgende inokulasie onderskeidelik. Monsters van hierdie gistings is met gereelde tydintervalle in 2011 geneem en analises is met FT-MIR en FT-NIR spektroskopie en ensiematiese verwysingsmetodes gedoen. Monsters is ook met gaschromatografiese vlam ionisasie-opsporing (GC-FID) en massaspektrometrie (GC-MS) op twee kritiese stadia tydens gisting geanaliseer, naamlik toe AMG 50% en 100% voltooid was, om ‘n profiel te verkry van die evolusie van die aromaverbindings wat met elke inokulasie-scenario verband hou. Drie duidelik gedefinieerde navorsingsdoelwitte is vir hierdie studie bepaal. Die eerste doelwit het die uitbreiding van die bestaande kwantitatiewe platform vir gistingsmonitering behels. Hiervoor is gedeeltelike kleinstekwadraat [partial least squares (PLS)] kalibrasiemodelle vir die voorspelling van appelsuur en melksuur in die gistende mos en wyne in ons studie op die basis van MIR- en NIR-spektra bepaal. Die modelle het in onafhanklike geldigheidsbepaling van die toetsstel uitstekende voorspellingsvermoëns getoon. Hierdie uitkoms het ‘n noemenswaardige bydrae gemaak tot ons bestaande PLS kalibrasiekapasiteit, veral om AMG te monitor. Die kwantitatiewe data wat met die PLS-modelle verkry is, is ook gebruik om die tempo van alkoholiese gisting in die verskillende lotte grafies uit te beeld deur middel van kromlynige passing van regressiepersele, waarmee dit maklik was om die algehele patrone van suiker- en etanolmetabolisme in die verskillende gistings te visualiseer. Die tweede navorsingsdoelwit het die kwalitatiewe monitering van gistings behels. Hierdie benadering het FT-MIR en FT-NIR spektra tesame met chemometrie gebruik om tendense tussen die verskillende gistingsbehandelings te identifiseer. Hoofkomponentanalise [principal component analysis (PCA)] het duidelik die tydtendens vanaf die aanvang van gisting, deur alkoholiese gisting en AMG, geprojekteer. Geen unieke bakteriese tendens is egter met spektroskopie waargeneem nie. Hierdie uitslae illustreer die potensiaal van hierdie tegnieke om vir die modellering van gistings in industriële situasies gebruik te word deur kritiese inligting oor die evolusie van die proses te verskaf. Verder verskaf hierdie tegnieke gereedskap vir die identifikasie van problematiese en afwykende gistings. ‘n Spektrale gelykvormigheidstoets gebaseer op eenvoudige berekeninge van die standaardafwyking tussen die absorbansie by elke aangetekende golfgetal in die spektra het ook die identifikasie van die kritiese gistingstadia bevestig. Hierdie tegniek omloop die noodsaak vir spektrale interpretasie en is ‘n baie nuttige byvoeging tot die portefeulje van metodes wat in hierdie studie bepaal is, veral vanuit ‘n bedryfsperspektief. Die derde navorsingsdoelwit het die behoefte aangespreek om tussen die verskillende prosesstadia en melksuurbakterie-behandelings te onderskei deur gebruik te maak van eenvariant- (ANOVA) en meervariant- chemometriese tegnieke soos PCA, PLSdiskriminantanalise en sagte onafhanklike modellering van klasanalogie [soft independent modelling of class analogy (SIMCA)] vir moontlike toekomstige verklarende en klassifikasiedoeleindes. Die ondersoekende gereedskap van PCA is gebruik om ooreenkomste en verskille tussen die chemiese voetspore van die verskillende behandelings te ondersoek. PCA het duidelike differensiasie tussen die twee prosesstadia getoon op grond van chemies gekwantifiseerde data. Differensiasie tussen die melksuurbakterie-behandelings was met PCA sigbaar, met ‘n meer prominente skeiding teen 50% voltooide AMG. Verder is die vermoë van spektroskopie vir potensiële klassifikasie met PLS-diskriminantanalise en SIMCA getoon. Die profielsamestelling in hierdie studie kan beskou word as ‘n voorlopige studie vir verdere diepgaande navorsing oor die profielsamestelling van verskillende melksuurbakteriebehandelings en inokulasiestrategieë.
Description
Thesis (MSc)--Stellenbosch University, 2013.
Keywords
Wine fermentations -- Bioprocess monitoring, Wine and wine making, Wine fermentations -- Chemometric modelling, Theses -- Wine biotechnology, Dissertations -- Wine biotechnology
Citation