Browsing by Author "Lambrecht, Kiera Nareece"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    In-Line Monitoring of Red Wine Fermentation
    (Stellenbosch : Stellenbosch University, 2021-03) Lambrecht, Kiera Nareece; Aleixandre-Tudo, Jose Luis; Du Toit, Wessel J.; Nieuwoudt, Helene; Stellenbosch University. Faculty of AgriSciences. Dept. of Viticulture and Oenology. Institute for Wine Biotechnology.
    ENGLISH ABSTRACT: Phenolic compounds may only account for a small percentage of the final composition of a finished red wine but are vital to its sensory attributes. During red wine making, extraction of these phenolic compounds takes place, whereby there is mass transfer from the solids of the grape into the liquid phase. The rate and the extent to which this extraction occurs is dependent on many factors. There are many different methods employed in the wine industry which can influence the composition of the wine. These techniques are varied and can involve manipulating process conditions such as temperature or the addition of certain oenological products. As the final composition of the wine is a major contributing factor to the quality of the wine, it is vital to be able to monitor and control this process. It has been demonstrated through a variety of studies that the use of infrared spectroscopy along with chemometrics provides an avenue for implementation of monitoring and control systems in wineries. However, the limiting factors in these studies are the extensive sample pre-treatment to remove solids before scanning as well as their discrete and off-sight sampling. In the contents of the first research chapter (Chapter 3), a system was designed for the purpose of automatic sampling directly from vessels containing fermenting wines. This was an extensive design process which required separate sampling pumps and sampling lines which delivered samples to a single instrument. Another requirement was automation of different components and synchronisation of these in an individual system. The resulting design was put through a series of stress tests to ensure functionality and reliability. The results showed that the automated system was capable of full-time operation without experiencing component failures and, therefore, it was applied to actual fermentations. For this, 24 hours of real time monitoring was achieved. The turbidity remained a challenge as a perfectly clarified sample was not achievable. This led to the development of partial least squares (PLS) calibrations for three different spectroscopy techniques where the samples used incorporated differing degrees of sample pre-treatment to reduce turbidity. The results of this endeavour compiled in Chapters 4 and 5 showed favourable results for samples with different levels of turbidity as well as for contactless methods of conducting analysis. With further optimisation of the models using spectral pre-processing and wavenumber selection, it was possible to develop models suitable for application in an industrial setting. Finally, in Chapter 6, these models were deployed for use with a series of fermentations, where the ability to monitor phenolic extraction of fermentations receiving different treatments was explored. The results show that the system can be used to monitor trends in phenolic extraction in an industrial set-up. In addition to this, the system has the capacity for updated models and different methods of process control, thereby allowing it to be tailored to each unique scenario.