Otimização da produção do ácido láctico através da fermentação do soro de queijo por Lactobacillus helveticus

Abstract

In this work the homolactic fermentation of cheese whey using Lactobacillus helveticus (ATCC 15009) was studied. The experiments were carried out in a batch reactor, with agitation, temperature and pH control during a period of 32 hours. The influences of four variables according to a composite design on the production of lactic acid were analyzed: temperature, pH, concentration of lactose and concentration of yeast extract, used as a supplement for the fermentation medium. The effects of these variables and its interactions were analyzed by the response surface methodology. The biggest influences were exerted by pH and by the concentration of yeast extract. Through a canonic analysis of the adjusted response surface, the optimal variables values that led to the biggest production of lactic acid were determined: 82 g/L of lactose, 23.36 g/L of yeast extract at 40 ºC and pH 6.8. In these conditions, the concentration of the lactic acid reached 59 g/L. Through a kinetic study, it was proved that the homolactic fermentation of whey is inhibited by the substrate as well as by the product. Models were tested to describe the microbial growth, the substrate consumption and the product formation. The logistic equation represented well the growth of the Lactobacillus helveticus. The biggest deviations occurred at the beginning of the fermentation and at the stationary phase, where the condition of instantaneous rate of growth equal to zero is not predicted by the model. The measures of bias from Box and of curvature from Bates and Watts referring to this model demonstrated that there is a statistic confidence for the least square estimators of the parameters. The adjustment of the growth model from Amrane (1999) for the experimental data was slightly better than the one obtained from the logistic equation. However, the measures of bias from Box and of curvature from Bates and Watts demonstrated that the statistical inferences on the parameters estimative are not valid. Therefore, the possibility of using this model was discarded. The models of substrate consumption from Pirt and of formation of product from Luedeking and Piret fitted well to the experimental data.