Background: The diauxic growth of Saccharomyces cerevisiae on glucose and xylose during cellulose-to-ethanol processes extends the duration of the fermentation and reduces productivity. Despite the remarkable advances in
strain engineering, the co-consumption of glucose and xylose is still limited due to catabolite repression. This work addresses this challenge by developing a closed-loop controller that is capable of maintaining the glucose concentration
at a steady set-point during fed-batch fermentation. The suggested controller uses a data-driven model to measure the concentration of glucose from ‘real-time’ spectroscopic data. The concentration of glucose is then automatically controlled using a control scheme that consists of a proportional, integral, differential (PID) algorithm and a supervisory layer that manipulates the feed-rates to the reactor accounting for the changing dynamics of fermentation.
Results: The PID parameters and the supervisory layer were progressively improved throughout four fed-batch lignocellulosic-to-ethanol fermentations to attain a robust controller able of maintaining the glucose concentration at the pre-defined set-points. The results showed an increased co-consumption of glucose and xylose that resulted in volumetric productivities that are 20–33% higher than the reference batch processes. It was also observed that fermentations operated at a glucose concentration of 10 g/L were faster than those operated at 4 g/L, indicating that there is an optimal glucose concentration that maximises the overall productivity.
Conclusions: Promoting the simultaneous consumption of glucose and xylose in S. cerevisiae is critical to increase
the productivity of lignocellulosic ethanol processes, but also challenging due to the strong catabolite repression ofglucose on the uptake of xylose. Operating the fermentation at low concentrations of glucose allows reducing the effects of the catabolite repression to promote the co-consumption of the two carbon sources. However, S. cerevisiae is very sensitive to changes in the glucose concentration and deviations from a set-point result in notable productivity
losses. The controller structure developed and implemented in this work illustrates how combining data-driven measurements of the glucose concentration and a robust yet effective PID-based supervisory control allowed tight