The world population is projected to reach 9.7 billion in 2050, which means that the Food and Feed industry is supposed to keep improving its productivity in order to provide all these people with enough food at the same pace. The current trend of regulatory authorities toward application of new process analytical technology tools to improve process understanding as well as reliability and ensure product quality during the production, has awakened the need of investing in novel analytics, especially in (bio)pharmaceutical industries, but is being extended to other fields. Moreover, the increasing acceptance of industrial companies that relevant concentration gradients affecting process performance as well as product quality appear in production vessels, is turning the scale down representation of conditions of the large scale in the lab indispensable. Furthermore, the actual digitalization transformation experienced in everyone´s life is becoming more and more relevant in industrial manufacturing, with the current tendency to develop a so-called digital twin, which simulates the (bio)process running in the plant in silico, thus minimizing out-of-specification batches and allowing near future personnel as well as materials/consumables planning. In this work, (i) electrooptical measurements of cell polarizability as well as size, (ii) single- and multicompartment scale down strategies and (iii) mechanistic modeling of macroscopic variables as well as population heterogeneity were applied to Streptococcus thermophilus fermentations for the first time. Firstly, the at-line determination of bacterial polarizability (i.e. orientation under the application of an electrical field) allowed the elucidation of different growth phases and resulted to be an early indicator of nutrient imbalance as well as growth cessation. Moreover, the analysis of the mean cell size without sample preparation with the same device also allowed the monitoring of qualitative morphological changes during growth. These were verified with parallel flow cytometric analyses, which revealed calibration issues in the equipment preparation, which should be addressed in future experiments. Secondly, pH shifts in the range from 5.5 until pH 8.0 (i.e. pH = +2.0;-0.5) were induced in singlecompartment reactor cultivations leading to a 48.5 % biomass productivity loss in the worst case scenario, while repeated pH pulses in a similar region were performed through ammonia addition in the plug-flow reactor of multi-compartment reactor experiments which yielded a 20 % less cell concentration at the end. Importantly, relevant morphologic changes under the different cultivation conditions were detected: increased chain length under alkali conditions and more homogenous cocci chain length distribution with shorter chains at low pH values. Nevertheless, computational fluid dynamic studies of a 700 L pilot scale fermenter revealed that those scale down conditions were exaggerated in terms of pH-gradients induced: only pH pulses up to 6.3 were monitored throughout a S. thermophilus fermentation under optimal growth conditions, while the pH never dropped below 5.8 far away from the base addition zone. However, extended mixing times and limited power input in the industrial scale may lead to higher pH, so that their effect on process performance and product quality was further assessed. Thirdly, a population balance model based on a mechanistic description of typical growth metabolites (namely biomass, lactose, lactic acid and galactose concentrations) was developed, being able to predict the evolution of certain populations (namely 1-coccus, 2-, 3-, 4- and 5 or more cocci chains) during S. thermophilus cultivation under optimal growth conditions and variable pH-gradients.The application of the first device (EloTrace, EloSystems GmbH, Berlin, Germany) in lactic acid bacteria large scale production would change the current quality by testing mindset to a quality by design/control approach, where the polarizability could be defined as a new critical quality attribute to be maintained inside a certain window by changing critical process parameters during the fermentation. The different scale down concepts applied in this study improved current process understanding of the industrial partner and should encourage the consideration of such lab scale simulators in early process development of new products or in optimization of existing bioprocesses. Finally, the hydrodynamic as well as population balance models developed in this work, if coupled to in situ microscopy technologies to determine cell size distribution in real-time, would enable the implementation of a model-based soft sensor strategy, where population heterogeneity could be minimized by changing critical process parameters, like the tip speed or base addition point.