Multi-objective versus single-objective optimization of batch bioethanol production based on a time-dependent fermentation model

Traditional process simulators, such as Aspen Plus, are inadequate for optimizing multiple-objective systems in fermentation-based processes. This work uses a novel integrated platform of the robust genetic algorithm optimization in MATLAB linked with an Aspen Plus unsteady-state batch fermentation simulation to optimize the batch ethanolic fermentation process with respect to initial substrate concentration, fermentation time, and in situ product removal. A time-dependent fermentation model that utilizes both glucose and xylose, the major sugars present in lignocellulosic hydrolysate, with Monod cell growth kinetics, substrate and product inhibitions, is used as a model system. The optimized design variables from the multi-objective optimization (MOO) and single-objective optimization (SOO) suggest the typical concentrations of sugars from lignocellulosic hydrolysate must be concentrated to optimize the performance of the batch fermentation process. Furthermore, time-dependent information from an unsteady-state simulation was used to design an integrated batch fermentation with in situ product recovery, allowing higher initial sugars concentrations to be used in the fermentation process (about 50%, for the best optimal solution in the MOO). This resulted in 15% ethanol productivity, 143% total ethanol produced, and 67% fraction of sugar converted improvements relative to the batch fermentation without product recovery. Unlike the single optimal solution from the SOO, MOO presents many equally optimal solutions that can be used as a decision-support tool to guide the choice of design variables for optimum process performance. This study creates a platform that can be used to optimize integrated biorefinery and refinery processes.