Data-Driven Optimization of Microwave-Assisted Continuous-Flow Microreactor for Biomass-Derived Carbohydrates' Conversion

There is an emerging interest in deploying microwave technology (MWT) for energy efficient and rapid, volumetric, and selective heating. The utilization of green energy, such as solar and wind, and of biomass to replace fossil fuels for chemicals production makes the technology more appealing. MWT has been applied to a wide range of chemical processes, such as organic synthesis, nanoparticle synthesis, reactive separation, reactive distillation, and catalytic process. However, fundamental understanding of microwave-assisted reactions is limited and this in turn hinders scalability, design and optimization of microwave reactors (MWR).

In this work, we combine computational fluid dynamics (CFD) simulations and experiments to develop understanding of a MW-heated microfluidic systems and leverage machine learning techniques to pave a path for process optimization using CFD models. We first develop a CFD model for a continuous flow MWR in a commercial single-mode microwave applicator and compare to experiments. The effects of various geometric and processing parameters on heating efficiency are investigated. A surrogate model is developed for multi-objective optimization, considering outlet temperature and energy efficiency simultaneously. Moreover, the studies are extended to include fructose dehydration, and the microreactor performance is estimated in terms of sugar conversion, product selectivity and energy efficiency. Importantly, this technology provides insights into and a framework for scale-up and optimization and opens up the possibility for small-scale, intensified and distributed processes for biomass utilization.