Modeling and optimization of the extraction of ylang-ylang essential oils using surrogate models from simulated data, coupled with covariance matrix adaptation evolution strategy

The plant Ylang-ylang is a tree native to the subtropical regions. Essential oils are typically extracted from their flowers to be used in several alternative medical treatment options. This extraction is typically performed using steam batch distillation. The literature reveals research efforts in the simulation and optimization of the production of essential oils; nevertheless, these efforts are limited to simplified modeling of the distillation process without a robust coupling with optimization algorithms. In this work, we performed the modeling and optimization of the extraction of Ylang-ylang essential oils using simulated data from a process simulation model to generate surrogate models coupled with a multiobjective optimization algorithm. Our base case simulation model, validated with experimental data, was used to generate process data, which was analyzed using a two-stage Exploratory Data Analysis approach. The ‘clean’ data was then fitted to surrogate models coupled with an optimization algorithm to minimize the duty in the steam batch distillation column and maximize the extraction yield. Economic analysis revealed that the optimized case increases the profit compared to the base case in 9.43% (+3.85 $/h), evidencing that this framework provides a reliable operational envelope of the extraction process that can be used for process design, monitoring, and optimization and potentially apply to other unit operations in the food industry.