Machine learning assisted reactor and full process optimization design for alcohol oxidation

The oxidation of 2-ethylhexanol (2-EHA) to produce 2-ethylhexanoic acid (2-EHAD) suffers from low efficiency and high energy consumption caused by industrial batch production process. To tackle this challenge, we proposed an AI-assisted design methodology for continuous reactor and process of 2-EHA oxidation to 2-EHAD to enhance problem-solving efficiency. Specifically, a precise reactor surrogate model is developed to accelerate the optimization of reactor internals and enhance the utility of experimental data, thereby overcoming the challenge of scarce continuous oxidation experimental data caused by long operating cycles and oxygen safety concerns. Based on optimal reaction parameters, an economic profit growth of 30% to 40% and carbon emissions reduction of 10% to 50% can be attained compared to traditional batch processes and butyraldehyde processes at the same production level. Our work not only propels continuous process design of alcohol oxidation production processes but also lays the groundwork for their widespread industrial application.