(693h) Effect of Scale-up on Mass Transfer and Flow Patterns in Liquid-Liquid Microchannels Using Experiments and Computations

In this work, we developed a correlation to predict the mass transfer rate and investigated the impact of the microchannel diameter and solvents on two-phase liquid-liquid flow patterns. To achieve this, we use machine learning (ML) models, including random forest¹ and symbolic genetic regression¹, to predict flow patterns and mass transfer rate, respectively. Experimental and Computational Fluid Dynamics (CFD) data were used to train the models to understand the impact as we increase the diameter. Active learning² was employed to minimize the number of CFD simulations and to improve model accuracy. A new methodology is proposed to train the ML model using a mixture of computational and experimental data while considering uncertainties. The uncertainty of both the models (random forest to predict flow patterns and correlation built to predict mass transfer rate coefficient) was also studied using the gaussian inference method.³

Keywords: Microreactors, scale-up, mass transfer, flow pattern, slug, dimensionless numbers, surrogate model, hybrid data set, active learning, symbolic regression

(1) Hastie, T.; Tibshirani, R.; Friedman, J. H. The Elements of Statistical Learning: Data Mining, Inference, and Prediction, 2nd ed.; Springer series in statistics; Springer: New York, NY, 2009.

(2) Active Learning; Settles, B., Ed.; Synthesis lectures on artificial intelligence and machine learning; Morgan & Claypool: San Rafael, Calif., 2012.

(3) Chen, W.; Cohen, M.; Yu, K.; Wang, H.-L.; Zheng, W.; Vlachos, D. G. Experimental Data-Driven Reaction Network Identification and Uncertainty Quantification of CO2-Assisted Ethane Dehydrogenation over Ga2O3/Al2O3. Chemical Engineering Science 2021, 237, 116534. https://doi.org/10.1016/j.ces.2021.116534.