Hassan Alasiri

Efficient and selective oxidative dehydrogenation (ODH) catalysts are crucial to advance the production of valuable petrochemicals. In this study, we leverage the power of machine learning to predict dehydrogenation (DH) product yield and unravel the factors influencing the product distribution. A comprehensive data set obtained from experiments conducted in a fixed-bed reactor under varying temperatures, feed ratios (O₂/n-butane), and metal oxide loadings (Ni, Fe, Co, Bi, Mo, W, Zn, and Mn) on an aluminum oxide support served as the basis for model development. Three supervised machine learning models, Boosted Tree (BT), Extreme Gradient Boosting Linear (XGBL), and Support Vector Machine Radial (SVMR), were evaluated. The ensemble technique of the three models showed remarkable accuracy, with an RMSE of 1.65 and MAE of 1.14 on the test data set, and it demonstrated robust generalization capabilities by capturing 87% of the variation in DH yield. In the feature importance analysis of the selected models, Mo, Co, Ni, and W emerged as critical factors influencing the DH yield. The practical significance of these findings lies in their potential to revolutionize catalysis research and industrial applications. The ability of the ensemble model to predict DH yields opens new avenues for optimizing DH products and designing more advanced catalysts. By providing essential insights into the influential variables governing the ODH reactions, researchers can make informed decisions to achieve higher yields and efficiencies.

The development of efficient and selective catalysts for the oxidative dehydrogenation (ODH) of n-butane to produce butenes and butadiene with high performance has been the subject of intense research in recent years. Herein, we report a novel approach for predicting the performance of mixed metal oxides supported on Al₂O₃ for ODH using artificial intelligence (AI). Specifically, artificial neural networks (ANNs), support vector regression with nu parameter (NuSVR), extreme gradient boosting regressor (XGBR), and gradient boosting regression (GBR) machine learning algorithms were trained with a dataset of consistent experimental data to build the chemometric models using reaction temperatures, feed ratios of O₂ : C₄, and catalyst composition as input features to predict the yield of ODH products as a measure of catalyst performance. The results show that the AI-based models can proficiently predict the performance of mixed metal oxide catalysts for ODH of n-butane, with a prediction accuracy of 82%, 89%, 92%, and 94% using ANN, NuSVR, XGBR, and GBR models, respectively. Feature importance analyses also revealed that the amount of Ni loading in the catalyst(s) has the greatest influence on the yield of butenes and butadiene. These findings demonstrate that accurate predictions of catalyst performance can be made even with simple and easily accessible features, thus paving the way for the development and discovery of more efficient catalysts.

The prediction of the yields of light olefins in the direct conversion of crude oil to chemicals requires the development of a robust model that represents the crude-to-chemical conversion processes. This study utilizes artificial intelligence (AI) and machine learning algorithms to develop single and ensemble learning models that predict the yields of ethylene and propylene. Four single-model AI techniques and four ensemble paradigms were developed using experimental data derived from the catalytic cracking experiments of various crude oil fractions in the advanced catalyst evaluation reactor unit. The temperature, feed type, feed conversion, total gas, dry gas, and coke were used as independent variables. Correlation matrix analyses were conducted to filter the input combinations into three different classes (M1, M2, and M3) based on the relationship between dependent and independent variables, and three performance metrics comprising the coefficient of determination (R²), Pearson correlation coefficient (PCC), and mean square error (MSE) were used to evaluate the prediction performance of the developed models in both calibration and validations stages. All four single models have very low R² and PCC values (as low as 0.07) and very high MSE values (up to 4.92 wt %) for M1 and M2 in both calibration and validation phases. However, the ensemble ML models show R² and PCC values of 0.99-1 and an MSE value of 0.01 wt % for M1, M2, and M3 input combinations. Therefore, ensemble paradigms improve the performance accuracy of single models by up to 58 and 62% in the calibration and validation phases, respectively. The ensemble paradigms predict with high accuracy the yield of ethylene and propylene in the catalytic cracking of crude oil and its fractions.