چكيده به لاتين
Oxidethylene is one of the most important consumables in the petrochemical industry, which is usually produced using silver/alumina catalyst, exposed to ethylene gas and oxygen. This catalyst is obtained by impregnating silver metal and other enhancers on alpha alumina base. In this research, using the machine learning approach, the most important features of the base and the final catalyst were identified and optimized. First, with the help of the analysis of 313 data collected from the sources, the optimal range of the most effective characteristics of the base was obtained on the selectivity of the catalyst, that is, the specific surface of the base and its strength, and based on this, 14 different formulations of the base were mixed and shaped and heat treated using the extrusion method. The optimal foundation had a specific surface area of 2.80 m2/g and a strength of 30.31 kgf (N 306.74). In the next step, four linear models, decision tree, random forest and gradient promotion, were designed and trained on the collected data in order to predict the selectivity of the catalyst. had 5 Then the proposer model was designed and the optimal amount of silver, the type of boosters and the optimal conditions of the reactor were obtained from the proposer model in order to achieve the highest selectivity. The final catalysts were made in two stages, in the first stage, the prepared bases (based on the optimal composition of the base) were inoculated in a solution containing silver, and in the second stage, they were activated with solutions containing cesium, chlorine and sodium boosters. Finally, the samples were analyzed by specific surface area (BET), phase analysis (XRD), adsorption/desorption analysis (TPD-O2), elemental analysis (AAS) and microstructure analysis (FESEM). After silver inoculation, the catalysts had an apparent density of 2.75 g/cm3 and a volume density of 1.42 g/cm3; Also, according to elemental and phase analysis, the amount of metallic silver and alpha-alumina was found to be 9.60% by weight and 90.12% by weight, respectively. Finally, according to the prediction of the model, the catalyst with cesium and chlorine boosters had the highest selectivity with a value of 5±76%. Also, the catalysts inoculated with silver with the times of thirty seconds and one hour had a specific surface of about 3.70 m2/g and 2.45 m2/g, respectively. Also, the catalyst inoculated with silver in one hour, in the TPD-O2 analysis, showed a better performance than the inoculation of thirty seconds. Finally, according to the TPD-O2 analysis, the catalysts containing chlorine and cesium boosters had the best performance and the highest molecular and atomic oxygen absorption/desorption compared to other catalysts. In the map obtained from the microstructural analysis, the presence of chlorine and cesium elements in the final catalyst, and the uniformity of their distribution, were confirmed.
