Real‐Time Detection of Atmosphere Composition in Three‐Component Gas‐Phase Copolymerization of Olefins

Abstract

The uniformity of copolymer composition has a great influence on its properties; however, it is very difficult to control. Especially in the laboratory research of the multi‐component gas‐phase copolymerization of olefins, the polymerization is usually conducted in a batch reactor, and the composition of the resulting copolymers is often distributed very widely. The samples of such copolymers are in fact not representative. Detection and control of the atmospheric composition in the batch reactor during olefin gas‐phase copolymerization is of great significance for the control of the uniformity of copolymer composition. In this work, a method based on three flow meters in series has been developed for real‐time detection of the atmospheric composition in olefin gas‐phase copolymerization with a delay time of less than 3 s. The method is applicable to copolymerization of up to a ternary system, including the binary copolymerization of olefins with hydrogen modulation of molecular weight or the ternary copolymerization of olefins. The consumption rate of each monomer is also online measured.

On‐Line Monitoring Device for Gas Phase Composition Based on Machine Learning Models and Its Application in the Gas Phase Copolymerization of Olefins

“…Consequently, the composition of the mixed gas can be determined by analyzing the reading ratio of the three flowmeters. [15,18] In our research group's previous work, the traditional fitting equation was used to calculate the gas composition from the readings of the three flowmeters, but the results lacked the desired accuracy. [15,18] In this work, machine learning algorithms, including decision trees, random forests, and XGBoost, were employed to significantly enhance the accuracy and generalization of prediction results.…”

“…Consequently, the composition of the mixed gas can be determined by analyzing the reading ratio of the three flowmeters. [ 15,18 ]…”

“…However, the sensor's complex principle and cumbersome usage limited its ability to effectively analyze multiple gases. Hu [14] and Zheng [15] from our research group adopted flowmeters with various detection principles to derive a fitting equation for gas composition based on the reading ratio of the outlet flowmeter. This approach enabled real-time determination of gas composition within the reactor.…”

“…This study builds upon our previous research [14,15] with three various gas flowmeters integrated in series in the exhaust gas pipeline of a gas-phase copolymerization reactor. By collecting a substantial amount of data from cold model experiments, machine learning models were developed to determine gas composition, resulting in a notable improvement in the accuracy of online gas detection.…”