The solubility of vapors within semicrystalline polyethylenes produced in the gas phase process is examined. Experimentally measured solubilities of ethylene, 1-butene, isobutane, isopentane, 1-hexene, and n-hexane as a function of temperature, pressure, and vapor composition are reported for a series of semicrystalline linear low density polyethylenes, and a model for predicting these results is presented in detail. This paper demonstrates that conventional thermodynamic methods, which accurately predict polymer phase behavior in fluid systems, fail to account for a critical concept associated with semicrystalline polymers; the elastic constraint. Through application of a simple thermodynamic model addressing this effect, vapor solubility in such systems can be accurately predicted using a limited number of adjustable parameters.

commercially using metallocene catalysts in a variety of different reactor technologies including gas-phase and solution reactors. [3] Metallocene catalysts are organometallic compounds with a transition metal such as Zr or Hf, surrounded by cyclic organic ligands. [3] Metallocene aluminoxane catalysts can produce PE with narrow molecular weight distribution (MWD) and comonomer composition distribution (CCD), which can result in excellent mechanical properties, but limited processability. The MWD of PE produced using metallocene catalysts can be broadened using several techniques. Using a combination of two or more metallocene catalysts in the same reactor can produce PE products with a tailored joint MW and composition distribution that leads to favorable processing and end-use properties. [3] Supporting a single type of metallocene catalyst on a solid support can also result in the formation of two or more types of active catalyst sites, which may have a beneficial influence on polymer properties. [2] Interactions with different activators or scavengers in the reactor system can also result in multi-site behavior. [4] When multiple types of active sites are present in a reactor, each site type instantaneously produces PE of a different MW and composition distribution. For example, in a two-site metallocene system, one site may tend to produce PE with higher molecular weight and higher comonomer incorporation than the other site, resulting in broad orthogonal composition distribution (BOCD). [4,5] The joint MW and composition distribution of ethylene/α-olefin copolymers produced using metallocene catalysts depends on the reactor operating conditions (e.g., temperature and the concentrations of ethylene, comonomer, and hydrogen, which is used as a chain-transfer agent to control average molecular weight). As a result, PE with tailored properties can also be produced using multiple reactors in series or a single reactor containing multiple zones with different operating conditions in each. [6] When designing new product grades and deciding how best to make them, PE producers benefit from mathematical models that predict product properties from reactor operating conditions. [6] The main difficulty associated with model development for PE processes is determining suitable values for the many kinetic A dynamic model is developed to predict detailed chain-length and comonomer incorporation behavior during gas-phase ethylene/hexene copolymerization using a supported hafnocene catalyst. The multi-site catalyst results in a copolymer with a broad orthogonal composition distribution (BOCD) where the high molecular-weight tail has high hexene incorporation. The model relies on gel permeation chromatography measurements obtained using multiple detectors (GPC-4D), so that the composition of the copolymer is determined for different chain-length fractions. Chainlength distributions are discretized into bins so that comparisons can be made between GPC-4D data and model predictions. Parameter estimation is aided by an estimabil...

This paper extends the previous article by the authors on the solubility of hydrocarbon vapors in semicrystalline polyethylenes produced in the gas phase process. That work demonstrates a computational model for solubility based on an activity coefficient modification of the Sako-Wu-Prausnitz equation of state. In that work, by fitting a key parameter, one related to the constraint of tie chains on polymer fluid behavior, to a single isopentane solubility isotherm, accurate predictions of hydrocarbon solubility in polymer granules over a range of temperature, pressure, and composition are reported. In the present work, additional experimental solubility data are reported, an error in the authors' previous article is corrected, and a useful parameterization method that improves the predictive capabilities of the model is demonstrated. By using the model to predict much of the authors' own experimental data, as well as that published by others in the field, it is demonstrated how the proposed parameterization method allows for accurate predictions using a limited amount of experimental measurements.

Front Cover: This article demonstrates a method to parameterize predictive models for hydrocarbon solubility in semicrystalline polyethylenes. The method reproduces solubility data at high loading, as well as in the dilute (Henry) regime. Such models are useful in predicting hydrocarbon absorption in polymer granules produced in the Gas Phase Polyethylene process, supporting kinetic modeling and purging processes. This is reported by Bruce J. Savatsky, Joseph A. Moebus, and Brian. R. Greenhalgh in article 1900003.

Improving properties of polymers can bring about tremendous opportunities in developing new applications. However, the commonly used trial-and-error method cannot meet the current need for new materials. We demonstrate the utility of Machine Learning (ML) algorithms in creating structure-processproperty models based on industrial data in polymer processing. In this study, ML algorithms were used to predict the optical and tensile strength of multilayer co-extrusion polyethylene films as a function of material structures and process parameters. The input features to predict the mechanical and optical properties are the composition of five-layer polyethylene film, polyethylene molecular properties like the amount of long chain branching LCB ð Þ, and the extrusion process conditions. Different data featuring steps are conducted to improve the quality of the input data: (1) feature importance scoring using an ensemble algorithm (XGBoost); (2) application of autoencoder to reduce the dimensionality;(3) replacing the categorical inputs with molecular characteristic properties. We then use this data to build an Artificial Neural Network. Finally, the prediction capability of the resulting model was investigated. This project demonstrates a successful end-to-end execution of a material data science project; from understanding material science, data engineering, algorithm development, and the model evaluation.