The experimental investigation and data–driven modeling for thermal decomposition kinetics of Green River Shale

Abstract

Given that kerogen is a source of vast amount of hydrocarbons and organically–bound inorganic elements, it’s important to understand the thermal decomposition kinetics of kerogen. Green River Shale contains a significant amount of immature kerogen (Type I), which can be an ideal source for the sample of experimental study. In this study, Thermogravimetric Analysis and Derivative Thermogravimetry (TGA/DTG) was used to quantify the weight loss during the pyrolysis process of the Green River Shale from Utah and subsequently establish the kinetic model of thermal decomposition of kerogen based on the Friedman method followed by the data–driven modeling approach. A two–step reaction mechanism and components of production during the pyrolysis were determined by implementing Thermogravimetry Analysis–Differential Scanning Calorimetry–Gas Chromatography (TGA–DSC–GC). The chemical bonds were analyzed with Fourier–Transform Infrared spectroscopy (FTIR) before and after the pyrolysis. From the experiments, we observed the two–stage reactions in the hydrocarbon evolution window when the heating rate was lower than 30 °C/min, while only one stage was observed with the higher heating rates. C14 hydrocarbon was generated continuously during the hydrocarbon evolution, which indicated that the Green River Shale contained a plentiful amount of it. The kinetic parameters were obtained for the decomposition of organic and inorganic mixture and the organic matter (kerogen) only. The Artificial Neural Network (ANN) method was implemented to train the kinetic parameters obtained from the TGA/DTG experiment. The prediction of extrapolated cases showed a good performance when the heating rate was smaller than 5 °C/min. The generated proxy model can be coupled with various physical models to simulate the thermal decomposition of kerogen with high accuracy.