Modeling the Nonisothermal Devolatilization Kinetics of Typical South African Coals Burgert B. Hattingh,* Raymond C. Everson, Hein W. J. P. Neomagus, John R. Bunt, Daniel van Niekerk, and Ben P. Ashton Research Focus Area for Chemical Resource Beneficiation, North-West University, Potchefstroom Campus, Private Bag X6001, Potchefstroom, 2520 North West, South Africa Energy Systems, School of Chemical and Minerals Engineering, North-West University, Potchefstroom Campus, Private Bag X6001, Potchefstroom, 2520 North West, South Africa Sasol Technology (Pty) Ltd., Research & Development, Coal & Gas Processing Technology, Box 1, Sasolburg, 1947 Free State, South Africa * S Supporting Information ABSTRACT: Multicomponent model fitting was conducted in order to evaluate the devolatilization rate behavior of four typical South African coals, with the aid of nonisothermal thermogravimetry. Rate evaluation was conducted at four different heating rates (5, 10, 25, and 40 K/min) by heating the samples under an inert N 2 atmosphere to 950 °C. Evaluation of the kinetic parameters of each coal involved the numerical regression of nonisothermal rate data in MATLAB 7.1.1 according to a pseudocomponent modeling philosophy. The number of pseudocomponents used ranged between three and eight, as larger values induced the risk of over fitting. Quality of fit(QOF) was found to decrease with decreasing heating rate as a result of improved separation of the individual component reactions at the lower heating rates. All four coals showed the occurrence of similar pseudocomponent reactions, although significant differences were observed in the fractional contributions of the different pseudocomponents to the overall reaction rates. Modeling results indicated that the assumption of eight pseudocomponents produced the lowest QOF values and subsequently the best fit to the devolatilization profiles of each coal. For the vitrnite-rich coals (G#5 and TSH), no remarkable decrease in QOF could be observed after 6 pseudocomponent reactions, suggesting that even 6 or 7 pseudocomponent reactions would have provided accurate experimental predictions. Activation energies determined from the selected number of pseudocomponents (between 3 and 8) were found to range between 20 and 250 kJ/mol. 1. INTRODUCTION Coal devolatilization plays an important role in not only the metallurgical industry for producing coke but also during coal gasification where it normally constitutes the initial step of the process. 1,2 It is therefore important to evaluate the devolatiliza- tion behavior of a coal feedstock in order to assess and optimize the production of valuable products such as char/coke, tar, and gas. Extensive research during the past few decades has advanced our knowledge of the kinetics and mechanisms of the devolatilization process. Furthermore, it has also provided us with valuable techniques for predicting, to a reasonable extent, the behavior of coals. 3-5 Devolatilization modeling is quite straightforward if the chemical reaction step is rate controlling and the fuels are of a simple characteristic nature. Kinetic models for describing thermal decomposition range in different levels of complexity from free radical mechanistic models for simple hydrocarbon species such as propane 6 to more complex reaction schemes. The latter involves a number of individual reactions, incorporating extra transport steps such as in the case of naphtha devolatilization. 7 The kinetic description of more complex poly- aromatic substances such as coal presents a challenging task, due to a vast amount of reactions involved. Kinetic evaluation of these substances is therefore normally conducted using pseudome- chanistic models, which attribute the overall measured reaction rate to the cumulative effect of a number of separate reactions. 8,9 A large number of possible modeling strategies are currently available, of which the simplest are empirical in nature and employ global kinetics. 8-11 Available models can be divided into either general weight-loss models or structural models. Typical weight-loss models include models employing a (1) single rate, (2) two rates, (3) multiple rates, and (4) distributed rates. 12-21 Although simplistic in nature, the validity of a single reaction rate mechanism is quite limited. Kinetic parameters derived at a single heating rate has been generally shown not to be appropriate to other heating rates. 22 Currently, the Distributed Activation Energy Model (DAEM) (Anthony-Howard model) has been shown to be the most power- ful model for predicting devolatilization behavior. 10,18,19,22-25 This complicated model was first proposed by Pitt 26 and assumes the devolatilization process to consist of an infinite series of independent parallel first-order reactions. Accordingly, coal devolatilization can be explained by a distribution of activation energies about some mean activation energy value (E a,0 ). The function within the model f(E a ) accounts for a distribution of activation energies and is assumed to be of Gaussian form. A simplistic approach for solving the DAEM considers a common, Received: October 25, 2013 Revised: December 25, 2013 Published: December 26, 2013 Article pubs.acs.org/EF © 2013 American Chemical Society 920 dx.doi.org/10.1021/ef402124f | Energy Fuels 2014, 28, 920-933