A systematically generated, pressure-dependent mechanism for high-conversion ethane pyrolysis. 2. Radical disproportionations, missing reaction families, and the consequences of pressure dependence.
A previous, companion paper of this work (preceding paper in this issue) systematically developed a pressure-dependent reaction mechanism for the high-conversion pyrolysis of ethane, as studied experimentally by Glasier and Pacey. By combining conventional equilibrium, sensitivity, and reaction pathway analyses, that study identified a complex set of reaction pathways governing the formation and destruction of the important minor products acetylene, propylene, 1,3-butadiene, and benzene, which are pyrocarbon deposition precursors. To more confidently understand, and potentially manipulate, the complex chemistry governing deposition precursors, this work examines in more detail (1) the role of large sets of radical disproportionation reactions in forming the minor products, (2) the consequence of ignoring certain reactions during mechanism construction, (3) the appropriateness of the plug flow assumption used to model the reactor, and (4) the importance of reaction pressure dependence to the predictions of the minor product concentrations. We find that the predicted benzene concentration is sensitive to the presence of a specific, large collection of radical disproportionation reactions, typically neglected in most modeling efforts and of consequence only in the aggregate. Reaction families allowing Diels-Alder reaction, ene reaction, and triplet ethylene formation are safely ignored during model construction for the experimental conditions, but two specific reactions proposed in the literature, which rapidly convert fulvene to benzene, potentially explain the underprediction of benzene by the generated model. However, the rates proposed for these reactions are either unrealistically fast, or unconfirmed. We find the plug flow assumption reasonable in most respects, but its neglect of H-atom diffusion could explain the mechanism's systematic underprediction of hydrogen concentration at longer residence times. Finally, we demonstrate that accurate predictions of the minor product predictions appear to require the systematic treatment of pressure dependence employed by our mechanism generation algorithm. Changing the pressure in the experimental reactor alters the distribution of the minor products, or deposition precursors, through changes in the pressure-dependent reaction rates involved in their formation. We close with suggestions for the future development of automated mechanism generation tools, based on the current results.