(45b) Maximizing Propylene Yield for Steam Cracking of C2-C6 Hydrocarbon Feeds: Experimental and Modeling Study

Steam cracking, being the principal industrial method for production of olefins, has constantly received substantial attention from the kinetic modeling community. However, recent changes in terms of feedstocks and cracker technologies demand a revisit to this otherwise well-established area of kinetic modelling. Historically, out of the steam cracking products, ethylene has been the most produced building block followed by propylene. The diversity of propylene derivatives, which includes polypropylene, epoxies, phenol formaldehyde resins, polycarbonates, acrylonitrile and propylene oxide, has resulted in a sharp rise in demand for propylene over the last decade. Contrary to this trend, the shift from heavy feed crackers to pure ethane crackers in response to the availability of inexpensive shale gas based ethane is forecasted to reduce propylene coproduction relative to ethylene. The newer pure-ethane fed crackers commissioned after 2015 also produce negligible or no propylene at all as biproduct.

To meet this increasing demand for propylene, crackers will have to either co-feed heavier fractions such as C3-C6 or rely on the upcoming on-purpose propylene (OPP) technologies. Being a stable technology, the former pathway demands minimum or zero investment compared to OPP technologies which are mostly catalytic in nature. This study focuses on developing a fundamental model for steam cracking of these hydrocarbon fractions with efforts in establishing the aromatic formation pathways as elaborate as possible. There has been several works on modeling steam cracking reactions starting from empirical models to detailed automatically generated microkinetic models and proprietary packages such as COILSIM1D and SPYRO with elaborate reaction networks. Most of these models with elaborate reaction networks for olefin and other non-aromatic species formation, lacks a fundamental nature in predicting aromatic yields. Having a comprehensive kinetic model which is true to chemistry, incorporating pathways leading to formation of major aromatic species including benzene, toluene, styrene, indene, xylenes and naphthalene apart from olefins will inherently aid in accurate understanding of the cracking mechanism and in turn is a valuable asset in targeting increased production of a particular product, i.e., propylene.

Figure 1: Comparison of experimental and simulated propylene and benzene yields for ethane and propane steam cracking experiments. (HC flowrate:130g/hr, dilution:0.4kg/kg, residence time:0.4s)

Genesys, the in-house automatic network generation tool, was used to generate the presented model and the model was validated against steam cracking experiments carried out in a benchscale steamcracker setup equipped with a refinery gas analyser (RGA) and two-dimensional gas chromatography (GCGC) for product quantification. Figure 1 compares propylene and benzene yield predictions of the Genesys model for ethane and propane steam cracking with the experimentally determined benzene yields. The mean absolute error for yield predictions for propylene was 0.1wt% and 0.4wt% for ethane and propane cracking respectively.  For benzene this was 0.1wt% and 0.2wt% respectively. Similar nominal errors in yield prediction were observed for all the aromatics species reported in this study.