Robert H. Harding

Catalytic cracking of alkylbenzenes: Modeling the reaction pathways and mechanisms

Abstract The catalytic cracking reaction pathways and mechanisms of 1-phenylhexane, 1-phenyloctane and 1-phenyldecane were studied. Experiments at 500°C with a rare earth Y (REY) catalyst indicated that the dominant reactions included dealkylation and cracking in the alkyl side chain. To describe these results, a mechanistic model of the catalytic cracking of 1-phenyloctane was developed using a novel mechanism-based lumping scheme that exploits the chemical similarities within reaction families. The formal application of 17 reaction family matrices, which correspond to 15 reaction family classes, to the matrix representations of the reactants and derived products generated the model. The reaction family concept was further exploited to constrain the kinetics within each reaction family to follow a quantitative structure/reactivity Polanyi relationship. Ultimately, eight Polanyi relationship parameters, one catalyst specific parameter and three coking/deactivation parameters were optimized using experimental data. The resulting model correlations were excellent, as the overall parity between experimental and model values was y Model =0.00131+0.994y Exp with a correlation coefficient of 0.998.

Rational assessment of FCC catalyst performance by utilization of micro-activity testing

Abstract The increasing demand for a quick and simple screening tool of FCC catalysts and feedstocks has propelled the well known micro-activity test (MAT) ASTM-D 3907-80 far beyond its original purpose. Where once this test was a tool for assessing the catalyst activity, it is now routinely used to predict the yields in commercial FCC units. However, the necessity of a thorough revision of this test is generally agreed upon to come up to its new requirements. To obtain data reflecting those observed in circulating riser pilot units, which provide the best smaller-scale simulation of commerical FCCU operation, a new MAT method was developed using an operating scenario quite different from those of the traditional ASTM-MAT test. Cracking of vacuum gas oil and heavy resid feeds over catalysts with quite different properties was performed with the ASTM-MAT technique, the modified MAT approach and in a riser pilot unit to evaluate the modified procedure. This paper shows that with the choice of appropriate experimental conditions for MAT, a good simulation of the riser pilot units can be obtained. It is demonstrated that the modified test eliminates the ranking reversals identified for the comparison of catalyst selectivities obtained by the ASTM-MAT and riser pilot units. Moreover the selectivity levels obtained by the modified MAT configuration are very close to riser data indicating similar reaction pathways in these very different reactor types.

Latest developments in microactivity testing: influence of operational parameters on the performance of FCC catalysts

An accurate assessment of catalyst performance is the most important goal in the testing of fluid catalytic cracking (FCC) catalysts since the catalyst is a key contributor to the overall unit performance. For this purpose small scale tests in the laboratory have been the workhorse in the industry because they are less expensive and time-consuming to operate than circulating riser pilot units. Two reactor types are common in microactivity testing, simple fixed bed (plug-flow) reactors and fixed fluidised bed reactors. Pitfalls have been identified for both experimental modes and this paper discusses the strengths and weaknesses of these two techniques. This work demonstrates that catalyst testing in small-scale fixed fluidised bed reactors can result in erroneous catalyst ranking, while the use of improved fixed bed reactors has a better agreement with riser pilot units. Furthermore, reactors of either type that employ in-situ regeneration result in unrealistically high coke and hydrogen yields due to the oxidation of contaminant metals, and therefore advanced deactivation procedures for metallated FCC catalysts cannot be utilised in small-scale tests with this technique.

Evaluation of sparse data sets obtained from microactivity testing of FCC catalysts

Abstract Although many mechanistic studies of gas–oil cracking on fluid catalytic cracking catalysts have been published, only a small amount of literature has described their application for the modelling of sparse data sets amenable to typical evaluations in catalyst screening studies. The well-known microactivity test is the most common approach to determine the activity and selectivities of fluid catalytic cracking catalysts and to predict their performance in commercial FCC units. For this purpose the yields obtained for different catalysts or different feedstocks have to be compared at constant values such as conversion, coke or gasoline. Generally, the values of interest cannot be obtained experimentally and therefore interpolation techniques have to be applied to gain knowledge on the individual yields. However, the small number of experiments usually performed for such studies and the large number of parameters which have to be estimated, if equations derived from reaction kinetics are used, render the utilisation of appropriate models for yield predictions difficult. In this paper a method is presented which enables the modelling of sparse data sets in kinetic terms. FCC catalysts have been characterised with this new approach and its effectiveness was compared with conventional methods. It is demonstrated that the utilisation of informative graphical representations and the application of reduced kinetic expressions lead to an efficient data analysis.

Catalytic cracking of alkylcyclohexanes: Modeling the reaction pathways and mechanisms

The cracking reaction pathways and mechanisms of ethylcyclohexane and 1-cyclohexyloctane with a rare earth Y (REY) catalyst were studied. Experiments at 500°C indicated that the dominant reactions were ring opening with subsequent secondary cracking, cracking in the alkyl side chain, isomerization, and hydrogen transfer. A kinetic model of the catalytic cracking of 1-cyclohexyloctane was developed using a novel mechanism-based lumping scheme that exploits the chemical similarities within reaction families. The formal application of 17 reaction family matrices, which correspond to 13 reaction family classes, to the matrix representations of the reactants and derived products generated the model. The reaction family concept was further exploited to constrain the kinetics within each reaction family to follow a quantitative structure/reactivity Polanyi relationship. Ultimately, nine Polanyi relationship parameters and three coking/deactivation parameters were determined by optimizing the model fit to the experimental data. The resulting model correlations were excellent, as the overall parity between experimental and model values was yModel=−0.000470+0.986yExp with a correlation coefficient of 0.971.

Slowing down near the critical point in optically bistable ZnSe

We measure relaxation rates near the critical point and the left hysteresis limit of an optically bistable system, a ZnSe interference filter. Our ZnSe system has an inhomogeneous geometry where a long, narrow illuminated region has boundaries at a temperature near and below that of the lower state. We determine the critical angle, the angle of incidence of light at which the hysteresis limits coalesce to form a critical point, and perturb the system by changing the input power beyond the critical point. For incidence angles equal to or slightly greater than the critical angle, we find that relaxation rates increase exponentially as the critical point is approached. The critical exponents for perturbations which increase the input power beyond the critical point are greater than those for perturbations which decrease the input power. In either case, the critical exponents increase as the angle of incidence approaches the critical angle. When the hysteresis region is large, we find slowing down near the left hysteresis limit in accordance with our calculations based on a one‐dimensional inhomogeneous model.