No abstract

No abstract

No abstract

Supporting information Heterocatalytic oxidation of HCl into Cl 2 (HCl conversion process) over a CeO 2-CuO catalyst in a fixedbed reactor has been optimized using four different objective functions and three different methods. For a given residence time, the HCl conversion could be maximized in various reactor configurations, namely with the application of a graded catalyst bed, with partitioning the reactor shell and using multistage temperature control, or with the combination of both methods. However, as the obtained temperature profiles differed considerably from each other, we considered three objective functions in order to smooth the reactor temperature profile. These implement a nonconventional approach since the proposed objective functions aim to deal directly with the temperature changes and HCl conversion is only taken into consideration as a nonlinear constraint. The different results from each method and objective function were compared using the apparent temperature gradients along the length of the reactor. The results show that reasonably low temperature gradients and a relatively smooth temperature profile can be achieved that also indirectly contributes to the stability of both the reactor and the catalyst.

The so-called "lumping approach" is widely used to study complex processes such as hydrocracking of vacuum residue. In order to describe the composition changes in such systems, not only must the kinetic parameters be determined, but the lumped reactions that occur should also be identified. In this study, the modeling of catalytic hydrocracking of vacuum gas oil has been carried out using six component lumps. Three different identification strategies have been developed to determine the reaction subnetwork containing a given number of reaction pathways that provides the data fit. The strategies were compared according to their tendency to provide increasingly better results, as a function of the number of reactions present. Although, in this way, 40% of the original reaction superstructure was eliminated from the system, the kinetic parameters of the remaining reactions still could not be identified with complete certainty. Hence, the linearized state-space model representations of the reaction networks have been further analyzed with the objective of identifying observable subsystems. The results show that there exists a five-reaction network that is observable and can be determined using the finally proposed identification strategy, while its curve fitting is also satisfactory. It has also been shown that it gives the best results from all possible fivereaction subsystems.