Kinetic mechanism for the cracking of Colombian heavy crude oil in refinery equipments

We present a methodology for developing lump-mechanisms for the main three refinery processes with heavy crude oil fractions. Hydrocracking, Fluid Catalytic Cracking and Delayed Coking. Given the available resources, Colombia has a strong trend to process heavy crude oil fractions instead of the light ones; thus, through simulation methods, this thesis focus on solving one of problems of heavy crude oil fractions associated with its simulation. It is well known that the heavy crude oil fractions are composed by a large number of species, and therefore, a detailed kinetic mechanism implies the modeling of a large number of reactions. Thus, lump-mechanisms were used to give an approximate representation of the cracking processes in each equipment. The lump-mechanisms were obtained through a regression of experimental data reported for each process. The experimental data collected in literatura were standardized in a eXtensible Mark-Up language database to facilitate the reading by the simulation softwares. Initially, we fitted the data to a mechanism reported in literatura through an optimization method. We calculated the best set of kinetic parameters using fmincon routine available in Matlab 2014. Using a wide range of process conditions from many different experiments instead of one or two experiments provides a global vision about each process. This thesis considered a wide range of operational conditions for each process. For hydrocracking such as: temperature, varying from 360_C to 450_C; Liquid Hourly Space Velocity (LHSV) varying from 0.9 h−1 to 1.36 h−1. This thesis also evaluated a wide range of the other processes, for Fluid Catalytic Cracking included experimental data obtained from experiments carried out in batch reactors and Plug-Flow Reactors (PFR). For delayed coking the operational conditions such as: temperature varying from 410_C to 475_C and residence times varying from 0.5 h−1 to 3 h−1. A regression analysis shows good accuracy between the data predicted by the model and experimental data. The regression coefficients were 0.96, 0.68 and 0.84 for the three processes respectively. A bootstrap methodology was used to calculate the confidence interval and standard uncertainty for the parameters associated to each process. Bootstrap consisted in a re-sampling with replacement of experimental data vector for each process. Once the re-sampled vector is generated, the optimization was carried out to get the set of kinetic parameters. Therefore, independent samples were obtained. The process is repeated 2000 times to create a big sample of the kinetic parameters and therefore calculate the mean value and the standard uncertainty. The results showed values of standard uncertainties around two magnitude or ders less than the mean value of the distribution. Kinetic results showed the compensation effect for those processes which were modeled using first-order and irreversible reactions such as hydrocracking and delayed coking. For each kinetic parameter, a sensitivity analysis was carried out to determine the success of the optimization routine and to determine the number of bootstrap re-samplings. Results showed that the calculated parameters corresponded to a global minimum in the optimization.

• Autores: López Pérez, Luis Carlos
• Fecha de publicación: 2016
• Institución: Universidad Nacional de Colombia
• Repositorio: Universidad Nacional de Colombia
• Idioma: spa
• OAI Identifier: oai:repositorio.unal.edu.co:unal/56569
• Acceso en línea: https://repositorio.unal.edu.co/handle/unal/56569; http://bdigital.unal.edu.co/52389/
• Palabra clave: 66 Ingeniería química y Tecnologías relacionadas/ Chemical engineering; Bootstrap; Kinetic parameter estimation; Refinery process; Sensitivity analysis; Uncertainty calculation; Cálculo de incertidumbres; Procesos de refinería; Bootstrap; Estimación de parámetros cinéticos; Análisis de sensibilidad
• Rights: openAccess
• License: Atribución-NoComercial 4.0 Internacional