Modeling and optimization for trans-stilbene expoxidation process.

• Main Author: Lin, Xin Ni.
• Other Authors: Chen Bin
• Format: Final Year Project
• Language: English
• Published: 2010
• Subjects: DRNTU::Engineering::Chemical engineering::Chemical processes
• Online Access: http://hdl.handle.net/10356/39556
• Institution: Nanyang Technological University

Data-base model can search for the optimal operating condition for a specific process without prior knowledge about the mechanism of reactions. However, a new model is required for a new process even if there exist a model of similar process, since data-based model is developed on data of corresponding process. On the other hand, if effective extraction of the process similarities is performed, it can reduce the number of experiments required to develop a new model, hence saving time and cost. Model migration is the technique, used in this study, which considers the hidden relationship between two similar processes. In this project, the author‟s effort lies mainly on the lab-scale experimentation that facilitates the migration of an existing Gaussian Process Regression (GPR) model to a similar process. The author conducted trans-Stilbene epoxidation using Co2+-NaX zeolite catalyst and tert-butyl hydroperoxide (TBHP) oxidant, considering four process factors: reaction temperature, initial trans-Stilbene concentration, stirring rate and reaction duration. Accurate process optimization was achieved by conducting experiments at the design point where predictive response is high and/or predictive variance is large. Results show that this model is accurate enough in high response region to give a convincing prediction of optimal point, with only 14 experimental points needed.