Please use this identifier to cite or link to this item:
|Type:||Artigo de evento|
|Title:||Dynamic Parameter Estimation In Polymerization Process|
|Abstract:||Mathematic modeling of chemical process is a very useful toll since it allows process knowledge without major investments, as is the case of pilot plants, for example. However, the model must be trustful, allowing predictions to be made in accordance with the process reality. Depending on the process considered, some of the model parameters are unknown, such as kinetic constants, heat capacity and so on. Thus, they must be estimated in order to allow the model to represent the main phenomena taking place in the system. The parameter estimation can be accomplished with experimental studies or using mathematic models, as in the case of this contribution. As a matter of fact, a parameter estimation procedure is an optimization problem: the process model has the same input variables as the real process and thus both output variables should be compared to give an error function which might be minimized. In the optimization problem the decision variables are the parameters to be estimated. The present study take into account an ethylene polymerization process with ZieglerNatta catalyst as a case study, for which real dynamic process data were used. In this process, many kinetic constants and physic parameters, around 30, were unknown and so must be estimated. As it is a dynamic optimization of a complex system, it was necessary to develop a robust optimization procedure based on the use of the Successive Quadratic Programming and the NAG (Numerical Algorithms Group) subroutines were used. The results show that the parameters were successfully estimated, with good agreement between process and model outputs.|
|Appears in Collections:||Unicamp - Artigos e Outros Documentos|
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.