(200a) Implementation and Validation of a Model-Centric Support System within a Pilot Plant Scale Packed Distillation Column

Separation of mixtures is an important stage in the production process of a considerable number of products. In most cases, the quality and purity of the final products depend on the degree of separation achieved. Therefore, special attention has been drawn to distillation of vapor/liquid mixtures as it is a typical operation in petrochemical and chemical industries. During the normal operation of a distillation unit at plant scale, temperature and flow measurements are collected through the distributed control system (DCS). This information may serve for many different purposes such as process control, parameter estimation, data reconciliation and optimization, among others. However, the majority of studies in the area show the use of such information for control and optimization purposes only.

The conventional equilibrium stage model has been commonly utilized. It assumes that the phases leaving each stage are at complete equilibrium, simplifying the model of tray distillation in comparison to the description of the process in terms of continuous mass and energy balances (Perry, R.H.). This approach has been also applied to packed distillation columns assuming they are composed by several sections, each of them being treated similarly to a stage in a tray column (Steiner et al., 1978; Krishnamurthy, R. and Taylor, R., 1985; Yang, L. and Chuang, K.T., 2000). Fortunately, new methodologies for the solution differential and algebraic equations (DAEs) and the increasing computational capability of computers have originated the development of models based on first principles for packed distillation units assuming that the mass transfer in continuous across the column are being developed (Karacan, S. et al., 1998; Fernández-Seara et al., 2002; Alpbaz et al., 2003). It is possible now to think in the application of other advanced model-based methodologies such as dynamic data reconciliation and parameter estimation to packed distillation processes using more fundamental modeling.

Concurrently, the consolidation of the CAPE (Computer Aided Process Engineering) community in the 1990's, the subsequent development of the CAPE-OPEN (CO) project and the advances in state-of-the-art modeling environments paved the way for a paradigm shift towards technologies based on models. The conceptual definition of a single and consistent model-centric framework for integrated decision support of process systems (IDSoPS) was proposed in previous work (Rolandi and Romagnoli, 2005; Aragon et al. 2006). It provides an innovative approach for simplifying the problem formulation of model-based activities (i.e. simulation, parameter estimation, data reconciliation and optimization) by incorporating the Problem Definition Environment (PDE), which takes advantage of the capabilities of state of the art modeling and solution environments (MSEs).

In this work, the mechanisms and methods employed in the IDSoPS and its corresponding PDE are presented and discussed through a case study of a pilot plant packed distillation unit. The backmixing model is assumed to develop a rigorous dynamic model based on first principles for the continuous operation of the packed distillation column with a ternary feed composed. In addition to mass and energy balances, mass and heat transfer along the tower are included in the model. A state-of-the-art modeling environment, capable of solving distributed and highly nonlinear equations on-line in non-stationary state will be used to implement and test a number of advanced operational activities such as: joint data reconciliation (dynamic) /parameter estimation, troubleshooting of past operating conditions and dynamic optimization studies (transition planning).