517 CHEMICAL ENGINEERING TRANSACTIONS Volume 21, 2010 Editor J. J. Klemeš, H. L. Lam, P. S. Varbanov Copyright © 2010, AIDIC Servizi S.r.l., ISBN 978-88-95608-05-1 ISSN 1974-9791 DOI: 10.3303/CET1021087 Please cite this article as: Manenti F., Lima N.M.N., Zuñiga Liñan L. and Colombo S., (2010), Exploiting C++ polymorphism for operational optimization of chemical processes, Chemical Engineering Transactions, 21, 517-522 DOI: 10.3303/CET1021087 Exploiting C++ polymorphism for operational optimization of chemical processes Flavio Manenti 1* , Nadson M. N. Lima 2 , Lamia Zuñiga Liñan 2 , Simone Colombo 1 1 CMIC Dept. “Giulio Natta”, Politecnico di Milano, Piazza Leonardo da Vinci 32 20133 Milano, Italy 2 State University of Campinas (UNICAMP), Dept. of Chemical Processes flavio.manenti@polimi.it Object-oriented programming is more and more spreading in engineering and scientific areas for some relevant benefits making it particularly appealing to write complex codes. Nevertheless, the use of such a programming philosophy still encounters large inertia in those areas characterized by a long programming experience and one of them is engineering. This is mainly due to a set of existing models and subroutines wrote in procedural (usually Fortran) language. The present paper is aimed at showing some benefits coming from object-oriented programming applied to the field of process optimization and specifically to the operational levels of supply chain management paradigm such as nonlinear model predictive control (NMPC). Polymorphism is exploited to provide a single solution for different NMPC techniques such as input blocking, offset blocking, and  -blocking. 1. Introduction The optimization of unit operations and chemical processes has recently seen the significant spreading of Nonlinear Model Predictive Control (NMPC) methodology against other control alternatives (Qin and Badgwell, 2003; Bauer and Craig, 2008). Success of NMPC is mainly due to the following reasons: it is intrinsically able to manage nonlinearities in process dynamics and in profits (Manenti and Rovaglio, 2008; Dones et al., 2010); it can be based on first-principles mathematical models and on nonlinear semi-empirical models as well (Lima et al., 2009a; Lima et al., 2009b); it allows solving simultaneously the predictive control and the dynamic optimization (Zavala et al., 2005; Manenti et al., 2010). The basic architecture of NMPC has been wide described in the literature (Morari and Lee, 1999; Rawlings, 2000; Findeisen and Allgower, 2003). The plant provides raw data, which is reconciled (Bagajewicz, 2003; Signor et al., 2010) and sent to NMPC at each sampling time. Specifically, the reconciled data are sent to an optimization routine, which includes an objective function, a dynamic model and, usually, according to the mathematical model type, a numerical integrator to solve specific differential or differential-algebraic systems