CHEMICAL ENGINEERING TRANSACTIONS VOL. 29, 2012 A publication of The Italian Association of Chemical Engineering Online at: www.aidic.it/cet Guest Editors: Petar Sabev Varbanov, Hon Loong Lam, Jiří Jaromír Klemeš Copyright © 2012, AIDIC Servizi S.r.l., ISBN 978-88-95608-20-4; ISSN 1974-9791 DOI: 10.3303/CET1229065 Please cite this article as: D‘Isanto M., Manenti F., Grottoli M. G., Altavilla M. and Di Marco R., (2012), Online superstructure optimization for energy saving of an industrial gas distribution system, Chemical Engineering Transactions, 29, 385-390 385 Online Superstructure Optimization for Energy Saving of an Industrial Gas Distribution System Matteo D’Isanto a , Flavio Manenti a, *, Maria G. Grottoli a , Marcello Altavilla b , Roberto Di Marco c a Politecnico di Milano, Dipartimento di Chimica, Materiali e Ingegneria Chimica “Giulio Natta”, Leonardo Chemical Engineering LAB, Piazza Leonardo da Vinci 32, 20133 Milano, Italy b Linde Gas Italia, Terni Plant, Viale Benedetto Brin 214, 05100, Terni, Italy c Linde Gas Italia, Headquarter, Via Guido Rossa 3, 20010 Arluno (MI), Italy flavio.manenti@polimi.it Mixed-integer optimization is a common approach to handle decision-making problems. Nevertheless, such an approach still presents certain operational limitations, especially for online superstructures. One of these limitations is given by Jacobian singularity, which arises for certain combinations of the set of Boolean variables and which easily leads to possible infeasible numerical solutions. This paper proposes a novel approach to overcome this problem, ensuring the use of mixed-integer optimization also to solve online issues. The case of nitrogen supply for the Thyssen-Krupp steel mill placed in Terni (Italy) is considered as validation case. 1. Introduction The optimization of gas transport network is a very important industrial problem (Martin et al., 2006; Ehrhardt and Steinbach, 2005; Domschke et al., 2011). Many industrial gases production plants are directly connected to tonnage users such as steel mill plants and refinery plants requesting thousands of Nm 3 per hour of gases at different pressures and without a previously accorded planning. Therefore a strategy to react rapidly and flexibly to the requests will be of significant importance for the gas company to reduce energy consumption and to better exploit plant performances. This leads to a very complex problem to solve because of the Boolean nature of the variables involved, which is challenging either for a mathematical point of view or to better tackle with the business challenge (Ierapetritou et al., 2002; Floudas et al., 2005; Glankwamdee et al., 2008; D'Isanto et al., 2011). Here it is presented a new simulation and optimization tool for this problem. The situation is basically the minimization of distribution costs (for nitrogen only) from the gas production site (Linde Gas Italia, Terni) to a high capacity steelworks plant (Thyssen-Krupp); to supply Thyssen-Krupp, an existing network is used. It is rather complicated since it has two different sources, consisting of two separate air separation units, with one of them that produces gas products only and the other one that produces both liquid and gas products. Moreover, the distribution line of a disposed air separation unit is still operating and it can be used whenever the energy cost makes it appealing. The gaseous nitrogen distribution network here considered includes several compressors dedicated to the supply of a high- capacity steelworks plant that requests nitrogen at three different pressures with a demand that significantly changes many times within the day due to different steel treatments and castings. The energy consumption for the entire distribution network is high due to the large amount of power for the compression work, but also to the fast-response required to adapt the distribution network to the