Development of mathematical model to optimize utility network within an industrial complex

Abstract

The economical and environmental performance of utility system is an important issue because of its central role in industrial activities. Recently, many companies have been trying to reduce energy consumption due to the high oil price. Moreover, several environmental regulations such as Kyoto Protocol impose significant restriction on companies’ productivity. Especially in chemical or petrochemical industries which consume huge amount of energy, there have been many attempts to optimize utility network. However, optimization of an individual utility system seems to have very few chances to be improved. Although there are a few utility networks between subsidiary companies, there still remain significant energy saving opportunities in the whole industrial complex point of view. In chapter 2, we suggested a useful mathematical tool, Steam Networking Matrices (SNMs) which mean source and sink companies’ matrices including steam connection information. Overall steam network can be easily represented by using the proposed tool. Illustrative example of steam network optimization shows that both total energy cost and $CO_2$ emissions of an industrial complex can be reduced by exchanging steam between source and sink companies. Although SNMs are intuitional tools which are able to express steam network information, it is difficult to model detail units or various utilities such as fuel or electricity. Therefore, we developed a novel mathematical model for optimization of utility network within an industrial complex. Chapter 3 consists of three sections. The first section is a unit modeling using thermodynamic principles, mass and energy balances and process data. The second section is a development of multi-period MILP model for the integration of different process systems. The last one is a comparative analysis of the optimization results. The approach was applied to the Yeosu industrial complex considering multi-period utility demands. The results show that not ...