Modeling, optimization and simulation of an LNG distribution network to the natural gas productive chain in the Brazilian Amazon region

Abstract

This project aims to propose a network for distributing liquefied natural gas (LNG) in the State of Amazon. There are two natural gas production sites, namely Urucu and Campo de Azulão, and a pipeline that transports the gas produced at Urucu to Manaus. If one considers the adoption of natural gas as a fuel by the fleet of inland river towboats, and as the main fuel of the thermoelectric plants found in 30 isolated areas (counties that aren't connected to the country's National Interconnection System), then a distribution problem arises related to the supply of LNG in the Amazon. It must be decided where to locate the liquefaction plants and the terminals for loading the vessels; an heterogeneous fleet must be defined to distribute the LNG, and routes must be devised in such a way that the customers (counties and ports) are periodically visited. This problem is a variation of the periodic location routing problem, having as a distinct feature the routes' duration, that can last up to 10 days, in contrast to the cases found in the literature, where the duration is usually limited to a workday in the context of urban distribution. The problem will be solved with the support of the column generation method, with the purpose of generating lower bounds, and heuristically (constructive, metaheuristics or hybrid heuristics), to generate upper bounds. The project will be done in partnership with the Federal University of Amazon, whose research team will be responsible for obtaining the necessary data, and for interacting with the State of Amazon authorities, industrial players and representatives from the navigation sector. UFAM will also devise a discrete event simulation model, to evaluate heuristic solutions with uncertainties. (AU)