Scheduling arrivals in an LNG Receiving Terminal

Abstract

The demand for Liquefied Natural Gas (LNG) has increased rapidly over the last 30 years. LNG is Natural Gas (NG) that is cooled down until it becomes liquid. LNG is easier to transport over long than NG distances and also in larger quantities. Before the LNG can be used it is made gaseous again, for example at an LNG receiving terminal. In this thesis we create multiple Mixed Integer Linear Programs (MILPs) to solve the problem of creating an Annual Delivery Plan (ADP) for an LNG receiving terminal. An ADP describes the times vessels should arrive at the terminal and when this LNG will be gasified and send-out to the network. We proof this problem is NP-Complete. First we create three MILPs that can solve for terminals with a single client. We have a discrete time formulation, a formulation that makes use of piecewise linear functions and an event based approach. Experimentation learns us the discrete time and event based formulations are solved the fastest. We then expand the discrete client formulation to work for multiple clients and also adjust this for- mulation to work with a rolling horizon approach, which gives up some optimality in return for solving speed. From experimentation and evaluation using a simulation model we learn that the rolling horizon approach is comparable with an heuristic algorithm currently used in practice. While the heuristic is a little faster our solution shows potential to improve the found solutions.