Liquefied Natural Gas (LNG) is natural gas that has been cooled to a liquid state for shipping and storage. Liquefaction makes it possible to transport natural gas from producing regions to markets where pipelines are not economically feasible. The LNG supply chain broadly consists of five entities corresponding to extraction, liquefaction, transportation, regasification and distribution. The supply chain originates with natural gas production and then it is transported via pipeline to a liquefaction plant. Subsequently, LNG is shipped to regasification terminals around the world, where it is regasified and distributed as natural gas to customers. This paper focusses on the procurement of LNG by the regasification terminal.
In literature, LNG procurement has been studied as a tactical decision primarily with a focus on selecting an optimal set of contracts through the perspective of ensuring energy security while minimizing the procurement cost. Biresselioglu et al. (2012) developed a MILP model which determines the optimal set of sellers for Turkey’s LNG demand while minimizing total import and inventory holding cost as well as levels of risks1. Similarly, Geng et al. (2017) developed a multi-objective optimization problem to generate optimal import portfolio by considering multiple risk factors such as the economic risk of exporting countries, maritime risk etc2. Shaikh et al. (2017) modelled a natural gas import scheme for China3. They developed a mathematical model which minimizes the import cost, transport distance, domestic and political instability associated with each supplier. Later, with a surge of spot market share in LNG trade, it was studied as the portfolio optimization problem that is determining the ratio of LTC and spot market in total imports. Kim and Kim (2018) developed an optimal portfolio for Korea4. In this study, they developed a two-step portfolio model mean-variance optimization model which determines the ratio of LTC and spot in imports followed by linear programming model which chooses the optimal set of contract for LTC procurement. Once these tactical decisions are made, the importer still has to make procurement decisions considering the factors such as price difference between spot market and LTCs, demand at the regasification terminals, and obligations under the LTCs. No model in literature has been proposed considering these factors. We seek to adress this gap in this work.
In this work, we develop an import strategy for an LNG importer under the terms of an ongoing long-term contract and a spot market available for procurement both LTC and spot market is considered as LNG cargoes available for procurement in the market. The goal of the importer is to choose an optimal set of the cargos from cargoes available for procurement such that the total cost of procurement is minimized. Toward this, a mathematical formulation for LNG procurement (LNGPM) is proposed. The objective of LNGPM is to minimize the procurement cost by utilizing the available spot cargoes while abiding by the contractual terms of LTCs. In LNGPM, we model all the LNG available in the form of cargoes. Each cargo belongs either to LTC or spot market, having its volume, price, the window of availability and transportation cost. Apart from the cargoes, we consider resources such as jetties and storage tanks at importing terminals. Importers are limited by finite resources such as the number of jetties and storage capacities.
For comparison of LNGPM, we also propose a heuristic-based methodology called LNG procurement heuristics (LNGPH). LNGPH procures cargos based on inventory levels at the terminals. First, it fulfils all the contractual obligations of all existing LTCs. Once the contractual obligations are met, and there is a need for additional procurement, it procures a cargo from the set of long and spot cargos available at that juncture that has a minimum unit cost.
Next, we present both the methodologies using an Illustrative example. This example involves an LNG importer having three regionally distributed terminals, having two jetties at each site. The planning horizon is 30 days. At each terminal, the minimum capacity that has to be maintained over the horizon is 10 % of maximum storage. The importer has a long-term contract to which all the three terminals have been nominated along with four LNG carriers. The contract terms dictate importer that it shall buy at least 40 % of demand from LTC in a month. The maximum volume an exporter can supply during the planning horizon is 65 % of the demand. Sixty spot cargoes are also available for procurement over this period, with each cargo being available in a different time window. Each spot cargo also has a different volume and different prices per unit. The illustrative example is solved using both the methodologies and results are compared. LNGPM was able to perform better than LNGPH and was able to save $ 67 million. It is found that both the models procure the same amount of LNG from LTC. However, for the remaining demand, they select a different set of cargoes which leads LNGPH to procure more volume as compared to LNGPM. To reduce the effect of additional inventory at the end of the planning horizon, we solve the illustrative example on a hopping horizon basis for a twelve-month period; this gives us a better comparison of the performance of models. Even on this twelve-month basis, the LNGPM was better than LNGPH by approximately 1%, in absolute terms; saving $ 114 million annually.
In the real world, uncertainty is prevalent in the LNG business, which can have a significant effect on the relative performance of the two solution methodologies, as well as their computational costs. To characterize the relative performance of the two solution methodologies under different conditions, we conduct a scenario-based study. We focus on three distinct uncertainties in this study, demand at the import terminals, the quantity of LNG procured using long-term contracts, and the difference between long-term and spot market price. Three values were considered for each uncertainty, which leads to twenty-seven different scenarios. All twenty-seven different scenarios were solved using both LNGPM and LNGPH, and results were compared. The LNGPM was able to perform better than LNGPH consistently and was able to save on an average of $ 84.25 million annually over the 26 scenarios. However, the LNGPM solver was unable to solve one scenario. In this paper, we will present the detailed mathematical formulation and the heuristic methodology for procuring LNG. We will also report our observations from the solutions to the twenty-seven scenarios modeling the different uncertainties.
|Journal||2020 Virtual AIChE Annual Meeting|