ONE-AND-TWO-LEVEL NATURAL GAS EQUILIBRIUM MODELS AND ALGORITHMS

Abstract

This dissertation consists of three parts; Part 1 provides two applied studies for the current issue of the global natural gas market, Part 2 presents the World Gas Model (WGM) 2014 version-a significant extension of WGM 2012, and Part 3 develops a novel Benders decomposition procedure with SOS1 reformulation to solve mathematical programs with equilibrium constraints (MPECs) and then is applied to several applications in natural gas and additional test problems.

Part 1 presents two applied studies related to the impacts of U.S. liquefied natural gas (LNG) exports on global gas markets as well as the influence of the Panama Canal tariff selection on global gas trade. The first study within Part 1 investigates the effect of the U.S. LNG exports on the global gas markets using the WGM 2012 (Gabriel et. al., 2012), a market equilibrium model for global LNG markets based on a mixed complementarity problem (MCP) format. The second study within Part 1 focuses on the influence of the Panama Canal tariffs on global trade using WGM 2012 as well. After a planned expansion, the Panama Canal waterway will accommodate more than eighty percent of LNG tankers, providing significant potential time and cost savings for LNG buyers and producers. The aim of the second applied study is to address how the Panama Canal tariffs affect global gas trades

In Part 2, a significant extension of the World Gas Model 2012 is developed. This new version called WGM 2014, distinguishes itself from the previous version in the sense of more detail for LNG markets including more market participants e.g., liquefiers, regasifiers, LNG shipping operators, and a canal operator as new players with separate optimization problems and market-clearing conditions. Moreover, the LNG shipping costs and congestion tariffs for canal transit fees are endogenously determined inside the model as opposed to being exogenously determined before. Also, WGM 2014 has flexible LNG routes. In particular, there are three route options for each LNG shipping operator: 1. Sending LNG via the Panama Canal, 2. the Suez Canal, or using a regular route without a canal. Moreover, WGM 2014 takes into account the limitations of maritime transportation by limiting the size of the LNG tankers that can pass through the Panama and Suez canals which itself is a major improvement for natural gas policy study.

In part 3, the method we develop uses an SOS1 approach based on (Siddiqui and Gabriel, 2012) to replace complementarity in the lower-level problem's optimality conditions. Then, Benders algorithm decomposes the MPECs into a master and a subproblem and solves the overall problem iteratively. This methodology is applied to small, illustrative examples and a large-scale MPEC version of the World Gas Model where the Panama Canal operator is a Stackelberg leader with a reduced version of the rest of the global gas markets considered as followers.