World Gas Model Franziska Holz Joint Work with

World Gas Model Franziska Holz Joint Work with Ruud Egging, Steve Gabriel, and Jifang Zhuang DIW Berlin University of Maryland Work in Progress Presented at the 4 th Ph. D Seminar of Natural Gas Oxford, November 10, 2006 -1 -

Outline Model Overview Optimization Problems of Selected Players - Producers - Transmitters - Pipeline operators - Storage operators - Marketers State of the Work and Future Plans -2 -

Model Overview - Static complementarity model of the world natural gas market - Daily flows seasonal patterns over one year - Detailed representation of the players in the natural gas business: - Producers - Transmitters - Liquefiers - Regasifiers - Storage operators - Marketers - Mixed complementarity approach instead of MPEC which would we hard to solve One stage with market power (one player) - Other players assumed to behave competitively, and to be linked via market-clearing conditions -3 -

Overview of the Natural Gas Industry LNG DISTRIBUTION SYSTEM CITY GATE STATION Marketer/Shippers INDUSTRIAL COMMERCIAL TRANSMISSION SYSTEM RESIDENTIAL National network: local transportation ELECTRIC POWER GAS PROCESSING PLANT UNDERGROUND STORAGE Impurities Gaseous Products Gas Well Liquid Products Cleaner Compressor Station Associated Gas and Oil Well GAS PRODUCTION -4 - International Gas Pipeline

Country 1 Country 3 Overall picture C 1 Producer C 3 Country 2 Transmitter TT 2 1 T 1 L 1 TT 3 1 Season 1 LNG Liquef S 1 Sectors R 3 Storage S 3 LNG Regasif M 3 M 1 Season 2, 3 K 1, 2, 3 Marketer -5 - K 1, 2, 3

Model Overview - Static complementarity model of the world natural gas market - Daily flows seasonal patterns over one year - Detailed representation of the players in the natural gas business: - Producers - Transmitters - Liquefiers - Regasifiers - Storage operators - Marketers - Mixed complementarity approach instead of MPEC which would we hard to solve One stage with market power (one player) - Other players assumed to behave competitively, and to be linked via market-clearing conditions -6 -

Producer’s Problem: Description Maximize production revenues less production costs s. t. - bounds on daily production rates - bounds on volume of gas produced in time-window of analysis (one year) Decision Variables - How much to produce in season and year (cubic meters/day) Market Clearing - Producers’ sales MUST EQUAL Transmitter’s purchases from Producer -7 -

Producer’s Problem: Formulation -8 -

Complementarity Formulation: KKT Conditions of the Producer‘s Problem -9 -

Transmitter’s Problem: Description Maximize selling revenues less purchase costs from “its” domestic producer s. t. - material balances, including international pipeline losses Decision Variables - How much to sell in season and year (cubic meters/day) - How much to buy from producers and neighboring transmitters (cubic meters/day) Market Clearing: - Sales MUST EQUAL Purchases of (domestic) Marketers, Storage and LNG Liquefaction - 10 -

Transmitter Characteristics Interfaces between producers and end-user markets (marketers) Separate entity Market mechanism vs. ‘dedicated trading companies for each producer’ Some real world counterparts (e. g. , Gazexport) Low/high calorific markets: not that interesting not included for the time being New concept, seems to work, non-conventional Here presented formulation: producer dedicated transmitters - 11 -

Incorporating market power in Transmitter’s Formulation • Market power in Europe: producers • Initial transmitter formulation: producers face only one buyer. Exerting market power only relative to this one buyer, no direct means to withhold gas from or bring it to specific markets. So what…? • multiple transmitters, dedicated for one producer • integrate transmitter into producer (more conventional) àone transmitter producer = the exporting subsidiary of a producer • The transmitter is exerting the market power vis-à-vis its customers, not the producer • Decision Criteria: • Usefulness • Num Variables ~ Solvability, Acceptance • Usefulness: there are examples of this type of agent in reality (e. g. , Gazexport) • Solvability: turns out not to be increased but we stick with this representation - 12 -

Transmitter’s Problem: Formulation - 13 -

Pipeline Operator’s Problem: Description Maximize congestion revenues s. t. - capacity bounds on flow Decision Variables - How much capacity to sell to Transmitters (in each season and year) Market Clearing - Capacity sold to Transmitters MUST EQUAL Capacity purchased by Transmitters - 14 -

Pipeline Operator’s Problem: Formulation - 15 -

Storage Reservoir Operator’s Problem: Description Maximize net revenues from marketers less injection costs, distribution costs, and purchasing costs from transmitter and LNG Regasification s. t. - volumetric bound on working gas - maximum extraction rate bound - maximum injection rate bound - annual injection-extraction balancing Decision Variables - How much gas to buy from Transmitters and LNG Regasifiers - How much gas to sell to Marketers Market Clearing - Storage Operators’ Sales MUST EQUAL Marketers’ Purchases from Storage - 16 -

Storage Reservoir Operator’s Problem: Formulation - 17 -

Marketer/Shipper’s Problem: Description 2 1 4 Marketer/Shipper 3 Maximize demand sector revenues less local delivered costs from transmitter, storage and LNG Regasification s. t. - Sales to Sectors MUST EQUAL Purchases from Transmitter, Storage, LNG Regasifier Decision Variables - How much to buy from transmitter, storage and LNG - How much to sell to each sector - 18 -

Marketer/Shipper’s Problem: Formulation 2 1 4 Marketer/Shipper 3 Marketers don‘t have a decision variable, but are determined by their demand function to the transmitters Market clearing must be satisfied - 19 -

Application Covers Europe and all LNG world-wide, one player each type in each country (when applicable. ) 51 countries (also outside of Europe) 28 producers • 15 large, 5 LNG only, 13 ‘domestic only’ 36 consuming countries • 3 sectors/countries, 6 LNG only LNG: • 10 Liquefiers, 15 Regasifiers, 150 LNG routes 20 Storage Operators 74 pipelines Programmed in GAMS using PATH - 20 -

State of the Work and Future Plans Currently: Running simulations for model with market power and for different scenarios The rich data input allows to investigate issues like international LNG flows, substitution effects between Russian, North African pipeline gas and LNG Later: extensions of the model • Stochasticity into players’ problems, for example with stochastic demand realizations • Alternative demand functions • Scenario Reduction & Decomposition • Other strategic behavior/market power for producers, marketers • Dynamic model with decisions on investments in transport infrastructure - 21 -

Thank you very much for your attention! For any comments and suggestions, please contact: fholz@diw. de - 22 -