Seasonal Flexibility in the European Natural Gas Market.

• Author: Riepin, Iegor

1. INTRODUCTION

   Seasonal demand swings (i.e. differences in gas consumption across seasons) constitute a fundamental element of the European gas market. Heating demand, which is primarily driven by temperature, is high in the winter but low in the summer, causing strong seasonal demand swingsthe aggregated European gas demand is typically more than twice as high in the winter than it is in the summer.1 European countries balance the demand variation with a mix of flexibility options, such as varying domestic gas production, varying pipeline or liquified natural gas (LNG) imports, and operating of underground gas storage facilities.

   These options differ in terms of both cost and availability. Varying domestic gas production requires free domestic production capacity. Varying imports requires free foreign production capacity and free capacity in transportation infrastructure. LNG imports are only available to certain European countries, as they require regasification terminals; however, increasingly integrated European gas markets allow cross-border gas transfers. Gas storages provide seasonal flexibility by shifting gas demand from the winter to summer; utilization of gas storages is also subject to available capacities and technical characteristics of storage facilities.

   The years leading up to 2018 saw a relative abundance of flexible capacity in the gas market. This was largely due to (i) low gas demand in almost all European countries over the past decade, (ii) investment in additional assets, (2) and (iii) the integration of European gas markets driven by the optimized utilization of existing assets. This abundance of flexible capacity was reflected by low seasonal gas price spreads at European gas hubs and the low utilization of European regasifi-cation terminals.

   However, this abundance of seasonal flexibility is not permanent. In the future, several factors will put significant downward pressure on the availability of flexibility options. First, market forces reflect an oversupply of such options amid lower price spreads between summer and winter months. Lower spreads make investments in additional flexibility less attractive and may even cause a shutdown of existing flexibility options. Second, both the Netherlands and the United Kingdomthe European Union's two largest gas producers-will provide less flexibility in the future. In response to seismic activity, the Dutch government announced a series of directives to limit maximum annual production from the Groningen field; (3) the annual-production cap of 42.4 bcm p.a., which had been in place since January 2014, was reduced to 21.6 bcm p.a. for the 2017-2018 gas year with the ultimate goal being to completely shut down the Groningen field by 2030 (Honore 2017; Snam, IGU, and BCG 2018). Furthermore, in 2016, the Dutch government established regulations to spread out natural gas production as evenly as possible throughout the year. In terms of monthly fluctuations, this regulation fixes gas extraction from the Groningen field each month to a range of plus or minus 20% (Honore 2017). Taken together (Figure 1), these two changes reduce the Groningen field's seasonal flexibility by around 85% (from a swing of 4-5 bcm between the winter and the summer in 2011-2013 to one of just 0.6 bcm in 2017-2018). (4) The UK government also indicates that there will be a rapid decline in domestic gas production. (5) The projected 2030 production volume is 17.8 bcm p.a., which constitutes a drop of more than 50% from the 2015 production volume. Consequently, gas import dependency is expected to increase significantly-from 44% in 2015 to 74 % in 2030 (The Oil and Gas Authority 2016). Europe must respond to this drop in domestic production and the associated decline in flexibility with alternative options and find a new cost-optimal way to cover seasonal demand swings.

   On the other hand, new infrastructure projects are expected to enter the market. The TYNDP infrastructure report (2018), published by ENTSOG, identifies plans for around 120 transmission and compressor stations, 27 LNG terminals, and 9 underground storage facilities. Forty-six of these projects have been approved for investment; almost 75% of the submitted initiatives are expected to be commissioned no later than 2022.

   Overall, the future need for seasonal flexibility remains unclear. Assessments must consider both regulatory and economic changes in the gas market structure. The application of an economic modeling framework can reveal market fundamentals and the evolving structure of flexibility options.

   This paper analyzes seasonal gas demand swings and the flexibility necessary to cover them using a fundamental modeling framework. We analyze how various flexibility options (domestic production, gas storage, and pipeline and LNG imports) cover European demand fluctuations in monthly resolution. We contribute to the existing discussion on seasonal flexibility by addressing the problem with a mathematical gas market optimization model. Our paper provides valuable empirical insights into the decline of gas production in northwestern Europe, much of which stems from the Groningen event and recent developments in the UK (see data section below). Such structural breaks are optimally addressed by fundamental models. Furthermore, we differentiate between LNG and pipeline imports to analyze the specific flexibility features of pipeline and LNG supplies. In terms of methodology, we construct a gas market model and publish the complete source code. (6) Furthermore, we propose a new metric to improve the quantification of supply sources' provision of seasonal flexibility. This metric extends the well-established coefficient of variation.

   The remainder of this paper is organized as follows. Section 2 details the methodology employed for this study. We present our modeling framework as well as a mathematical description of our market optimization model and its associated data. Section 3 presents and interprets the modeling results. We begin by illustrating modeling results in the form of monthly gas demand profiles to lay a background for the analysis. We continue with an investigation into supply sources' quantitative contributions to cover gas demand and determine which supply source offers the most flexibility in covering seasonal demand fluctuations. Finally, Section 4 concludes with major findings and outlines our ideas for future work.

2. METHODOLOGY AND DATA

   We construct and apply an optimization model covering the European gas market and its neighboring regions. The model is formulated as a deterministic linear programming problem with perfect foresight. This allows us to solve the large-scale optimization model with intertemporal constraints and high temporal granularity over a large timespan. As such, decision variables (e.g. gas production, trade, and storage) have a time resolution of 12 consecutive months for each modeled year. We simulate market operations over a long time period (from 2018 to 2030). This enables us to explore future market developments driven by changing supply and demand fundamentals. The model's spatial coverage encompasses European countries and major non-European gas exporters (Norway, Russia, United States, Algeria, Libya, Nigeria, and Qatar). The dataset, which includes all necessary economic and technical data, is taken from publicly available sources. We discuss our assumptions regarding gas demand and supply structures as well as transmission infrastructure elements below. The model is formulated in GAMS v25.1 (7) and solved with a CPLEX solver with default solver settings. (8) The applied GAMS code, associated input data, and processing of the results are available in a public GitHub repository: https://github.com/Irieo/SeasonalFlex

   2.1 Related Work

   The mathematical modeling of gas markets has a long history. Mathiesen et al. (1987) were among the first to model the European gas market. Interest in model-based analysis of the European gas market increased significantly at the end of the 1990s when the European Commission initiated liberalization policies. The growth of computing power and the advancement of mathematical models-paired with the challenges of energy transitions-facilitated the widespread employment of elaborated mathematical models in this field. Since then, a considerable amount of research has been oriented to the economic modeling of the European gas market.

   Most of the studies in this research stream have adopted one of two methodological approaches. Studies using the first approach analyze the operation of the gas market using mixed complementarity-based equilibrium models, which allow for the incorporation of individual players' strategic decisions (e.g. Chyong and Hobbs 2014; Egging 2010; Gabriel et al. 2005; Hecking 2012; Holz 2009; Holz et al. 2016; Lise and Hobbs 2009). Those that use the second approach employ bottom-up optimization models in which the whole system is optimized with regard to the costs of the gas supply and relevant constraints. While these models must rely on the assumption of perfectly competitive market operation, they benefit from the use of optimization solvers that allow for the incorporation of a high level of spatial and temporal resolution as well as more detailed representation of complex gas infrastructure (e.g. Dieckh[eth]ner 2012; Eser et al. 2019; Kiss et al. 2016; Neumann et al. 2011; Petrovich et al. 2016; Weigt and Abrell 2016).

   Most recent studies that use a bottom-up optimization model to analyze flexibility in the European gas market focus on short-term flexibility, often dealing with supply security (e.g. the ability of the gas system to sustain operation under shock scenarios). REKK (2014) analyzes the flexibility of the European gas market with a focus on how interconnectivity, gas storage, and demand-side adjustments impact the resilience of the gas system during supply shocks. T[eth]th et al. (2017)...