Applied Energy 238 (2019) 816–830
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Impact of Nord Stream 2 and LNG on gas trade and security of supply in the European gas network of 2030
T
⁎
P. Eser , N. Chokani, R. Abhari Laboratory for Energy Conversion, Institute for Energy Technology, ETH Zurich, 8092 Zurich, Switzerland
H I GH L IG H T S
gas system model simulates impact of Nord Stream 2 and LNG on Europe in 2030. • Novel Nord Stream 2, gas transit through Poland and Ukraine reduced by 23% and 13%. • With loses 40% market share for any gas bypassing Ukraine via Nord Stream 2. • Russia import strategy requires 17% of European pipelines to be bidirectional. • LNG-focused • Short-term disruption of LNG increases gas prices in southern Europe by up to 40%.
A R T I C LE I N FO
A B S T R A C T
Keywords: Nord Stream 2 LNG EU Energy Union Gas system simulation Security of supply
By 2030, the projected decrease in domestic European gas production will result in a shortfall of 12% of the EU’s gas demand. In this work, two different strategies to cover this shortfall are investigated: Increased imports of liquefied natural gas (LNG) from different global sources versus increased deliveries of Russian pipeline gas via Nord Stream 2. A novel gas system model, which captures both the market behavior of gas traders and gas system operators, is developed and applied. The model is highly detailed, as it simulates all individual components of the Europe-wide gas transmission system with hourly resolution. Simulations show that Nord Stream 2 impacts the gas transit through Poland more severely (23% loss of transit flows compared to 2014) than the transit through Ukraine (13% loss). Completely cutting off Ukraine is found to be a detrimental strategy for Russia, because only 40% of Ukrainian transit can be re-routed via Nord Stream 2 over a short timeframe. Increased imports of LNG are found to require 17% of European gas pipelines to be bidirectional, which requires significant investments into the European gas infrastructure. Additionally, the penetration of LNG is found to be highly sensitive to the price of LNG, with LNG losing 50% market share when priced 20% more expensively than pipeline gas. Hence overall, the choice between Nord Stream 2 and LNG exposes Europe to either a) the political risk of being more dependent on Russia or b) the technical and financial risks of importing the globally traded commodity LNG.
1. Introduction In its Energy Strategy 2030, the EU targets a 40% cut in greenhouse gas emissions compared to 1990 levels [1]. As well as achieving an overall reduction in per-capita energy consumption, a shift from coal to natural gas is planned. Thus in its ten-year network development plan, the European Network of Transmission System Operators for Gas (ENTSOG) predicts that the EU’s overall annual gas demand will remain approximately constant until 2030 [2]. Over this period, the domestic gas production of the UK and the Netherlands, the two largest gas producers within the EU, is projected to reduce by 60% (UK) and 50%
⁎
(Netherlands) by 2030 [2]. Since alternative gas sources such as shale gas are not considered to be financially competitive in Europe by 2030 [3], the reductions in domestic gas production will result in a shortage of gas in the EU by 2030, which will need to be covered with additional imports. Two diametrically different strategies of gas imports are investigated in this work: (1) Increased imports of liquefied natural gas (LNG) versus (2) the construction of Nord Stream 2 (NS2), an additional pipeline connecting Russia and Germany through the Baltic Sea. The first import strategy, an increased dependence on LNG, will diversify gas supplies, yet potentially at the expense of increased import prices. As Wood [4] argues, the projected increase in natural gas consumption
Corresponding author. E-mail address:
[email protected] (P. Eser).
https://doi.org/10.1016/j.apenergy.2019.01.068 Received 10 July 2018; Received in revised form 4 January 2019; Accepted 11 January 2019 0306-2619/ © 2019 Elsevier Ltd. All rights reserved.
Applied Energy 238 (2019) 816–830
P. Eser et al.
Nomenclature
Tav Tstd Vgas,max Vs Zgas η γ ρstd
Sets a c n nb s t
pipeline of gas system compressor station within gas system node of gas system node at system boundary (subset of n) storage of gas system hour of year
average temperature of gas in pipelines [310 K] standard ambient temperature [288.15 K] maximum flow velocity of gas in pipelines [15 m/s] volume of storage s [m3] compressibility of nat. gas [0.9] efficiency of compressors [0.8] heat capacity ratio of natural gas [1.3] density of natural gas at pav [55 kg/m3]
Variables arcflowa,t mass flow of gas through pipeline a at time t [ton/h] cpowc,t compressor power usage of compressor c at time t [MW] cpowss,t compressor power usage at storage s at time t [MW] importnb,t import of gas at system boundary nb at time t [ton/h] leftflowa,t gas flow into left end of pipeline a at time t [ton/h] lpa,t line packing in pipeline a at time t [ton] ms,t mass of gas stored in storage s at time t [ton] pn,t gas pressure at node n at time t [MPa] Pnodal nodal price of gas at node n at time t [€/MWh] n,t Pstorage price of gas stored in storage s at time t [€/MWh] s,t Plinepack price of gas stored in pipeline a at time t [€/MWh] a,t psss,t gas pressure in storage s at time t [MPa] rightflowa,t gas flow into right end of pipeline a at time t [ton/h] stoflows,t mass flow of gas into/out of storage s at time t [ton/h]
Parameters dn,t gas supply/demand at node n at time t [ton/h] diama diameter of pipeline a [m] epipe pipeline efficiency [1.0] fn,t flexible demand at node n at time t [ton/h] g specific gravity of natural gas [0.65] gascostnb,t cost of gas at import node nb at time t [€/ton] importFlowLimnb maximum hourly import at node nb [ton/h] importLimnb maximum annual import at node nb [ton] lena length of pipeline a [km] lpa,min minimum line packing in pipeline a [ton] lpa,max maximum line packing in pipeline a [ton] lprel relative line packing optimization target [65%] target Mmolar molar mass of natural gas [20 g/mol] nVs maximum volume of storage s [million Nm3] pav average pressure of natural gas in pipelines [6.0 MPa] pmean mean pressure in pipeline a at hour t [MPa] a,t pstd standard atmospheric pressure [0.1013 MPa] pipeUsageCost cost of pipeline usage [2.7 ct/100 km ∗ MWh] R gas constant [8.314 J/kg K]
Abbreviations ENTSOG European Network of Transmission System Operators for Gas LNG Liquefied Natural Gas NS2 Nord Stream 2
impacts on European gas prices by exerting more market pressure on piped gas imports from Russia. Despite this finding, Egging et al. [9] predict a reduced role of LNG imports for Europe after 2020 using a mixed complementarity model for the global gas market. Using a welfare maximization model, the role of Nord Stream on inner-European gas prices was investigated by Neumann et al. [10], who find that Poland and Slovakia incur the highest increases in gas prices if gas is routed through Nord Stream instead of through Poland and the Ukraine. Similarly, a detrimental effect of NS2 on gas transit flows through Ukraine and Slovakia is seen by Kotek et al. [11] using an equilibrium model. Using a Hotelling model, Orlov [12] finds the projected increase in domestic Russian gas prices to increase the export of gas to Europe, which would further increase the gas transit flows from Russia to Europe. The direct competition of LNG and Russian pipeline gas in the European gas market was investigated in other works that showed different outcomes. Using a multi-period equilibrium model, Holz et al. [13] find that the current gas infrastructure in Europe is sufficient in the future due to the projected reduction of the overall gas demand. By applying a partial equilibrium model with monthly resolution, Abrell et al. [14] find that NS2 has financially beneficial impacts for European gas consumers compared to increased LNG imports. Using a global gas market model, Lochner and Bothe [15] find Russia to be the main supplier of natural gas for Europe in 2030. On the other hand by applying a more detailed gas infrastructure model, Lochner et al. [16] find cheap LNG imports in southern Europe result in reduced gas flows through the Nord Stream pipeline. Apart from the overall supply of natural gas for Europe, several studies have also looked at questions of security of supply arising from short-term disruptions of gas flows. Using partial equilibrium models, both Baltensperger et al. [17] and Richter et al. [18] find that eastern
in Asia, especially in China, will attract short-term shipments of LNG to Asia, both jeopardizing the EU’s security-of-supply and increasing the cost of LNG in Europe. On the other hand the second import strategy, the construction of NS2, is a highly debated issue within the EU for several reasons. While both Germany and Russia have fared well with the construction of a first pipeline (termed Nord Stream) through the Baltic Sea in 2011, several countries in Central and Eastern Europe, especially Poland, Slovakia and Ukraine, have suffered losses in gas transit fees due to diversion of gas from their pipelines to the Nord Stream pipeline. Since the construction of NS2 would further divert gas, several EU member states argue that NS2 contradicts the goals of the EU energy policy, because NS2 causes conflicts of interest between Western and Eastern European countries, while increasing the dependence on Russian natural gas [5–7]. Many previous works have looked at gas supply strategies for Europe with a large variety of different methodologies. Most of these prior works can be attributed to one of two dominant branches of models that have been applied to model the European gas market. In the first branch, (partial) equilibrium models are used to describe the market forces within the gas markets. These models are often formulated as mixed complementarity problems to incorporate the optimality conditions of strategic players in oligopolistic markets, similar to Cournot models. In such equilibrium models, the individual players of the gas value chain (producers, traders, transmission system operators, etc.) are oftentimes modeled individually to incorporate their profit-optimizing incentives. In the second branch, linear optimization models are used to simulate the physical gas infrastructure. While these models are more detailed with regards to the restrictions of the physical gas infrastructure, they place less emphasis on the market power of individual actors within the gas markets. Using an oligopoly model, Dorigoni et al. [8] find increased LNG imports to have beneficial 817
Applied Energy 238 (2019) 816–830
P. Eser et al.
shown, and then outcomes of the different 2030 scenarios are discussed in-depth. Section 4 summarizes the findings and highlights the most significant conclusions.
European countries are most adversely affected by supply disruptions from Russia. Using a Monte Carlo based approach, Monforti et al. [19] confirm these findings. By applying a mixed complementarity model, Egging et al. [20] identify that increased LNG imports are cost-intensive replacements of disrupted gas flows from Russia. Most of the aforementioned works use simplified network topologies in which there is one representative system node per country; such simplified models do not resolve the real pipeline network. As Lochner et al. [21] and Holz et al. [22] identify, simulation of the real gas pipeline network is crucial in order to be able to identify bottlenecks in the gas system for future scenarios. Additionally, Deane et al. [23] stress the necessity of simulating line pack, which is the amount of gas stored within a pipeline, to investigate the dynamic responses of the gas system to disruptions of supply. Since disruptions of supply usually happen on the time scale of days to weeks, the monthly or annual time steps used in most of the prior works are insufficient to analyze such disruption scenarios in detail. An identification of bottlenecks within a country further necessitates a high spatial resolution of demand modeling, as explained in Dieckhöner et al. [24]. In this regard, Bouwmeester et al. [25] stress the necessity of differentiating the gas consumer sectors in order to realistically model the spatial distribution of gas demand. Given the aforementioned concerns around Nord Stream 2’s impact on European security-of-supply, and the potential to replace Russian gas with LNG by 2030, a gas system model with high spatial and temporal resolutions is necessary to identify the technical, economic and political risks associated with Europe’s gas sourcing strategy in 2030. None of the aforementioned works uses such a highly resolved model to address the question of Europe’s gas supply in 2030. Hence to conduct a comprehensive analysis of gas sourcing strategies and security-of-supply scenarios for Europe in 2030, a novel simulation approach is developed in this work. Compared to prior works, this novel approach adds to the literature in the following regards:
2. Methodology 2.1. Bottom-up modeling of Europe’s gas infrastructure A novel approach to simulate the European gas system is developed in this work, thereby extending our simulation framework EnerPol, which hitherto focused only on electric power systems [26]. In this approach, the physical and economic aspects of gas demand, sourcing and transportation are modeled. In this study, the interconnected gas transmission system of Central and Western Europe is simulated with hourly time resolution for a full year. The geographical extent covers Austria (AT), Belgium (BE), Switzerland (CH), the Czech Republic (CZ), Germany (DE), Spain (ES), France (FR), Great Britain (GB), Italy (IT), the Netherlands (NL), Poland (PL) and Portugal (PT). A Geographical Information System (GIS) database of the natural gas transmission system was developed from publicly available sources [2,27–29]. In this database, the individual components of the real high-pressure gas transmission system are modeled. The model comprises 500 network nodes and 52,000 km of high-pressure gas pipelines, as shown in Fig. 1, as well as 150 compressor stations and 80 gas storages, as shown in Fig. 2. The structure of the simulations consists of two steps, as shown in Fig. 3. In the first step, the annual sourcing of natural gas is optimized over a full year with a Monte Carlo approach. 30 optimizations of the gas sourcing, each with stochastically varied gas prices at the boundaries of the network, are solved. The resulting imports of natural gas at the system boundaries are then used as boundary conditions in simulations of the gas system operation, which optimizes the operation of pipelines, compressors and storages for every hour of a year. The underlying model equations are implemented in the Python-based optimization modeling language Pyomo [30], and are mathematically described below in Sections 2.2 and 2.3.
1. The gas transmission system of Europe is modeled in a bottom-up approach with all physical components of the system. Pipelines, compressor stations and gas storages are modeled individually; compressibility of gas is fully accounted for. 2. The gas system is simulated with hourly resolution for the whole year of interest. Combined with the simulation of line pack, this enables a detailed analysis of disruption-of-supply scenarios with hourly time resolution. 3. The demand of natural gas is modeled hourly with high spatial resolution. The gas demand is differentiated by end-user sectors of industry, commerce, power generation and households. The spatial distribution of each sector is used to derive realistic profiles of gas demand, which enable a more realistic identification of system bottlenecks in each country. 4. To capture the real-world behavior of actors in the gas market, a two-step modeling procedure is applied in this work: First, the annual imports of natural gas are financially optimized, which captures the market behavior of gas traders. Second, the nominations of the gas traders are passed into an hourly gas system operation model, which physically optimizes the short-term operation of the gas network with hourly resolution, hence capturing behavior of the gas system operators. 5. The annual gas sourcing model is applied within a Monte Carlo framework, in which stochastically varied hourly import prices of gas are applied at the boundaries of the simulated network. This stochasticity captures the hedging strategies of gas traders in their year-ahead gas purchases.
2.2. Annual gas sourcing model The annual gas sourcing model financially optimizes the gas imports at the boundaries of the simulated gas network. In this work, the gas imports are from Algeria, Libya, Morocco, Norway, Russia and globally traded LNG. The annual gas sourcing model hence captures the
The remainder of the paper is organized as follows: In Section 2, the methodology of the simulation framework is explained in detail. Furthermore, the assumed conditions for the scenarios of the year 2030 are described. In Section 3, a validation of the gas flow methodology is
Fig. 1. Model of network nodes and gas pipelines in Central and Western Europe used in this work. 818
Applied Energy 238 (2019) 816–830
P. Eser et al.
Fig. 4. Example of the stochastically varied import prices that are used to capture gas traders’ uncertainty in gas sourcing strategies. Fig. 2. Model of compressor stations and gas storages in Central and Western Europe used in this work.
natural gas is assumed to be incompressible. Hence in the annual gas sourcing model no compressor stations are modeled to transport the gas through pipelines, and no gas can be stored as line pack in the pipelines. The model does however include the flow restrictions within each individual pipeline, as well as the volume constraints of the gas storages. The hourly reference price profiles of gas at each boundary of the simulated gas network are derived from [31]. To incorporate the gas traders’ uncertainty when deciding on a year-ahead purchasing strategy, a Monte Carlo approach is used. Specifically, the annual gas sourcing is simulated 30 times with the gas prices stochastically varied relative to the reference price profile in each simulation. The variability of the stochastic deviations is specified such that 95% of the price profiles are within the maximum annual spread of the corresponding reference price profile at the end of the year. An example of the stochasticity of the import prices is shown in Fig. 4. Overall, the model is solved to minimize the cost of annual gas sourcing across all countries for all hours of a year simultaneously (1). Simultaneous optimization of gas sourcing across all simulated countries assumes a perfectly coordinated international gas market. This is in line with the EU’s Third Energy Package [32], which mandates the creation of an EU-wide gas market that includes EU-wide gas trading hubs and internationally coordinated planning and operation of the gas system.
⎛ min ⎜∑ ⎝ t
∑ nb
⎞ importnb, t ·gascostnb, t ⎟ ⎠
(1)
The maximum storage of the existing gas storage sites is calculated from the storage volume and the storable gas mass using the ideal gas law (2). The minimum storage is set to be 50% of the maximum capacity (3), in order to simulate risk-averse operation of the storage sites. The amount of gas flowing into or out of a storage site is defined with (4). The amount of gas stored at the end of a year is set to be equal to the amount stored at the beginning, in order to simulate an annually repetitive operation of storages. The maximum flow rate of gas is scaled with the diameter of the pipelines using a continuity Eq. (5). The import of gas at the import nodes is limited with two constraints: First, the amount of gas that can be imported within an hour is limited to the maximum historically observed import at the node (6). Second, the annual sum of gas to be imported at a given node is limited to the maximum historically observed annual import at the node (7). Eqs. (6) and (7) therefore capture the fact, that gas from external sources (e.g. Norway, Russia, Algeria) is limited, both hourly and annually. The historical import data are obtained from [28]. Domestic gas production is treated as an import in the model, and thus is also subject to the constraints (6) and (7). At every node of the gas system, the hourly
Fig. 3. Flow chart of simulation structure used in this work.
behavior of gas traders: Assuming perfect foresight for a full year, the least-cost combination of gas imports from different sources that meet the domestic European gas demand is determined. The model captures the gas flows and their restrictions with hourly resolution, while optimizing the imports for all hours of an entire year in one optimization. Since such a model is computationally complex and resource-intensive, 819
Applied Energy 238 (2019) 816–830
P. Eser et al.
physical flows of gas are balanced following a mass conservation Eq. (8). In (8), the gas flows from pipelines (a → n denotes all pipelines a connecting to node n), storages (s → n denotes all storages s connecting to node n) and imports are balanced with the end-user’s gas demand at the node for the given hour. In (8), the own-price elasticity of gas demand is accounted for with the parameter f, which reduces the gas demand of the industrial and power sectors based on price elasticity statistics [33].
ms, max =
nVs·pstd ·Mmolar
∀s
Zgas·R·Tstd
ms, min = 0.5·ms, max
flowlima =
(2)
∀s
stoflows, t = ms, t − ms, t − 1
allowable line pack is set as 30% of the maximum to avoid that the pipelines empty completely (15). The line pack at hour t is calculated as the line pack at hour t-1 plus the difference between inflow and outflow of the pipeline at hour t (16). Using the adiabatic compressor power equation, the power of each compressor station is calculated from the flow rate through the compressor and the desired pressure ratio (17). For the storages, the maximum storable volume and the pressure of gas within the storage are calculated with the ideal gas laws (18) and (19) respectively. The compressor power needed to feed gas into storage is calculated based on the flow rate and the pressure ratio of gas within the storage compared to the pressure at the adjacent network node with an adiabatic compressor power Eq. (20). The mass flow balances applied in the annual gas sourcing model (2)–(4) are also applied hourly in the hourly operation model. At each node, the mass flows of inflow and outflow gas, the imports (which are determined in the annual gas sourcing model) and the hourly gas demands of end-users are balanced using the mass-conservation Eq. (21).
(3)
∀ s, t
(4)
Π ·diama2 ·ρstd ·Vgas, max ·3600 4
∀a
(5)
− importFlowLimnb ≤ importnb, t ≤ importFlowLimnb
∀ nb, t
(6)
pamean = ,t
8760
− importLimnb ≤
∑
importnb, t ≤ importLimnb
∀ nb (7)
t=1
0=
ρa, t =
∑ arcflowa→n,t − (dn,t − fn,t ) − ∑ stoflows→n,t + importn,t
The hourly import profiles from the annual sourcing model, which represent the nominations of gas flows from the gas traders, are used as the boundary conditions for the hourly gas system operation model. In the gas system operation model, the compressibility of gas is fully accounted for. Hence compressor stations are modeled to transport the gas through the pipelines and to store the gas in gas storages. The compressed gas may also be stored as line pack in pipelines. With hourly resolution, the gas system operation is optimized to minimize the cost of operation while maintaining safe line pack levels in the pipelines (9). Therefore, the hourly gas system operation model captures the behavior of the gas system operators.
∑ s
cpowsss, max
1.0788
Tstd pstd
∀ a, t
·
∑ a
lpa, t ⎞ ⎞ ⎛ ⎜lptarget − ⎟ ⎟ lp a, max ⎠ ⎝ ⎠
⎝
Vs =
∀t
( )·( γ γ−1
γ−1 γ Zgas ⎛ pright , t Tav· η ·⎜ ⎛ p ⎞ left , t
⎝
⎠
ρstd
⎞ − 1⎟ ⎠
ps, max ·Mmolar Zgas·R·Tstd·ms, t Vs·Mmolar
(13)
(14) (15)
leftflowa, t
Zgas·R·Tstd·ms, max
psss, t =
∀a
∀a
lpa, t = lpa, t − 1 + (leftflowa, t − rightflowa, t )
cpowc, t = 4.0639·
(12)
( )
·diama2.6182 ·ρstd /24
lpa, min = 0.3·lpa, max
2
+
∀a
pn, max ·Mmolar Π lpa, max = ⎛ ·diama2 ·lena⎞· ⎝4 ⎠ Zgas·R·Tstd
2.3. Hourly gas system operation model
cpowsss, t
6730·(pleft , t − pright , t ) (g 0.8539·Tav·lena·Zgas )0.5394
(10)
(11)
Π ·diama2 ·ρa, t ·Vgas, max ·3600 4
leftflowa, t = 4.5965·10−3·epipe ·
∀ a, t
∀ a, t
Zgas·R·Tav
flowlima =
Each Monte Carlo simulation yields an hourly, cost-optimal import profile at each of the boundary nodes of the simulated network. The average of these import profiles is taken as the risk-reduced gas importing strategy from the gas traders’ perspective, and therefore is used in the hourly gas system operation model.
cpowc, t ⎛ min ∑ + ⎜ cpow c, max ⎝ c
pamean ·Mmolar ,t
∀ n, t (8)
pleft , t ·pright , t ⎞ 2 ⎛ · p + pright , t − 3 ⎜ left , t pleft , t + pright , t ⎟ ⎝ ⎠
∀ a, t
(16)
)
·24
∀ c, t (17)
∀s (18)
∀ s, t
(19)
(9) Since gas is compressible in this model, the gas flows and flow rate limits within each pipeline depend on the pressures of the gas at both sides of the pipeline. The minimum and maximum allowable pressures at each node are specified, and the hourly flow rate limit in each pipeline, calculated from a continuity Eq. (12), is based on the average gas pressure and corresponding gas density, which are respectively calculated from (10) and the ideal gas law (11). The hourly flow rate of gas through each pipeline is calculated based on the pressure at both sides of the pipeline, using the Panhandle A equation [34], and is constrained not to exceed the hourly flow rate limit. The Panhandle A equation is an empirical representation of the flow of compressible gas through long pipelines, and is commonly used to calculate the pressure loss in natural gas pipelines. To avoid non-convexities in the model equations, the Panhandle A equation is linearized, with the assumption that pressure gradients along the pipelines are small (13). Using the ideal gas law, the maximum line pack is derived from the volume of the pipeline and the maximum allowable pressure (14). The minimum
cpowsss, t = ·
Zgas η
( )·stoflow ·T
γ R · γ−1 Mmolar
⎛ pss · ⎜ ⎛ p s, t ⎞ ⎝ s → n, t ⎠ ⎝
γ−1 γ
⎞ 1 − 1⎟· 3600 ⎠
av
s, t
∀ s, t (20)
0 = ∑ rightflowa → n, t − ∑ leftflowa → n, t − dn, t − ∑ stoflows → n, t + importn, t
∀ n, t
(21)
The hourly gas prices are calculated for each pipeline, node and storage with (22)–(24). In (22), the price of line pack within a pipeline is calculated as the mean of the nodal prices of the two adjacent network nodes, plus a price factor of 2.7 ct/(100 km ∗ MWh) derived from [35–37], which accounts for the infrastructure cost of transporting gas through the pipeline. The nodal gas price for non-boundary nodes is calculated as the mass-weighted average of the line pack prices of all incoming pipelines (23). For the import nodes at the boundaries of the simulated network, the nodal price is specified from the annual gas 820
Applied Energy 238 (2019) 816–830
P. Eser et al.
sourcing model. The price of gas in a storage is derived from the nodal price of the gas that is fed into the storage (24).
Palinepack = ,t
nodal nodal Pleft , t + Pright , t
2
∑ Pnnodal = ,t
+ lena·pipeUsageCost
a→n
∑
leftflowa → n, t
AT BE CH CZ DE ES FR GB IT NL PL PT
∀ n, t
nodal Psstorage , t − 1 · ms, t − 1 + Ps → n, t · stoflows, t
ms, t
(23)
∀ s, t
Annual gas demand 2014 [bcm]
Annual gas demand 2030 [bcm]
Relative change
7.8 16.0 2.6 8.0 79.0 27.0 37.0 71.0 62.0 40.0 18.0 4.0
7.7 18.2 2.8 10.1 73.5 37.0 37.0 80.9 67.0 36.0 23.6 5.8
−1% +14% +9% +26% −7% +37% 0% +14% +8% −10% +31% +45%
(22)
leftflowa → n, t ·Palinepack → n, t
a→n
Psstorage = ,t
∀ a, t
Table 1 Annual gas demand in 2014 and 2030 used in this work.
(24)
2.4. High-resolution hourly gas demand modeling 2.5. Scenarios for the European gas system in 2030 Similar to our previous electricity simulations with the EnerPol framework [26], the gas demand, compressor stations and gas storages are aggregated to the nearest node of the gas system using a Voronoi tessellation. The hourly profiles of gas demand are derived from [28] by calculating the difference between gas imports, exports and changes in gas storage levels for every hour, and are scaled for each country to match the country’s annual gas demand that is given in [38]. For each hour, the countrywide gas demand is disaggregated to the end-user categories of households, industry, commerce and power generation based on the relative consumption shares given in [39]. Using the spatial distribution of each end-user sector, which is derived from [40] for population data, [41] for industrial and commercial sites, and previous work by the present authors [42] for gas power plants, the gas demand is allocated across each country and aggregated to the geographically closest node of the gas system, using a Voronoi tessellation of the gas system nodes. An example of this approach is shown for northern Germany in Fig. 5. The allocation of gas demand from household and commercial sectors is further refined by taking into account the hourly ambient temperature, such that colder areas of a country have a higher share of the country’s hourly gas demand. The temperatures are derived from inhouse meso-scale weather simulations, which are performed with the open source weather simulation tool WRF [43]. More information about the demand model can be found in [44].
The year 2014 is used as the reference year in this work. The annual domestic gas demand of each country for 2014 is derived from [38], and is shown in Table 1. For 2030, the predicted demand from ENTSOG’s “slow progression” scenario [2] is assumed. The resulting gas demand for 2030 and the change compared to 2014 are summarized in Table 1. Several infrastructure upgrades are assumed in accordance with ENTSOG’s ten-year network development plan [2]. Of all gas infrastructure projects listed in [2], only those projects with a successful final investment decision and projects in advanced planning stages are considered to be completed by 2030. Due to legal, financial and political uncertainties, all other infrastructure upgrades reported in [2] are not considered to be completed by 2030 in this work. The upgraded pipelines for the 2030 scenarios are shown in Fig. 6. In accordance with ENTSOG’s predictions [2], the annual domestic gas production is reduced for Great Britain (−60% compared to 2014) and the Netherlands (−50%) for all 2030 scenarios. In combination with the changes in gas demand that are shown in Table 1, this reduction in the domestic productions of the UK and the Netherlands results in an annual Europe-wide gas shortage of 50 bcm. To close this import gap in 2030, four scenarios, shown in Table 2, are simulated in this work:
Fig. 5. Voronoi tessellation of gas system nodes and allocation of population and industrial areas for northern Germany. 821
Applied Energy 238 (2019) 816–830
P. Eser et al.
Fig. 6. Upgraded pipelines for 2030 simulations. Table 2 Changes relative to 2014 in domestic gas production and available gas imports in the simulated scenarios.
NL domestic production GB domestic production Available gas imports via LNG Available gas imports via NS2 All other import sources
2030-REF
2030-LNG
2030-NS2
2030-COMP
−50% −60% +25% +25% +25%
−50% −60% +170% 0% 0%
−50% −60% 0% +200% 0%
−50% −60% +170% +200% 0%
Fig. 8. Comparison of predicted and measured relative average gas prices for the year 2014.
1. In the reference case for 2030 (termed 2030-REF), no active gas sourcing policy is assumed to be implemented. Hence, the supply gap is closed by uniformly increasing the maximum available imports from all sources by 25% compared to 2014. This includes LNG as well as pipeline imports from Algeria, Libya, Morocco, Norway and Russia to Europe. 2. The second scenario (2030-LNG) assumes that the import gap is completely covered by additional LNG imports to Italy, France and Spain. To accomplish this, the maximum available imports at LNG terminals are increased by 170% compared to 2014. 3. The third scenario (2030-NS2) assumes the import gap to be fully covered by Nord Stream and the newly built NS2 pipeline. This requires not only a full utilization of the NS2 pipeline, but also an increased utilization of the existing Nord Stream pipeline, which was only used at 65% of its maximum capacity in 2014 [45]. Therefore, the available gas imports via Nord Stream and NS2 to Germany are assumed to increase by 200% compared to 2014. 4. The fourth scenario (2030-COMP) increases imports of for both LNG and NS2 in order to investigate the competition between the two gas sourcing strategies. Combining the cases 2030-LNG and 2030-NS2, the available gas via LNG is increased by 170% and simultaneously, the available import via NS2 is increased by 200%. This means, that gas is available excessively, which enables the model to find the most cost-competitive combination of LNG and pipeline gas.
Fig. 7. Comparison of predicted and measured annual crossborder gas flows for the year 2014.
To further investigate the interchangeability of LNG and NS2 as sources of gas import for Europe in 2030, a LNG price sensitivity study 822
Applied Energy 238 (2019) 816–830
P. Eser et al.
Fig. 9. Comparison of annual imports of natural gas in 2014 and in the reference scenario of 2030.
is conducted for the 2030-COMP scenario. In the 2030-COMP scenario, the average LNG price is the same as that for gas imported through NS2. A low-price scenario (2030-COMP−), and a high-price scenario (2030COMP+), in which LNG is respectively 20% cheaper and more expensive on average than gas imports via NS2, is simulated additionally in this work.
Fig. 10. Comparison of intra-European gas trades for 2014 and for reference scenario in 2030.
2.6. Disruption of supply test cases for 2030 In the scenarios described above, questions of overall sourcing of gas and impacts on the inner-European gas trade are assessed. In addition, questions of short-term security of supply are also of importance in Europe’s energy strategy. To investigate the vulnerability of the gas system in the two diametrically different strategies that are described in scenarios 2030-LNG and 2030-NS2, an additional case study with a short-term disruption of gas imports is simulated for both scenarios. During the four-week period with highest gas demand, weeks 3–6 of the reference year 2014, a disruption of gas imports is imposed on the profiles of gas imports determined in the annual gas sourcing model. For the disrupted 2030-LNG scenario, all LNG imports are disrupted in the four-week period, whereas for the disrupted 2030-NS2 scenario, all imports from Russia via the Ukraine are disrupted. The profiles with disrupted imports are used as boundary conditions in the hourly gas system operation model, and thus alternative strategies of system operation to cover the supply shortage are determined by the model. 3. Results
Fig. 11. Pipelines with a reversed flow direction in 2030 reference scenario compared to 2014.
3.1. Model validation for 2014 annual gas trades amongst the simulated countries are derived from [28]. In Fig. 7, the predicted annual crossborder flows are compared to the measurement data. It is evident, that the simulation correctly identifies the main flow directions for 16 out of 18 simulated
In order to validate our novel simulation approach, which is comprised of a simulation of the annual gas sourcing followed by a simulation of the hourly gas system operation, the simulation of the reference year 2014 is compared to measurement data. The measured 823
Applied Energy 238 (2019) 816–830
P. Eser et al.
Fig. 14. Annual gas transits through Ukraine for the 2030-REF, 2030-NS2 and 2030-LNG scenarios compared to 2014.
crossborder pipeline connections in central and western Europe. Due to the absence of pipeline directionality constraints in the model, the connection between the Czech Republic and Poland is incorrectly predicted to transport gas from Poland to the Czech Republic. This reduces the Czech gas imports needed via the Ukraine, which frees more Ukrainian transit flow to be transported to Austria. In the measurement data, the same amount of gas is transported from Poland via Germany to Austria, which means that the overall flow balance between Germany, Poland, Ukraine and Austria is predicted correctly, despite the mismatch in gas flows seen in Fig. 7. It should further be noted, that the predicted flow direction at the Polish-Czech interconnection correctly reflects the long-term target of the interconnection: That is to transport gas from Poland to the Czech Republic and further on into central Europe. Therefore, the predicted gas flow at this interconnection correctly reflects the predominant gas flow between Poland and the Czech Republic in the future scenarios. The mass-averaged difference between the predicted and actual annual crossborder gas flows is 22%. The predicted and actual average nodal gas prices are shown in Fig. 8. The gas prices are normalized relative to the European average. The actual gas prices are derived from [31]. It can be seen that the trends in gas prices across Europe are well predicted, with the gas prices increasing from northern to southern Europe. The gas price in France is overestimated by 15%, which causes the downstream gas prices in Switzerland and Italy to be similarly overestimated. The average difference between the predicted and actual nodal gas prices is 8%. The validation of gas crossborder flows and gas prices for 2014 should be evaluated especially with consideration, that the present model does not require any pipeline directionality constraints or other non-physical calibrations. This is crucial, because only a non-constrained model without non-physical calibration can correctly assess the operation of the European gas system in 2030, even if the market environment is substantially different than today’s. In light of this argument, the good agreement between the predicted and actual physical gas flows and gas prices validates the suitability of our novel methodology to predict physical and financial gas trades within Europe in scenarios of future years.
Fig. 12. Annually averaged gas price for 2014 and 2030 reference scenario.
3.2. The European gas system in 2030 In the reference scenario for 2030, the reduced domestic gas productions of the Netherlands and Great Britain are compensated with additional imports from Norway, Russia and LNG, as shown in Fig. 9. In the absence of active policy measures that are either construction of Nord Stream 2 or increased LNG imports, Norway profits most from the reduced European gas production. Of the additional gas imports to Europe in 2030, Norway attains a market share of 48%, whereas Russia has 32% and LNG only accounts for 20%. It is thus evident, that both Russia and LNG suppliers would need to lobby for active policy measures in order to increase their market share in Europe 2030.
Fig. 13. Comparison of annual imports of natural gas for 2030-LNG, 2030-NS2 and 2030-REF scenarios.
crossborder connections. This means, that the novel simulation approach correctly predicts the overall large-scale flow of natural gas throughout Europe, even when considering the complexity of 824
Applied Energy 238 (2019) 816–830
P. Eser et al.
Fig. 15. Average gas flows for 2030-NS2 (15a, left) and 2030-LNG (15b, right) scenarios.
still predominant. Notwithstanding, the flow directions around the Benelux countries are changed in 2030-REF compared to 2014, because both Great Britain and the Netherlands transition from being net exporters to being net importers of natural gas due to their reduced domestic productions. This change in general flow direction is also shown in Fig. 11, which highlights the pipelines that have a reversed flow direction in the 2030 reference scenario compared to 2014. While most pipelines across Europe continue to operate with flows in the same direction as today, the pipelines connecting the Benelux states to France, Germany and Great Britain would need to be technically upgraded to allow for reversed flow directions. Such upgrades would require additional investments in the adjacent compressor stations. The additional flow reversals in France, Poland, Germany and Italy can mostly be attributed to be reactions to the pipeline upgrades that are assumed for 2030 (as seen in Fig. 6). Due to the moderate changes in gas imports and intra-European gas trades, the overall average gas prices change by no more than 6% between 2014 and 2030-REF, as shown in Fig. 12. Overall, the 2030 reference scenario shows no significant change in gas trading or pricing compared to 2014. This is due to the fact, that the shortfall in gas supply due to the reduced gas production in the Netherlands and Great Britain is assumed to be equally covered from all available sources of imports, hence there is no substantial shift in the overall trends of gas trade within Europe. 3.3. Nord Stream 2 and LNG as gas suppliers of Europe in 2030 Fig. 13 compares the scenarios when the shortfall in 2030 gas production is entirely covered either by LNG imports (2030-LNG) or gas pipeline imports through Nord Stream 2 (2030-NS2); shown also in Fig. 13 is the 2030 reference scenario (2030-REF). With Nord Stream 2 built, the imports of gas from Russia to Germany are increased by 49 bcm/year compared to the 2030-REF scenario at the expense of reduced gas transits through Poland and Ukraine of 18 and 16 bcm/year respectively. Furthermore, the increased penetration of Russian gas into central Europe pushes Norwegian gas westwards: While less Norwegian gas is imported into Belgium, Germany and the Netherlands, more Norwegian gas is imported into Great Britain instead. Hence Norway does not lose market share if Nord Stream 2 is built, but the diversity of its buyers is reduced. If, on the other hand, more LNG is imported in southern Europe (2030-LNG), then Norway and Russia lose respectively 30% and 20% of their market shares. Especially in Great Britain, the
Fig. 16. Intra-European gas crossborder flows for 2030-LNG and 2030-NS2 scenarios, compared to 2030-REF scenario.
Since the main sources of gas imports are not changed significantly between 2014 and the 2030-REF scenario (Fig. 9), the crossborder gas trade within Europe is also not changed significantly, as shown in Fig. 10. The main flow directions of north-to-south and east-to-west are 825
Applied Energy 238 (2019) 816–830
P. Eser et al.
Fig. 17. Reversed pipelines for 2030-NS2 (17a, left) and 2030-LNG (17b, right) scenarios compared to 2014.
Fig. 18. Average gas prices in 2030-NS2 and 2030-LNG scenarios, compared to 2030-REF scenario.
Fig. 19. Annual imports of gas for 2030-COMP scenarios with different LNG prices.
increased LNG imports replace pipeline gas imports from Norway. The Europe-wide market share of LNG is thus increased to 34% for the 2030-LNG scenario, up from 16% for 2030-REF. One of the main topics of the discussion regarding the political acceptability of Nord Stream 2 is the potential loss in gas transit fees for Ukraine, which is analyzed in Fig. 14. In Fig. 14, only the gas transits entering into Western Europe through Poland, the Czech Republic and Austria are considered, while the gas that is delivered to Eastern Europe and the Balkans through Ukraine is excluded. As can be seen for the 2030-REF scenario, without Nord Stream 2 and increased LNG imports, the annual gas transits through Ukraine are increased by 22% compared
to 2014 due to the increased gas demand across Europe. Successful construction of Nord Stream 2 (2030-NS2) reduces the gas transits, but only to levels that are 13% below the 2014 transit. In financial terms, the losses in transit fees amount to 170 m€ for Ukraine compared to the 2014 transit fees of 1.3b€, which is less severe than anticipated by other works [11,18]. Interestingly, the strategy of increasing the LNG imports to Europe also has detrimental effects on the Ukrainian gas transit, with a slight reduction of 3% (40 m€) compared to 2014. The significant impacts of the 2030-NS2 and 2030-LNG strategies on intra-European gas trade are shown in Figs. 15 and 16. In the 2030-NS2 scenario (Fig. 15a), the predominant flow direction of gas from north826
Applied Energy 238 (2019) 816–830
P. Eser et al.
Fig. 21. Average domestic gas prices for 2030-COMP scenarios.
Fig. 20. Impact of LNG prices on intra-European gas trade.
east to south-west that is seen is similar to that in today’s European gas market. Compared to the 2030-REF scenario, the gas transit through Poland is reduced by 23%, which means that Poland experiences a higher loss of transit volume than Ukraine due to the construction of Nord Stream 2. In the 2030-LNG scenario, a more pronounced change of flow directions is visible across Europe (Fig. 15b). Since LNG is imported in geographical proximity to the consumers in south-west Europe, the large-scale transportation of gas from north-east to southwest is reduced, and the gas flow direction between France and Spain is reversed (Fig. 16). Overall, the intra-European crossborder trades are reduced by 33% on average in the 2030-LNG scenario compared to the 2030-REF scenario, because France, Italy and Spain are more self-sufficient in their gas supply due to increased LNG imports. The change in the large-scale gas flows across Europe has further implications for the gas pipeline infrastructure, as shown in Fig. 17. Since the established flow direction of east-to-west across Europe is partially reversed in the 2030-LNG scenario, 120 pipelines accounting for 17% of the length of the simulated European gas network need to accommodate reversed gas flows compared to 2014 (Fig. 17b). Thus higher investments in compressor stations upgrades are required compared to the 2030-NS2 scenario, where 91 pipelines accounting for 10% of the network length have reversed flow directions compared to 2014 (Fig. 17a). The sensitivity of gas prices to the diametrically opposite gas sourcing strategies in scenarios 2030-NS2 and 2030-LNG is shown in Fig. 18. It can be seen that despite the significant differences in European gas flows, the impact on average gas prices is relatively low, if LNG is similarly priced to pipeline gas, as is assumed in the 2030-LNG
Fig. 22a. Sources of replacement gas for disrupted imports in the 2030-LNG scenario.
scenario. For both 2030-NS2 and 2030-LNG scenarios, the gas prices differ by less than 5% compared to the 2030-REF scenario. In the 2030LNG scenario, countries in northern Europe see price increases of up to 3%, because gas imported as LNG in southern Europe has to be transported over longer distances, and hence incurs higher infrastructure costs than in the 2030-NS2 scenario. It is therefore evident that LNG must be very competitively priced compared to pipeline gas if there is to be successful penetration of LNG into the central European market. 3.4. Sensitivity to LNG prices The impact of LNG prices on LNG imports, when both LNG and Nord Stream 2 are available sources for gas imports, is shown in Fig. 19. The market share of LNG imports is most sensitive to LNG prices in Spain, where pipeline gas from Morocco and Algeria is available to replace 827
Applied Energy 238 (2019) 816–830
P. Eser et al.
LNG shipments. For this reason, LNG imports to Spain are reduced by 80%, if LNG is 20% more expensive than the average pipeline gas (2030-COMP+scenario). Similar, but less pronounced effects are visible in France, Italy and Great Britain, where LNG imports are reduced in the range of 30–60% if LNG is 20% more expensive than pipeline gas. This high sensitivity to LNG prices underlines the need for LNG to be priced very competitively if an increased market share in the European gas market by 2030 is to be achieved. It can also be seen in Fig. 19, that expensive LNG shipments are replaced by pipeline gas from Norway and Russia, which gain 15% and 19% market share respectively, if the LNG price is 20% above the average pipeline gas price. This shift in gas sourcing has profound impacts on the intra-European gas trades, as shown in Fig. 20. While Spain is independent of gas imports via France when LNG is priced 20% below pipeline gas (2030-COMP- scenario), Spain heavily relies on imports via France when LNG is 20% more expensive than pipeline gas. The additional gas flows via France to Spain are sourced on the one hand from Nord Stream 2 and routed via Germany to France, and on the other hand from Norway and routed via Great Britain and Belgium to France. This means, that the price difference between LNG and pipeline gas not only determines the import strategies of individual countries, but also impacts the pan-European gas flow. Due to the fluctuating nature of the global LNG market, this price difference can switch several times per year. Therefore a sourcing strategy that is more reliant on LNG must include more investments to make the European gas infrastructure flexible enough to accommodate such large-scale gas flow reversals. The impact of the LNG price on the average countrywide gas prices is shown in Fig. 21. As one would expect, the countries most sensitive to the LNG price are the countries that import LNG directly. However, since pipeline gas from other sources is available to displace expensive LNG, the fluctuations in domestic gas prices are damped compared to the LNG price fluctuations. The domestic gas prices in Spain show the largest dependence on LNG prices, with domestic prices in Spain increased by 16% when LNG prices are increased by 20%. In other countries that import LNG, the availability of pipeline gas to replace the more expensive LNG shipments further reduces the price sensitivity: For a 20% increase in LNG prices, the gas price is increased by 10% in France, and by 5% in Italy and Great Britain. This shows that amongst European countries, Spain is the most sensitive to price fluctuations in the LNG market, because Spain’s share of domestically consumed gas that is imported as LNG is the highest.
Fig. 22b. Sources of replacement gas for disrupted imports in the 2030-NS2 scenario.
Fig. 23. Relative contribution of each country's storage to close the import gap in 2030-LNG disruption scenario.
Fig. 24. Increase in local gas price due to import disruption for 2030-LNG (24a, left) and 2030-NS2 (24b, right) scenarios. 828
Applied Energy 238 (2019) 816–830
P. Eser et al.
3.5. System resilience to supply disruptions for LNG and Nord Stream 2
•
In the case of disruption of gas imports, gas supplies must be obtained from alternative sources. As described above, disruptions during the four-week period of highest gas demands were simulated for the 2030-LNG and 2030-NS2 scenarios. For the 2030-LNG scenario, 77% of the disrupted gas supplies are replaced with extractions from storage sites across Europe, as shown Fig. 22a. As can be seen in Fig. 23, the storage sites in LNG importing countries Great Britain, Spain, France and Italy account for 81% of this replacement gas. This shows, that the gas storages in the countries that directly import LNG are sufficient to accommodate a four-week disruption of all LNG imports. In the case of a four-week disruption of all gas flows through Ukraine (Fig. 22b), 40% of the gas shortfall is compensated by re-routing Russian gas through Poland and Nord Stream 2. The remainder 60% is sourced from western European storages, Norway, and temporarily increased domestic gas productions in Great Britain and the Netherlands. This is the case, because the circumvention of Ukraine causes the flows of Russian gas into southern Europe, especially into Italy, to incur significantly higher transit costs since being routed through Nord Stream 2, Germany and Switzerland. This finding shows, that despite Nord Stream 2 being available to circumvent Ukraine as gas transit hub, Russia incurs a 60% loss in market share for any gas that is re-routed to avoid Ukraine. The impact of the supply disruptions on the local gas prices is shown in Fig. 24. Since the unavailable LNG in the disruption of the 2030-LNG scenario is mostly derived from storages that have higher prices, the local gas prices in the LNG importing countries Spain, France and Italy are increased by up to 40% (Fig. 24a). For the disruption of the 2030NS2 scenario however, the disruption does not impact gas prices substantially beyond Poland, the Czech Republic and Austria, in which prices are increased by up to 20% (Fig. 24b). This is, because 40% of the unavailable gas is replaced by comparably priced Russian gas that is imported via Nord Stream 2 and Poland. Due to the existence of Nord Stream 2, the price impact is geographically less dispersed than in [46], where the impact of the 2009 Ukraine gas disruption on the European gas market has been investigated in an ex-post analysis. Overall, the scenarios of gas supply disruption show: Because more gas can be sourced via alternative routes in north-eastern Europe, the disruption of the Ukrainian flows has less impact on the operation of the European storages and gas prices in the 2030-NS2 scenario. If LNG supplies are disrupted in the 2030-LNG scenario, few alternative sources are available in southern Europe, for which reason stored gas must be used to close the supply gap at increased cost.
•
•
•
The decision whether to rely more on LNG imports or Russian gas via Nord Stream 2 in 2030 not only impacts the countries that import the gas, but also determines the overall flow direction of gas through central and western Europe. While Nord Stream 2 negatively impacts Ukraine’s and Poland’s roles as gas transit countries, it offers a resilient low-cost option to fulfill Europe’s growing demand for gas imports. Relying more on LNG diversifies Europe’s gas supplies, but with several detrimental effects: The gas infrastructure needs to be more flexible to handle more than one predominant flow direction, and the sensitivity of gas flows through Europe to LNG prices is comparably high. Hence while building Nord Stream 2 entails more political risk due to the increased dependence on one gas supplier, importing more LNG means more technical and financial risks in the European gas system. For all scenarios investigated in this work, the availability of pipeline gas from North Africa was assumed to be identical to today’s gas importation levels. Also consistently across all scenarios, the demand of natural gas across Europe was assumed according to ENTSOG’s “slow progression” scenario. Since the future development of gas supply and demand is one of the fundamental drivers for the results shown in this study, further sensitivity studies regarding these assumptions need to be conducted. For example in future work, the present European gas system model shall be coupled with a model of the global LNG trade to assess the impact of ongoing trends in global LNG supply and demand on the European gas supply in 2030.
4. Conclusion In this work, our bottom-up energy system simulation framework EnerPol was used for the European gas system in 2030 to assess the impacts of increased LNG imports versus increased Russia pipeline gas through the construction of Nord Stream 2. A novel approach to simulate the European gas network with all individual components of the gas system was developed. The findings are as follows:
• Due
•
impacted more severely by Nord Stream 2, losing 23% of its gas transit compared to 2014. A major shift in flow directions across Europe is found if more LNG is imported. This causes 17% of the pipelines to be operated with flows in the opposite direction compared to today’s operation, which requires substantial investments into compressor stations across Europe. LNG imports have a large sensitivity to LNG prices, with LNG losing 50% market share if priced 20% more expensively than pipeline gas. This shows, that LNG must be priced very competitively to push Norwegian and Russian pipeline gas out of the European market. Since LNG only penetrates the European market if priced comparably to pipeline gas, the impact of the importing strategy on the final gas price is within 5%. This interchangeability of LNG and pipeline gas shows, that the European pipeline network is generally sufficient to enable both LNG and Nord Stream 2 import strategies, assuming all necessary compressor stations are upgraded to be bidirectional. While Nord Stream 2 increases Russia’s capability to bypass the Ukraine, stopping all gas flows through the Ukraine nevertheless has a detrimental impact on Russia’s market share. Only 40% of disrupted Ukrainian gas flows can be re-routed through Poland and Nord Stream 2, the remaining 60% are delivered from other sources. Due to the abundance of other gas sources in northern Europe, a short-term disruption of Ukrainian gas flows has only minor operational and financial impacts on the European gas system. A short-term disruption of LNG imports has significant impacts on the gas prices in south-western Europe, with prices increasing by up to 40%. This is because no alternative opportunities for gas imports are locally available, and hence 77% of disrupted imports must be extracted from local storages.
to the reduced gas production in Great Britain and the Netherlands, 12% of the European gas demand in 2030 needs to be sourced from alternative suppliers. If no active energy strategy is pursued until 2030, Norway will capture 48% of this new market segment, while Russia and LNG only capture 32% and 20% respectively. This shows, that both Russia and the global LNG suppliers need to lobby active policies and infrastructure measures to maintain their market shares in the European gas market. If Nord Stream 2 is built, it will re-route Russian gas directly to Germany, instead of transiting it through Poland or the Ukraine. For the Ukraine, the re-routing amounts to a moderate reduction in transit volume of 13% compared to 2014 transit volumes, which translates to a 170 m€ loss in transit fees. Poland however is
Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.apenergy.2019.01.068. References [1] European Commission, 2030 Energy Strategy, [online, accessed 20.04.18], < https://ec.europa.eu/energy/en/topics/energy-strategy-and-energy-union/ 2030-energy-strategy > .
829
Applied Energy 238 (2019) 816–830
P. Eser et al.
[25] Bouwmeester M, Oosterhaven J. Economic impacts of natural gas flow disruptions between Russia and the EU. Energy Policy 2017;106:288–97. [26] Eser P, Singh A, Chokani N, Abhari R. Effect of increased renewables generation on operation of thermal power plants. Appl Energy 2016;164:723–32. [27] Gas Infrastructure Europe, Maps & Data. [online, accessed 21.04.18] < https:// www.gie.eu/index.php/maps-data > . [28] ENTSOG, Transparency Platform. [online, accessed 21.04.18] < https:// transparency.entsog.eu/ > . [29] Gas Storage Europe, Aggregated Gas Storage Inventory. [online, accessed 21.04. 18], < https://agsi.gie.eu/#/ > . [30] Hart W, Laird C, Watson J, Woodruff D. Pyomo – optimization modeling in python. Springer optimization and its applications, vol. 67; 2012. [31] PEGAS gas trading platform. [online, accessed 22.04.18] < https://www. powernext.com/spot-market-data > . [32] European Commission. Third Energy Package, [online, accessed 06.10.18], < https://ec.europa.eu/energy/en/topics/markets-and-consumers/marketlegislation > . [33] Bernstein M, Griffin J. Regional differences in the price-elasticity of demand for energy. NREL/SR-620-39512; 2006. [34] Shashi Menon E. Pipeline planning and construction field manual. Gulf: Professional Publishing; 2011. [35] SNAM RETE Gas. Gas transmission tariffs, year 2017. [online, accessed 22.04. 18] < http://www.snam.it/export/sites/snam-rp/repository-srg/file/ENG/ Thermal_Year_20162017/Gas_transmission_tariffs/Tariffe_di_Trasporto_Anno_2017_ Inglese_1.pdf > . [36] Fluxys TENP. Tariffs of Fluxys TENP GmbH – January 2017. [online, accessed 22. 04.18] < http://www.fluxys.com/tenp/en/services/tarrifs/~/media/FluxysTENP/ Files/Tarrifs/2017/161222_Preisblatt_Fluxys_TENP_2017_en.ashx > . [37] GRT Gaz. Gas transmission tariffs 2017. [online, accessed 22.04.18] < http://www. grtgaz.com/fileadmin/clients/fournisseurs/documents/en/2017-Transmissiontariff.pdf > . [38] Enerdata. Global energy statistical yearbook 2017. [online, accessed 21.04. 18] < https://yearbook.enerdata.net/natural-gas/gas-consumption-data.html > . [39] Eurogas. Statistical report 2014. [40] United Nations. World population prospects, the 2012 revision; 2013. [41] OpenStreetMap Data Extracts. [online, accessed 21.04.18] < https://download. geofabrik.de/ > . [42] Eser P, Chokani N, Abhari R. Optimal RES portfolio to achieve 45% renewables electricity in central Europe by 2003. In: Power & energy society general meeting, Chicago; 2017. [43] Skamarock W, et al. A description of the advanced research WRF version 3. NCAR technical note; 2008. [44] Singh A, Eser P, Chokani N, Abhari R. Improved modelling of demand and generation in high resolution simulations of interconnected power systems. In: 12th international conference on the European energy market, Lisbon; 2015. [45] Nord Stream. Another record year for Nord Stream – 39 bcm of natural gas delivered to the European Union. Press release; 2016. [46] Lochner S. Modeling the European natural gas market during the 2009 RussianUkrainian gas conflict: ex-post simulation and analysis. J Nat Gas Sci Eng 2011;3:341–8.
[2] ENTSOG. Ten-year network development plan 2017. [3] Cooper J, Stamford L, Azapagic A. Economic viability of UK shale gas and potential impacts on the energy market up to 2030. Appl Energy 2018;215:577–90. [4] Wood D. A review and outlook for the global LNG trade. J Nat Gas Sci Eng 2012;9:16–27. [5] Lang K, Westphal K. Nord Stream 2 – A political and economic contextualization. SWP Research Paper; 2017. [6] Munteanu D, Sarno C. South Stream and Nord Stream 2 – Implications for the European energy security. Analise Europeia 2016;2:60–96. [7] Riley A. Nord Stream 2: A legal and policy analysis. CEPS Special Report, no. 151; 2016. [8] Dorigoni S, Graziano C, Pontoni F. Can LNG increase competitiveness in the natural gas market? Energy Policy 2010;38:7653–64. [9] Egging R, Holz F, Gabriel S. The World Gas Model – A multi-period mixed complementarity model for the global natural gas market. Energy 2010;35:4016–29. [10] Neumann A, Viehrig N, Weigt H. InTraGas – A stylized model of the European natural gas network. MPRA paper no. 65652; 2015. [11] Kotek P, Selei A, Toth B. The impact of the construction of the Nord Stream 2 gas pipeline on gas prices and competition. Regional Center for Energy Policy Research; 2016. [12] Orlov A. Effects of higher domestic gas prices in Russia on the European gas market: a game theoretical Hotelling model. Appl Energy 2016;164:188–99. [13] Holz F, Richter P, Egging R. The role of natural gas in a low-carbon Europe: Infrastructure and supply security. Energy J 2016;37:33–59. [14] Abrell J, Chavaz L, Weigt H. Pathways for the European natural gas market. In: 13th international conference of the European energy market, Porto; 2016. [15] Lochner S, Bothe D. The development of natural gas supply costs to Europe, the United States and Japan in a globalizing gas market – Model-based analysis until 2030. Energy Policy 2009;37:1518–28. [16] Lochner S, Dieckhöner C, Lindenberger D. Model-base analysis of infrastructure projects and market integration in Europe with special focus on security of supply scenarios. EWI; 2010. [17] Baltensperger T, Füchslin R, Krütli P, Lygeros J. European Union gas market development. Energy Econ 2017;66:466–79. [18] Richter P, Holz F. All quiet on the eastern front? Disruption scenarios of Russian natural gas supply to Europe. Energy Policy 2015;80:177–89. [19] Monforti F, Szikszai A. A Monte Carlo approach for assessing the adequacy of the European gas transmission system under supply crisis conditions. Energy Policy 2010;38:2486–98. [20] Egging R, Gabriel S, Holz F, Zhuang J. A complementarity model for the European natural gas market. Energy Policy 2008;36:2385–414. [21] Lochner S, Bothe D. From Russia with Gas – An analysis of the Nord Stream pipeline’s impact on the European gas transmission with the Tiger-model. EWI Working Paper, No 07.02; 2007. [22] Holz F, von Hirschhausen C, Kemfert C. A strategic model of European gas supply (GASMOD). Energy Econ 2008;30:766–88. [23] Deane J, Ciarain M, Gallachoir B. An integrated gas and electricity model of the EU energy system to examine supply interruptions. Appl Energy 2017;193:479–90. [24] Dieckhöner C, Lochner S, Lindenberger D. European natural gas infrastructure: the impact of market developments on gas flows and physical market integration. Appl Energy 2013;102:994–1003.
830