ARTICLE IN PRESS Energy Policy 38 (2010) 2486–2498
Contents lists available at ScienceDirect
Energy Policy journal homepage: www.elsevier.com/locate/enpol
A MonteCarlo approach for assessing the adequacy of the European gas transmission system under supply crisis conditions F. Monforti n, A. Szikszai European Commission, JRC—Institute for Energy, Energy Security Unit, P.O. Box 2, 1755 ZG Petten, The Netherlands
a r t i c l e in f o
a b s t r a c t
Article history: Received 7 October 2009 Accepted 16 December 2009 Available online 13 January 2010
Europe’s dependency on non-EU countries’ energy supply is sharply increasing. Recently, sudden supply disruption caused by international disputes outside the EU have created serious problems for some EU countries and raised concern in many others. In these situations, it is highly desirable to have a tool to assess possible outcomes of supply disruptions. This paper presents a newly developed model, MC-GENERCIS, aimed to assess the robustness of the EU transnational gas transmission system during both normal and special operating conditions, including high-demand situations and/or a supply shortage. The model has a country-by-country resolution and examines all possible dispatching choices of national TSOs on the basis of a probabilistic MonteCarlo approach. The preliminary validation of the model through its application to the ‘‘normal’’ conditions for the winter 2008–2009 and to the recent supply disruption involving Ukrainian gas transit is also described. & 2010 Elsevier Ltd. All rights reserved.
Keywords: Security of gas supply MonteCarlo modeling European gas transmission system
1. Introduction Natural gas is projected to become more and more important in the energy mix of European countries over the next decades (PRIMES, 2008, baseline scenario), with a share of gross energy consumption in the EU27 increasing from 23% (393 417 ktoe) in 2000 to 25.7% (516 210 ktoe) in 2030. This share, however, is unevenly distributed throughout EU27 countries, ranging from no natural gas use in Cyprus and Malta or 2% of gas in the Swedish energy mix to the current share of 40% in the Netherlands, a producer country and of 38% in Italy, a main consumer country (expected to reach 41.5% in 2030). According to PRIMES projections, domestic EU27 production of natural gas is not expected to satisfy this increasing demand, but on the contrary is expected to decrease from 207 559 ktoe in the year 2000 to 84 761 ktoe in 2030, with the import share of natural gas consequently increasing from 48% to 86% over the same period. Therefore, in many European countries security of gas supply has become a key issue, which is reflected both in legislation and the scientific literature. 1.1. Recent key studies on EU security of gas supply. The security of EU gas supply has been analyzed thoroughly in the literature, taking into account the various related aspects: n
Corresponding author. Tel.: + 31 224 565141; fax: +31 224 565630. E-mail address:
[email protected] (F. Monforti).
0301-4215/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.enpol.2009.12.043
from the geopolitical and strategic point of view, studies have mainly focused on Russian influence on European energy security (see, e.g., Weisser, 2007; Heinrich, 2008; Baker Schaffer, 2008; Finon and Locatelli, 2008). Market aspects have been extensively ¨ examined (Kjarstad and Johnsson, 2007), often with the support of specifically developed market models of the gas supply chain in Europe (Lise et al., 2008; Holz et al., 2008; Egging et al.,2008), while Lochner and Bothe (2009) have shown the added value of inserting European gas market in a wider global vision of gas trade. Detailed studies are also available on single aspects of the general security of supply problem, e.g., the consumer willingness to pay for an increased gas security (Damigos et al., 2009) The suitability of some possible ‘‘solutions’’ to the problem have also been widely discussed: some authors have focused their attention on LNG technology (Dorigoni and Portatadino (2008); Barroso et al., 2008; Ball and Roberts, 2007) while other have discussed the security of gas supply in the general context of the quest for an increased energy efficiency (Barrett et al., 2008; Warnig, 2008). 1.2. Security of gas supply legislation in the EU Within the general approach of the EU Energy Security and Solidarity Plan (EC, 2008), the cornerstone of EU legislation on Security of gas supply in Europe is Directive 2004/67, establishing a common and transparent framework for Member States to set up measures for security-of-supply policies. The main instrument provided by the directive consists of the definition of a set of ‘‘security of supply standards’’ in order to
ARTICLE IN PRESS F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
protect household customers (the most vulnerable population group) in the event of: (a) a partial disruption of national gas supplies during a period to be determined by Member States taking into account national circumstances; (b) extremely cold temperatures during a nationally determined peak period; (c) periods of exceptionally high gas demand during the coldest weather periods statistically occurring every 20 years. While the directive has proven to be a solid tool for customer protection, the new challenges for the European gas market described above and the recent gas supply crisis have prompted a revision that is currently under way under the coordination of the European Commission Directorate-General for Energy and Transport (EC, DG-TREN, 2009). Following the recent crisis a stronger approach to security of supply issues is required both at a policy and societal level and Member States seem ready to accept stronger security of supply standards facing a public opinion that is more and more frightened by the risk of running out of gas in very cold periods.
1.3. The need for a systemic view. The January 2009 events have shown that crises not always arise due to a lack of gas: according to data provided by Member States, around 300 Mcm/day did not reach the EU27 borders at a moment where the daily consumption was estimated around 2000 Mcm/day. In principle, such a shortage could have been dealt with, as the capacity existed to increase gas supply from import, production, storage withdrawal and LNG. Furthermore, while the peak crisis lasted quite a limited period of time (2 weeks), available market solutions were used quite rapidly to provide extra LNG to, e.g., Greece and Turkey. Nevertheless, it has not been technically feasible, at least in the first days of crisis, to bring such a potential supply from the regions where gas was abundant to EU areas where customers were almost left without supplies. The EU gas transmission system has not provided the necessary back-up for some member states to allow them to switch to other gas providers in times of crisis and in the most affected countries, the lack of gas was balanced by demand side management (interruptible contracts, fuel switching, extra electricity imports, and so on). Again, considering these circumstances it is evident that serious supply disruptions cannot be coped with through a single country approach: around ten countries were involved, with different levels of impact and the lack of gas rapidly propagated creating a kind of domino effect, which is difficult to manage by national emergency plans only. During the crisis a lot of bilateral and multilateral consultations took place, sometimes providing good solutions in due time, sometimes providing good solutions, sometimes providing a solution that was good but not in time, and sometimes not providing a solution at all. A careful assessment of concerns about the adequacy of energy transmission networks considering the expected increase in demand and/or crisis situations requires the set-up of a solid analytical tool designed to test the sufficiency of both current infrastructures and its possible future evolution. Probably the most detailed, publicly available studies on the present and future robustness of the European gas infrastructure system have been produced by Lochner and Bothe (2007) with the TIGER model, a dispatching model with monthly resolution describing more than 1000 infrastructure elements.
2487
Nevertheless, a complete technologically rich model of the EU gas transmission system still does not exist, even if on a regional scale, the large majority of TSOs have models in support of their daily operational needs. Briefly, the main reasons for this gap in the knowledge of the EU gas system can be attributed to, among others, the high commercial sensitivity of most of the infrastructure data, in the presence of a large number of TSO (sometimes several per country) and to the lack of a strong association of European TSOs willing to afford a real coordination effort. Even if commercial data companies provide quite good databases on EU gas infrastructures, these data are often rather difficult to harmonize in order to base a model on them. In the following we suggest a different approach to infrastructure adequacy problem through a new methodology where the reliability of the EU gas system is assessed on a systemic level through the study of a large set of gas paths through the system. If applied on the continental level, such an approach needs a limited amount of aggregate data to produce significant insights.
2. The MC-GENERCIS model. 2.1. Core ideas of MC-GENERCIS. MC-GENERCIS (MonteCarlo-based Gas Energy Network for Europe, Russia, and the Commonwealth of Independent States) is a numerical tool aimed at assessing the adequacy of gas transmission networks. Given a general description of the gas transmission system and appropriate figures for production, consumption, and storage withdrawal capacities, MC-GENERCIS explores the flow configurations allowed by the system in order to assess the feasibility of safely providing gas to every country considered in the scenario. The main novelty of the model consists in the use of random numbers to model the different options available to gas transmission operators. Indeed, MC-GENERCIS does not look for the ‘‘optimal’’ configuration of the flows, but explores a large number of possible dispatching strategies offered by the system to the national TSOs in order to find a ‘‘solution’’, i.e., a set of individual TSO strategies that together are able to provide safely the required amount of gas to all the countries. More details on the generation of dispatching strategies are provided in Section 2.5. In case many different successful strategies are found MCGENERCIS simply enumerates them without putting any priority scale on them. Unlike other gas system models, MC-GENERCIS is not based on an optimization approach as exploring all possible choices available to TSOs is expected to capture the operational possibilities of the system even under less normal conditions, including also situations where emergency needs could become more stringent than other constraints, like, e.g., market laws. 2.2. System description. A gas transmission system is described in MC-GENERCIS by means of nodes and links. In the current version nodes represent countries and links describe transnational gas pipelines and LNG terminals (see Fig. 1). For every link its Maximum Technical Capacity (MTC) has to be provided, currently in Mcm/day. Fig. 1 shows an example of a theoretical gas system where arrows coming from ‘‘outside’’ the group of countries modeled represent the import pipelines and/or LNG terminals capacity. 2.3. Scenario definition MC-GENERCIS can assess the adequacy of the gas system previously defined under different scenarios. In this framework we
ARTICLE IN PRESS 2488
F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
define a scenario as a set of data associated with all the countries and with the pipelines originating ‘‘outside’’ the system (supply routes). More in detail, for every country three figures need to be provided, stating production, consumption, and Storage Withdrawal Capacity (SWC) to be assumed while for each supply route the actual flow has to be provided with the constraint of being less than or equal to the maximum technical capacity supply route considered. Unit measure should be the same for all data and currently Mcm/day is used. Fig. 2 shows an example of a possible scenario involving the system defined in Fig. 1. Such an example scenario describes a situation where country A is receiving from outside the system the maximum amount of technically feasible gas, while country B only receives 50% of the maximum possible gas, and country G is not receiving any gas at all from outside the system. The associated table shows example figures for production, consumption, and storage withdrawal capacity. In practical MC-GENERCIS applications, the definition of an appropriate scenario is crucial for obtaining meaningful results. The category of data to be used in setting a scenario depends on the question to be answered by the model. As an example, if MCGENERCIS is used to assess the adequacy of the gas transmission system in ‘‘typical’’ winter, average winter data for production and consumption have to be used, and gas flows into supply routes could be supposed to be close to 100% of MTC, even if not necessary at the maximum. On the contrary, as a further example, if a ‘‘crisis’’ situation is modeled, it is reasonable to suppose that production data should be set at its known maximum capacity
20
and pipelines not involved in supply crisis should work 100% to try to supply lacking gas. For SWC, different scenarios could be run with different values ranging from 100% to lesser amount of the known maximum flow capacity in order to assess different time slices along the year. Indeed, storages can be full (or almost full) at the beginning of the winter, when pressure is at its maximum and then maximum flow is available. On the contrary, pressure and flow could decrease substantially moving towards the end of the winter. 2.4. Model rules Once a scenario is defined, MC-GENERCIS explores the whole set of theoretically possible gas dispatching strategies in order to find how many ‘‘solutions’’ the transmission system allows, where by ‘‘solution’’ we mean a configuration of gas flows throughout the system able to provide enough gas to every country to cover its consumption. To achieve this goal, the current version of MC-GENERCIS applies the following rules in the following order to every country: Rule 1. Domestic production is used at first to cover domestic demand Rule 2. If Rule 1 does not provide enough gas, withdrawal from stocks is used (if available) to cover domestic demand. Rule 3. If Rules 1 and 2 do not provide enough gas input flows are used to cover domestic demand. Rule 4. The extra gas originating from production and import remaining after application of previous Rules 1 and 3, if any, is dispatched downstream the gas network whenever possible.
10 A
C 15
5
15
5
Of course, the rules stated here are not the only possible ones given the general structure of the model, these rules have been considered the most suitable to describe a ‘‘close to crisis’’ situation, but of course reality could be more complex. Just to cite one of the main limits, market rules are not included in this picture and market laws could complicate the supposed interoperability of the system under many different points of view. As an example, contractual constraints decrease the freely usable gas capacity while prices considerations could rule out certain ‘‘good’’ strategies in terms of security of supply. Furthermore, also nonhomogeneous regulations could represent a barrier to access certain infrastructures. It is worth noticing that also some policy choices could not be represented by the present rules: in the present version a kind of common cooperative behaviour is assumed, but this is not
G
10
8 B
D 7
9 E
6
8
F
4
Fig. 1. Example of gas transmission system as described in MC-GENERCIS. A–G boxes represent countries. Arrows represent transnational pipelines connecting countries or coming from ‘‘outside’’ the system (i.e., external supplier and/or LNG terminals). Red numbers associated with pipelines represent maximum technical capacity of each pipeline. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
20/20
10
A
C 15
5
5
G
10
4/8
B
D 6 8
7
9
E
4
F
Country
Prod
Cons
SWC
A
50
65
5
B
45
60
5
C
10
12
5
D
25
15
5
E
0
10
5
F
15
10
5
G
12
10
5
0/15
Fig. 2. Example of a scenario defined for the system shown in Fig. 1. In the right table data for production, consumption, and storage withdrawal capacity are listed. Black numbers associated to pipelines represent the amount of gas actually flowing in the pipeline from outside the A–G system.
ARTICLE IN PRESS F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
necessarily a realistic case while a more ‘‘selfish’’ behaviour could be also possible. And the reality could be even more complex, with different countries implementing different market/regulatory/policy environments. In the following sections all these possible refinements are not investigated and the discussion will focus mainly on crisis or almost-crisis situations, i.e., special ‘‘abnormal’’ situations at the border of market failure where considerations on technical feasibility of solutions can prevail on their market or strategic suitability. Nevertheless, future model versions are expected to provide the user the additional grade of freedom of customizing his favorite rules. 2.5. MonteCarlo approach Rules 1–3 state how countries are supposed to act in order to satisfy internal demand. On the contrary, Rule 4 states how surplus gas is dispatched to countries downstream in the network whenever possible. In MC-GENERCIS approach the surplus of gas available in each country is split randomly between downstream countries, if any and if pipeline workload allows it, with some countries acting as bifurcation points where the gas can follow different roads, in a totally free way. Repeating the simulations a large number of times (customizable by the user) MC-GENERCIS could span the set of all technically feasible gas flows configurations that comply with Rules 1–4 and can assess how often these flow configurations are successful in providing enough gas to every country in the system. In other words, MC-GENERCIS performs a so-called adequacy analysis with the goal of answering the question: ‘‘Given a scenario hypothesis, how many pathways for satisfying gas demand does my system configuration allow?’’1 Clearly, such an analysis does not allow to find which one, among the possible flow distribution pathways is the most likely or the more ‘‘realistic’’ one and all strategies, both successful and unsuccessful are enumerated without any weighting without introducing any kind of optimization. Nevertheless, it could be possible in principle to scan the randomly generated strategies in order to find the ‘‘best’’ one according to some pre-defined key indicators. For the moment this aspect has not been investigated in detail because of the problem of deciding what is meant by ‘‘best strategy’’ in this context. As an example, in case of lack of an important share of gas, is it not clear if one should prefer a ‘‘common damage’’ strategy where many countries lack some gas or a strategy leading to a gas shortage concentrated in a few countries, maybe the ones with the best switching capabilities. The present version of the model does not rank strategies in any way and only the nonweighted statistical indicators described in the next section are computed.
2489
defined as SðCÞ ¼ SDðCÞ=N Where SD(C) is the number of MC-GENERCIS runs in which enough gas is provided to country C. S(C) can be interpreted as a measure of the difficulty or the ease of providing gas to C country: a value of S(C) close or equal to one means that whatever is the strategy followed by TSOs the needs of C country will be fully covered. Such a situation is typical when gas is abundant and/or no bottlenecks are present in the system. On the contrary, a low value of S(C) implies that only a limited number of dispatching strategies are successful in providing gas to country C with the extreme situation of an S(C) value equal to zero meaning that whatever TSOs do there is no technical solution to the problem of providing gas to country C. S(C) value lies between 0 and 1, but it should not be interpreted as the ‘‘probability’’ of successfully providing gas to country C because in reality TSOs do not act on a random basis: providing gas is not a stochastic process for which probabilities make sense. On the contrary, S(C) should be considered as a risk indicator to be interpreted by the final user. As an example, values of S(C) around 0.5 could not be considered ‘‘risky’’, because they mean the TSOs can follow 5000 different strategies to comply with their duty and even if the MonteCarlo analysis found 5000 unsuccessful strategies, TSOs are not expected to follow them in reality. On the contrary, values of S(C) closer and closer to 0 should be considered more and more risky because these mean that the freedom of TSOs is reducing and more coordination between them is needed for success. Given such a more constrained situation, the risk of not providing enough gas becomes higher and higher with the limit of S(C) =0 for which supplying gas to C country becomes technically not feasible. In such a context the interpretation of values of S(C) between the limit values of S(C)= 1 (total security) and S(C)=0 (unfeasibility) depends mainly on the risk vocation or aversion of the user. In Section 4.5.4, as an example, an arbitrary value of S(C) =0.2 will be considered as the border between security and insecurity of gas supply under crisis situations. 3.2. Gas supply margin MC-GENERCIS can also compute a set of values MðC; RÞ ¼ ðSupplyðC; RÞDemandðCÞÞ=DemandðCÞ where Supply(C,R) is the amount of gas that reaches country C in the Rth MonteCarlo run (with R ranging from 1 to 10,000 in the present version) and Demand(C) is the demand for country C in the scenario considered. More in detail, the supply term side is defined as: SupplyðC; RÞ ¼ ProdðCÞ þ WithdrawnðC; RÞ þ InflowðC; RÞ
3. Main model outputs. 3.1. Share of ‘‘successful’’ strategies Probably the most interesting result that can be obtained with MC-GENERCIS is the share S(C) of successful dispatching strategies for each country C. Given a fixed number of MC-GENERCIS runs (currently fixed to 10,000 but fully customizable), S(C) is 1 Strictly speaking, the method described is not a classical MonteCarlo, because no probability distributions or functions of merit are supposed for the randomly generated quantities in order to avoid assumptions difficult to be justified in front of the limited data availability. Here the term ‘‘MonteCarlo’’ is used with the broader meaning of ‘‘based on the generation of random numbers’’.
where Prod(C) is the domestic production of gas in C, Withdrawn(C,R) is the amount of gas withdrawn from storages in country C in the Rth simulation, and Inflow(C,R) is the gas provided to country C through incoming pipelines in the Rth simulation. For a ‘‘successful’’ supply strategy one has M(C,R) Z0 while whenever it is impossible to fully satisfy gas demand one has 1rM(C,R)o0. When smaller than zero, the M(C,R) values represent the distribution of ‘‘gaps’’, i.e., the amount of needed gas not reaching the country providing an estimation of how serious the gas shortage expected could be. On the contrary, a value of M(C,R) larger than zero means that enough gas can reach the country. In this case, the absolute value of M(C,R) has to be considered a measure of the robustness of gas supply in the Rth
ARTICLE IN PRESS 2490
F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
simulation: the higher its value, the more secure is the gas supply to country C. In the following paragraphs the average value of M(C,R) for R= 1, y, 10,000 will be shown and referred as AM(C).
3.3. Pipelines use Another result MC-GENERCIS is able to provide is the amount of capacity actually employed (or workload) for each pipeline involved in the simulation: WðP; RÞ ¼ FlowðP; RÞ=CapacityðPÞ where Flow(P,R) is the gas flow (in Mcm/day) through the Pth pipeline in the Rth MonteCarlo run and Capacity(P) is the maximum technical capacity of the Pth pipeline involved in the simulation. Such a value could be important to identify areas of the network possibly subject to bottlenecks even if, again, the complete reality is more complex because physical congestion can differ from contractual congestion with situations where even in case of spare capacity the network could in reality be unavailable to transfer further gas.
4. Test case: gas crisis originating in Ukraine under different storage fillings MC-GENERCIS has been tested against the gas crisis caused by lacking of gas reaching Europe through the Ukrainian pipelines. Moreover, the model was not only used to reproduce the situation Europe had to face in January 2009, but has been exploited to assess the impact of a crisis under different possible conditions (e.g., different demand patterns, storages filling, remaining pressure in Ukrainian pipelines, and so on) Detailed assumptions and results are described in the following sections.
4.1. The EU-27 extended gas transmission system. Table 1 shows the countries considered in the definition of the system. All EU-27 countries were included in the system analysis with the exception of Cyprus and Malta that are not connected with the rest of Europe through pipelines. Moreover, other countries were included in the definition of the system namely Turkey, Switzerland, and Balkan countries with the exception of Albania. In order to simplify the analysis, in some cases couples of countries known to be strictly linked to each other were considered as a single country. For this reason Sweden has been unified with Denmark and Luxembourg with Belgium. Pipelines connecting these countries were seen as ‘‘internal’’ pipelines and not considered in the analysis. Table 2 shows origin and destinations of the pipelines contained in the system definition. Pipelines originating outside the system are highlighted. Table 3 shows the countries for which the presence of LNG terminals has been considered and their total maximum send-out capacity (EC, 2009) Actual data for pipelines technical capacity have been obtained from both public and commercial information sources and gathered to the model through the Energy Market Observation System developed and maintained by Directorate General for Energy and Transport of European Commission (EC DG TREN, 2008).
Table 1 EU (left) and non-EU (right) countries included in the EU gas system analysis. EU countries
Non-EU countries
Austria Belgium Bulgaria Czech Republic Germany Denmark Estonia Spain Finland France Greece Hungary Ireland Italy Lithuania Latvia Netherlands Poland Portugal Romania Slovenia Slovakia UK
Bosnia and Herzegovina Switzerland Croatia FYROM Serbia Turkey
4.2. Setting the scenarios: demand and production MC-GENERCIS is a flexible model that in principle could be applied to very different scenarios in order to assess the reliability of the system given a certain situation of demand, production capacity, and stocks availability. For production, maximum values have been considered (Table 4, second column, Source: EC, DG TREN, 2009a) while two demand scenarios have been elaborated: Average winter (AW): data for a ‘‘typical’’ winter gas demand have been calculated on the basis of the share of heating period consumption extracted from EuroStat database and yearly consumption coming from (besides EuroStat) other public access. Data are shown in Table 4, third column. (For countries not having heating period data the arithmetic average of those having it has been applied.) 1-in-20 winter (20W): demand data estimated for the socalled ‘‘1-in-20 winter’’ have also been used (see Section 1.2). These data are based on the official estimates produced by EU Member States (EC, DG TREN, 2009a) and are shown in Table 4, fourth column. Both AW and 20W demand scenarios have been investigated with the main goal of providing a validation of the model: AW is a typical ‘‘business as usual’’ scenario where no major problems are expected, while in 20W countries are requested to deal with high but not totally unexpected demand. It is also worth noticing as the 20W scenario is probably tougher than a real ‘‘1-in-20’’ winter scenario because high-demand levels will take place in the same time all over the modeled countries, while in the reality such a huge ‘‘cold wave’’ is unlikely to actually happen.
4.3. Setting the scenarios: storages withdrawal capacity Maximum known storage withdrawal capacity data employed in the simulations are shown in Table 4, fifth column (Source: EC, DG TREN, 2009a). Values therein are the maximum possible flow of gas available from storages and they represent a situation where storages are completely filled. On the contrary, it is well known that pressure and then available gas flows decrease with decrease in working storage gas. In MC-GENERCIS such an effect is
ARTICLE IN PRESS F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
2491
Table 2 Internal (left) and external (right) pipelines included in the EU gas system analysis. Internal pipelines
Pipelines originating from outside the system
Origin
Destination
Name
Origin
Destination
Name
AT AT
DE DE
Oberkappel
BY BY
LT LT
Kotlovka –
AT AT AT BE BE BE BE BE BE BE BG BG BG CH CZ DE DE DE DE DE DE DE DE DK DK ES ES ES FR FR GR HU IT IT IT IT
HU IT SL DE DE FR FR NL NL UK FY GR TR IT DE AT BE BE CH CZ FR NL PL DE NL FR PT PT CH ES BG SR AT CH CR SI
¨ Burghausen/Uberackern Mosonmagyarovar Tarvisio Murfeld/Cerˇsak Eynatten Eynatten Blaregnies (H)/Taisnie res (H) Blaregnies (L)/Taisnie res (L) Zelzate (Zebra) Zelzate (GTS) Zeebrugge IZT Zidilovo Sidirokastron (frmr. Kula) – Griespass Waidhaus Kiefersfelden Eynatten Eynatten – Hora Svate´ Kateriny (CZ)–Deutsch Neudorf Medelsheim/Obergailbach Bunde–Oude Statenzijl Lasow Ellund – Biriatou Badajoz/Campo Major Valenc- a do Minho/Tuy – Larrau – Kiskundorozsma Tarvisio – – ˇ Sempeter/Gorizia
BY BY DZ LY MA NO NO NO NO NO NO NO RU RU RU RU UA UA UA UA UA UA UA UA
PL PL ES IT ES BE DE DE FR UK UK UK EE FI LV TR HU PL RO RO SK SK SK SK
Wysokoje Tieterowka – Gela Tarifa Zeebrugge ZPT Emden (NPT) Dornum/NETRA Dunkerque Easington Tampen Link St. Fergus (Vesterled) Varska Imatra – South Stream(?) Beregdaroc Drozdowicze Isaccea Mediesu Aurit Velke Kapusany Velke Kapusany Velke Kapusany Velke Kapusany
LV LV NL NL NL NL NL NL NL NL PL PT PT RO SI SI
EE LT BE BE BE BE DE DE DE UK DE ES ES BG CR IT
Karksi Kiemenai Zandvliet Hilvarenbeek/Poppel Obbicht/Dilsen s Gravenvoeren Bocholtz Zevenaar Winterswijk Julianadorp (H-gas) Mallnow Badajoz/Campo Major Valenc- a do Minho/Tuy Negru Voda Rogatec ˇ Sempeter/Gorizia
SK SK SR TN TR UK UK
AT CZ BH IT GR BE IE
Baumgarten Lanzhot Zvornik Mazara del Vallo ITG Zeebrugge IZT Moffat
not modeled endogenously (see also the following discussion on model limitations) and the previously described AW and 20W demand scenarios were studied in the case of different values of SWC, ranging from 100% of maximum withdrawal capacity to 0% in decreasing steps of 25%. Such a sequence of simulations is intended to broadly represent different times during a typical winter season, with
storage reserves progressively decreasing from full (or almost full) at the beginning of the winter to empty (or almost empty) at the end of the cold season. It has also to be mentioned that in the studied scenarios SWC decreases of the same percentage in all countries: of course this is a rough estimation because in reality every country is expected to manage storages in a different way, responding to different needs and in dependence of different
ARTICLE IN PRESS 2492
F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
Table 3 LNG maximum send-out capacity in EU (Source EC, DG TREN, 2009a). Country
LNG send-out capacity (mcm/day)
Belgium Spain France Greece Italy Portugal UK
24.7 160.9 42.44 13.69 10 14.2 84.56
strategies. Nevertheless, finding reliable data on storage use is quite difficult, and the assumption of equal percentage SWC decreasing was taken as a first approximation for analyzing the EU gas system. 4.4. Other hypotheses To complete the definition of the test cases investigated the following hypotheses were also added:
All pipelines originating outside the studied area are used 100% of their capacity
LNG terminals are also employed 100% of their capacity It is worth noticing that these assumptions do not describe the typical daily situation of the European gas system because it is well known that both transnational pipelines and LNG terminals are rarely exploited 100% of their nominal capacity. On the contrary, these supplementary hypotheses have been specifically stated to represent possible countermeasures to be taken in a crisis situation. In summary, the cases studied represent an ‘‘upper limit’’ of the system reliability in dealing with a gas crisis: model results should be considered an answer to the following question: ‘‘If some gas is lacking and Europe exploits at the maximum level its current alternative supply routes, to what extent the current transmission system could be able to provide the needed gas to every country?’’. Results shown in the next paragraph are intended to provide an answer to this question. 4.5. Results 4.5.1. Baseline ‘‘no crisis’’ scenarios At first, simulations describing a baseline scenario where gas flows freely from supplier to all countries without any restriction were run. In such a situation one should expect S(C), the share of successful gas strategies described in Section 3.1 to have a high value irrespective of the storage filling situations.2 Indeed Table 5 shows values for S(C) (in percentage) and AM(C) for all countries with available storage withdrawal capacity decreasing form 100% to 0%. Almost all countries show values of S(C) equal to 100%, except for FYROM that shows a more than safe value of S(C) equal to 95.7%. It is also worth noticing as such a very unlikely lack of gas in FYROM found in only 4.5% of simulations it is also expected to be of almost negligible value with only the 3% of gas demand not covered. This result is in full agreement with the common experience that in ‘‘normal’’ situations every European country can satisfy its need for gas. From Table 5 one could also notice as the security of gas supply margin, numerically represented by AM(C), is decreasing when 2 Indeed, gas crisis usually do not happen even at the end of the cold season, with storages usually far from being completely filled.
the amount of gas available from storages decreases. Both AM(C) and its decreasing rate are quite different in different countries but AM(C) remains safely larger than zero for most countries. On summary, the EU gas system has been confirmed to be very robust when facing a typical wintry demand of gas, with all countries showing a satisfactory security margin for gas supply whatever the level of storages filling. Table 6 shows the same results as Table 5 in the case of 20W demand scenario, i.e., a ‘‘1 in 20’’ gas demand taking place simultaneously all over Europe. Even if the first column data confirm a safe situation when storages are totally filled3, it is evident that security margins are much smaller in this case comparing with the previous AW scenarios. Indeed, simulations performed with decreasing levels of gas available in the storages show as more and more countries move from a safe situation to a more difficult one, with S(C) decreasing more or less steeply from 100% to lower values and the average of AM(C) moving from positive (safe) to negative (gas gap expected). On summary, the EU gas system has been shown to be also robust enough when facing a ‘‘1 in 20’’ gas demand provided that storages are filled. On the contrary, some countries could experience lack of gas in case of storages partially or totally empty. Lack intensity and its likelihood is different in different countries. Also this last modeling result is in line with the real functioning of the system: EU countries are requested by current legislation to be ready to manage ‘‘1 in 20’’ gas demand level and storing gas is the usual way states protect their consumers. Table 6 confirms such a strategy is successful because no crisis is expected when storages are totally filled. Nevertheless, problems could arise with gas storages not filled for one or more of the many possible reasons (gas not available during the summer to fill up storages, lack of financial resources to buy gas, exceptionally long cold peak, a previous supply crisis forcing to overuse storages and so on). Even in the reality, European countries know very well such a risk and try to act in a way to keep their storages as filled as possible. 4.5.2. ‘‘Crisis’’ scenarios. Tables 7 and 8 show the same results as Tables 5 and 6 with the same demand scenarios (AW and 20W, respectively) if gas supply from Ukraine is completely switched off. The effect of supply disruption is immediately evident in countries closest to the origin of disruption with a group of countries showing the total impossibility of covering gas demand whatever the storage situation even in a ‘‘normal’’ winter (Table 7, second columns). Moving to a higher demand and less storage available situation further increases the number of countries affected. Even if the worst scenario could be considered very improbable (exceptional high demand, empty storages, and supply cut all together), the number of countries potentially affected is a clear demonstration of the high level of interconnectivity in the European gas transmission system. Model results under supply crisis hypothesis will be further discussed in Section 4.5.4. 4.5.3. Model comparison with January 2009 gas crisis. From the point of view of model validation it could be useful to compare the results obtained with MC-GENERCIS with the actual outcomes of the January 2009 crisis, once realistic data for at least gas demand and filling of storages are provided. Unfortunately data for expected gas demand in that days are still not available in the open literature, but it is well known that the crisis took place 3 The gas lack expected in FYROM is likely to be caused by some overestimation of demand data
ARTICLE IN PRESS F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
2493
Table 4 Production, consumption for AW and 20W and maximum SWC. All data are in Mcm/day. See text for details on data sources. Country
Production
Consumption (AW)
Consumption (20W)
Storages maximum withdrawal capacity
AT BE BG BH CH CZ DK EE ES FI FY FR DE GR CR HU IE IT LT LV NL PL PT RO SK SI SR TR UK
12.16 0 0.3 0 0 0.3 29.9 0 0 0 0 2.4 45 0 6.9 9 1 24 0 0 440 6.48 0 34.3 0.3 0 0.5 2.44 231
25.67 49.80 11.94 0.98 9.90 27.09 16.96 3.16 86.07 1.00 2.00 126.83 289.24 13.57 8.43 37.34 20.30 238.10 9.59 5.49 175.87 50.87 13.56 57.06 15.44 3.51 6.85 97.82 296.69
49.41 139.20 15.60 1.80 9.90 67.60 31.70 4.30 160.20 1.00 3.00 370.00 400.00 14.00 11.50 92.50 22.73 425.00 16.00 9.00 235.00 59.71 19.30 75.00 29.90 5.80 10.00 100.00 536.00
48 22.80 4.20 0 5.00 55.00 19.40 0 10.54 0 0 231 463.32 0 5 47.50 2.60 298.85 0 14.69 15 34.20 7 25.80 34.87 0 0 12 126.50
Table 5 Percentage of successful gas delivery strategies S(C) and average margin AM(C) under the average winter demand scenario and for different values of storages withdrawal capacity available. Country Storage withdrawal 100%
AT BE BG BH CH CR CZ DE DK EE ES FI FR FY GR HU IE IT LT LV NL PL PT RO SI SK SR TR UK
Storage withdrawal 75%
Storage withdrawal 50%
Storage withdrawal 25%
Storage withdrawal 0%
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
100 100 100 100 100 100 100 100 100 100 100 100 100 95.7 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
6.70 6.41 5.57 0.97 7.95 0.54 7.91 2.72 1.91 6.28 1.77 19.68 2.41 0.23 1.45 1.93 0.48 1.56 3.08 6.16 2.07 2.70 1.54 2.03 1.90 20.77 0.83 0.25 1.30
100 100 100 100 100 100 100 100 100 100 100 100 100 95.7 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
6.23 6.30 5.49 0.97 7.82 0.54 7.40 2.32 1.62 6.28 1.74 19.68 1.95 0.23 1.45 1.61 0.62 1.24 3.08 5.49 2.05 2.53 1.41 1.92 1.92 20.21 0.83 0.20 1.20
100 100 100 100 100 100 100 100 100 100 100 100 100 95.63 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
5.77 6.18 5.40 0.97 7.69 0.54 6.89 1.92 1.33 6.28 1.71 19.68 1.50 0.23 1.22 1.29 0.59 0.93 3.08 4.82 2.02 2.36 1.28 1.81 1.91 19.64 0.83 0.16 1.09
100 100 100 100 100 100 100 100 100 100 100 100 100 95.44 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
5.30 6.07 5.31 0.97 7.57 0.65 6.38 1.49 1.05 6.28 1.68 19.68 1.04 0.23 0.92 0.86 0.56 0.65 3.08 4.15 2.00 2.20 1.15 1.69 3.14 19.08 0.83 0.12 0.98
100 100 100 100 100 100 100 100 100 100 100 100 100 95.73 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100
4.83 5.95 5.22 0.97 7.44 0.58 5.88 1.08 0.76 5.81 1.65 19.68 0.59 0.23 0.62 0.52 0.53 0.32 2.93 3.48 1.98 2.03 1.02 1.58 2.07 18.51 0.83 0.08 0.88
ARTICLE IN PRESS 2494
F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
Table 6 Percentage of successful gas delivery strategies S(C) and average margin AM(C) under the 1-in-20 winter demand scenario and for different values of storages withdrawal capacity available. Country Storage withdrawal 100%
AT BE BG BH CH CR CZ DE DK EE ES FI FR FY GR HU IE IT LT LV NL PL PT RO SI SK SR TR UK
Storage withdrawal 75%
Storage withdrawal 50%
Storage withdrawal 25%
Storage withdrawal 0%
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
100 100 100 93.44 100 100 100 100 100 100 100 100 100 0 100 100 100 100 100 100 100 100 100 100 100 100 99.98 99.75 100
3.00 1.55 4.03 0.04 7.95 0.64 2.57 1.63 0.56 4.35 0.49 19.68 0.17 -0.16 1.31 0.18 0.66 0.43 1.44 3.37 1.27 2.15 0.78 1.31 1.51 10.24 0.24 0.09 0.26
100 100 100 0 100 100 100 100 100 100 100 100 100 0 100 100 100 100 100 100 100 100 100 100 100 100 17.46 88.55 100
2.76 1.51 3.96 0.97 6.07 0.47 2.37 1.31 0.40 4.35 0.43 19.68 0.01 -0.22 1.05 0.08 0.62 0.27 1.44 2.96 1.25 2.01 0.69 1.22 1.09 9.95 0.19 0.05 0.20
100 100 100 0 100 99.62 100 100 100 100 100 100 0 0 100 0 99.96 100 100 100 100 100 100 100 99.86 100 0 65.8 100
2.52 1.26 3.90 1.00 5.65 0.31 2.16 0.98 0.25 4.35 0.41 19.68 0.14 0.32 0.84 0.03 0.53 0.07 1.44 2.55 1.21 1.87 0.60 1.14 0.83 9.66 0.95 0.02 0.11
100 100 100 0 100 1.21 100 100 100 100 100 100 0 0 100 0 52.27 0 100 100 100 100 100 100 44.35 100 0 40.04 100
2.08 0.85 3.83 1.00 4.87 0.24 1.88 0.59 0.10 4.02 0.39 19.68 0.31 0.42 0.70 0.21 0.05 -0.16 1.36 2.15 1.09 1.70 0.51 1.05 0.11 9.37 0.95 0.02 0.04
100 100 100 0 99.99 0 100 100 0 100 100 100 0 0 100 0 0 0 100 100 100 100 100 100 34.09 100 0 16.62 0.01
1.80 0.38 3.76 1.00 3.62 0.36 1.13 0.21 0.06 3.45 0.37 19.68 0.60 0.50 0.64 0.35 0.95 -0.40 1.28 1.74 0.88 1.47 0.42 0.96 0.22 9.08 0.95 0.06 0.09
in a specially cold month of a cold winter when gas demand was expected to be close to a peak value, even if probably the demand is not to be considered as high as in a ‘‘1 in 20’’ winter. Also the storages filling at crisis time is not known in detail, but one can notice that the crisis did not take place at the very beginning of the winter, so probably storages were not 100% full. Nevertheless, it was estimated that at least at the beginning of the crisis a considerable amount of gas was still present in the storages and SWC can be assumed equal to 70–80% of the maximum SWC. In conclusion, the January crisis can be supposed to be correctly represented by a scenario somewhere in between the ones presented in Table 7, third columns, and Table 8, third columns (in italic). In reality (EC, DG TREN, 2009b), the countries most affected were Bulgaria, Romania, Serbia, FYROM, Bosnia and Erzegovina, and Turkey, i.e., all and only the countries that show a value of S(C) equal to zero AW crisis scenario. Slovakia was also affected and the model is able to forecast such a crisis in the case of 20W scenario. Hungary, Slovenia, and Croatia4 show a different grade of security in the two scenarios, but it is difficult to assess which one better represents their actual conditions at the beginning of the crisis without more detailed data. In summary, notwithstanding data uncertainties, model results show a very good match with what happened in January 2009. It is also worth noticing that MC-GENERCIS can be correctly employed to evaluate the gas supply security at the very beginning of the crisis. All the countermeasures put in place in
4 The very small gap expected in Greece (2%) for 20W demand scenario is easy to be managed with demand control measures and could also be caused by uncertainties in input data.
the following two weeks are not modeled in the current MCGENERCIS version. 4.5.4. ‘‘Turning points’’ to gas crisis. Comparing Table 5 with Table 7 or Table 6 with Table 8, it is clear that in case of lack of gas supply some countries move from a ‘‘safe’’ situation (i.e., S(C)E100% and AM(C)4 0) to a problematic situation for which S(C)= 0 and AM(C)o0. To better investigate the details of such a transition, some additional runs of MCGENERCIS were performed with same scenario inputs already described, with gas flow from Ukraine decreasing from 100% to 0% in steps of 20%. In other words some supplementary scenarios were studied to answer the question: ‘‘Given a certain demand level and storage filling situation, to what extent can a country safely manage a decrease in gas flows coming from the Ukraine?’’ As an example, Table 9 shows the value of S(C) for Slovenia under the 20W demand scenario for the full set of simulations performed. One can see how for each value of storage filling the ‘‘security’’ of Slovenia decreases with decrease in gas flow from the Ukraine. If a country for which S(C)420%,5 is arbitrary assumed as ‘‘secured’’, one could see how under the 20W demand hypothesis, total storage filling provides Slovenia full security whatever the gas flow from the Ukraine. On the contrary, a ‘‘turning point’’ between security and insecurity appears for decreasing values of storages availability: e.g., for 50% of storage flow available Slovenia becomes insecure when Ukrainian gas 5 As stated in Section 3.2, the choice of this value depends basically on the risk aversion of the model user, and in definitive, of the emphasis decision makers put on energy security issue. Furthermore, the definition of a ‘‘secured’’ country could also involve a threshold value for AM(C) in order to exclude small easy manageable gaps. (e.g., S(C) 420% and AM(C) 4 0.05).
ARTICLE IN PRESS F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
2495
Table 7 Percentage of successful gas delivery strategies S(C) and average margin AM(C) under the average winter demand scenario for different values of storages withdrawal capacity available in the hypothesis gas supply from Ukraine are completely cut. Country Storage withdrawal 100%
AT BE BG BH CH CR CZ DE DK EE ES FI FR FY GR HU IE IT LT LV NL PL PT RO SI SK SR TR UK
Storage withdrawal 75%
Storage withdrawal 50%
Storage withdrawal 25%
Storage withdrawal 0%
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
100 100 0 0 100 100 100 100 100 100 100 100 100 0 100 100 100 100 100 100 100 100 100 100 100 100 0 0 100
1.79 6.41 0.61 1.00 7.95 0.20 2.67 2.45 1.91 6.28 1.77 19.68 2.41 1.00 0.01 0.54 0.48 1.14 3.08 6.16 2.07 2.39 1.54 0.05 0.50 1.28 0.77 0.40 1.30
100 100 0 0 100 100 100 100 100 100 100 100 100 0 100 100 100 100 100 100 100 100 100 0 100 100 0 0 100
1.33 6.30 0.70 1.00 7.82 0.20 2.16 2.05 1.62 6.28 1.74 19.68 1.95 1.00 0.01 0.22 0.62 0.82 3.08 5.49 2.05 2.23 1.41 0.06 0.50 0.71 0.77 0.44 1.20
100 100 0 0 100 100 100 100 100 100 100 100 100 0 100 100 100 100 100 100 100 100 100 0 100 100 0 0 100
0.86 6.18 0.79 0.94 7.69 0.14 1.65 1.65 1.33 6.28 1.71 19.68 1.50 1.00 0.01 0.15 0.59 0.48 3.08 4.82 2.02 2.06 1.28 0.17 0.28 0.15 0.85 0.47 1.09
100 100 0 0 100 100 100 100 100 100 100 100 100 0 100 0 100 100 100 100 100 100 100 0 100 0 0 0 100
0.39 6.07 0.88 1.00 7.57 0.14 1.14 1.20 1.05 6.28 1.68 19.68 1.04 1.00 0.01 -0.20 0.56 0.17 3.08 4.15 2.00 1.89 1.15 0.29 0.28 0.42 0.93 0.50 0.98
0 100 0 0 100 0 100 100 100 100 100 100 100 0 100 0 100 0 100 100 100 100 100 0 0 0 0 0 100
-0.43 5.66 0.97 1.00 6.30 0.18 0.60 0.74 0.76 5.80 1.65 19.68 0.53 1.00 0.01 -0.76 0.53 0.15 2.93 3.48 1.90 1.68 1.02 0.40 1.00 0.98 0.93 0.53 0.88
Table 8 Percentage of successful gas delivery strategies S(C) and average margin AM(C) under the 1-in-20 winter demand scenario for different values of storages withdrawal capacity available in the hypothesis gas supply from Ukraine are completely cut. Country Storage withdrawal 100%
AT BE BG BH CH CR CZ DE DK EE ES FI FR FY GR HU IE IT LT LV NL PL PT RO SI SK SR TR UK
Storage withdrawal 75%
Storage withdrawal 50%
Storage withdrawal 25%
Storage withdrawal 0%
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
Share of successful strategies
Average margin
100 100 0 0 100 100 100 100 100 100 100 100 100 0 0 0 100 100 100 100 100 100 100 0 99.12 100 0 0 100
0.45 1.55 0.71 1.00 7.95 0.38 0.47 1.40 0.56 4.35 0.49 19.68 0.17 1.00 0.02 0.31 0.66 0.18 1.44 3.37 1.27 1.89 0.78 -0.20 0.48 0.18 -0.95 -0.42 0.26
100 100 0 0 100 1.31 100 100 100 100 100 100 100 0 0 0 100 56.49 100 100 100 100 100 0 0 0 0 0 100
0.03 1.51 0.78 1.00 5.57 -0.06 0.27 1.07 0.40 4.35 0.43 19.68 0.01 1.00 0.02 0.52 0.62 0.00 1.44 2.96 1.25 1.75 0.69 0.28 0.94 0.12 0.95 0.45 0.20
0 100 0 0 100 0 100 100 100 100 100 100 0 0 0 0 100 0 100 100 100 100 100 0 0 0 0 0 100
0.22 1.27 0.85 1.00 5.14 -0.18 0.06 0.72 0.25 4.35 0.41 19.68 0.14 1.00 0.02 0.65 0.52 0.19 1.44 2.55 1.21 1.60 0.60 0.37 1.00 0.41 0.95 0.48 0.11
0 100 0 0 99.56 0 0 100 100 100 100 100 0 0 0 0 2.41 0 100 100 100 100 100 0 0 0 0 0 30.52
-0.47 0.42 0.91 1.00 2.38 -0.29 0.29 0.25 0.10 4.02 0.39 19.68 0.43 1.00 0.02 0.77 0.80 0.43 1.36 2.15 0.89 1.35 0.51 0.46 1.00 0.70 0.95 -0.51 -0.01
0 99.99 0 0 0.31 0 0 18.11 0 100 100 100 0 0 0 0 0 0 100 100 100 100 100 0 0 0 0 0 0
0.75 0.37 0.98 1.00 0.95 0.40 0.99 0.03 0.06 3.45 0.37 19.68 0.69 1.00 0.02 0.90 0.95 0.66 1.28 1.74 0.87 1.21 0.42 0.54 1.00 0.99 0.95 0.54 0.09
ARTICLE IN PRESS 2496
F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
Table 9 Percentage of successful strategies S(C) in Slovenia in the case of 1-in-20 winter demand under different assumption for gas storage withdrawal capacity and amount of gas flow from Ukraine. The minimum amount of Ukraine gas flow for which S(C) 420% (‘‘security turning point’’) is also shown in the last table line (see text for details).
Table 11 Minimum amount of gas flow from Ukraine that European countries need in order avoid a possible gas supply crisis ‘‘security turning point’’. Results are shown for different values of storage withdrawal availability in the case of 1-in-20 winter demand scenario (see text for details). Storage withdrawal capacity available
Ukraine gas flow
Storage withdrawal capacity available 100%
75%
50%
25%
0%
0% 20% 40% 60% 80% 100%
100 100 100 100 100 100
1.31 95.29 97.93 100 100 100
0 0.45 3.94 19.1 56.17 99.62
0 0 0.15 0.43 0.99 1.21
0 0 0 0 0 0
Turning point
Secure
0–20%
60–80%
Critical
Critical
Table 10 Minimum amount of gas flow from Ukraine that European countries need in order avoid a possible gas supply crisis ‘‘security turning point’’. Results are shown for different values of storage withdrawal availability in the case of Average Winter demand scenario (see text for details). Storage withdrawal capacity available
AT BE BG BH CH CR CZ DE DK EE ES FI FR FY GR HU IE IT LT LV NL PL PT RO SI SK SR TR UK
100%
75%
50%
25%
0%
Secure Secure 0–20% 0–20% Secure Secure Secure Secure Secure Secure Secure Secure Secure 40–60% Secure Secure Secure Secure Secure Secure Secure Secure Secure Secure Secure Secure 0–20% 40–60% Secure
Secure Secure 0–20% 0–20% Secure Secure Secure Secure Secure Secure Secure Secure Secure 40–60% Secure Secure Secure Secure Secure Secure Secure Secure Secure 0–20% Secure Secure 0–20% 40–60% Secure
Secure Secure 0–20% 0–20% Secure Secure Secure Secure Secure Secure Secure Secure Secure 40–60% Secure Secure Secure Secure Secure Secure Secure Secure Secure 0–20% Secure Secure 0–20% 40–60% Secure
Secure Secure 20–40% 40–60% Secure Secure Secure Secure Secure Secure Secure Secure Secure 60–80% Secure 0–20% Secure Secure Secure Secure Secure Secure Secure 0–20% Secure 0–20% 40–60% 60–80% Secure
0–20% Secure 20–40% 40–60% Secure 20–40% Secure Secure Secure Secure Secure Secure Secure 60–80% Secure 40–60% Secure 20–40% Secure Secure Secure Secure Secure 20–40% 20–40% 0%-20% 60–80% 60–80% Secure
drops between 60% and 40% of maximum flow. Such a turning point is summarized in the last row of Table 9, where ‘‘secure’’ means S(C)420% for all values of Ukraine gas flow and ‘‘critical’’ means S(C)o20% for all values of Ukraine gas flow. Tables 10 and 11 show the outcomes of applying such a procedure to assess turning points between security and insecurity applied to all the countries modeled in both AW and 20W demand scenarios. These results will be further discussed in next section.
5. Discussion The newly created model MC-GENERCIS has been applied to assess the security of gas supply of the European countries under
AT BE BG BH CH CR CZ DE DK EE ES FI FR FY GR HU IE IT LT LV NL PL PT RO SI SK SR TR UK
100%
75%
50%
25%
0%
Secure Secure 20–40% 80–100% Secure Secure Secure Secure Secure Secure Secure Secure Secure Critical 20–40% 40–60% Secure Secure Secure Secure Secure Secure Secure 0–20% Secure Secure 60–80% 60–80% Secure
Secure Secure 20–40% Critical Secure 0–20% Secure Secure Secure Secure Secure Secure Secure Critical 20–40% 60–80% Secure Secure Secure Secure Secure Secure Secure 0–20% 0–20% 0–20% Critical 60–80% Secure
0–20% Secure 20–40% Critical Secure 60–80% Secure Secure Secure Secure Secure Secure Critical Critical 20–40% Critical Secure 60–80% Secure Secure Secure Secure Secure 20–40% 40–60% 0–20% Critical 80–100% Secure
0–20% Secure 40–60% Critical Secure Critical 0–20% Secure Secure Secure Secure Secure Critical Critical 4–60% Critical 60–80% Critical Secure Secure Secure Secure Secure 20–40% 60–80% 0–20% Critical 80–100% Secure
20–40% Secure 40–60% Critical 40–60% Critical 20–40% 0–20% Critical Secure Secure Secure Critical Critical 40–60% Critical Critical Critical Secure Secure Secure Secure Secure 20%-40% 80–100% 0–20% Critical Critical Critical
crisis conditions. The methodological bases of the model have been shown and some significant scenarios have been studied. In the following, model potential and limits are briefly summarized before discussing the meaning of its results for the European gas system.
5.1. MC-GENERCIS 5.1.1. Model potential and validation MC-GENERCIS is a model based on a MonteCarlo approach that is completely new in the field of gas system modeling. On top of such a scientific novelty, the stochastic approach offers the opportunity to test the whole spectrum of possible ‘‘dispatching strategies’’ available to transmission systems operators to satisfy the expected gas demand. The model successfully provides a systemic view of the gas infrastructures as it is able to show which combination of single TSOs choices can result in an overall ‘‘successful strategy’’. Furthermore, the stochastic nature of the approach naturally provides the model user the opportunity to base his/her final judgment about the ‘‘security of gas supply on his/her personal risk aversion and/or risk tolerance (see Section 4.5.4). MC-GENERCIS offers a unique methodology for analyzing the reliability of gas transmission systems for present and future gas needs. Model outcomes have been compared with actual system operation (both in normal and crisis situations) and crisis indicators have been shown to be in line with reality. Small deviations could be attributed to the limited accuracy of data available. In particular a better benchmark with the actual January 2009 crisis features could be possible in case of availability of more reliable data on expected demand in those
ARTICLE IN PRESS F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
days and further investigation in this direction is currently under way. In any case, considering input uncertainties, results shown in Section 4.5.3 should be considered fully satisfactory from the point of view of the model validation.
5.1.2. Model limits and sphere of application. Although being able to provide robust result, MC-GENERCIS shows some limitations mainly related to both its time and spatial resolution and its overall design. At first, MC-GENERCIS is not designed to provide insights into national systems and neither to consider possible countermeasures to gas disruptions other than gas withdrawal from storages. As an example, fuel switching, one of the most typical crisis countermeasures, is not modeled in MCGENERCIS. Furthermore, as already noticed, MC-GENERCIS does not apply any kind of weighting when exploring the space of available dispatching strategies. Strategies are only divided among successful and unsuccessful (see Section 2.5) and they are not distinguished among realistic and improbable, for example from the market point of view. Finally, another main limitation of the model consists in the lack of a temporal dimension. The current version of MCGENERCIS can be basically applied to a single time slice. Even if in principle it could be possible to apply the model to a sequence of scenarios representing a time evolution, these scenarios should be developed externally to the model because the model will be not able to choose the ‘‘best’’ evolution step from a time frame to the following one. For all these reasons, the best sphere of application of MCGENERCIS is in providing early warnings on the possibility of a gas lack arising in a certain country or area, given a certain scenario. Such an information could be also useful to redesign the system in order to minimize the calculated crisis likelihood. As an example, the model could be applied to the study of planned infrastructures (pipelines, storages, or interconnectors) to investigate how much these infrastructures can improve the situation summarized in Tables 10 and 11.
5.2. The European gas transmission system. Tables 10 and 11 show the security turning point of European countries in case of different levels of gas storage availability and for the demand scenarios studied, i.e., the minimum amount of gas flow from the Ukraine that assures the country security of its gas supplies, here defined as a value of S(C) larger than 20%. European countries can be broadly divided in three main groups. A first group contains countries not worried, to which a ‘‘secure’’ score is always awarded, namely Belgium, Baltic countries, Finland, Netherlands, Poland, Spain, and Portugal. Security of gas supply in these countries is insensitive to the actual gas flow in Ukrainian pipelines, for different reasons6 . A second group contains the countries that are sensitive to Ukraine gas flow whatever their storage filling: Bulgaria, Bosnia and Herzegovina, FYROM, Serbia, and Turkey and also, in case of high gas demand, Romania, Hungary, and Greece. The third group of countries contains all other ones: they show a transition from security to partial and sometimes total insecurity when storage availability decreases. It is worth noticing how all major European countries belong to this group, some of them only marginally (e.g., UK and Germany are not totally secure in very extreme conditions only) while others are in principle 6 E.g., Baltic countries are directly supplied by Russia while Spain and Portugal are geographically far away form crisis epicenter.
2497
more sensitive to what happens in eastern pipelines (e.g., Italy and France). Such a picture, even simplified, confirms the current common ideas on the European gas system and again the robustness of the approach applied insofar. Obviously, a number of improvements are still like including the expected amount of lacking gas AM(C) in the definition of ‘‘security’’ and/or introducing economic reference points in order to rank possible crisis in function of their likely cost. Nevertheless, as a first application of a totally new methodology, results have shown to be robust enough to be considered a solid starting point for further developments.
6. Conclusions A new model for providing early warnings on possible gas crisis in Europe has been developed, validated, and applied to the case of a decrease or shut down of gas flows to Europe from its eastern borders, specifically from the Ukraine. The model, based on a MonteCarlo approach, has been shown to reliably in provide early warnings. As a first application, the model has been used to systematically analyze the adequacy of the European gas transmission system in case of partial or total lack of gas supply under different conditions of storage filling. Even if a number of limitations prevent its use as an emergency or optimization tool, the model and the methodology employed are nevertheless likely – if properly fed with reliable data – to provide an important added value in analyzing the reliability and the adequacy of present and planned gas systems at least as an ‘‘early warning’’ tool for a potential gas supply crisis.
Acknowledgments The authors would like to thank their former colleague Russel Pride from JRC-IPSC for long and useful discussions. For the same reason they are also in debt with many colleagues from DG TREN: Dinko Raytchev, Kitty Nyitrai, Miroslav Marias, and Mathieu Samyn. The continuous support of Christine Berg and Jean-Arnold Vinois has also to be warmly acknowledged together with the management and the staff of Energy Security unit in JRC-IE that made possible to create MC-GENERCIS. This work was financially supported by JRC-IE in the frame of its institutional duties. Disclaimer The views expressed in this paper are purely those of the writers and may not in any circumstances be regarded as stating an official position of the European Commission. References Baker Schaffer, M., 2008. The great gas pipeline game: monopolistic expansion of Russia’s Gazprom into European markets. Foresight 10, 11–23. Ball, J., Roberts, P., 2007. LNG as a new force in global energy security. Gas Technology Institute—15th International Conference and Exhibition on Liquified Natural Gas, 34–40. Barrett, M., Lowe, R., Oreszczyn, T., Steadman, P., 2008. How to support growth with less energy. Energy Policy 36, 4592–4599. Barroso, L.A., Rudnick, H., Mocarquer, S., Kelman, R., Bezerra, B., 2008. LNG in South America: the markets, the prices and the security of supply. IEEE Power and Energy Society 2008 General Meeting: Conversion and Delivery of Electrical Energy in the 21st Century PES, art. no. 4596232. Damigos, D., Tourkolias, C., Diakoulaki, D., 2009. Households’ willingness to pay for safeguarding security of natural gas supply in electricity generation. Energy Policy 37, 2008–2017. Dorigoni, S., Portatadino, S., 2008. LNG development across Europe: infrastructural and regulatory analysis. Energy Policy 36, 3366–3373. Egging, R., Gabriel, S.A., Holz, F., Zhuang, J., 2008. A complementarity model for the European natural gas market. Energy Policy 36, 2385–2414.
ARTICLE IN PRESS 2498
F. Monforti, A. Szikszai / Energy Policy 38 (2010) 2486–2498
European Commission, Directorate-General for Energy and Transport, 2008. Commission staff working paper on the implementation of the Energy Market Observation System (EMOS) SEC(2008) 2898. European Commission, Directorate-General for Energy and Transport, 2009. Proposal for a Regulation of the European Parliament and of the Council concerning measures to safeguard security of gas supply and repealing Directive 2004/67/EC. COM(2009) 363. European Commission, Directorate-General for Energy and Transport, 2009. Accompanying document to the proposal for a regulation of the European Parliament and of the Council concerning measures to safeguard security of gas supply and repealing Directive 2004/67/EC—Impact assessment. SEC 2009, 980. European Commission, Directorate-General for Energy and Transport, 2009b. Accompanying document to the proposal for a regulation of the European Parliament and of the Council concerning measures to safeguard security of gas supply and repealing Directive 2004/67/EC—The January 2009 Gas Supply Disruption to the EU: An Assessment. SEC(2009) 977. Finon, D., Locatelli, C. 2008. Russian and European gas interdependence: could contractual trade channel geopolitics? Energy Policy 36 (1).
Heinrich, A. 2008. Under the Kremlin’s Thumb: does increased state control in the Russian gas sector endanger European energy security? Europe – Asia Studies 60. Holz, F., von Hirschhausen, C., Kemfert, C., 2008. A strategic model of European gas supply (GASMOD). Energy Economics 30, 766–788. ¨ Kjarstad, J., Johnsson, F., 2007. Prospects of the European gas market. Energy Policy 35, 869–888. Lise, W., Hobbs, B.F., van Oostvoorn, F., 2008. Natural gas corridors between the EU and its main suppliers: simulation results with the dynamic GASTALE model. Energy Policy 36, 1890–1906. Lochner S., Bothe D., 2007. From Russia with gas, an analysis of the Nord Stream pipeline’s impact on the European gas transmission system with the TigerModel. EWI Working Paper, no 07.02. Lochner, S., Bothe, D., 2009. 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 37 (Issue 4), 1518–1528. Warnig, M., 2008. A link for security of energy supply and climate protection in Europe. OECD Observer 267, 76–78. Weisser H., 2007. The security of gas supply – a critical issue for Europe? Energy Policy 35.