New Spain: France Transmission Line: A Cost-Benefit Analysis

Abstract

As the costs for the new planned underground high-voltage direct current (HVDC) line between Spain and France exceed the ones of alternating current (AC) overhead line alternatives it raises the question if the high costs can be justified by adequate benefits for society like increased security of supply. For this reason, the economic effects of the planned HVDC underground transmission line are evaluated through a cost-benefit analysis and the results are compared to an alternative new AC overhead line along the same route as well as an upgrade of a nearby AC overhead line. For an evaluation of the economic welfare effects, investment and operating costs as well as benefits from increased security of supply are quantified using the electricity network model ELMOD. All considered line options result in overall positive discounted net welfare gains for society with the new AC line option showing the highest ones. However, it is stressed that the HVDC solution holds certain advantages over the AC technology that cannot be explicitly quantified in a line assessment.

Zusammenfassung

Die Kosten der geplanten unterirdisch verlegten Hochspannungs-Gleichstrom-Übertragungsleitung (HGÜ) zwischen Spanien und Frankreich übersteigen die Investitionskosten einer vergleichbaren Wechselstrom-Freileitung. Die Frage stellt sich, ob die Mehrkosten mit ausreichendem Zusatznutzen für die Gesellschaft, wie zum Beispiel erhöhter Versorgungssicherheit, zu rechtfertigen sind. Aus diesem Grund wurden die ökonomischen Effekte der geplanten unterirdischen HGÜ-Leitung durch eine Kosten-Nutzen-Analyse untersucht und die Ergebnisse sowohl mit einer alternativen neuen Wechselstrom-Freileitung entlang derselben Route als auch mit einer Erweiterung einer bestehenden Wechselstrom-Freileitung in der Nähe verglichen. Um die ökonomischen Wohlfahrtseffekte zu berechnen, wurden die Investitions- und Betriebskosten sowie die Vorteile einer erhöhten Versorgungssicherheit mit Hilfe des Netzmodels ELMOD quantifiziert. Alle untersuchten Leitungsalternativen ergaben einen positiven abgezinsten Nettowohlfahrtsgewinn für die Gesellschaft unter Berücksichtigung einer gesteigerten Versorgungssicherheit, wobei der Neubau einer alternativen Wechselstrom-Freileitung den höchsten Wert erzielte. Es muss jedoch hervorgehoben werden, dass eine Gleichstromleitung bestimmte Vorteile gegenüber der Wechselstrom-Technologie besitzt, die jedoch nicht explizit in der Analyse quantifiziert werden konnten.

1 Introduction

Within the EU, cross-border transmission lines (interconnectors) are needed to secure electricity supply, support the liberalization of the internal EU energy market and to integrate renewable energy resources. Especially the installation of renewable energy resources in EU-27 has increased significantly over the last years. However, EU member countries will have to increase this share even further to meet their 2020 targets. Hence, significant amounts of renewable energy must then be transported from areas with large renewable resources to the demand centers through the electricity network.

Looking at the existing interconnections in Europe, several Members States are weakly interlinked with their neighbors. In continental Europe, Spain shows the lowest interconnection capacity share of 3.1 % of their installed production capacity (EC DG Energy 2010). This is below the target for electricity interconnections for EU Member States of at least 10 % (European Council 2002, p. 15). Spain will also serve as the feed-in point for the exported renewable energy from North Africa which is planned for instance by the Desertec project. Furthermore, Spain wants to increase its own, already large installed renewable energy capacity by 12.4 GW until 2015. This growth will increase the need for additional interconnection capacity, especially into North direction to France. As shown in Fig. 1, Spain and France are connected via four alternating current (AC) overhead lines resulting in a net transfer capacity in summer of 1,200 MW from France to Spain and only 500 MW in the opposite direction (ENTSO-E 2011a).

Fig. 1

Existing transmission lines between Spain and France. Own illustration

Over the last 3 decades the French and Spanish Transmission System Operators (TSO) have been trying to realize a new line across the borderline but faced significant public resistance. The planned line has meanwhile also been prioritized and declared as line of ‘European interest’ by the European Commission (Decision No 1229/2003/EC) and in 2008 the EU even sent a coordinator to overcome the outstanding difficulties. He was successful with the proposal to bury the line underground and along existing infrastructure on the Eastern Pyrenees between Baixas and Santa Llogaia (see Fig. 1). The new line is planned to be operational in 2015 and will have a transmission capacity of 2 × 1,000 MW. As the longest applied AC underground cable so far is an installation in Japan with a length of only 40 km¹, the TSOs decided to use the high voltage direct current (HVDC) technology instead for the planned length of 65 km. This technology choice will result in total investment cost of EUR 700 million which is nearly eight times higher than the EUR 90 million originally planned for a double circuit AC overhead line. Investment costs for underground cables are regarded as being up to 12 times more expensive than overhead lines depending mostly on the terrain (ICF Consulting 2003). Additionally, HVDC technology requires expensive converter stations with large conversion losses which explain the higher investment costs than AC lines. However, due to their low line losses compared to AC, HVDC lines are advantageous for long distance transmission above 600 km as the benefits of low transmission losses compensate for the higher investment costs and losses of the converter stations. Following Wu et al. (2006, p. 956), “expansion planning focuses on the selection of least-cost alternatives” and thus on solutions that are most economical. As the new HVDC line between Spain and France is only 65 km long, it is therefore questionable if the potential benefits will exceed and justify the higher investment cost. Despite the fact that interconnections at the France-Spain borderline are regarded as crucial for integrating the Iberian Peninsula to the rest of Europe, there is no detailed assessment in literature available yet. Thus, the article addresses this issue by an appropriate cost-benefit analysis. In particular, the costs and benefits of the new HVDC line in 2015 are assessed and compared to a new AC overhead line, which is the most economical technological transmission line solution, on the same route and to an alternative upgrade of an existing AC line. The last one is included as option as public resistance was a major obstacle in the planning phase. Upgrading an existing transmission line faces usually less opposition as the ‘visual impact’ already exists. The comparison aims to answer (1) whether the planned HVDC solution is the most cost-effective transmission line design choice for society to extend the interconnection capacity at the Spain-France borderline and (2) if and to what extent the Spanish and French consumers gain welfare through the installation of a new interconnector.

The remainder of the article is organized as follows. Section 2 briefly outlines and discusses how interconnector installations are evaluated through the methodology of a cost-benefit analysis. Section 3 describes the electricity network model, which is used to calculate the welfare effects of the line for the main European electricity market. The section also outlines the data used, the different transmission line options and the calculated representative hours which are upscaled in order to get the results for a total characteristic year. The model results are described and discussed in Sect. 4 before the article concludes with the final Section.

2 Methodology to Evaluate Interconnectors

Eijgenraam et al. (2000) provide a guideline for cost-benefit analyses to evaluate large infrastructural projects. They propose to apply a quantitative macro-economic simulation model to derive the welfare effects. Similarly, the Californian Independent System Operator (CAISO) (2004, p. 48) developed a Transmission Economic Assessment Methodology (TEAM) specifically for regulated transmission lines which uses classical economic surplus measures to define the welfare of society. Specifically, economic welfare consists of consumer rent, congestion rent and producer rent (see Fig. 2 ²).

Fig. 2

Welfare gain through installation of additional interconnection capacity. Own illustration

Consumer rent can be described as the utility for electricity consumers by being able to purchase electricity for a price ${{P}_{con\,\,q(1)}}$ that is less than the highest price $p{{\,}_{con\,\max }}$ that they would be willing to pay for (according to their price sensitivity). In the example in Fig. 2, it would be calculated as

$$\frac{1}{2}q(1)(p{{\,}_{con\,\max }}\,-\,{{P}_{con\,\,q(1)}})$$

(1)

Likewise, producer rent is the amount that electricity producers earn by selling their generated electricity at a market price ${{p}_{prod\,q(1)}}$ that is higher than the least price they would be willing to sell for. In the case of perfect competition this least price is set by the marginal generation cost $c(q(n))$ of the marginal generator. The producer rent is calculated as

$${{p}_{prodq(1)}}\,q(1)\,-\,\int\limits_{n=0}^{1}{(c(q(n))dq(n)}$$

(2)

The third rent in the market according to CAISO is the congestion rent and is earned by the TSOs. They gain through electricity flows between regions utilizing the electricity price arbitrages. In the example in Fig. 2 it is calculated as the price arbitrage between two nodes times the respective transmission flows

$$q(1)\left( {{P}_{con\,q(1)}}\,-\,{{P}_{prod\,q(1)}} \right)$$

(3)

In general, increasing the interconnection capacity between two electricity markets or countries from $q(1)$ to $q(2)$ leads to price convergence. This means, that one country will face lower prices, whereas the other country’s prices will increase. Therefore, some consumers would be worse off whereas producers who export their electricity to neighboring countries that have higher prices would win. Overall, “investments [in additional interconnection capacity] should positively address social welfare” (ENTSO-E 2010a, p. 16), i.e. gains in rents of market participants across countries must exceed losses of others. Moreover, CAISO suggests considering the additional benefits from increased security of supply facilitated through the installation of a new interconnector. However, due to software limitations and difficulties of quantification, they did not include them in their TEAM approach. Also Kapff and Pelkmans (2010, p. 6) strongly suggest to consider the benefits of higher grid reliability in a cost-benefit analysis. They argue that interconnections allow for more solidarity between countries in case of disruption. Though, the authors do not provide an explicit method on how to calculate these effects. Malaguzzi Valeri (2009) studied the welfare effects of an interconnection between Ireland and Great Britain. She identified a positive but decreasing welfare effect for society trough new interconnection capacity between the two countries. However, she concludes that there is the aspect of enhanced security of supply and lower amount of reserves needed to maintain a secure system which are likely to positively affect social surplus but are not taken into account in her work. We agree with Joskow (2005), who states that the increase of reliability of the transmission grid is often the main reason behind new line investments. In our view, the increased security of electricity supply is a major benefit for society and therefore we regard the consideration of these benefits as crucial in a transmission line assessment. Thus, in contrast to previous research the benefits through increased security of supply will be a core part of our cost-benefit analysis next to the applied economic welfare calculation as outlined by CAISO (2004). We will add the calculated reliability benefits to the modeled economic welfare effects and compare them with all related investment and operating costs for the different transmission alternatives at the France-Spain border. The evaluation of additional benefits for society is theoretically possible, e.g. the benefit of reduced public opposition through the installation of an underground line. Due to less public resistance, an underground solution could theoretically be implemented in a shorter period of time which leads to additional benefits for society, especially for TSOs. However, the quantification of such effects is complex and would go beyond the scope of this paper. Thus, additional social costs and benefits are not considered in the cost-benefit analysis of this work.

3 European Grid Simulation Model

3.1 Model Description

In order to evaluate the economic welfare gains of the different interconnection options the grid model ‘ELMOD’ developed by Leuthold et al. (2005) is used. ELMOD is selected as it is today “one of the larger engineering-economic models, with a very high granularity, allowing differentiated spatial price and flow analysis” and can be classified as “a large scale (…) optimization model maximizing social welfare under the assumption of perfect competition taking into account technical constraints” (Leuthold et al. 2010). ELMOD employs the so called DC load flow (DCLF) approximation to determine the AC line flows which is a widely used approximation in techno-economic large-scale electricity network models. The detailed model description can be found in Leuthold et al. (2010). ELMOD effectively assumes real-time scheduling and dispatch with a nodal pricing approach, i.e. it optimizes the dispatch and transmission flows according to the respective nodal prices in the grid in a timeframe of 1 h³. As the different interconnector options at the Spain-France border are assessed, the electricity grids of both countries need to be modeled in detail. All neighboring countries are included as aggregated nodes in order to account for potential loop flows (see Fig. 3).

Fig. 3

Regional model scope and transmission grid

The model objective is to maximize the total economic welfare across all grid nodes. This is determined through a node specific linear inverse demand and a supply function. The individual demand function is derived by a reference price, reference demand and price elasticity⁴. The supply function is equal to marginal cost, representing the so called ‘merit order’. Thus, ELMOD’s objective function maximizes welfare W across all nodes n with the following equation:

$$\underset{g,q}{\mathop{\max W}}\,=\sum\limits_{n}{\left( \int\limits_{0}^{q(n)}{np(q(n))dq(n)}-\sum\limits_{p}{\left( c(p,n)g(p,n) \right)} \right)}$$

with p power plant type

$q\left( n \right)$ demand at node n

$np\left( q\left( n \right) \right)$ linear inverse demand function

$c\left( p,n \right)$ marginal generation cost of power plant p at node n

$g\left( p,n \right)$ generated electricity of power plant p at node n

In order to make sure that demand, supply, electricity inflows and outflows are equalized at each grid node, a nodal energy balance (Eq. 5) for each grid node n is defined. With that equation the model determines the transmission line flows in a way that the electricity demand $q\left( n \right)$ and the electricity consumption ${{q}_{psp}}\left( n \right)$ of pumped hydro power plants at a node n are met by the sum of electricity generation (wind ${{g}_{wind}}(n)$, solar PV ${{g}_{pv}}(n)$, all other power plants $g(p,n)$), the $netinput(n)$ (net effect of all power in- and outflows of electricity at node n) as well as the net in- or outflow of a connected HVDC line ${{P}_{HVDC}}(n,nn)\le {{P}_{HVDC\,\max }}(n,nn)$.

$$\begin{aligned}& 0=q\left( n \right)+{{q}_{psp}} \left( n\right)-{{g}_{wind}}\left( n \right)-{{g}_{pv}}\left( n \right)\\& \,\,\,\,\,\,-\sum\limits_{p}{g\,\,\left( p,n\right)+netinput\,\,\left( n \right)}\quad \quad \quad \quad \quad\quad nodal \, energy \, balance \\&\,\,\,\,\,\,+\sum\limits_{nn}{{{P}_{HVDC}}\,\,\left( n,nn\right)-\sum\limits_{nn}{{{P}_{HVDC}}\,\,\left( n,nn \right)}} \\\end{aligned}$$

(5)

with ${{q}_{psp}}(n)$ electricity consumption of Pumped Hydro Storage Plants (PSP) at node n

$q(n)$ demand at node n

${{g}_{wind}}(n)$ wind generation at node n

${{g}_{pv}}(n)$ solar PV generation at node n

$g(p,n)$ generation of all other power plants p at node n

$netinput(n)$ injection or withdrawal of electricity at node n

$$\left| P(l) \right|\le {{P}_{\max }}(l)\quad \quad \text{capacity constraint of AC lines}$$

(6)

$${{P}_{HVDC}}(n,nn)\le {{P}_{HVDC\,\max }}(n,nn)\quad \quad \text{capacity constraint of DC lines}$$

(7)

$$g(p,n)\le {{g}_{\max }}(p,n)\quad \quad \text{maximum generation constraint}$$

(8)

Electricity flow P on an AC line l is subject to capacity restriction ${{P}_{\max }}$ (Eq. 6). Due to their technical characteristics, HVDC lines cannot be modeled with the DC load flow approach. Thus, the HVDC lines are considered in the model as point-to-point connections by including their directed positive flows ${{P}_{HVDC}}\,\,\,(n,nn)$ in the energy balance for both the import and export direction (Eq. 5). Power flows on the HVDC lines ${{P}_{HVDC}}$ are positively defined and limited to ${{P}_{HVDC\,\max }}$ (Eq. 7) that includes a capacity reduction of 3.4 % per line (Cole et al. 2006, p. 23) considering the high electricity conversion loss of the required converter stations. HVDC line losses are neglected due to very short line distance of 65 km. Finally, power plant output g needs to remain below the maximum generation capacity ${{g}_{\max }}$ (Eq. 8).

The complete model is then coded and solved as non-linear problem in the General Algebraic Modeling System (GAMS) with a CPLEX solver.

3.2 Input Data for Grid Model

3.2.1 Generation Data

National data for conventional power plants in 2015 are taken from the European Commission Directorate-General for Energy (2010). For Spain and France all power plants need to be allocated to the respective grid nodes. Plant location and capacity are derived from Meller et al. (2008) that include all operational plants which have a generation capacity greater than 100 MW. These are about 400 and are classified as the following different plant types: hydro, PSP, nuclear, lignite, coal, open cycle oil turbine (OCOT), combined cycle oil turbine (CCOT), oil steam, open cycle gas turbine (OCGT), combined cycle gas turbine (CCGT) and gas steam. For generation capacities that could not be assigned to single grid nodes in France or Spain their predicted electricity generation is subtracted from the national demands to still consider them in the model.⁵

To consider PSPs despite modeling only single representative hours, PSP plants are included with the status of charging during each off-peak hour and with the status of discharging when modeling a mid-peak or peak hour, which is in line with a typical load curve of a PSP plant. The planned shut-down of several German nuclear power plants until 2015 that was politically triggered after the recent nuclear accident in Fukushima, Japan is also considered. Planned national installation capacities of renewable energies in 2015 are taken from the National Renewable Energy Action Plans (NREAPS) (ECN and EEA 2011). For the detailed French and Spanish grid it is not possible to identify the location of each and every wind and photovoltaic (PV) system in order to assign it to its respective grid node. For wind, this is solved by assessing wind resource data and available information on total installed capacities per region. With this information, a wind share per grid node is calculated. Similarly for solar PV the mean global horizontal irradiation (GHI) is calculated per NUTS-3⁶ region with a Geographic Information System (GIS) and data from European Commission Directorate-General Joint Research Center (2011). Together with available information of the regional breakdown of already installed PV capacity in France and Spain (Emerging Energy Research 2009) the PV share per NUTS-3 region is derived. This share is distributed equally on all grid nodes within each single NUTS-3 region. If there is no node within a certain region, the PV share is equally distributed to all neighboring districts with nodes. Finally, the total national installed PV capacity in 2015 can be multiplied with the individual PV share per node.

Regarding concentrated solar power (CSP), a manageable amount of large-scale power plants will exist in 2015—almost exclusively in Spain. These plants are identified through available plans and maps, satellite and aerial photos and allocated to the closest grid node. Additionally for solar, the first Desertec reference project is expected to be online in 2015 and to have a total capacity of 500 MW with the split of 100 MW PV and 400 MW CSP. It was announced that 20 % of the PV and 80 % of the CSP generated electricity will be exported to Spain (van Son 2011). To ensure this dedicated export in the model, 340 MW of Desertec power are directly fed in the Spanish grid node that is connecting to Morocco.⁷

Finally, country specific capacity factors are calculated for 2015 in order to get more realistic model results. The factors can be found in Table 7 in Appendix A. The detailed overview of the generated electricity, installed capacity and demand per country is provided in Appendix B. As ELMOD optimizes the dispatch, marginal generation costs of the power plants are used. Marginal cost is defined as “the cost of producing an additional unit of electricity (…) [and] mainly reflect the fuel costs” (IEA and NEA 2010, p. 169). All renewable energy resources except biomass are thus assumed to have marginal cost of zero as sun, wind, running water and geothermal energy do not face direct fuel costs. PSPs are charging at night at cost of 35 EUR/MWh (EEX 2011a). With a conversion efficiency of 75 % (Mueller 2001) the marginal costs result in 47 EUR/MWh. The generation costs of biomass and conventional power plants are calculated via their respective efficiencies, fuel cost and their related cost for emission allowances. For 2015 a price of EUR 24/tCO₂e is used according to Deutsche Bank (2010). The detailed marginal generation costs per plant type are shown in Table 1. For simplification purposes, it is assumed that the marginal costs of power plants do not differ across the regions⁸.

Table 1 Marginal generation costs of power plant types. (Source: Own calculations based on Mueller (2001) and IEA and NEA (2010))

3.2.2 Demand Data

First, to derive the reference demand curve a reference price is approximated through the average current annual spot price of the respective European electricity exchanges, which is 60 EUR/MWh. The average price reduction that will be caused through increasing integration of renewable energy until 2015 can be assumed to be compensated by corresponding inflation. The nodal demand share calculated from the electricity consumption forecast for 2015 is used as reference demand (ECN and EEA 2011). We furthermore assume a price elasticity of − 0.25 which is in line with the price elasticizes of electricity analyzed and compared in the work of Lijesen (2007).

Second, to model power flows in the detailed grids in Spain and France, a node-specific allocation is not only required for electricity generation but also for its consumption. The gross value added is used as a proxy for regional consumption and is available on NUTS-3 level for Spain and France (Eurostat 2011). The level specific share is distributed equally across all nodes within a specific NUTS-3 area. Via these node specific shares the national demand is broken down to the single grid nodes. As already mentioned, PSPs consume electricity during off-peak hours. Their demand needs to be considered as well. The forecasted national electricity demand figures for 2015 already include PSP demand, which has to be subtracted first before the total demand is distributed to the grid nodes. The PSP consumption is then allocated according to the installed capacity and the calculated national capacity factors of the respective PSP plants which are already assigned to specific nodes in France and Spain. All remaining countries are modeled as aggregated nodes and therefore do not need a node specific split of national demand but separation of PSP demand. Total demand data for 2015 is derived from the NREAPS for the EU countries (ECN and EEA 2011). In case of Switzerland and Morocco several other sources are consulted⁹. Demand varies throughout the day, the week and the season. Thus, the hourly demand variations of a whole year are obtained from the ENTSO-E database (ENTSO-E 2011b).

3.2.3 Transmission Grid Data

The electricity transmission grid in ELMOD is based on the ENTSO-E grid map from 2009 (ENTSO-E 2010b) and contains extra high-voltage lines of the two voltage levels 220 and 380 kV. As the year of scope is 2015, the grid in that year might look slightly different. Although new planned grid installations from 2010 to 2014 can be found in ENTSO-E’s recent Ten-Year-Network-Development-Plan (ENTSO-E 2010a) the ELMOD grid is not upgraded accordingly. The reason is that there is an uncertainty if planned transmission lines will be realized in time or even realized at all. As a good example for that the DENA I grid study by the German Energy Agency (DENA) can be mentioned. In their study published in 2005, DENA identified a total of 850 km of additionally required transmission lines in Germany until 2015. By the end of 2010 only 90 km of that have been realized which accounts for roughly 11 % of the proposed plans with not much more to be realized until 2015 (DENA 2010, pp. 2–3). As the European Commission just has recently pointed out, “experience in a majority of EU Member States over the last decade has shown that the construction of energy infrastructure is becoming increasingly difficult as its development conflicts with other policy objectives such as protection of landscapes and/or the environment, spatial planning, public health and safety, tourism and other forms of local economic development”. As the focus of this research lies on the interconnection rather than the detailed national grids, ELMOD is used with its current network data for application. The only lines added in the model are two 400 kV AC subsea cables between Tarifa (Spain) and Fardioua (Morocco) with a length of 31 km, a 2,000 MW (270 kV) HVDC land/subsea cable between Sellindge (UK) and Bonningues-lès-Calais (France) with a length of 73 km (46 km subsea) and the following three interconnector options for evaluation between France and Spain (see also Fig. 4):

• The planned 2 × 1,000 MW HVDC VSC underground cable with a length of 65 km between Baixas (France) and Santa Llogaia (Spain).

• A new two circuit 380 kV AC overhead line with a length of 65 km between the same substations as the planned HVDC line and an approximate maximum transmission capacity of 2,720 MW (typical capacity of one 380 kV AC circuit is around 1,360 MW)

• An upgrade of the existing 380 kV AC overhead line (single-circuit) between Baixas (France) and Vic (Spain) by another identical circuit. Total length is 90 km and the approximate maximum capacity increase can be estimated with 1,360 MW. The capacity upgrade of this line is limited to an additional circuit. Thus, the capacity addition of this option is significantly less than the other two previous alternatives.

As shown in Fig. 4, additional 119 km of AC overhead lines are currently being built from Vic to Santa Llogaia on the Spanish side to link the planned HVDC line and also to integrate the coastal area. These lines are also included into the model. All new nodes lie within one NUTS-3 area. To adjust the demand share per node, the former share of this area is split equally across the new amount of nodes. The same is done for the wind and solar generation accordingly.

Fig. 4

Different interconnection options for Spain—France border. Own illustration

Thus, the final grid for the simulation consists of 1,169 nodes, 1,665 extra-high voltage AC transmission lines above 220 kV and two HVDC lines. To approximate the n-1 reliability criterion in the model we include a transmission reliability margin of 20 % which means that all AC lines within the modeled grid will be stressed with a maximum of 80 % of their thermal capacity limit (Leuthold et al. 2010).

3.3 Scenarios for Calculating a Representative Year

As mentioned above, due to computational challenges ELMOD can only model a timeframe of 1 h for such an extensive grid. In order to still get the results for a whole year the optimization problem is broken down. For that the annual 8,760 h are classified into several representative hours which can be modeled by ELMOD in a reasonable amount of time. In total, 36 different representative hours can be identified. Distinctions are made for seasons, load schemes and for different sun and wind situations (see Fig. 5). The set of representative hours is calculated for four individual scenarios. These are the several interconnector options at the France—Spain borderline:

• No additional interconnector (0)—as reference scenario

• New HVDC line (H)

• New alternative AC line (N)

• Upgrade of existing AC line (U)

Fig. 5

Model scenarios and representative hours. Own illustration

Hence, 144 representative hours are calculated overall. In order to get the total annual values the results from the representative hours are then multiplied with their frequency of occurrence within an average year according to ENTSO-E (2011b) and EEX (2010, 2011b). 2015 is used as the year of scope as this will be the first year in which the new HVDC interconnector is fully operational. Furthermore, the first Desertec reference project will export electricity from Northern Africa to Spain in 2015 for the first time.

4 Cost-Benefit Analysis of Transmission Line Alternatives

In order to evaluate and compare the different transmission line options, the economic welfare effects are modeled with ELMOD for a representative year, discounted over the lifetime¹⁰ of the transmission line and compared to the respective discounted investment, operation and maintenance costs. As outlined above, we consider also the discounted benefits from increased security of supply. If the net present value (NPV) of welfare gains and benefits exceeds direct costs, a line is beneficial for society and should be built.

The simulations with ELMOD based on the explained input data revealed the economic welfare gains which are listed in Table 2.

Table 2 Total annual economic welfare effects of several interconnection options at France-Spain border

The HVDC solution shows a slight positive welfare effect. In contrast to that, the two AC alternatives result in small welfare losses. The different results of the AC lines compared to the HVDC option can be explained on the one hand by the different transmission capacity of each option and on the other hand by the different transmission technology characteristics. When integrating the new HVDC line, the first model run with ELMOD showed significant loop flows at the Eastern France-Spain borderline. This was mainly caused by the limited transmission capacity in the hinterlands of the border between France and Spain. In reality the TSOs would aim to avoid any loop flows as they unnecessarily consume line capacity and cause additional transmission losses. They also lead to uneconomical transmission flows (e.g. from a low price zone to a high price zone) and thus to contrary welfare effects. The advantage of the HVDC VSC technology is that power flows can be fully controlled. The TSOs can actively reduce the voltage difference between the two HVDC line nodes and decrease power flows until loop flows are eliminated. Thus, a HVDC line can be operated in a way to maximize economical transmission profits whereas AC lines as integral part of the meshed grid and their electricity flows depend on the laws of Kirchhoff. Any change made in one section of an AC grid (e.g. changing the load demand at a node) will spread immediately to other sections of this interconnected grid and will change the load conditions on all other transmission lines (Wu et al. 2006). The AC line flow can only be controlled to a very limited extent, e.g. by the additional installation of a flexible AC transmission system (FACTS), which cannot be taken into account in our model. As the DCLF model approach does not consider reactive power and voltage changes, we reduced the HVDC line capacity manually for each calculated hour until loop flows were eliminated at the Eastern border between France and Spain. As already mentioned, for the AC lines a manual control was not possible in the model.

Economically, congested transmission lines lead to less efficient results as consumers cannot access cheap electricity generation to the full extent. In turn, more expensive electricity generation is dispatched due to network congestion which increases the respective nodal prices. As a consequence, some consumers have to pay more for electricity. When increasing transmission capacity between two areas or nodes with a price difference, the prices will converge as more consumers in the high price zone can import electricity from the lower price zone. All analyzed line options show consumer rent gains and producer rent losses. Moreover, all line options have positive congestion rents with the AC options showing higher values compared to the HVDC line. This is because the HVDC line capacity was manually reduced to the optimal transmission capacity in order to avoid loop flows whereas the AC lines accounted for higher transmission flows including significant loop flows. The overall slightly negative welfare results of the two AC transmission line options mean that the benefits of consumers, who on average have to pay less, and the surplus of TSOs, who earn higher congestion rents due to increased transmission capacity, do not exceed the losses of producers, who on average have to sell at lower price levels.

When analyzing the welfare effects of the countries Spain and France only (see Table 3), the model results show positive welfare effect sums for all three line options. This means, that in the total European context, some neighboring countries lose welfare due to the price changes in France and Spain.

Table 3 Annual welfare effects of countries Spain (ES) and France (FR) through installation of different interconnector options^a

Spain loses welfare from the installation of a new interconnector whereas France benefits from it. This is because the higher French electricity prices converge to the lower Spanish level. The comparison of different line options for the two countries shows that the French consumers benefit from every installation with the new AC line being the best option for them. Spanish producers win as they then can sell their electricity for higher market prices whereas the French electricity generators lose as their market price deteriorates. In contrast to the HVDC line, the two AC options result also in a slight consumer rent gain on the Spanish side. This can be argued by the technology characteristic of AC lines as they are subject to loop flows and cannot be optimally directed to the extent a HVDC line can be. The additional benefits for Spanish consumers are at the cost of French producers and Spanish TSOs. Their losses are much higher for the AC lines than for HVDC.

However as mentioned already above, only considering the pure welfare effects of new transmission lines is not sufficient in evaluating transmission expansion options. The increase in security of supply needs to be also taken into account in a cost-benefit analysis. Thus, the calculated welfare effects are discounted with 8 % (Golder Associates and ECOFYS 2008, p. 178) and adjusted for inflation with a rate of 3 % (DENA 2010, p. 189). For the lines a lifetime of 40 years is chosen (Golder Associates and ECOFYS 2008, p. 178). All related costs for investment (CAPEX), operation and maintenance (OPEX) are presented in Table 4. Costs for line losses are already included in the welfare effects as transmission losses for the AC and HVDC lines are directly considered in the model.

Table 4 Costs associated with interconnector options. (Sources: Siemens (2011), Migliavacca (2011) and Golder Associates and ECOFYS (2008))

Regarding security of supply, several ways can be found in practice for estimation. For example, ICF Consulting (2003) calculates the cost of a power outage with the consumers’ willingness to pay, which “gives a measure of the ‘cost of reliability’ of electrical services”. They approximate the willingness to pay as 100 times the retail price of electricity. This price is multiplied with the lost load during the blackout. The Dutch TSO Tennet follows the same approach in their cost-benefit analysis of the NorNed seacable between Norway and the Netherlands. They assume a 2 h blackout every 20 years and calculate the benefit as the value of lost load (de Nooij 2011). In contrast to that, the Irish TSO EirGrid estimates increased security of supply in the assessment of a new interconnector between Ireland and the UK as potential savings from the reduced need for peaking plants. This is calculated by the cost for providing open cycle gas peaking-plant capacity equal to the new transmission capacity (de Nooij 2011). As the situation of a large national blackout is statistically very unlikely and occurred only rarely in the past in Europe, we regard the latter calculation approach as more appropriate to estimate the benefits of increased security of supply. However, it still needs to be considered as approximation. In general, it cannot be stated, whether this is an under- or overestimation as it strongly depends on the kind, location and duration of a possible blackout. While it is probably underestimated for heavy industry sites (as also investments can be damaged in the case of a longer non-supply) it is mostly an overestimation for rural areas, where a higher flexibility to execute other workings without the need for electricity exist.

The Commission for Energy Regulation (2011, p. 30) outlines the associated costs of providing a peaking plant as the fixed cost of a best new entrant with EUR 74.87/kW (2012) per annum. The nominal capacity of avoided backup power is assumed to be 30 % of the interconnector capacity increase for every connected country¹¹. We are aware that this assumes that the additional generation capacity is always available in the neighboring states. Further research and detailed dynamic analyses would be required to assess whether this capacity can be really guaranteed throughout all 8,760 h of a year, i.e. also in hours with simultaneous peak loadings in the countries.

In Table 5, the NPV W_gross shows the discounted welfare effects of Table 2. CAPEX is highest for the HVDC line due to the high investment costs of this technology. The same holds true for NPV OPEX. The discounted security of supply aspect is calculated according the different line capacities increases, i.e. 2,000 MW for the HVDC line, 2,720 MW for the new AC line and 1,360 MW for the AC line upgrade. After considering all related costs and benefits the new AC line option has the highest net discounted welfare gain (NPV W_net) of EUR 691.8 million, followed by the AC upgrade and the HVDC line.

Table 5 Discounted welfare gains, related costs and further benefits of interconnector options

As the individual lines increase the interconnection capacity differently the welfare gains are put indicatively in relation to the individual capacity expansion in Table 6. Indicatively because AC transmission capacity cannot be regarded as fully available for active power flows due to reactive power limitations. It shows that the new AC line option also has the highest discounted welfare gain per increased megawatt as well as per euro of cost. It is closely followed by the AC line upgrade.

Table 6 Indicative^a discounted welfare gains per capacity increase and costs

Although the HVDC technology might seem to be the least option here, it needs to be stressed that the assessment does not consider the benefits of its controllability. HVDC allows for control of line flows and is not subject to loop flows within a meshed network structure. Especially for interconnections between countries it is essential that power flows can be managed actively, for example to reduce loop flows or synchronize grids. Also in terms of grid stability, HVDC technology has an advantage over AC. In case of a failure in the neighboring grid it prevents cascading. Nevertheless, all transmission line options result in an overall positive discounted welfare gain that justifies their construction (when taking the security of supply aspect into account).

5 Conclusion

This article assesses the welfare gains of the new HVDC underground line between Spain and France in 2015 compared to two AC overhead line alternatives. As the investment costs for a HVDC line exceeds the investment costs of an AC overhead line alternative, it raises the question if its welfare gain justifies the higher investment cost. The model demonstrates positive discounted welfare gains for all line alternatives after considering the benefits of increased security of supply. It can be concluded that according to the model results the installation of a new AC overhead line between Vic (Spain) and Baixas (France) reveals to be the most effective interconnection solution as it offers the best balance of benefits and related costs for society and eases the grid bottlenecks on the Eastern and middle borderline mostly. But it should be mentioned that the benefits of controllability that the HVDC technology offers could not be considered in the evaluation but might also play a role in the TSOs’ choice. The French and Spanish TSOs’ faced significant public resistance when planning the transmission extension at the border line. Thus, an underground cable was more or less the only solution for a successful grid extension¹². Due to a length of more than 40 km the TSOs decided to use the HVDC technology.

The second research question was to find out whether and to what extent the Spanish and French consumers face welfare gains through the installation of a new interconnector. Interestingly, model results show that both Spanish and French consumers would benefit from both overhead line options. In contrast, Spanish consumers would lose welfare from their requested HVDC underground line. Thus, research should find out if the people would rather accept overhead lines which they are opposing if the TSOs present such a kind of cost-benefit analysis and showing them that a proposed line is the best alternative to release congestion and implies economic welfare gains for the consumers.

The applied approach and methodology is limited to the following extent which needs to be considered when interpreting the findings: We use a static approach modeling a single representative year to evaluate the investment decision of different transmission line options. Thus, our static approach does not consider e.g. any potential power plant investments over time which might be triggered by congested transmission capacity.

Furthermore, the calculation of welfare gains from increased security of supply as the avoided standby cost of equivalent installed open cycle gas peaking-plant capacity needs to be considered as approximation. Finally, we are aware that nodal pricing does not exist in the included European countries. In principal, a two-step approach would be needed, i.e. modelling the spot market clearance followed by calculating the consequential transmission flows and likely redispatches. Nevertheless, we used nodal pricing in our model for simplification reasons. Although the dispatch of power plants might be different to some extent, the nodal pricing approach leads to an optimal solution from a system planning perspective which would only be possible in a two-step approach with additional assumptions.

6 Appendix A

Table 7 Country and power plant type specific capacity factors in 2015

7 Appendix B

Table 8 Generation capacities, electricity production and demand in 2015