ENERDAY Workshop

Dresden, April 21st 2006

Applied Congestion Management in the European

Context Christian Todem*)and Florian Leuthold

February 2006

*) Corresponding author: Dr. Christian Todem VERBUND Austrian Power Grid AG

Am Hof 6a 1010 Wien Austria tel.: +43(0)1-53113-52283 mail: Christian.Todem@Verbund.at

Abstract EU Regulation (EG) 1228/2003 (EU, 2003) regulates the framework for cross-border

transactions within the electricity sector. The regulation follows the general efforts of the EU

to understand the electricity sector economically as an integrated market that needs specified

rules and regulations. The paper describes the development of European congestion

management methods and proposals from currently applied bilateral contracts towards flow-

based multilateral coordinated auctions. Additionally, the incentive systems of different

allocation methods for auction revenues within the scope of flow-based coordinated auctions

for South-East-Europe (SEE) are analyzed. The model bases on a zonal approach and

embraces eight zones linked by twelve border lines. The problem is perceived as a linear

optimization problem which is solved in GAMS. Through our analysis, evidence can be

identified that the allocation methods proposed by ETSO (2001) can provide false incentives

to TSOs regarding an efficient transmission capacity usage.

2

1. Introduction In times of vertical integrated monopolistic energy companies, the task of congestion

management was to minimize generation cost while guaranteeing the highest level of network

security (Christie et al., 2000). In the course of deregulating the electricity sector, the

requirements to efficient congestion management changed. The task of the regulated system

operator still consists in the warranty of network stability. However – considering most of the

applied congestion methods – the TSO does not any more have influence on the location of

generation. The decision about which provider satisfies the existing demand is made by the

liberalized energy market. Accordingly, the system operator only has the opportunity to

define a framework and prepare appropriate actions in order to achieve a sufficient network

security (Christie et al., 2000). Moreover, cross-border flows become more and more

important. Therefore, two main reasons can be identified: On the one hand, the liberalization

of the electricity market facilitates cross-border trade of electrical energy – an important

example is the establishment of energy exchanges for standardized products. On the other

hand, differences in energy prices between different countries within the EU provide an

economic incentive for transporting electrical energy from one country to another. The

fundamental reason for these differences consists in different age structures and generation

technologies of the EU countries’ power plants. In accordance to this tendency towards higher

trading activities as well as due to a lack of transmission capacities within few control areas,

the necessity for a coordinated cross-border congestion management increases.

2. Survey of congestion management methods

2.1. Difference between flow-based and not flow-based methods

In general, flow-based and not flow-based market based congestion management methods can

be distinguished. Not flow-based methods do not take into account physical flows that result

from demand1 for cross-border transmission capacity – i.e. in form of bids for transportation

from a source node/zone to a sink node/zone. Among not flow-based methods, mainly

bilateral auctions take place as the effort for coordination is limited to the two TSOs at the

both ends of a cross-border line. Centralized coordination does not take place in most cases.

Regarding an energy transport from one zone to any other zone, a bidder may have to award

1 Hence, it is assumed that electricity can be transported from one specific zone to another specific zone. In reality, however, each input spreads over the entire meshed network according to Kirchhoffs’ laws.

3

in several bilateral auctions in order to transmit across several borders. Moreover, there are

also coordinated not flow-based auction.2 Thereby, bids across more than one border are

allowed but their physical impact is not considered. It is a pure commercial consideration.

In both cases, commercially available capacity is set by the TSOs such that network stability

is secured for each possible load case. This can lead to a remote utilization of the physically

available capacity.

Regarding load-based methods, however, physical flows that result from commercial bids are

calculated individually and assigned to each line capacity. Then, the optimal network usage is

calculated by means of demand and willingness to pay – while also taking into account

physical constraints. Hence, this calculation of an optimal resource allocation is

mathematically an optimization problem. The given capacity constraints define constraints

within the optimization. Ideally, the capacity allocation should be administrated by a

coordination center.

2.2. Nodal pricing

The economical and technical optimal congestion management method is nodal pricing

(compare i.e. Schweppe et al, 1988; Stoft, 2002, S. 391ff.). Often also referred to as locational

marginal pricing (LMP). Nodal pricing means that each node within the network is considered

separately. The price for energy at a node represents the incremental cost that incur for

delivering one more MWh of energy to exactly this node. The energy price is, consequently,

the best scarcity signal for the commodity electrical energy at a specific node. Disregarding

transmission losses and assuming sufficient generation capacity, the energy price is made up

by incremental cost of generation and scarcity prices of transmission. The prices are equal for

each node in case congestions do not occur. In case of congested lines, transmitting electricity

becomes a scarce commodity. The price for this commodity produces an extra charge on top

of the energy price according to supply and demand per each node. Accordingly, nodal prices

vary between different nodes.

2 An example is the CEPS coordinated auction for Poland, the Czech Republic and Germany/Vattenfall.

4

2.3. Market splitting

Within the market splitting approach, injections and withdrawals of several nodes are

represented mutually by a zone; they are assigned to one specific zone. Hence, there is only

one energy price per zone. Zones can be interpreted as sub-markets that, ideally, form due to

network congestions. In most cases, however, zones are defined by political borders (mostly

control areas). Basically, there are two alternative ways to merge nodes together as a zone.

The first alternative is similar to the nodal pricing approach. The entire network is considered

physically; also the lines within a zone. The second alternative does not consider the entire

network. The nodes of a zone are treated as they would be located upon one big cupper plate.

Hence, congestions within a zone are not regarded. Zones are linked by interconnections.3

Ehrenmann and Smeers (2004, p. 14ff.) refer to the first alternative as the ideal market

splitting, whereas, the second one can only be a second-best approach.

2.4. Market coupling

The difference between market coupling and market splitting consists in differing starting

conditions (Ehrenmann and Smeers, 2004, p. 23ff.). The market splitting approach assumes a

market that is split up in sub-markets. The market coupling approach, in contrast, assumes

that sub-markets already exist and cannot be merged to one integrated market in a short or

medium term. Therefore, market coupling tries to interlink sub-markets as far as possible.

Interpreting, again, a sub-market as a zone, market splitting and market coupling do

theoretically not differ significantly from each other; both can be referred to as zonal pricing

or zonal price models, respectively. However, implications for a practical implementation are

different.

2.5. Explicit coordinated auctions

Similar to the market coupling approach, explicit coordinated auctions interlink different

markets. The difference between a coordinated auction approach to the approaches mentioned

above consists in the chronology of the market clearing process and, accordingly, also in the

traded commodities. The traded commodity of the methods described above is electrical

energy at a specified location – a node or a zone. The market clearing price includes

3 The aggregation of the network bears is a potential source of error as the entire highly meshed grid can hardly be represented by a much smaller model.

5

generation cost as well as congestion costs.4 There is no separate price for transmission; that

is why those methods are often also referred to as implicit auctions. The first step of an

explicit coordinated auction is the allocation of transmission capacity. Only after the award of

transmission capacity, the energy market opens (Ehrenmann and Smeers, 2004, p. 31) and the

required quantities of energy can be bought and transmitted according to the awarded

transmission rights.

2.6. Appraisal and current developments

Although nodal pricing is already applied on several markets5, an implementation in Europe

within the next future is not realistic. One weakness of this method is the high degree of

complexity that may lead to problems concerning market transparency (Alaywan et al, 2004,

p. 1) and market liquidity (Ehrenmann & Smeers, 2004, p. 13). Regarding the EU region,

there is also the problem of political divergence. The different degrees of market liberalization

within the EU-states and the fact that there are many established system operators of which

each is responsible for one control area, constrain an implementation of nodal pricing in

Europe.

A study commissioned by the European Commission (Consentec and Frontier Economics,

2004, p. 90-97) examined different congestion management methods regarding three

benchmark criteria (Consentec and Frontier Economics, 2004, II):

• Compatibility to EU regulation 1228/2003,

• Economic welfare, and

• Practical feasibility.

The study recommends a stepwise implementation of hybrid auctions6 as congestion

management method. Furthermore, the introduction of flow-based explicit auctions is

considered to be a purposeful first step towards the right direction (Consentec & Frontier

Economics, 2004, IV).

4 Again disregarding transmission losses and assuming sufficient generation capacities. 5 For example in New Zealand and in the USA (PJM). 6 A hybrid auction allows explicit and implicit bids. That means that the from implicit bids resulting energy price differences between zones make up the benchmark whether to accept an explicit bid for transmission between two specific zones or not.

6

Parallel to the European Commission, regulators of the member states currently tend to

evaluate and adjust existing proposals for a market based congestion management. For this

purpose, discussion meetings have been holding once to twice a year since 1998 – the so-

called ‘Florence Fora’. In order to harmonize market structures on a regional basis, the

foundation of so-called regional ‘Mini Fora’ was decided at the ‘11th Florence Forum’ in

September 2004. In the course of this process, the EU-states were merged to seven regional

sub-markets in order to introduce regionally coordinated allocation systems. Accordingly,

Austria was assigned to the ‚Central East Europe’ (CEE) and to the ‚Northern Borders of

Italy’ (NBoI) Mini Forum. Within the CEE Mini Forum not flow-based coordinated auctions

were implemented for Poland, the Czech Republic and Germany/Vattenfall already in 2005. It

is planned to enlarge this auction to the Slovak Republic and Germany/E.ON in 2006. A

participation of Verbund APG – the regulated Austrian system operator – was also discussed.

However, it was not supported by the Austrian regulator E-Control due to the fact that these

auction are not flow-based and, therefore, do not price physical but commercial congestions.

The allocation of capacity takes place through a market based procedure but it is not fair

according to the input involved. The actual resource utilization is not economically adequate

reflected. Hence, an efficient resource utilization is not guaranteed.

3. Flow-based coordinated auctions within the SEE-region: The

pilot project

Alternatively to a not flow-based approach, a pilot project regarding flow-based coordinated

auctions for South-East-Europe (SEE) was initiated within the Organization of South-East-

European System Operators (SETSO). The pilot project evaluates the feasibility of flow-

based coordinated auctions – also considering the proposals presented by ETSO (2001, p. 22).

The model (Figure 1) embraces eight zones that are interconnected by 13 borders (=

aggregated lines).

7

Figure 1: SEE region

3.1. Flow-based modeling

A flow-based analysis includes physical flows resulting from commercial bids. On this basis,

the respective line usages are calculated (compare Section 2.1). Within the scope of the

aforementioned pilot project, the network is represented by so-called PTDF7 matrices R

which are based on a DC load flow model (compare Schweppe et al., 1988; Stigler and

Todem, 2005). Thereby, only real power is modeled. The coefficients of the matrix represent

technical parameters that define which flow Ld over a border line l result from an input g at

zone i in relation to a reference zone (hub):

Ldl = gi * Ri,l (1)

7 PTDF = power transfer distribution factor.

8

Basically, there are two possible notations for PTDF matrices: a zone-to-zone- and a zone-to-

hub-notation (ETSO, 2001, p. 9). Both notations provide the same information. Within the

zone-to-zone-matrix, for each pair of zones – i.e. input at a zone C and withdrawal at a

different zone A – and for each border line is stated how the resulting flows split up on the

respective border lines. The elements of the zone-to-hub-matrix constitute relative values

referring to a hub.

3.2. Determination of the PTDF matrices

Considering the pilot project, monthly auctions are implemented for the time being. Hence,

PTDF matrices have to be calculated in advance and are presumed to be constant for the

allocation period. Consequently, these matrices can only be considered to be a best possible

approximation for each allocation period. In order to take into account trading that took place

already, the following procedure for determining the PTDF matrices is applied (compare

ETSO, 2005, p. 11):

• Increasing of generation in zone A while at the same time decreasing generation in

zone B by 100 MW.

• The resulting flows provide the PTDF values for combination A-B.

• Repeating this action for each pair of zones (both directions) yields the complete

PTDF-matix in zone-to-zone-notation.

In order to determine maximum transmission capacities, an empirical approach according to

ETSO (2001) is applied.

3.3. ‘With netting’ versus ‘without netting’

The term ‘netting’ refers to the handling of flows with different algebraic signs. There is the

possibility that due to a specific bidding structure resulting flows over one line have different

signs. Hence, they flow into different directions – against each other. Physically, the line is

then only loaded by the sum of all flows. Thus, flows with different signs cancel out. The

latter is called netting. If netting is considered to be not allowed – for example in long-term

auctions where netting seems to be problematic due to network security reasons – flows with

different signs are not netted out but considered separately. Both cases, with and without

9

netting, require different organizational designs. In case with netting, it must be secured that

the – through the auction allocated – capacities are actually entirely used or, respectively,

regulation intervention costs that incur due to the non-use of capacity are compensated.

Consequently, in this case transmission rights must be obligatory. In the case without netting,

congestions cannot occur due to the non-use of capacity. Hence, in this case transmission

rights can be optional. Our paper only considers the case with netting.

4. Mathematical formulation

4.1. Optimization problem

The mathematical formulation of the problem is geared to the a proposal of ETSO (2001).

The conducted analyses are based upon a model with Z zones that are interconnected by L

lines. Explicit demand for transmission capacity (quantity in MW) and respective willingness

to pay (price in €/MW) is represented by bids placed by bidders within the auction. Resulting

flows are calculated on basis of these bid quantities and the optimization problem can be

written as follows8:

)]};,,(*),,([max{ ,, byxdbyxp abidBYXΣ (1)

s. t. (2) );,,(),,( byxdbyxd bida ≤

(3) ;0),,( ≥byxda

).()]},,([*),,({, lBCbyxdlyxr aBYX ≤∑∑ (4)

Whereby: x = zone as source

y = zone as sink

b = bid within the auction

l = line within the network

),,( byxda = accepted quantity per bid

= required quantity per bid )

)

,,( byxdbid

= bid price per bid ,,( byxpbid

= element of the PTDF matrix ),,( lyxr

8 For an alternative formulation of the optimization problem see ETSO (2001, Appendix 3).

10

4.2. Revenue determination and allocation

The revenue calculated in equation (1) of the aforementioned optimization problem does not

yield the actual auction revenue as the market clearing price (MCP) has to be considered

instead of the respective bid prices.9 In accordance to the marginal pricing approach under

perfect competition, the scarcity price can be obtain as marginal price mp(l) per each line. The

marginal price equals the dual variable – or shadow price – for equation (4) of the

optimization. In case of an overstepping of the maximum allowable transmission capacity of a

border line, the accepted bid quantity has to be cut according to the following formalism:

pbid(x, y, b) / r(x, y, l) (5)

The price pa(x, y) that has to be paid for the transport from a source x to a source y and total

auction revenue are calculated as follows:

. (6) )],,(*)([),( lyxrlmpyxp La ∑=

Whereby: = accepted price from x to y ),( yxpa

)]},,([*),({, byxdyxpTAR aBaYX ∑∑= (7)

Whereby: TAR = total auction revenue

In order to allocate this revenue to the TSOs, there are several deceivable options. Thereby,

the incentive system of different methods has to be assessed carefully – beyond pure

economical (= mathematical) considerations. The analysis of incentive systems is one purpose

of this paper. ETSO (2001) proposed two allocation methods. The first one recommends an

allocation in equal shares to those TSOs that are source or sink of a bid, respectively (Method

1). The second method recommends an allocation in equal shares to those TSOs in-between a

congested line (Method 2).

9 This paper applies a marginal pricing approach. Also, a ‘pay-as-bid’ approach would be possible. In the latter case, the auction revenue would equal the revenue calculated in equation (1).

11

Method1:

)],,(*),([21)( ,1 byxdyxpxAR aaBY∑= (8)

)],,(*),([21)( ,1 byxdyxpyAR aaBX∑= (9)

Whereby: = auction revenue allocation according to Method 1 to TSOs )(1 xAR

that act as bid source

= auction revenue allocation according to Method 1 to TSOs )(1 yAR

that act as bid sink

Method2:

)]}([/)],(*)([{**21)(2 lmplxIMlmpTARxAR LL ∑∑= (10)

)]}([/)],(*)([{**21)(2 lmplyIMlmpTARyAR LL ∑∑= (11)

Whereby: = auction revenue allocation according to Method 2 to TSOs )(2 xAR

that act as source of a congested line

)(2 yAR = auction revenue allocation according to Method 2 to TSOs

that act as sink of a congested line

IM = incidence matrix of the network10

In addition to the methods proposed by ETSO, further methods were analyzed. Therefore, a

quotient (Ratio) was calculated on basis of line usages. The quotient can be related to the

maximum offered border line capacity (BC) or to the thermal limit (TL).11 One share of the

total auction revenue (TAR) is then allocated according to this quotient. Another share is

allocated according to Method2. The factor p defines the weight of the share of Method 2:

)(*)(**)1(*21)( 2 xARpxRatioTARpxAR +−= (12)

)(*)(**)1(*21)( 2 yARpyRatioTARpyAR +−= (13)

Whereby: = auction revenue allocation according to further methods to )(xAR

TSOs that act as source of a congested line

= auction revenue allocation according to further methods to )(yAR

10 The incidence matrices IM(x, l) and IM(y, l) described in equations (10) and (11) state which zone serves as source x or sink y of a line l. The coefficients of this matrices can only adopt values of one or zero, exclusively. 11 For a verbal formulation of the Ratio determination see Section 5.2.

12

TSOs that act as sink of a congested line

Different revenue allocation proportions per TSO are obtained by the means of varying

quotient Ratio and parameter p. Accordingly, different incentives signals to TSOs for fixing

their BCs are sent.

5. Data and basic assumptions

5.1. Network and bidding structure

The optimization problem outlined in Section 4.1 is solved by the optimization software

GAMS12 based on a constant PTDF matrix – related to the technical parameters of the

network (Table 1) – as well as on the given demand structure (Table 2).

Border Lines From To BC

forwardBC

reverse Thermal

Limit [MW] [MW] [MW]

L1 B A 350 -150 600 L2 B C 400 -420 600 L3 B E 300 -1200 1200 L4 B F 1300 -460 1500 L5 B G 1000 -600 1200 L6 B H 200 -300 600 L7 C A 190 -210 600 L8 C H 480 -270 600 L9 D A 100 -250 600 L10 E D 200 -600 600 L11 F D 500 -1200 1200 L12 G F 260 -900 1200

Table 1: Network

Bids that have the same bid prices – in relation to a line [compare equation (5)] – are cut in

accordance to the order of placement (‘first come, first served’-method; that means the

chronologically last received order is cut first).

12 Compare www.gams.com.

13

Bids From To Bid Quantity

Bid Price

[MW] [€/MW]13

Bid1 B C 200 1,5 Bid2 B C 400 1,5 Bid3 B D 200 3 Bid 4 D B 200 1,5 Bid 5 F D 200 1,75 Bid 6 F D 300 1,76 Bid 7 G D 280 1,6 Bid 8 C H 350 1,65

Table 2: Bidding structure

5.2. Individual profit maximization and scenarios

Starting from a base scenario, two variations are analysed (Table 3). Further BCs of selected

borders are – ceteris paribus – varied and the differences concerning revenue allocation to the

TSOs – responsible for the BC change – are analyzed. The basic assumption is that a TSO

aims to maximize its individual revenues and does have an economic incentive to change its

BCs such that individual revenue is maximized.

Scenario1 Base scenario BCL6 from 200 MW to 0 MW BCL6 from 200 MW to 100 MW

Scenario2: Varying BCL6

BCL6 from 200 MW to 600 MW BCL11 from 500 MW to 0 MW BCL11 from 500 MW to 200 MW BCL11 from 500 MW to 400 MW BCL11 from 500 MW to 800 MW

Scenario3: Varying BCL11

BCL11 from 500 MW to 1200 MW Table 3: Scenarios

As already mentioned, the different variations are compared in terms of change of revenue

allocation to the single TSOs. Thus, the different methods of revenue allocation shall be

shortly described verbally:

13 These prices refer to hourly values.

14

Method3L: Allocation according to: p times Method2 plus

(1-p) times “Line usage in respect to BCs weighted linearly“ Method3NL: Allocation according to: p times Method2 plus

(1-p) mal “Line usage in respect to BCs weighted quadratically“

Method4L: Allocation according to: p times Method2 plus

(1-p) times “Linearly weighted line usage in respect to BCs, to TLs, and to capacity disposal BC/TL“

Method4NL: Allocation according to: p times Method2 plus

(1-p) times “Quadratically weighted line usage in respect to BCs, to TLs, and to capacity disposal BC/TL“

Method5L: Allocation according to: p times Method2 plus

(1-p) mal “Line usage in respect to TLs weighted linearly“ Method5NL: Allocation according to: p times Method2 plus

(1-p) mal “Line usage in respect to TLs weighted quadratically“

5.3. Information and incentive issues

Due to existing asymmetric information, a TSO is able to stipulate BCs independent from

other TSOs and is able to change them for network security reasons. Hence, the BC becomes

a variable which is in contrast to the thermal limit of a line that is derived as a physically

given parameter. Aiming for an efficient resource allocation, it is economically necessary to

put as much capacity at the disposal as possible – by also taking into account security of

supply issues. Consequently, a revenue allocation method is favourable if it stimulates TSOs

– through respective individual revenue allocation – such that they set their BCs close to or

equal to the thermal limit, respectively. Technical necessities can likewise be incorporated by

adjusting physical thermal limits by a (n-1)-security margin. Thus, economical and technical

goals can be conformed quite easily.

15

6. Results and analysis

6.1. Scenario 1: Base scenario

Regarding the base scenario and under the given bidding structure, it can be stated that a

congestion between zones E and D occurs (L10 in Table 4).

Border Lines

From To Load Flow

Dual Variable

[MW] L1 B A 56.68 0 L2 B C 226.76 0 L3 B E 200 0 L4 B F -105.19 0 L5 B G -65.46 0 L6 B H 112.89 0 L7 C A 13.77 0 L8 C H -37.01 0 L9 D A -70.45 0 L10 E D 200 5.03 L11 F D 429.24 0 L12 G F 34.43 0

Table 4: Line flows base scenario

For there is a congested line, the dual variable (shadow price) for this line greater to zero

(Table 4). Hence, accepted prices for all bids are calculated according to their impact on this

line (Table 5) – via the respective PTDF coefficient.

Bids From To Accepted Quantity

Accepted Price

[MW] [€/MW] Bid1 B C 200 -0.08 Bid2 B C 400 -0.08 Bid3 B D 200 1.89 Bid4 D B 200 -1.89 Bid5 F D 200 1.43 Bid6 F D 300 1.43 Bid7 G D 199.69 1.6 Bid8 C H 350 0.05

Table 5: Auction results base scenario

16

Regarding revenue allocation, it must be stated that we observe negative revenues (=

payments) for Method1. This can be explained by the fact that this paper considers the case

with netting. The interpretation is as follows: Bidder whose bids result in a counter flow

concerning congested lines help to release a congestion or, respectively, at least enlarge

available border capacities. Hence, bidders are paid for their bids as long as the benefit from

releasing a congestion is greater than these payments.14 However, in the scope of Method1,

those TSOs receive payments that act as source or sink of a bid. Here, it is possible that a

TSO has to make payments to bidders in order to release a congestion on a line in which it

does not have a stake (Table 6). Being aware that each of the TSO is a stand-alone company,

the aforementioned fact creates an external effect and cannot be favourable as the revenue

allocation is not adequate. Also, it is further down shown that in the scope of Method1, TSOs

have an incentive to reduce or, respectively, at least not to increase BCs in order to create

higher revenues or, respectively, not losing them (Figure 3 and Figure 4).

Revenues [€]

Method1 Method2 Method3L MethodNL Method4L Method4NL Method5L Method5NL

A 0 0 27,17 13,09 26,24 11,72 31,53 17,82

B -22,64 0 120,93 185,57 132,37 188,72 127,81 195,34

C -13,83 0 40,87 17,33 67,66 60,81 62,1 47,7

D 516,98 503,14 364,21 424,05 359,59 411,48 360,13 408,87

E 0 503,14 339,29 308,56 297,42 265,53 318,71 284,51

F 357,23 0 64,19 32,31 85,89 57,35 61,3 30,74

G 159,75 0 12,71 1,2 15,79 2,91 11,17 1,14

H 8,81 0 36,92 24,18 21,33 7,75 33,54 20,17

Table 6: Revenue allocation base scenario

6.2. Scenario 2: Varying BCL6

The decisive goal is the identification of some influencing factors that help to generate

incentives for an efficient usages of network resources (compare Section 5.2). Basically, two

influencing factors were considered and analyzed in the scope of this paper: Share p of

Method2 and BC values.

14 Or of course as well: As long as counter flow producing bid quantities are available.

17

Thereby, it became evident that in case of varying factor p, the revenue change may differ in

absolute terms but the sign remained the same (Figure 215). Hence, in the first place varying

factor p can help to adjust the revenue change.

However, if the BC of border line L6 is redefined from 200 MW to 100 MW, one produces

ceteris paribus a new congested line. Consequently, both of the TSOs administering this line

receive higher payments (B and H) in all of the analyzed revenue allocation methods (Figure

2). The highest increase for both TSOs happens according to Method2 and the lowest

according to Method1.

0

100

200

300

400

500

600

B.1 B.2 B.3L B.4L B.5L H.1 H.2 H.3L H.4L H.5L

Revenue allocation methods

Cha

nge

in r

even

ues

[€]

Delta at 100 / p = 0.25 Delta at 100 / p = 0.5 Delta at 100 / p = 0.75

Figure 2: Scenario 2, variation of p in case of linear weighting and BCL6 = 100MW

6.3. Scenario 3: Varying BCL11

The results for the variation of p in scenario 2 are also valid for scenario 3. Thus, we leave

them aside for the time being and focus on different revenue allocation schemes. Comparing

those revenue allocation variants in which capacity usage and capacity disposal of each line

are weighted linearly (Method4L) with the quadratic weighted method (Method4NL), one

observation must be highlighted.

15 In order to interpret the figures, note that the abscissa shows considered zone (= TSO) and revenue allocation method. For example, B.1 represents revenue allocated to B according to Method1, H.2 represents revenue allocated to H according to Method2, and so on.

18

-800

-600

-400

-200

0

200

400

600

800

D.1 D.2 D.3L D.4L D.5L F.1 F.2 F.3L F.4L F.5L

Revenue allocation methods

Chan

ge in

reve

nues

[€]

Delta at 0 / p = 0.25 Delta at 200 / p = 0.25 Delta at 400 / p = 0.25 Delta at 800 / p = 0.25 Delta at 1200 / p = 0.25

Figure 3: Scenario 3 revenue allocation linear

TSO D denotes a higher revenue increase (Figure 3) in case the BC value approaches the TL

value (from 500 MW to 800 MW- and from 500 MW to 1200 MW-cases) than in case the BC

is reduced. There is no incentive to deliberately create an additional congestion – nor to shift

the congestion to L11, respectively (from 500 MW to 0 MW-, from 500 MW to 200 MW-,

and from 500 MW to 400 MW-cases).

-800

-600

-400

-200

0

200

400

600

800

D.1 D.2 D.3NL D.4NL D.5NL F.1 F.2 F.3NL F.4NL F.5NL

Revenue allocation methods

Chan

ge in

rev

enue

s [€

]

Delta at 0 / p = 0.25 Delta at 200 / p = 0.25 Delta at 400 / p = 0.25 Delta at 800 / p = 0.25 Delta at 1200 / p = 0.25

Figure 4: Scenario 3, revenue allocation quadratic

Whereas, the aforementioned incentive does not exist for TSO F. However, if one now

considers a quadratic weighting (Method4NL), the revenues for TSO F (Figure 4) in case of a

19

capacity provision close to the thermal limit (1200 MW case) tend to approach the revenue in

case of an artificial congestion (compare F.2 to F.4NL). In this case the revenue allocation

method provides an compatible incentive concerning an efficient usage of resources.

7. Conclusions

The analyzes shows that the revenue allocation methods for a flow-based explicit coordinated

auction proposed by ETSO (2001, p. 21f.) do not generate sufficient incentives in terms of

revenues to single TSOs in order to achieve a maximum provision of existing transfer cross-

border capacity to the market. This paper shows that the ETSO proposals rather give

incentives to restrict capacities and provoke congestions.

In order to mitigate such behavior, different revenue allocation schemes were analyzed

regarding their incentive signals to TSOs. Basic results show that those revenue allocation

schemes are favorable – from a system-wide view – that also reward line usage and capacity

provision to the market. An adequate design of a method according to the aforementioned

conclusions is in process.

20

References

Alaywan, Ziad, and Tong Wu (2004): Transitioning the California Market from a Zonal to a Nodal Framework: An Operational Perspective. In: Proceedings of the Power Systems Conference and Exposition, 2, 862-867. Retrieved November 3rd, 2005 from: http://www.caiso.com/docs/09003a6080/32/2c/09003a6080322c6d.pdf

Christie, Richard D., Wollenberg, Bruce F., and Ivar Wangensteen (2000): Transmission Management in the Deregulated Environment. In: Proceedings of the IEEE, 88, 170-195.

CONSENTEC and Frontier Economics (2004): Analysis of Cross-Border Congestion Management Methods for the EU Internal Electricity Market. Study commissioned by the European Commission, Directorate-General Energy and Transport, Final Report.

Ehrenmann, Andreas, and Yves Smeers (2005): Inefficiencies in European Congestion Management Proposals. Retrieved November 3rd, 2005 from: http://www.ksg.harvard.edu/hepg/Papers/Ehrenmann.Smeers_Inefficiences.European.Congestion.Mgmt.pdf

ETSO (2001): Co-ordinated Auctioning, A market-based method for transmission capacity allocation in meshed networks. European Transmission System Operators, Final Report.

ETSO (2005): Coordinated auctions in SEE region, Activity report. European Transmission System Operators, SETSO/SUDEL Sub working Group: Network Access, Congestion Management and Power Flows.

EU (2003): Regulation (EC) No 1228/2003 of the European Parliament and of the council of 26 June 2003 on conditions for access to the network for cross-border exchanges in electricity. In: Official Journal of the European Union, L 176/1. Retrieved November 16th, 2005 from: http://europa.eu.int/eur-lex/pri/en/oj/dat/2003/l_176/l_17620030715en00010010.pdf

Hsu, Michael (1997): An Introduction to the Pricing of Electric Power Transmission. In: Utilities Policy, 6, 257-270.

Schweppe, Fred C., Caramanis, Michael C., Tabors, Richard D., and Roger E. Bohn (1988): Spot Pricing of Electricity. Boston, Kluwer.

Stigler, Heinz, and Christian Todem (2005): Optimization of the Austrian Electricity Sector (Control Zone of VERBUND APG) under the Constraints of Network Capacities by Nodal Pricing. In: Central European Journal of Operations Research, 13, 105-125.

Stoft, Steven (2002): Power System Economics: Designing Markets for Electricity. Piscataway, NJ, IEEE Press, Wiley-Interscience.

21