Congestion Management in Deregulated Electricity Markets

K.L. Lo Y.S. Yuen L. A. Snider

k.lo@eee.strath.ac.uk cherry.yuen@strath.ac.uk eesnider@polyu.edu.hk
Dept Electronic and Electrical Engineering Department of Electrical Engineering
University of Strathclyde Hong Kong Polytechnic University
Glasgow, Scotland Hong Kong

Abstract: This paper evaluates congestion management in is essential in the calculation of optimum power flow: how
different deregulated electricity industries. Based on existing optimal the solution is depends on how close the bid curves
methods, it explains and illustrates how congestion is managed are to the true cost functions. Congestion is embedded in
under privatisation. A numerical example is utilised to manifest the uniform price by the addition of an ‘uplift charge’.
the principle involved. Their effects and implications on social
optimum and economic efficiency are also discussed. ‘Uplift’ composes unscheduled availability payments and
the cost from differences between forecast and actual
Keywords: congestion management, privatisation, energy demand. It is added to the Pool Purchase Price to work out
markets, social optimum. the final Pool Selling Price for settlement purposes for the
customers on the demand side. However, generators are not
1. Introduction charged for congestion explicitly and therefore this model
does not provide correct signals for the location of new
The success of privatisation of industries such as transmission facilities and power plants.
airline, telecommunications, etc. motivated the
deregulation and re-structuring of the electricity industry. In some parts of the U.S, e.g. Califomia, nodal pricing
Unlike many other commodities, electricity cannot be is adapted as the pricing method. The locational price
stored easily, and the transportation of electricity is differentials result in a net income or surplus for the system
constrained by physical laws which have to be satisfied at operator. The surplus from nodal price differentials can be
all times in order to maintain the reliability and security of used to pay the holders of ‘contract rights’ which are
the power system. The presence of constraints generally designed to hedge the risk caused by congestion to the
leads to higher marginal costs and reduced revenues. market participants. A contract right gives the right holder
the right to inject power at one node and remove it at
The transmission network presents its own set of another in the transmission network. The right holder is
constraints, mostly related to overloading, or congestion. paid compensation when the transaction is not feasible
Congestion management deals with the relief of congested physically. The main advantage of this model is that it
transmission networks in modem deregulated electricity gives correct signals for the siting of new transmission
markets. In the deregulated environment, generation, lines and power plants. However, as shown in the
transmission and generation fall into different legal entities. following section, it encourages strategic gaming behaviour
Energy market participants responsible for congestion of participants who seek to increase their revenue. This can
should be required to pay a premium and a congestion be detrimental to the achievement of a social optimum.
charging mechanism has to be devised in order to assign
the premium in a discreet manner. In the Nordic spot market, participants who help
alleviate congestion are credited the congestion charge
II. Background while those who worsen congestion are debited the charge.
The premium, so called Capacity Fee is calculated
In this paper, the charging methodologies employed in according to supply and demand characteristics of the areas
the U.K., California, Nordic countries (Norway, Finland affected by the bottleneck.
and Sweden) are evaluated and illustrated. These places are
chosen because their electricity industry has been In the multilateral trade paradigm [l], participants are
extensively privatised. In addition, multilateral trade is free to arrange trades among themselves without the
chosen to show how an alternative theoretical approach can interference of the system operator. The system operator
be applied to congestion management. only curtails trades when a line constraint is violated. This
is done by using the information obtained. from the
In a non-discriminatory auction, such as the U.K. pool, injection-to-flow sensitivity matrix. Additional trades need
all bidders are paid the same. The bidders have the to be arranged according to this information announced by
incentives to reveal their true costs, and therefore the bid the system operator.
curves can be regarded as close to true cost functions. This
The upcoming section serves to manifest these methods
in detail and in each section a numerical example is
included.
Paper accepted for presentation at the International
Conference on Electric Utility Deregulation and 111. Methods for Congestion Management
Restructuring and Power Technologies 2000,
City University, London, 4-7 April 2000. A. England & Wales Pool (E&W Pool)
02000 IEEE.
0-7803-5902-XJ00/%10.00 1. Uplift Cost
47
 In the E&W Pool, the cost of transmission constraints is
embedded in the Uplift cost. Uplift cost is added to the
Pool Purchase Price (PPP) to work out the Pool Selling
Price (PSP). Uplift is composed of different components
and they are:
Figure 1. Example.Network for E&W congdstion management
Transmission Services Uplift WMI
0 Energy Uplift
0 Reactive Uplift
Unscheduled Availability Payments

In general these costs account for the difference

be'tween the cost of 'on the day' operation and the costs
associated with the Unconstrained Schedule'. Energy
Uplift (EU) represents the costs of demand forecast errors
and generator shortfalls and redeclarations assuming 190 MHI
perfect foresight and ignoring the impact of transmission Figure 2. Equilibrium Point without Line Constraints
constraints. Reactive Uplift (RU) is the cost required to
maintain system voltage within limits prescribed in the
Grid Code. Unscheduled Availability Payments represent
the capacity payment paid to gensets that are available but

9-p
are not required to run while Transmission Services Uplift
(TSU) is the cost imposed by the physical limitations of the 40MW I SOMW
network. The exact calculation of the different components Figure 3. Optimal Dispatch without Line Constraints
can be found in [2]. TSU arises because of necessity of
activities such as re-scheduling of plant, the holding and 140MW

utilisation of frequency response capability and the holding

and utilisation of reserve.

2. Problems arising from Uplift Incentive Schemes 40MW IJOMW

Figure 4. Final Dispatch with Line Constraints.
National Grid Company (NGC), being the monopoly
over transmission, takes the roles of System Operator (SO)
and Transmission Operator (TO) and is responsible for
maintaining system security at all time. Since Uplift costs G1
are directly influenced by the action of NGC, there exist
different incentive schemes to encourage NGC to keep
these costs as low as possible. There has been significant
decrease in the total Uplift costs over the last six years2.As
a matter of fact, due to the possible loopholes of the
existing incentive schemes [3] and the implementation of
NETA3 in autumn 2000, it is likely that the existing
incentive schemes will be amended or replaced.

3. Numerical Example
The network as shown in Figure 1 with 2 generators Mw
and 2 loads are used to illustrate the congestion Figure 5.
management of the E&W Pool. G1 and G2 have marginal
cost functions MC(PGI)= 3 + 0.02PGI and MC(F'G2) = 8 +
0.07PG2 respectively and their maximum generation is
200MW. The perfectly inelastic loads L1 and L2 are
40MW and 15OMW respectively.

' Unconstrained Schedule is defined as the half hour by half hour schedule
of generating units notionally required to meet forecast demand and
reserve, which is produced in the day ahead of hading, ignoring
transmission constraints.
The six-year period refers to 1993 - 1999.
NETA, New Electricity Trading Arrangements for England and Wales;
are based on proposals published by OFFER, July 1998.

Figure 6.

Without line constraints, the system is optimised at

equilibrium point (6.8E/MWh, 19OMW) which sets the

48
 System Marginal Price (SMP). The generators’ bids Nodal pricing results in net merchandising surplus
fimctions are shown in Figure 2. G1 becomes the marginal during congestion. In the pool system, allowing the pool
generator and G2 becomes the out-of-merit generator. The operator to retain the surplus will create perverse
ideal unconstrained dispatch is shown in Figure 3. incentives. It is therefore proposed that the surplus should
be used to pay off holders of Transmission Congestion
Now consider the constraint of lOOMW over each line. Contracts (TCCs)/ contract rights. As a matter of fact,
The original dispatch violates the security of the system nodal pricing gives rise to potential strategic behaviour on
and therefore re-dispatch is necessary. The resulting part of generators which can eliminate nodal price
dispatch is shown in Figure 4. The Pool Purchase price differences 151. These generators do not tend to mitigate
(PPP) and Pool Selling Price (PSP) are calculated congestion since this could increase their revenue. It is
according to the following formulae: because all consumption at a given node is priced the same:
PPP = SMP + Capacity Payment with the highest nodal price chosen.
PSP = PPP + Uplifts
Capacity Payment provides the incentives for generators to 1. Contract Rights
maintain adequate margin over the level of demand. Its A contract right gives the right holder the right to inject
value depends on the Loss Of Load Probability (LOLP). In power at one node and remove it at another in the
this case, it is assumed that LOLP is zero and therefore PPP transmission network. The right holder is paid
is equal to S M P . Uplifts represent the cost of security and compensation when the transaction is not feasible
its value is the difference of the total cost of supply in.the physically. He is also reimbursed for any congestion
. unconstrained and constrained cases and they are shared charges he has to pay under nodal pricing. In both cases,
proportionally by both L1 and L2. The costs of supply for the amount he will get is the congestion charge, or the
the different scenarios are shown in Figures 2, 5 and 6. “rent”. Contract rights are designed for participants who
Because of the change of dispatch, demand charges and are indifferent to the difference between physical delivery
generator payments according to SMP do not balance each and financial compensation. Therefore the right is only
other anymore. Their difference accounts for the necessity useful for financial hedging of the price risk due to network
of Uplifts to balance the payments in and out of the Pool. congestion, but not for guaranteeing physical delivery.
Adjustments are therefore required to make up for the
difference and they are revealed as the areas under the bid 2. Numerical Example
curves as marked in Figures 5 and 6 . Different payments Using the network shown previously, the utilisation of
and charges are listed in Table 1. nodal pricing for congestion management is illustrated. The
network with the net flow and nodal prices is shown in
Demand Payments
Demand Charges (MI)
Total Charge ( E h )
Generator Payments
Generating Costs (Light-shaded
Areas) (fh)
Generating Payments from Pool
(Output*PPP) (fh)
L1

G1
308

616
I LZ
I
I G2
I
1156
1464

481
Table 2.

9-p
Figure 7. Different payments and Charges are listed in
Ncds 1:
f 5.8/Mwh
140MW
NcdC2
f I1.5IMwh

Adjustments (Dark-shaded Areas) 40MW 150MW

(fh) Figure 7 Network with Nodal Prices
Total Payment (Sum of Generating
I Payments and Adjustments) (fh) Node 1 Node 2
Table 1 Payments and Charges
Nodal Price ( M W h ) 5.8 11.5
Load (MW) 40 150
B. Nodal Pricing Demand Charges (fh) 232 1725
Generation (MW) 140 50
Nodal pricing is used for congestion management in Generation Paments 812 515
California. Its variations are adapted in different parts in Congestion Surplus
= Total Demand Charges - Total Generation Payments
the United States such as New York, Pennsylvania, New = 570 fnl
Jersey and Maryland. It is based on the theory of spot 1
Table 2
I

pricing [4]. Nodal prices are typically calculated as dual

variables or Langrange multipliers from Optimal Power Now consider the same example with a Contract of
Flow (OPF) calculations performed to compute the optimal Difference at a certain price of lOOMW between G1 and L2
dispatch. In the absence of binding inequality constraints, and at the same time G1 and L2 hold the capacity right to
all nodal prices, or Lagragian multipliers, are equal to A. transfer l O O M W from node 1 to node 2. With the difference
When there are binding constraints, the nodal price at the of nodal prices of the two locations, they are entitled to
reference bus is equal to h while the nodal prices at all have the compensation of lOOMW*(11.5-5.8) = 570Uh. In
other buses vary according to sensitivities of the network this case the congestion surplus is used to reimburse G1 and
flow constraints and the relative contribution of power L2 for the difference in nodal prices due to congestion.
injections at those buses to the congested lines. Since these
prices vary according to the geographic locations, nodal
prices are also called locational-based marginal prices
(LBMPs).

49
 3. Possibilities of Gaming in the Elspot Market, e.g. Sweden and Norway. For any
In order to illustrate how generators can game with intemal bottlenecks occurring inside individual areas, they
nodal pricing. An additional generator G3 is added to the are handled differently depending on the area itself.
same network shown in Figure 8. G3 has the highest bid
among the three generators and its bidding cost function is 1. Pricing Mechanism
equal to 12 + 0.08PG. At first a system price (PJ is calculated, based on
participants' overall bids and offers from all areas. This
Node I:
f 5.8MWh
Node 2:
PIZ.b/MWh
price does not take transmission constraints into
consideration. If calculation shows that power flow
q14OMW
- p 4 o M w
between bidding areas exceeds line limits, area prices (Po)
G3 IOMW
are calculated. The difference of Ps and Po is the capacity
fee. In the surplus area, the area price (PI) is found by the
right shifting of its demand curve by an amount equal to
40MW ISOMW
Figure 8 Higher Nodal Price resulting from Gaming the line capacity while in the deficit area, the area price (Ph)
is found by the right shifting of its supply curve by the
Under normal circumstances, G3 will not be dispatched same amount.
as shown previously. But assume that G2 is aware that G3
is a more expensive generator. It could declare less capacity In economics terms, the area price in the surplus area is
in order that G3 needs to be dispatched to serve the required set up in such a way that it should stimulate an extra
quantity of demand. If G2 declares a maximum capacity of demand which has the quantity equal to the capacity of the
40MW rather than 200MW, G2 will operate at 10.8fNWh constrained line. On the other hand, in the deficit area, the
while G3 will operate at 12.8fMWh. Under nodal pricing, area price is set up so that suppliers are encouraged to
at the same node, the highest nodal price is chosen and in supply an additional amount equal to the capacity of the
this case, 12.8EMWh is taken as the nodal price at node 2. line. The following example serves to illustrate the
Assume that the generators are bidding at their marginal mechanism.
cost curves, the exact cost of generation and profit can be
worked out. In this situation, G2 can increase its net profit 2. Numerical Example
by more than 50%. The network as shown in Figure 9 is used to illustrate
the pricing mechanism of the Nordic spot market. The two
Another gaming possibility exists for G1. The quantity generators have bidding fimctions MC(PGI) = 3 + 0.02P~1
of power required fiom G1 is determined solely by the line and MC(Pc2) = 8 + 0.07P~2respectively as mentioned
limit, as long as the bidding price of G1 is lower than G2. before and the loads L1 and L2 have cost functions UL,(PL~)
Therefore, G1 do not have to bid at its marginal cost = 7 - 0.03PLI and uL2(PL2) = 15 - 0.05PL2respectively.
function in order to get the maximum profit. Instead, it can Figure 9 shows the unconstrained dispatch and Figure 10
bid as high as G2 does. In this case, congestion surplus will shows the constrained dispatch.
be reduced, to the theoretical minimum of zero. Under this
scenario, the system does not operate at the optimal point At first the system price Ps is calculated by finding the
anymore and the profit allocation between generators and interception of the demand and supply curves formed by all
. the system operator deviates from the one as defined in spot the bids supplied by both areas. This is shown in Figure 11.
pricing theory [4]. The system price is the price at the equilibrium point,
which is equal to 6.52€/MWh. However, line transmission
C . Nord Pool capacity is exceeded and therefore the area prices for Area
1 and Area 2 have to be found. The principle is revealed
The Nordic spot market, also called The Elspot Market, graphically in Figures 12 and 13. As shown in Figure 12, in
in the Nord Pool is one of the products provided by The the surplus area, which is also the low price area, the area
Nordic Power Exchange. Similar to the E&W Pool, this is price is found by shifting the purchase curve by IOOMW
where electric power is traded on daily basis for delivery in (the capacity of the constrained line) to the right (from Do
the following day. Also, the System Price is set up in the to D1) and finding its new intercept with the sale curve.
same fashion: it is the price where demand curve meets the Similarly, Figure 13 shows the calculation of the area price
supply curve. But rather than adapting half-an-hour bids as in the high price area by shifting the sale curve to the right
in the U.K. Pool, bids are placed for every hour during the by 1OOMW (fiom So to SI). Various prices are revealed in
day. Figure 14 and Table 3 . The settlement price for all
participants is the system price and participants are charged
In case of transmission congestion, the system operator for the use of the transmission system. L1 and G2 will be
in Sweden, Norway and Finland use the price mechanism credited the Capacity Fee since they help alleviate the
in the Spot Market to adjust power flow in times of bottleneck while G1 and L2 will be debited the Capacity
bottlenecks between bidding areas. This will result in a Fee since they exacerbate the situation. This is reflected in
reduction of price in surplus areas and an increase of price the area price. Also, because of the physical flow of
in deficit areas until the transmission need falls down to the lOOMW on the constrained line, there is always a revenue
capacity level. Market participants incur additional cost and of 100MW*(Ph - PI) shared by the grid companies at the
this charge is called Capacity Fee [ 6 ] . This congestion borders of the two areas. Table 3 shows how the difference
management is only applied to the different bidding areas

50
 4-p
of the total charge credited to L1 and G2 and the total D. Multilateral Trades
charge debited to G1 and L2 can make up to this figure.
A multilateral trade is a generalisation of a bilateral
trade. This paradigm allows suppliers and consumers
176MW primarily to seek profit on their own while the IS0
guarantees security [l]. In fact, it is shown that it is
essential to have coordinated trades involving three or more
parties to relieve congestion or to ensure security of
16MW I&MW
Figure 9 Ideal Dispatch without Constraints
operation. This model has minimal role for the IS0 and
IS0 has no data on costs or financial arrangements. The
Surplus Arcr Dcficit Arm duties include verifying feasibility of trades 24 hours
ahead, dispatching and monitoring trades in real time and

o-tA--o
116MW
eliminating imbalances and charging commitment
violations. The trades are arranged by brokers. A broker
can be a generator or consumer involved in the trade, but
16MW 160MW can also be an unrelated third party.
Figure 10 Dispatch with Binding Constraint
In this trading platform, congestion management is the
main duty of ISO. IS0 intervenes to make a decision on
curtailment only when security is threatened. To maintain
security, the IS0 passes on information, so called Loading
Vector, to the generators and loads based on which
generators and loads structure their trades so as not to cause
security problems. All security information is open to the
public since maintaining security is a community effort.
Suppliers and consumers themselves are responsible for
carrying out the economic fhction to make the decisions
conceming the price and the amount at which to buy and
sell.

The mechanism of Loading Vector is elaborated as

follows. Whenever trades are infeasible, the system
L1 operator uses power flow Jacobian matrix to work out the
Figure 11 System Price Calculation Loading Vector. The amount of curtailment is calculated
by the sensitivity information provided by the Loading
Gl.Ll:-aWJ Vector. The operator then works out the initial curtailed
trades and announces the results to all parties. The parties
are responsible for finding out profitable trades themselves
1208
9.17 based on the Loading Vector.

1. Numerical Example
Congestion management associated with multilateral
ms w trade is illustrated using the network shown in Figure 15.
Figure 13 Area Price for Area 2
G1 and G2 have the cost h c t i o n s MC(F'G1) = 3 + 0.02PG1

:;F,
and Mc(p~2)= 8 + o.07PG2respectively while the load is
250MW. Initially participants seek maximum profit among
themselves and a broker arranges two contracts:
(i) TA: GI-L: 200MW at E7IMWh
(ii) TB:G2-L: 50MW at €11.5/MWh
The prices are set at the corresponding marginal costs of
the two generators. In general, price is set through
negotiation among the participants. The contractual
Figure 14 Area Prices and System Price arrangement is shown in Figure 15. Now assume that the
Capacity Fee in Surplus Area, C, = Ps - PI=0.72 f/MWh subsequent physical flow on the line fiom G1 to the load
Capacity Fee in Deficit Area, Cd = Ph - Ps 2.65 f/MWh
exceeds the capacity limit of 1 O O M w . The system operator
then works out the Loading Vector for the constrained line
Settlement Price P, = 6.52 f/MWh
using the power flow Jacobian matrix. Assume that the
Charge Credited to L1 and G2 (Mc) +.PGt*Cd=
=PLI*~ 170.52 fh Loading Vector for the constraint is FA, 'A, 0). It reveals the
Charge Debited to GI and L2 (Md) = PGI*C,+ PU*cd = 507.52 fh flow sensitivity of the constrained line to nodal injection of
Net Income of Grid Company =&-Me the three nodes of the network, i.e. % of the power from G1
= Capacity * (PI, - PI)= 337 f/h plus % of the power from G2 will flow on the Line. In other

51
words, the flow on the line from G1 to the load is govemed complex, Loading Vector can reveal different possibilities
by the equation F~,-L=3/4P~1+1/4PG2. of making profitable trades, finding the best trade can
become a tedious and unmanageable process and therefore
In this example, there exist many different ways of it is unlikely that the optimal operating point will be
curtailing the trade(s) in order to keep the system operating achieved within a reasonable period.
in the security margin. If the system operator aims to
curtail the trade (being TA in this case) which impacts the In addition, there is a contradiction from the paradigm.
constraint the most, the curtailment of trade TA can be One of the main duties of IS0 in the multilateral trade
calculated according to the Loading Vector: platform is to work out curtailments in order to satisfy
% * A P G I = 162.5 - 99.5 z APGI = 84MW network constraints. However, in order that IS0 has more
At the same time, the system operator announces the choices of curtailing trades, he needs to have reliable cost
Loading Vector to help participantshroker arrange any data. This is contradictory to one of the main advantages of
additional trades. In this case, the Loading Vector this platform over the E&W Pool that it is more socially
represents a set of trading rules that TA and TB have to efficient for IS0 not to have any data on costs or financial
follow so that any additional trades will not violate the line arrangements.
limit. The resulting flow and the trading rules are shown in
Figure 16. Now L is 84MW short of the desired demand IV. Conclusions
after the initial curtailment, the broker will arrange trades
for L with both G1 and G2 in order to get the rest of the In this paper, congestion management using different
demand according to the trading rules announced by the methods is illustrated and evaluated. Numerical examples
system operator. The final trades are illustrated in Figure are employed to manifest the principles of different
17. methods. Different gaming possibilities or strategies are
also investigated under different situations to show how
system equilibrium can deviate from the social optimum.

V. Acknowledgement

This research is funded by John Swire & Sons Ltd

under their James Henry Scott Scholarship for Engirieering
Studies at the University of Strathclyde.
Tw Figure 15
Reference

Felix F. Wu, “Coordinated Multilateral Trades for

Q 116MW
Electric Power Networks”, 12” Power Systems
Computation Conference, Dresden, August 19 - 23,
1996.

for L to get one additional MW.

The National Grid Company plc, Transmission
L has to sell back IRMW lo GI
and buy 3RMW bin G2
Licence Condition 10: Statements of Charges for Use
of System and Connection to the System for the Year
T W 1999/2000,April 1999.
Figure 16
Office of Gas and Electricity Markets, NGC Incentive

9 74MW Q 176MW
schemes from April 2000: An initial consultation,
August 2000.

Fred C. Schweppe, Michael C. Caraminis, Richard D.

Final stas: Tablors and Roger E. Bohn, Spot Pricing of
TA = 74MW
fmm GI m L a1f4.48lMWh Electricity, Kluwer Academic Publishers, 1988.
1B = 176MW
fmn GZ lo L aI€Z0,32/MWh

.(250MW Harry Singh, Shangyou Hao and Alex Papalexopoulos,

Figure 17
“Transmission Congestion Management in
Competitive Electricity Markets”. IEEE Transactions
The difficulty of implementing this congestion on Power Systems, Vol. 13, No. 2, May 1998, pp. 672-
management is that the congestion charge is shouldered by 680.
all participants and therefore it does not give any signals
for locating or installing new transmission facilities. Also, “The Elspot Market, Version as of October 1, 1998”,
it does not panelise any participants who constantly cause Nord Pool, available to public at web page:
congestion to the network. The concern of fairness arises http:\www.nordpool.com.
about how curtailment of trades is done initially.
Furthermore, in reality, power system networks are

52