Allocation of transmission loss cost using game theory

Allocation of transmission loss cost using game theory

Yuri P. Molina, Ricardo B. Prada and Osvaldo R. Saavedra, Member, IEEE

Abstract This paper presents a novel procedure for the

allocation of active and reactive power losses among
generators and loads in the transmission grid. The
procedure considers the grid as a block, making it
unnecessary to analyze the elements within that block. The
method allocates the active and reactive power losses
jointly by using the circuit theory combined with the game
theory (Aumann-Shapley). Arbitrarily, half the losses are
allocated to the loads and half to the generators. The
method was tested on a 5-bus sample system and on the
IEEE 30-bus system. The results show that the method
provides a consistent assignment of responsibilities,
reflecting the generator and load locations in the grid and
in agreement with the circuit laws.1
Index Terms loss allocation, transmission, circuits,
game theory, active power, reactive power.

I.

INTRODUCTION

HE restructuring of electric power systems was

accompanied by a redefinition of the obligations, rules
and practices observed by the traditional industrial sector.
This process met with difficulties, successes and failures and
even now several areas are still being explored and developed.
The most complicated tasks include the creation of openaccess transmission grids and the allocation of costs among
participants in the different markets [1].
Loss allocation has become a major issue. The active and
reactive power loss problem is very complex because of its
nonlinearity and nonseparability features [2]. The active losses
in the grid represent from 5% to 10% of the total system
generation. Therefore, a fair allocation of system losses has a
very important impact on all the agents involved. On the other
hand, an unfair allocation may result in cross-subsidies.
Several loss allocation methodologies have been proposed
in the literature, focusing particularly on active losses. The
most important studies for transmission loss allocation to
agents in the pool and bilateral contract markets are briefly
reviewed below.
The pro-rata techniques is one of the most common
methods. In such procedures, the losses allocated to the
1
Y. P. Molina and R. B. Prada with Department of Electrical Engineering,
Catholic University, Rio de Janeiro, 22453 900, Brazil, prada@ele.puc-rio.br /
molina@ele.puc-rio.br and O. Saavedra Mendez are with the Power System
Group, Federal University of Maranho, So Lus, MA 65085 580, Brazil,
o.saavedra@ieee.org.
The first author received financial support from CNPq and CAPES.

generation or load buses are proportional to the power levels,

but their relative position within the grid is not considered.
Consequently, the remotely located generators and loads
benefit from this at the expense of the others [2]-[3].
Allocation based on incremental loss coefficients has the
following characteristics: i) it may be positive and negative, ii)
it depends on the choice of the slack generator, and iii) it
typically results in over-recovery of losses (when the
coefficients are applied directly) [2].
The methods based on the electrical circuit theory present a
different characteristic. The fact that they explore the
electrical circuit laws, which are simple and easy to
understand, makes them an interesting alternative [2]-[5].
The main problem with allocating losses to loads, to
generators or to bilateral contracts is that although
approximations are made, the final allocations always contain
a certain degree of arbitrariness. This occurs because system
transmission losses are a non-separable and nonlinear function
[3].
This paper presents a method for allocating active and
reactive power losses jointly by using the circuit theory in
combination with the game theory (Aumann-Shapley). The
method considers that half of the system losses are allocated
to the loads proportionally to their respective currents, while
the other half are allocated to the generators. The AumannShapley theory is used for determining the participation of
these agents.
The paper is organized as follows: In section II, the
cooperative game theory is reviewed and the Shapley Value
and Aumann-Shapley method are explained in sequence. In
section III, the proposed method is developed based on the
circuit and game theories. In section IV, the method is
validated with a 5-bus sample system and the IEEE 30-bus
system. Finally, section V contains relevant additional
comments and conclusions regarding the procedure and the
result of its application.
II. COOPERATIVE GAME THEORY
The Cooperative Game Theory is applicable to cost
allocation problems among participants that make use of the
same service and its objective is to obtain the fairest solution.
Two methods used in game theory for cost distribution are
presented below: the Shapley Value method and the AumannShapley method, whose development is more generic.

Allocation of transmission loss cost using game theory

A. Shapley Value
The Shapley value is an approach to the fair allocation of
gains obtained by cooperation among several agents. In order
to formalize this situation, the notion of coalition is used. Let a
group N (of players) and a function v : 2N R be called
"value function" with the following properties:

v() = 0
v( S T ) v( S ) + v(T )

(1)
(2)

If S and T are disjoint subsets of N, the interpretation of the

function v is: if S is a coalition of players which agree to
cooperate, then v(S) describes the total gain predicted from
this cooperation, independently of what the agents outside the
coalition of S may decide. The superadditive condition of (2)
expresses the fact that the coalition can only help, but never
hinder the other players. The Shapley value is a way in which
to distribute the total gains among the players, on the
assumption that they have all collaborated. It is a "fair"
distribution in the sense that it is the only type of distribution
that presents certain desirable properties which are presented
below. The quantity gained by the agent i is the average of the
gains obtained in the coalitions in which it participates and is
expressed by (3):

i (v) =

| S |!(n | S | 1)!
v( S {i}) v ( S )
n!
S N \{i}

(4)

This means that the incremental cost due to i is

approximately the same as the marginal cost. By means of
some mathematical conditions [7], it may be observed that the
average of all the marginal costs, when the value of bi
increases from zero up to its maximum value, converges to a
value known as the "Aumann-Shapley unit cost",
mathematically expressed as:

i =

C (tb)
dt
bi
t =0
1

(5)

Finally, the cost allocated to each user i is:

B. Aumann- Shapley
The Aumann-Shapley method is presented as a natural
result of the Shapley method. This method was created based
on the notion of "dividing" the resources of each agent into
infinitesimal increments and applying the Shapley method to
each one as if each increment represented an individual agent.
At first glance the computational difficulties would be even
greater than those of the Shapley method because the number
of combinations would increase considerably. The AumannShapley method makes it possible to find an analytical
solution to the cost distribution problem when the agents are
divided into infinitesimal parts. It is the only method that
satisfies the following axioms which are fundamental to a fair
allocation among agents:
it recovers the total service cost,
it is additive,
it is invariant with respect to scale changes,
it is monotonic.

C (b* , i ) C (b* ) C (b* )

i
bi b =b*

(3)

where n is the total number of players and the agents of all

the subsets S of N do not contain player i. The formula was
derived from the assumption that the coalition is being formed
by one player at a time, and each player demands its
contribution [v (S {i}) v (S)] as compensation for its
participation. A more detailed explanation is presented in [5].

For this reason, several methods based on this concept have

been developed for cost allocation problems in the new
economic model of the electric power sector [10]-[11]-[12].
The development of this allocation method is more generic
than that of the Shapley Value method, according to [8]-[9].
As has been said, this method resolves the problem by
dividing each agent into infinitesimal parts (i0). In this
manner, the effect on the function (cost, gain or participation)
of the new subagent i takes the form of:

Ci = bi i

(6)

where:
Ci : cost allocated to the agent i
bi : quantity corresponding to the agent i
i : Aumann-Shapley unit cost of the agent i
III. LOSS ALLOCATION METHOD
The operation point is known and the grid is considered as
a block (Figure 1). Consequently, it is not necessary to
calculate the losses in each line. The superposition principle is
used for determining the relation between the load currents
and the current sources (linear function). Therefore, the losses
(nonlinear function of the current sources) are allocated with
the use of the game theory, specifically, the Aumann-Shapley
method. In this method, half the system losses are allocated to
the loads proportionally to their currents, while the other half
are allocated to the generators.

transmission system

Figure 1: Representation of the Power System

Allocation of transmission loss cost using game theory

In order to apply the method, it is essential to model the

power system with its sources and current and the loads as
impedances [4].

Loss Load F 1
F 2
=
F 2 + F1

I Gka
I G ka
I Gka
F 1
F 3

F 3 + F1

I Gka
G ka

Power supply:

S Supp = I Z Bus I

(7)

F1 =

Power consumption:

Load j
j = NL

j =1

SCon =

j = NL

S
j =1

(8)

S Load j = ( I Load j )( I Load j ) Z Load j

Z Load

j = NL

S
j =1

(9)

i =1

Z Load

I Gi

Load j

The unit participation of the reactive currents which flow from

the generator Gk to the load j in the losses is given by:

U PGkr Load j =

(10)

t =0

LossLoad j (tI G )

(15)

I Gkr

where:

where:

Z ( Load , i )

Z ( Load , i )

I Load

F 2 = I * Z Bus * I ;

F3 =

The load current is a linear function of the current sources:

I ZLoad =

Load j

where S Load j is the apparent load power j given by:

i = NG

(14)

F 1
L o ss L o a d
=
F 2 + F1
I G kr
I G kr
F 1

F 3 + F1
I G kr

: impedance matrix element

: load impedance

F 2

I G kr
F 3

I G kr

(16)

The power system loss is given by:

S Supp SCon

(11)

Half of the total active and reactive losses are allocated to

the loads proportionally to their currents, as is shown in (12).

LossLoad j

1
Load j
S Supp SCon
= j = NL

2
I Load j

j =1

t=0

where:

Loss Load j ( tI G )
I G ka

UPGkLoad j = IGka UPGkaLoad j + IGkr UPGkrLoad j

(17)

IV. NUMERICAL RESULTS

(12)
In this paper, two systems are used for testing the allocation
method: a 5-bus system and the IEEE 30-bus system.

The same equation is used for allocating the remaining half

of the losses to the generators. With a view to determining the
loss associated with the current injections which flow from the
generators to the loads, the Aumann-Shapley theory [7] is
applied. The Unit Participation (UP) of the active currents
which flow from the generator Gk to the load j in the losses is
given by:
U PG ka Load j =

The total participation is the sum of the participations of the

active and reactive currents which flow from the generator Gk
at a specific load j:

(13)

A. 5-Bus Test System

This test system was extracted from [13] and its unifilar
diagram is shown in Figure 2, where part of the data is
presented. The system has two generators and three load
buses. The active and reactive line power flows are presented
at the extremes of the lines in MW and Mvar, respectively.
According to the method proposed, the generating agent is
divided into two subagents, the real agent (real component of
the generator current) and the imaginary agent (imaginary
component of the generator current).

Allocation of transmission loss cost using game theory

V=1.015(p.u)
6.8

V=1.01(p.u)
Real Power

31.2

17.6

20

Reactive Power

Bus 1

14.49
7.44

Bus 2

14.39

16.73
11.16
0.61

1.12

0.05+j0.25

24.5

9.69
5.38

0.09+j0.35
9.58
6.3

16.44

0.06+j0.25

0.25+j0.6
6.3

4.64
0.15+j0.5

16.44

0.1+j0.45

6.44

6.36

2.3

2.05

Bus 3

0.94

0.93

3.32

3.38

Bus 4

10+04j

Bus 5

15+10j

25+06j

Figure 2: 5-Bus Test System

The losses associated with the reactive currents which flow
from the generators to the loads are presented in Table 1. The
table may be interpreted as follows:
For example, when generators 1 and 2 serve load 3, active and
reactive losses are incurred. The loss attributed to the currents
which flow from generator 1 to load 3 corresponds to 0.0109
+0.0415i, half of which is allocated to each agent.
A similar interpretation is given to the losses associated
with the active currents which flow from the generators to the
loads, presented in Table 2. Finally, in Table 3 presents the
losses attributed to the complex currents which flow from the
generators to the loads.
Table 1: Losses associated with the reactive currents which
circulate from generators to loads
Generator 1
Load 3 0.0109 + 0.0415i
Load 4 0.0331 + 0.0633i
Load 5 0.0129 + 0.0983i
Total 0.0569 + 0.2031i

Generator 2
0.0289 + 0.1111i
0.0959 + 0.1797i
0.0420 + 0.2873i
0.1668 + 0.5781i

Total
0.0399 + 0.1526i
0.1290 + 0.2430i
0.0549 + 0.3856i
0.2238 + 0.7812i

In Table 1, it may be observed that the losses associated

with the reactive currents which flow from generator 2 to the
loads are relatively greater than those associated with the ones
which flow from generator 1 to the loads. This occurs because
generator 2 injects more reactive power, while load 4
consumes most of this reactive power, as can be seen in
Figure 2. Therefore, the loss associated with the reactive
currents which circulate from generator 2 to load 4 is the
greater one. This allocation is logical and expected in terms of
order of magnitude.
In Table 2, it may be observed that the losses associated
with the active currents which flow from the generators to
load 4 are the greater ones due to the remote location of this
bus, as is revealed by the line impedances in Figure 2. It may
also be observed that bus 5 has the highest active power load
but that the losses associated with this load are smaller than
those of bus 4 because the location of bus 5 is better in
relation to the generators (shorter electrical distance). The
impedance between generator 2 and bus 5 is small in
comparison with that of other lines. In this manner, since most
of load 5 is served by generator 2, the losses allocated to load
5 are smaller, even though it consumes more active power
than the others.
Another interesting aspect to be pointed out has to do with
the losses incurred by the currents which flow from generator
2 to load 3. It may be observed that the location of this load is
remote with respect to generator 2. This being the case, it is to
be expected that the losses associated with these two agents
will be minimal. Actually, Table 1, Table 2 and Table 3 prove
this to be a fact.
B. IEEE 30-Bus Test System
1

28

8
11

Generator 2

0.6941 + 2.6896i 0.3002 + 1.0028i 0.9943 + 3.6924i

Table 3: Losses associated with the complex currents which

circulate from generators to loads
Generator 1

Generator 2

14

19

4
13

17

16

20
10
26

23

25

21

22

27

29

24

30

Figure 3: Unifilar Diagram of the IEEE 30-Bus Test System

Total

Load 3 0.1278 + 0.5814i 0.0869 + 0.3105i 0.2146 + 0.8918i

Load 4 0.4514 + 0.9052i 0.2658 + 0.4909i 0.7171 + 1.3961i
Load 5 0.1718 + 1.4061i 0.1144 + 0.7796i 0.2863 + 2.1857i
Total

18

Total

Load 3 0.1169 + 0.5399i 0.0579 + 0.1994i 0.1748 + 0.7392i

Load 4 0.4183 + 0.8419i 0.1698 + 0.3112i 0.5881 + 1.1531i
Load 5 0.1589 + 1.3078i 0.0725 + 0.4923i 0.2314 + 1.8001i
Total

15

12

Table 2: Losses associated with the active currents which

circulate from generators to loads
Generator 1

0.7510 + 2.8927i 0.4671 + 1.5809i 1.2181 + 4.4736i

When the reactive currents of the generators circulate

through the transmission system, they do not necessarily
dissipate power (due to the inductive and capacitive
characteristics of the transmission system). The results in
Table 4 prove that the losses associated with the reactive
currents which flow from the generators to the loads dissipate

Allocation of transmission loss cost using game theory

active power (0.3734 MW), but provide reactive power
(9.4247 Mvar).
Table 4: Losses associated with the reactive currents which
circulate from generators to loads
load
buses
2
3
4
5
7
8
10
12
14
15
16
17
18
19
20
21
23
24
26
29
30
total

-0.0090 0.0455 + .1207i 0.3044i

-0.0013 0.0049 + .0131i 0.0333i
-0.0122 0.0359 + .0363i 0.0905i
-0.1115 0.3583 + .4554i 1.1326i
-0.0022 0.0252 + .1246i 0.3153i
0.0623 + -0.1282 0.1976i 0.5098i
-0.0023 0.0100 + .0289i 0.0729i
0.0104 + -0.0176 0.0622i 0.1591i
-0.0047 0.0162 + .0300i 0.0754i
-0.0045 0.0173 + .0406i 0.1022i
0.0011 + -0.0002 0.0186i 0.0472i
0.0083 + -0.0138 0.0502i 0.1283i
-0.0019 0.0072 + .0158i 0.0399i
-0.0024 0.0131 + .0485i 0.1226i
-0.0010 0.0041 + .0110i 0.0278i
0.0164 + -0.0274 0.0979i 0.2506i
0.0011 + -0.0003 0.0172i 0.0437i
0.0136 + -0.0270 0.0513i 0.1321i
0.0038 + -0.0065 0.0204i 0.0522i
-0.0005 0.0030 +0.0126i 0.0319i
-0.0108 0.0351 +0.0526i 0.1318i
-0.0472 0.3548 +1.5054i 3.8035i

generator buses
5
8
0.0008 0.1922i
0.0006 0.0209i
0.0159 0.0583i
0.2232 0.6726i
0.0048 0.1924i
-0.1106 0.3061i
0.0015 0.0456i
-0.0214 0.0969i
0.0050 0.0475i
0.0039 0.0642i
-0.0032 0.0291i
-0.0167 0.0782i
0.0019 0.0251i
0.0002 0.0765i
0.0007 0.0174i
-0.0328 0.1528i
-0.0030 0.0269i
-0.0251 0.0797i
-0.0073 0.0317i
-0.0001 0.0199i
0.0133 0.0834i
0.0515 2.3174i

-0.0017 0.2361i
0.0011 0.0255i
0.0205 0.0710i
0.1703 0.8915i
-0.0034 0.2397i
-0.1088 0.3753i
0.0031 0.0556i
-0.0240 0.1189i
0.0072 0.0579i
0.0063 0.0783i
-0.0032 0.0356i
-0.0184 0.0959i
0.0029 0.0306i
0.0021 0.0933i
0.0013 0.0212i
-0.0360 0.1873i
-0.0030 0.0329i
-0.0284 0.0979i
-0.0079 0.0389i
0.0005 0.0242i
0.0191 0.1011i
-0.0004 2.9088i

11

-0.0029 0.0798i
0.0002 0.0086i
0.0065 0.0240i
0.0494 0.3010i
-0.0029 0.0806i
-0.0448 0.1256i
0.0025 0.0173i
-0.0069 0.0385i
0.0029 0.0186i
0.0029 0.0251i
-0.0005 0.0114i
-0.0035 0.0307i
0.0014 0.0097i
0.0026 0.0295i
0.0009 0.0067i
-0.0065 0.0598i
-0.0005 0.0106i
-0.0074 0.0316i
-0.0023 0.0127i
0.0002 0.0079i
0.0062 0.0331i
-0.0025 0.9627i

0.0001 0.0768i
0.0005 0.0083i
0.0070 0.0231i
0.0564 0.2919i
-0.0015 0.0787i
-0.0426 0.1235i
0.0017 0.0175i
-0.0024 0.0369i
0.0044 0.0174i
0.0046 0.0237i
0.0000 0.0110i
-0.0041 0.0303i
0.0017 0.0094i
0.0027 0.0290i
0.0008 0.0066i
-0.0088 0.0595i
-0.0001 0.0103i
-0.0074 0.0312i
-0.0022 0.0125i
0.0002 0.0078i
0.0063 0.0325i
0.0174 0.9377i

total
0.0328 0.7687i
0.0059 0.0834i
0.0737 0.2305i
0.7461 2.8343i
0.0200 0.7820i
-0.3727 1.2426i
0.0164 0.1799i
-0.0618 0.3882i
0.0310 0.1868i
0.0305 0.2528i
-0.0060 0.1159i
-0.0483 0.3133i
0.0131 0.0987i
0.0182 0.3024i
0.0069 0.0687i
-0.0951 0.6120i
-0.0058 0.1072i
-0.0816 0.3212i
-0.0226 0.1276i
0.0033 0.0791i
0.0694 +
0.3292i
0.3734 9.4247i

Table 5 presents the losses associated with the active

currents which flow from the generators to the loads. It may
be observed that the losses associated with these currents
dissipate a total active and reactive power of 17.2638 MW and
18.4933 Mvar, respectively. The losses associated with the
active currents are the most representative ones.
Table 6 presents the losses associated with the complex
currents which flow from the generators to the loads. It may
be observed that the total losses are basically due to the active
currents which flow from the generators to the loads.
It is also possible to evaluate the generators and loads
individually through an analysis of the tables presented above.
For example, the greatest amount of loss is allocated to
generator 1 because it injects 261 MW into the system, while
generator 2, which injects 40 MW, has the second highest loss
allocation.
The other generators are synchronous compensators which
basically help to reduce the system losses. The reason such

5
generators participate in the active currents is that the power
supplied by them is transformed into current sources by means
of the bus voltage variables, and since the angle of the
voltages at these buses is different from zero, they have an
active component.
Table 5: Losses associated with the active currents which
circulate from generators to loads
load
buses
2
3
4
5
7
8
10
12
14
15
16
17
18
19
20
21
23
24
26
29
30
total

1.2561 +
1.2846i
0.1333 +
0.1438i
0.2826 +
0.4941i
3.8961 +
5.7798i
1.3534 +
1.2670i
2.8000 +
1.2884i
0.2952 +
0.3134i
0.7886 +
0.5050i
0.2834 +
0.3510i
0.4018 +
0.4544i
0.2159 +
0.1727i
0.6352 +
0.4090i
0.1552 +
0.1795i
0.5103 +
0.5106i
0.1116 +
0.1208i
1.2422 +
0.7968i
0.2000 +
0.1593i
0.7010 +
0.3624i
0.2621 +
0.1616i
0.1349 +
0.1308i
0.4734 +
0.6430i
16.1321
+15.527ii

0.0958 0.0797i
0.0107 0.0090i
0.0359 0.0183i
0.4164 0.2609i
0.0930 0.0922i
0.0974 0.1959i
0.0230 0.0204i
0.0378 0.0551i
0.0257 0.0194i
0.0333 0.0276i
0.0128 0.0150i
0.0305 0.0444i
0.0131 0.0107i
0.0375 0.0352i
0.0089 0.0077i
0.0594 0.0867i
0.0118 0.0139i
0.0274 0.0491i
0.0121 0.0182i
0.0096 0.0092i
0.0468 0.0319i
1.1391 1.1005i

generator buses
5
8
-0.0022
+ .1315i
-0.0005
+ .0143i
-0.0108
+ .0403i
-0.0803
+ .4880i
0.0067 +
0.1341i
0.0814 +
0.2128i
-0.0007
+0.0319i
0.0154 +
0.0677i
-0.0033
+ .0333i
-0.0024
+ .0449i
0.0024 +
0.0203i
0.0123 +
0.0546i
-0.0011
+ .0175i
0.0005 +
0.0535i
-0.0004
+ .0122i
0.0243 +
0.1066i
0.0023 +
0.0188i
0.0182 +
0.0555i
0.0054 +
0.0221i
0.0003 +
0.0139i
-0.0083
+ .0583i
0.0594 +
1.6322i

-0.0055
+0.0905i
-0.0008
+ .0097i
-0.0084
+0.0270i
-0.0780
+ .3455i
0.0004 +
0.0927i
0.0499 +
0.1414i
-0.0012
+ .0215i
0.0087 +
0.0461i
-0.0031
+ .0224i
-0.0028
+ .0303i
0.0011 +
0.0138i
0.0070 +
0.0371i
-0.0012
+ .0118i
-0.0010
+ .0361i
-0.0006
+ .0082i
0.0137 +
0.0724i
0.0011 +
0.0127i
0.0109 +
0.0378i
0.0031 +
0.0150i
-0.0001
+ .0093i
-0.0070
+ .0389i
-0.0137
+ .1203i

11

-0.0047
+ .0613i
-0.0006
+0.0066i
-0.0060
+ .0183i
-0.0576
+ .2341i
-0.0010
+ .0633i
0.0318 +
0.1000i
-0.0008
+ .0142i
0.0056 +
0.0311i
-0.0022
+ .0150i
-0.0020
+ .0203i
0.0007 +
0.0092i
0.0046 +
0.0246i
-0.0008
+0.0079i
-0.0007
+ .0240i
-0.0004
+ .0054i
0.0091 +
0.0479i
0.0007 +
0.0085i
0.0072 +
0.0253i
0.0020 +
0.0101i
-0.0002
+ .0063i
-0.0051
+ .0264i
-0.0203
+ .7601i

-0.0049
+ .0446i
-0.0006
+0.0048i
-0.0048
+ .0132i
-0.0479
+ .1696i
-0.0023
+ .0463i
0.0207 +
0.0741i
-0.0010
+ .0105i
0.0036 +
0.0217i
-0.0017
+ .0105i
-0.0017
+ .0143i
0.0004 +
0.0066i
0.0028 +
0.0182i
-0.0007
+ .0057i
-0.0010
+ .0174i
-0.0004
+ .0040i
0.0054 +
0.0357i
0.0003 +
0.0061i
0.0046 +
0.0186i
0.0012 +
0.0075i
-0.0003
+ .0046i
-0.0044
+ .0192i
-0.0327
+ .5532i

total
1.3346 +
1.5328i
0.1415 +
0.1703i
0.2885 +
0.5746i
4.0487 +
6.7562i
1.4502 +
1.5112i
3.0813 +
1.6210i
0.3144 +
0.3711i
0.8597 +
0.6166i
0.2988 +
0.4128i
0.4263 +
0.5366i
0.2333 +
0.2077i
0.6924 +
0.4991i
0.1644 +
0.2118i
0.5454 +
0.6063i
0.1188 +
0.1429i
1.3542 +
0.9727i
0.2162 +
0.1916i
0.7693 +
0.4505i
0.2860 +
0.1980i
0.1443 +
0.1557i
0.4954 +
0.7539i
17.2638
+18.4933i

V. CONCLUSION
This paper has presented a new way in which to allocate
active and reactive losses simultaneously. In addition to being
one of the few methods that performs this type of allocation
simultaneously, this new method makes it possible to analyze
the losses associated with active and reactive current
components flowing from the generators to each load.
Although final loss allocations involve a certain degree of
arbitrariness with regard to loads, no assumptions need to be
made when loss allocation is performed in generators.
The allocation obtained reflects the order of magnitude of
the sources and loads as well as their location in the grid. The
use of the Aumann-Shapley method should be explored as a
solution to be proposed for allocation problems because this
method is able to resolve problems which have

Allocation of transmission loss cost using game theory

nonseparability and nonlinearity properties, as is the case of
active and reactive power loss allocation.
Table 6: Losses associated with the complex currents which
circulate from generators to loads
load
buses
2
3
4
5
7
8
10
12
14
15
16
17
18
19
20
21
23
24
26
29
30
total

1.2471 +
1.4053i
0.1320 +
0.1569i
0.2705 +
0.5304i
3.7847 +
6.2352i
1.3512 +
1.3916i
2.8622 +
1.4860i
0.2929 +
0.3423i
0.7990 +
0.5672i
0.2786 +
0.3810i
0.3972 +
0.4950i
0.2170 +
0.1913i
0.6435 +
0.4591i
0.1532 +
0.1954i
0.5079 +
0.5591i
0.1107 +
0.1318i
1.2586 +
0.8947i
0.2011 +
0.1765i
0.7145 +
0.4137i
0.2659 +
0.1819i
0.1344 +
0.1434i
0.4627 +
0.6956i
16.0848
+17.033i

0.1413 0.3841i
0.0156 0.0422i
0.0719 0.1088i
0.7747 1.3935i
0.1182 0.4074i
-0.0308 0.7057i
0.0329 0.0932i
0.0202 0.2142i
0.0419 0.0948i
0.0506 0.1298i
0.0126 0.0622i
0.0167 0.1727i
0.0203 0.0506i
0.0506 0.1579i
0.0129 0.0355i
0.0321 0.3373i
0.0115 0.0576i
0.0004 0.1811i
0.0056 0.0704i
0.0126 0.0412i
0.0819 0.1637i
1.4938 4.9040i

generator buses
5
8
-0.0014 0.0608i
0.0001 0.0065i
0.0051 0.0180i
0.1429 0.1846i
0.0115 0.0582i
-0.0292 0.0932i
0.0009 0.0137i
-0.0060 0.0292i
0.0017 0.0143i
0.0015 0.0193i
-0.0008 0.0088i
-0.0044 0.0236i
0.0007 0.0075i
0.0006 0.0230i
0.0004 0.0052i
-0.0085 0.0462i
-0.0007 0.0081i
-0.0069 0.0242i
-0.0019 0.0097i
0.0002 0.0060i
0.0050 0.0251i
0.1108 0.6851i

-0.0072 0.1457i
0.0003 0.0158i
0.0121 0.0440i
0.0922 0.5460i
-0.0030 0.1470i
-0.0589 0.2339i
0.0018 0.0341i
-0.0153 0.0727i
0.0042 0.0355i
0.0036 0.0480i
-0.0021 0.0218i
-0.0115 0.0588i
0.0017 0.0188i
0.0011 0.0572i
0.0008 0.0130i
-0.0223 0.1149i
-0.0019 0.0202i
-0.0175 0.0601i
-0.0048 0.0240i
0.0004 0.0149i
0.0121 0.0622i
-0.0141 1.7885i

11

-0.0076 0.0185i
-0.0004 0.0020i
0.0005 0.0057i
-0.0082 0.0670i
-0.0039 0.0173i
-0.0129 0.0256i
0.0017 0.0031i
-0.0012 0.0074i
0.0007 0.0035i
0.0009 0.0048i
0.0003 0.0022i
0.0011 0.0061i
0.0006 0.0018i
0.0018 0.0055i
0.0005 0.0012i
0.0026 0.0119i
0.0002 0.0020i
-0.0002 0.0063i
-0.0003 0.0025i
0.0000 0.0016i
0.0011 0.0067i
-0.0229 0.2026i

-0.0048 0.0322i
-0.0002 0.0035i
0.0022 0.0099i
0.0085 0.1223i
-0.0038 0.0324i
-0.0218 0.0494i
0.0007 0.0070i
0.0013 0.0152i
0.0027 0.0068i
0.0029 0.0094i
0.0004 0.0044i
-0.0013 0.0121i
0.0010 0.0037i
0.0017 0.0115i
0.0004 0.0026i
-0.0034 0.0238i
0.0003 0.0041i
-0.0027 0.0126i
-0.0010 0.0050i
-0.0001 0.0032i
0.0019 0.0133i
-0.0153 0.3845i

total
1.3675 +
0.7641i
0.1474 +
0.0869i
0.3622 +
0.3440i
4.7948 +
3.9218i
1.4702 +
0.7292i
2.7085 +
0.3783i
0.3309 +
0.1912i
0.7979 +
0.2284i
0.3298 +
0.2260i
0.4568 +
0.2838i
0.2274 +
0.0918i
0.6441 +
0.1858i
0.1775 +
0.1130i
0.5637 +
0.3039i
0.1257 +
0.0742i
1.2591 +
0.3607i
0.2104 +
0.0844i
0.6876 +
0.1293i
0.2635 +
0.0704i
0.1476 +
0.0766i
0.5648 +
0.4247i
17.6372 +
9.0686i

REFERENCES
[1]

[2]
[3]

J. M. Zolezzi and H. Rudnick, Transmission cost allocation by

cooperative games and coalition formation, IEEE Trans. Power Syst.,
vol. 17, no. 4, pp. 10081015, Nov. 2002.
A. J. Conejo et al., Z-bus loss allocation, IEEE Trans. Power Syst.,
vol. 16, no. 1, pp. 105110, Feb. 2001.
A. J. Conejo, J. M. Arroyo, N. Alguacil, and A. L. Guijarro,
Transmission loss allocation: A comparison of different practical

[4]
[5]

[6]
[7]

[8]

[9]

[10]

[11]

[12]

[13]

algorithms, IEEE Trans. Power Syst., vol. 17, no. 4, pp. 571576, Nov.
2002.
W. D. Stevenson, Jr., Power system analysis McGraw-Hill, 1995.
C. Unsihuay and O. Saavedra. Transmission loss unbundling and
allocation under pool electricity markets IEEE Trans. Power Syst., vol.
21, no. 01, pp. 7784, Nov. 2006.
From Wikipedia, the free encyclopedia. Shapley value. In
http://en.wikipedia.org/wiki/Shapley_value. Accessed on Feb. 2007.
L. J. Billera, C. D. Heath, J. Raanan, Internal telephone billing rates - a
novel application of non-atomic game theory, Operations Research,
Vol. 26, No. 6, pp. 956-965, (Nov. - Dec., 1978).
J. L. Billera, D. C. Heath, and R. E. Verrecchia. A unique procedure for
allocating common costs from a production process. Journal of
Accounting Research, 19:185196, 1981.
J. L. Billera and D. C. Heath. Allocation of shared costs: a set of
axioms yielding a unique procedure. Mathematics of Operations
Research, 7:3239, 1982.
X. V. Filho, S. Granvile, M. V. F. Pereira, B. G. Gorenstin, J. C. Mello,
and A. C. G. Melo. Ancillary services: how to allocate costs in an
optimization based framework. CIGR Neptune Symposium,1997.
E. L. Silva, J. C. Morales, and A. C. G. Melo. Generation cost
allocation: a methodology based on optimal power flow and cooperative
game theory. In Proceedings of the IEEE Porto Power Tech, pages
312316, Porto-Portugal, 2001.
P. A. Kattuman, J. W. Bialek, and N. Abi-Samra. Electricity tracing and
cooperative game theory. In Proceedings of the Power System
Computation Conference, pages 238243, Sevilha, 2002.
W. Chen, C. Chen. And C. Liao Allocating the costs of reactive power
purchased in an ancillary service market by modified Y-bus matrix
method. IEEE Trans. Power Syst., vol. 19, no. 01, pp. 174179, Feb.
2004
Yuri P. Molina was born in 1978, obtained Diploma
in Electrical Engineering (2003) from the National
University of Imagineering , Lima, Peru. M.Sc in
Power System (2005) from Federal University of
Maranho. He is currently working toward the Ph.D.
degree in Electrical Engineering at the Catholic
University of Rio de Janeiro, Brazil.
Ricardo B. Prada was born in 1951, obtained
Diploma in Electrical Engineering (1975) and M.Sc.
in Power Systems (1977) both from Catholic
University of Rio de Janeiro, Brazil, Ph.D. in
Electrical Engineering (1980) from Imperial College
of Science and Technology, London, U.K. He is
Associate Professor at Department of Electrical
Engineering, Catholic University of Rio de Janeiro,
Rua Marqus de So Vicente, 225, Rio de Janeiro,
RJ, Brazil, phone: +55 21 31141214, fax: +55 21
31141232, e-mail: prada@ele.puc-rio.br
Osvaldo R. Saavedra (S88M94) received the
M.Sc. and Ph.D. degrees from the State University
of Campinas, Campinas, Brazil, in 1988 and 1993,
respectively. From 1983 to 1986, he was with
Inecom Engineers Ltd, Arica, Chile. From 1994 to
1997, he was a Visiting Lecturer at the Federal
University of Maranho, Maranho, Brazil. Since
1997, he has been a Professor in the Electrical
Engineering Department and the Head of the Power
System Group, Federal University of Maranho.