See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/3976437

Use based allocation methods for payment of electricity transmission systems

Conference Paper · February 2002

DOI: 10.1109/ICPST.2002.1047531 · Source: IEEE Xplore

CITATIONS READS
25 262

4 authors, including:

Hugh Rudnick J.M. Zolezzi

Pontificia Universidad Católica de Chile 13 PUBLICATIONS 392 CITATIONS
222 PUBLICATIONS 3,248 CITATIONS
SEE PROFILE
SEE PROFILE

David Watts
Pontificia Universidad Católica de Chile
37 PUBLICATIONS 335 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

CORFO project 13CEI2-21803, Research Grant CONICYT/FONDAP/15110019 and CONICYT/FONDECYT/1181136 View project

Study on the key technologies of new generation ADN in Xiong'an apecial area View project

All content following this page was uploaded by Hugh Rudnick on 04 June 2014.

The user has requested enhancement of the downloaded file.

 Use Based Allocation Methods for Payment of
Electricity Transmission Systems
Francisco Danitz, Hugh Rudnick, Juan Zolezzi and David Watts

The bilateral approach aims at assigning the transmission

Abstract-- Three known use-based allocation methods for cost of any specific transaction between two buses of the
payments of electricity transmission systems are compared. system, commonly known as “wheeling” methods, among
They are the Bialek tracing method, the Kirschen them the "postage stamp", the "contract path" and the "MW-
proportionality method and the Rudnick distribution factor
mile allocation" methods.
method. Main characteristics of the methods are summarized.
A numerical comparison is performed. Results do not allow In this work we assess the schemes based on the use of
concluding which one is the best, as it is not possible to the network. The difficulty arises in identifying the
demonstrate the validity of the key principles that support each transmission lines used by each agent of the system.
of the methods. Electricity flows in accordance with physical laws, which
depend on network parameters and generator dispatch, not
Index Terms-- Deregulation, transmission pricing, related to commercial contracts among participants. In this
transmission use, wheeling allocation, distribution factors.
paper we study three methods that attempt to give an
approximation to how electricity flows in the grid, and
I. INTRODUCTION
assign responsibility to each agent for those flows. These

T he worldwide deregulation of the electrical sector has

changed the electricity business from a centralized and
vertically integrated industry to a segmented industry based
methods aim to track electricity flows from where they are
generated until they reach each consumer.

on competition among participants [1]. The new scheme II. ALLOCATION METHODS
established free access to the transmission lines and
The alternative methods compared aim at determining the
compulsory connection to the grid, that have allowed
contribution of each agent to the flow of each line, where
generators to transport their energy to the main consumer
the first one uses the superposition principle while the other
centers, boosting a competitive environment among
two use the proportionality principle. Main characteristics of
generators and consumers. In that environment, the
each method are presented.
transmission network is considered to be the key factor of
the electricity markets. A. Rudnick’s method
Electricity transmission has economies of scale and The first intent to relate transmission pricing to use of
scope, making the transmission sector a natural monopoly system factors was made by Hugh Rudnick, Rodrigo Palma
that has to be regulated. A price system has to be defined and Enrique Fernandez, and published in the IEEE in 1995
that does not distort market decisions, related with the as “Marginal Pricing and Supplement Cost Allocation in
operation of the existing capacity, investments in new Transmission Open Access” [2]. Their proposal works with
generating capacity and new loads. Marginal cost pricing, distribution factors previously defined for the study of
which is implemented in most of the new electricity power system security [3]. The method pretends to obtain
markets, does not allow recovering the total cost of parameters that indicate the level of utilization of the
transmission investments, mainly because transmission network by generators and loads.
marginal costs are lower than average costs. In order to There are three types of distribution factors. The A or
recover all costs, it is necessary to add a complementary GSDF distribution factors (called generalized shift
payment. The way to allocate this complementary charge distribution factors) consider the incremental changes in
among system’s users has been a main topic of discussion in power injection or withdrawal at all buses, except the
academic, business and legal fields [2]. Each country has reference bus. The GGDF distribution factors (generalized
reached its own solution in accordance with the reality of its generation distribution factors) measure the total impact
transmission systems. (not incremental) of power injection by generators over the
Several schemes have been developed to solve the flow in a given line. Finally, the GLDF distribution factors
allocation problem. The use of system approach aims at (generalized load distribution factors) measure the total
determining the contribution from all users of the grid, impact of negative injections, which correspond to loads,
based on the physical and economical use of the network, over the flow in a certain line.
independent from commercial contracts among participants.
1) A or GSDF factors
Thanks to Fondecyt. The authors are with Department of Electrical A Ai-k,b factor, defined by means of a sensitivity analysis,
Engineering, Universidad Católica de Chile, Casilla 306, Correo 22, relates a variation in the injected power ∆PI b in a bus b to
Santiago, Chile (email [email protected])
a variation ∆Fi − k in the flow through line i-k. It considers (D' i − k ,b ⋅Gb )
FPi − k ,b = (7)
that a change in the injected power in any bus is absorbed
by a similar negative change ∆PI R in the reference bus R,
∑ D'
g
i −k , g ⋅G g

while injections from generators and loads in other buses, as

B. Bialek’s method
well as losses, are maintained constant. Therefore, for a
variation in all the injections, the variation in the flow in a It was proposed by J. Bialek and published in the IEEE
particular line will be given by: in 1996 as “Tracing the flow of electricity" [4]. The method
aims at tracing the flows of electricity through power
∆Fi − k = ∑ Ai − k ,b ⋅ ∆PI b (1) networks. It allows quantifying how much of the active or
b≠ R
reactive power flows from a particular source to a specific
fulfilling load. It also allows quantifying the contribution from each
generator or load to flows and losses in a given line.
∑ ∆PI
b≠ R
b + ∆PI R = 0 (2)
It uses a proportionality principle, which states that for
any bus there are lines that inject power and others that
The A factors are obtained from a DC power flow, which evacuate power. If it is assumed that the origin of the
only models reactances, and are calculated as follow: incoming power is known (the proportional contribution of
each “upstream” generator to the flow in every line is
known), the method states the hypothesis that the
X i −b − X k −b
Ai − k ,b = (3) proportionality of the flows is maintained in the departing
X i ,k flows. So, this allows tracking the flow of electricity from
where Xi-b and Xk-b are elements of the reactance matrix (the each generator to every consumer. The proposed algorithm
inverse of the nodal admittance matrix, without the uses a non-loss power flow as a base. However, the author
reference bus column and row) and Xi,k is the reactance of presents different alternatives to allocate the losses in the
line between buses i and k. These factors are independent of lines.
the operational conditions of the system (generation and If the problem is analyzed from the perspective of
load distribution), but depend on the grid configuration and generation, the power injections in each bus of the system
the reference bus chosen. are given by:

2) D or GGDF factors
A Di-k,g factor relates the total injection from a generator
Pi = ∑ Pi − j + PGi ∀ i = 1,2,..., n (8)
j∈α i ( u )
Gg in a bus g, with the total flow Fi-k over a line i-k. These
(u )
factors emerge from the following equations: where Pi is the total flow through bus i, αi is the set of
buses that directly supply bus i (the flow must go from other
Fi − k = ∑ Di − k , g ⋅ G g (4) buses to bus i), PGi is the generation in bus i and Pi-j is the
g flow in line j-i, where

These factors are independent of the election of the

Pi − j = Pj −i
.
reference bus, but depend on the system configuration and Using the proportionality principle, the flow in a line can
on its operational conditions. GGDF factors are obtained be written as
with A factors:
Pi − j = c ji ⋅ Pj , where c ji = Pi − j / Pj
Di − k , g = Ai −k , g + Di − k , R (5)
replacing in (8) we obtain:
where Di-k,R is defined as:
 
Pi = ∑c ji ⋅ Pj + PGi ∀ i = 1,2,..., n (9)
 Fi − k − ∑ Ai − k , p ⋅ G p  j∈α i ( u )
 
= 
p≠R and arranging it:
Di − k , R (6)
 
 ∑ Gg 
  Pi − ∑c ji ⋅ Pj = PGi o Au P = PG (10)
 g  j∈α i ( u )
where Au is an (nxn) distribution matrix per injected
If some D factors turn out to be negative, they are powers, P is the vector of bus flows and PG is the vector of
considered nil for the allocation. Thus, to determine the bus generations.
level of contribution of a generator b to the flow Fi-k of a The elements of matrix Au are defined as follow:
particular line i-k, equation (7) can be used, where D'i-k,b 1 for i = j
would be equal to Di-k,b if the factor has the same sign as the 
−P
flow and would be nil if the factor has an opposite sign. [Au ]ij = − c ji = j −i for j ∈ α i (u ) (11)
 P j
0 e.o.c

where j must be a bus that supplies power to i. common are defined as the power exported through the
−1 links to other commons of higher rank.
If Au -1
exists, then vector P = Au ⋅ PG and its
It uses the same proportionality principle as Bialek, but
elements are given by: with a difference. While Bialek applies it to each bus of the
[ ]
n
Pi = ∑ Au system, Kirschen applies it to the commons. Doing this, the
−1
ik
⋅ PGk for i = 1,2,..., n (12)
k =1
following assumption can be made: “for a given common, if
The last equation shows that the contribution from the proportion of internal flow associated to generator i is

generator k to bus i is equal to Au [ ] −1

ik
⋅ PGk .
xi, then the proportion of external flow associated to
generator i is also xi”. With this assumption, a recursive
A withdrawal of power in line i-l from bus i can be method is built to determine each generator’s contribution to
calculated as: supply the loads in each common. The following variables

∑ [A ]
Pi −l Pi −l n are defined:
−1
Pi −l = ⋅ Pi = u ik
⋅ PGk Cij: contribution by generator i to the load and external
Pi Pi k =1 (13) flow of common j
n
(d ) Cik: contribution by generator i to the load and external
=∑D G
i −l , k ⋅ PGk for l ∈ α i flow of common k
k =1
(d )
Fjk: flow from common j to common k through the link
where α i is the set of buses directly supplied by bus i Fijk: flow from common j to common k through the link,

and D G i −l , k =
[ ]
Pi −l ⋅ Au
−1
ik
is a topological generation
coming from common i
Ik : internal flow of common k.
Pi
distribution factor, indicating the proportion of power that The following relations are defined:
generator k contributes to line i-1. Fijk = C ij ⋅ F jk
These D factors are the ones that permit to allocate the (14)
actual use of the transmission lines.
I k = ∑ F jk (15)
C. Kirschen’s method j
It was proposed by Daniel Kirschen, Ron Allan and
Goran Strbac and published in IEEE in 1997 as ∑Fj
ijk
"Contributions of Individual Generators to Loads and Cik = (16)
Flows" [5]. The proposed algorithm is based on the solution Ik
of a series of load flows, identifying buses that are reached
by the power generated in each generator. Then, the method These recursive equations are the ones that will be used to
groups all the buses supplied by the same generators. calculate each generator’s contribution in each common.
Assuming proportionality, it is possible to calculate the The calculation begins with the “root” commons, those with
contribution of each generator to the loads and to the line rank 1, where their internal flows are entirely produced by
flows. generators inside the common. The next step is to calculate
It is first necessary to define some concepts to properly the external flows for these commons and then continue
understand the method. with the others of higher rank. In short, what is obtained is
Generator’s domain: Generator’s domain is defined as the proportion of the contribution of each generator to each
the group of buses that are reached by the power common and therefore, to each line inside this common, as
generated by a given generator. The power of a well as the proportion of the flow that leaves each common.
generator is capable of reaching a particular bus only if
it is possible to find a way through the grid that links III. NUMERICAL COMPARISON
them, where the direction of the trip is consistent with
The comparison between the three methods was done
the direction given by a load flow.
over several systems of varied sizes. This paper reports on
Commons: A common is defined as a group of
the study of an 8-bus system (Figure 1), a reduced model of
neighboring buses supplied by the same generators. The
the Chilean Central Interconnected System (SIC). The
number of generators that supply a common is defined
system has a strong concentration of load in bus 4
as the rank of the common, and can be between one and
(Santiago), and that means that considerable part of the flow
the total number of generators in the system.
is in that direction (Table 1). The results obtained, shown in
Links: The lines that connect two different commons
Figure 2, correspond to the distribution of the
are defined as links. The flows in the lines of a
complementary payments that result from the contribution
particular link are in the same direction, always from a
of each generator to the flows in the lines.
common of rank N to other of rank M, where M>N
Initially it was observed that the results obtained using
always.
the two proportional methods where not substantially
changed using a DC analysis or an AC analysis, and given
The method calculates the internal flows of each
the fact that the allocation of losses is not the objective of
common as the addition of the power injected by the
this comparison, it was preferred to carry out the
generators in common buses, plus the imported power from
calculations only considering a DC power flow.
others commons through the links. The external flows of a
 use of system perspective, as generator G6 supplies a local
load, it should be liberated from paying for the use of any
transmission line. However, Figure 2 shows that all the
compared methods assign responsibility to this generator
over the use of lines L3-2, L3-4, L6-3 and L7-5 , the last in the
case of Rudnick. This was one of the discussion points that
followed the publication of the Kirschen method. The
author stated that there are two ways to apply the
contribution method, one through the net injections to each
bus or taking the loads and generators independently. This
last case is the one that has been presented in the previous
results.
If the method were applied considering net injections in
each bus, the calculations should be done considering a load
equivalent to 360-300 [MW] in bus 6. However, net
injections of power must be calculated in each bus, in which
case the model would not reflect the reality of the system,
and the differences between physical and commercial use of
the grid would be incremented.
Another important characteristic of the results obtained
for this system and for others analyzed, is that the GGDF
. factors assign more responsibility to the generators than the
Figure 1. 8-bus SIC system proportional methods, over the use of transmission lines.
This does not necessarily mean that the allocation to agents
TABLE I. SYSTEM’S FLOWS
Line R [p.u.] X [p.u.] Flow [MW] is greater.
1 2 0.0009 0.0013 150.00
3 2 0.0274 0.0390 200.00
IV. CONCLUSIONS
3 4 0.0001 0.0027 1100.00
5 3 0.0008 0.0170 1075.82 Conclusions are achieved from the study of systems of
6 3 0.0788 0.3920 34.18
7 5 0.0050 0.0361 455.82
different sizes. The first comment to be made relates to the
7 6 0.0682 0.2270 94.18 simplicity achieved using the DC load flow in the three
8 7 0.0310 0.1529 120.00 methods, when no losses are to be calculated
The two proportional methods provide similar results, but
From the results it can be observed that there is a this depends on the network configuration. The smaller the
similitude between the Bialek and Kirschen methods to commons of the Kirschen method, the more similar the
assign the flows to the generators. This is not the case with results of those two methods.
the Rudnick method, whose outcomes differ from the other The Rudnick method presents more differences, given it
two methods. There are lines with significant differences, utilizes a different principle, even resulting in negative flow
where responsibility is assigned to generators the other contributions.
methods do not include. However, it was observed that the In general, larger differences among the three methods
similitude between the Bialek and Kirschen methods arise on lines with lower allocations, which correspond to
depends on the configuration of the system. If both methods lines less used by generators.
None of the three methods provides a logical physical
are compared over more complex and meshed systems,
solution to the sunken generator condition. Net flow
differences increase, particularly when more generators use
assessment may be used, implying other limitations.
the lines simultaneously (smaller allocation percentages to
Nevertheless, the three methods provide reasonable
each generator). results to allocate transmission charges. Through
The difference of the Rudnick method to the proportional approximations, they try to measure the user of the network,
methods is due to the different principle (superposition) on “chasing” the power flows from generators to consumers,
which it is based. Besides, Rudnick’s calculations may thus allowing to “identify” those responsible for the line
result in negative flows that must be considered as nil for flows. Evidently, this is a major challenge, as the laws of
the allocation (Eq. 7). On the other hand, the similitude physics do not allow “coloring” the electrons of each
between the Bialek and Kirschen methods are reasonable generator to tag on their path. Besides, transmission flows
due to the use of the same principle, even though it is are interrelated with other system variables. Nevertheless,
applied in a different way. Bialek uses the principle of the approximations provided by the three methods are
proportionality bus by bus and Kirschen in a whole common reasonable enough for transmission pricing purposes.
area. The smaller the areas and the fewer the buses, the Finally, the assessment of the three methods did not
more similar the results, which is precisely the case of the provide enough evidence to argue for one method over the
simulated system.
other nor demonstrate the validity of the key principles that
In this system there is a special situation that is of
support each of the methods.
interest. There is a generator in bus 6 (Itahue) whose
injection is G6=300 [MW], with a local of 360 [MW], so G6
is a sunk generator. Following a “logical” analysis, from a
 V. REFERENCES system security evaluations”, IEEE Transactions on Power Apparatus and
Systems, Vol. PAS-100, Nº3, March 1981, pp 1001-1005.
[4] Bialek J., "Tracing the Flow of Electricity". IEEE Proceedings on
[1] Rudnick H, "Pioneering electricity reform in South America". IEEE Generation, Transmission and Distribution, Vol. 143, Nº4, July 1996, pp
Spectrum, pp. 39-44, August 1996. 313-320.
[2] Rudnick H., Palma R., Fernández J., "Marginal Pricing and [5] Kirschen D., Allan R., Strbac G., "Contributions of Individual
Supplement Cost Allocation in Transmission Open Access". IEEE Generators to Loads and Flows". IEEE Transactions on Power Systems,
Transactions on Power Systems, Vol. 10, Nº2, May 1995, pp 1125-1142. Vol. 12, Nº1, February 1997, pp 52-60.
[3] Ng, W. Y., “Generalized generation distribution factors for power

G1 G3

120 50

45
100
40
[%]
[%] 80 35

30

60 25

20
40
15

10
20
5

0 0
1-2 3-2 3-4 5-3 6-3 7-5 7-6 8-7 1-2 3-2 3-4 5-3 6-3 7-5 7-6 8-7

Line Line
Bialek Kirschen Rudnick Bialek Kirschen Rudnick

G5 G6

70 90

80
60
70
[%]
50
[%] 60

40 50

30 40

30
20
20
10
10

0 0
1-2 3-2 3-4 5-3 6-3 7-5 7-6 8-7 1-2 3-2 3-4 5-3 6-3 7-5 7-6 8-7

Line Line

Bialek Kirschen Rudnick Bialek Kirschen Rudnick

G7 G8

100 120

90
100
80
[%] [%]
70
80
60

50 60

40
40
30

20
20
10

0 0
1-2 3-2 3-4 5-3 6-3 7-5 7-6 8-7 1-2 3-2 3-4 5-3 6-3 7-5 7-6 8-7

Line Line
Bialek Kirschen Rudnick Bialek Kirschen Rudnick

Figure 2. Generator percentage contributions for 8 bus system

View publication stats