Natural Gas Network Pricing and Its Influence

on Electricity and Gas Markets

M. S. Morais and J. W. Marangon Lima, Member, IEEE
1


Abstract-- The introduction of competition at the electricity
generation and commercialization has been the main focus of
many restructuring experiences around the world. The open
access to the transmission network and a fair regulated tariff are
the keystones of the development of the electricity market. The
natural gas business has a great interaction with the electricity
market in terms of fuel consumption and energy conversion.
Given that the transmission and distribution monopolistic
activities are very similar with the natural gas transportation
through pipelines, economic regulation related to the natural gas
network should be coherent with the transmission counterpart.
This paper will show the main wheeling charge methods used at
transmission and their application to the gas network. Linear
stead-state equations are developed to adequate the various
pricing methods. Some examples will clarify the results, in terms
of investments for thermal generation plants and end consumers,
when combined pricing methods are used to transmission and gas
networks. The paper shows the synergies that should be
adequately used, otherwise wrong economic signals are sent to
the market players.

Index TermsEconomic Regulation, Electricity Market,
Natural Gas Network, Transmission Network, Wheeling
Charges.
I. INTRODUCTION
T
HE international environment has changed dramatically
since the beginning of the 90s. The introduction of
competition at the electricity generation and
commercialization has been the main focus of many
restructuring experiences for the electrical sector. The main
goal has been to achieve greater economic efficiency than the
past centralized and often-monopolistic environment.

The open access to the transmission network and a fair
regulated tariff are the keystones of the development of the
electricity market. Many methodologies have been proposed
to price transmission networks in order to give reasonable
economic signs to the electricity market players [1]. These
methods such as the MW-mile, bus marginal cost, and others,
usually incorporate the spatial nature of the transmission
systems given the opportunity to the generation and consumer
agents to place their generation and load units at the most
appropriated sites. To those agents that are already placed,
there is an opportunity to influence on the transmission
expansion plan to minimize the wheeling charges. The market
rules and the flexibility given to each agent vary from country
to country and also depend on the degree of the restructuring
development.

At the generation side, especially for the thermal units,
locating the assets where their production will be valued at
best is of uttermost importance for the companies future
return on investment. Besides the transmission fare and the
plant investment cost, one important portion of the total cost is
the fuel cost. For the natural gas plants, the fuel cost can be
split into two parts: the production cost and the transportation
cost. The gas transportation is usually performed by gas
pipelines, which have similar characteristics with the
electricity transmission network. Therefore, fuel supply
conditions, as well as generation and transmission capacity
constraints, have to be taken simultaneously into account in
the investment decision process. Synergies between electricity
and natural gas systems have to be identified and
economically quantified so that integrated decisions could
bring in an edge to the investment company. In the long-term
phase, the decisions are highly inter-dependent in gas and
electricity sub-systems, which justify an integrated analysis
[8]. Therefore, the economic regulation of electricity
transmission and gas transportation must be performed
together.

A methodology to charge gas pipeline networks is proposed in
this paper taking into account the transmission pricing
method. Transmission wheeling charge methods usually
consider the load flow equations, i.e., the static behavior of the
electrical systems. Therefore, a coherent method to price gas
network should also use steady-state equations of the gas flow
through the pipelines. Based on the linearization of such
equations, wheeling charge methods are compared for the gas
network. The methods need to match with the transmission
pricing methods. Therefore, a set of combined pricing
methods is suggested.

Some examples will demonstrate the importance of the
wheeling charge regulation both at the gas and electricity
systems. This is crucial for the thermal units, which have the
gas as the input and the electricity as the output of the
production process. Both products use physical network
structures as transportation. Therefore, the influence of the
wheeling method on the decision making process in terms of
sitting the assets will be showed.

1
This work was supported in part by CAPES and CNPq (#450656/02)

M. S. Morais and J. W. Marangon Lima are with Engineering System
Group (GESis) at Federal Universtiy of Itajub (UNIFEI), Av. BPS 1303,
Itajub, MG 37500-903, Brazil (email: marangon@iee.efei.br).
0-7803-7967-5/03/$17.00 2003 IEEE
Paper accepted for presentation at 2003 IEEE Bologna Power Tech Conference, June 23th-26th, Bologna, Italy

( ) | |
(
1 3
5 2
2
2
1

= s m
fSLTZ
D p p
p
T
C Q
n
n
n
) (2)
Another important result from the consumer point of view is
that electricity and gas may be interchange in some
applications like, for instance, in the generation of heat. The
electricity market and the gas market will provide the
commodity prices, but the transportation may influence on the
decision about the better fuel to use. In Brazil, where hydro
plants provide 92% of the electricity generation, the wheeling
charge method plays an important role.
where: where:
64
2
air
R
C

= = constant
The above general flow equation involves simplifying
assumptions which are:

1) isothermal flow due to insignificant temperature changes;
II. GAS STEADY-STATE EQUATIONS
2) negligible kinetic energy change and constant
compressibility across the pipe;
The steady-state flow of gas in a pipeline may be described by
equations that change according to the gas working pressure
and friction. These factors govern the gas flows that can vary
from small values, in low-pressure distribution systems, to
vary large values, in high-pressure transportation systems. The
effects of friction are difficult to quantify and are the main
reason for variations in the flow equations. The friction factor
is not a constant for a given section of a pipeline and it is
dependent on the roughness of the internal pipe surface, gas
velocity, gas density, gas viscosity and the internal diameter of
the pipe.
3) validity of the Darcy friction loss relationship across the
pipe;
4) constant friction coefficient along the pipe length.

Based on the working pressure of the network and the friction
factor, Eq. (2) can be simplified. Three categories are
provided in this paper, which can encompass a great variety of
situations [6]: low-pressure, medium-pressure and high-
pressure networks, with ( ) ( ) ( ). , ,
1 3
h m Q m L mm D
A. Low-pressure (0-75 mBar gauge)

For this pressure range, the Laceys equation may be derived:
After defining the gas operating conditions, the problem of
static simulation is to estimate the values of pressure at the
nodes and the flows in the individual pipes for known values
of sources pressures and of gas consumption in the nodes. The
pressures at the nodes and the flow in the pipes must satisfy
the flow equation, and together with the values of loads and
values of sources must fulfill the similar Kirchhoffs laws for
the electrical systems.
( )
(

=

fSL
D p p
Q
n
5
2 1 4
10 72 . 5 (3)
where
(

+ =
D
f
276 . 0
12
1 0044 . 0
In general, f may be set equal to 0.0065 givin the Poles
equation:
The relationship between pressure, volume and temperature of
gas allows the development of an empirical steady-state
pipeline flow equation, derived from Bernoullis equation.
The gas flow is given by [6]:
n
Q
( )
SL
D p p
Q
n
5
2 1 3
10 1 . 7

=

(4)
Assuming S=0.589, and the equation (4) can be
rearranged:
(mbar p )
( )
fSLTZ
D
T ZR
Sgh p
p p
p
T R
Q
air
av
n
n air
n

)

=
5
2
2
2
2
1
2
2
64

(1)
5
3
2
2 1
10 7 . 11
D
L
K for
KQ p p
n
=
=
(5)
where:
n n
p T ,
R
quantities at standard conditions of temperature and
pressure (288 K and 0.1 MPa);

B. Medium-pressure (0.75 7.0 Bar gauge)
air
constant of air ( )
1 1
K NmKg ;
S (
air
R

= specific gravity of the gas )
gas
R
tor;
;
f
For this pressure range, the Polyflos equation may be
derived:
dimensionless friction fac
D internal diameter o ; f pipe
;
( ) m
L
( )
(

=

fSLT
D p p
p
T
Q
n
n
n
5 2
2
2
1 4
10 57 . 7 (6)
length of the pipe ( ) m
K T temperature do gas ( ) ;
Z dimensionless compressibility factor;
where: ( ) bar p
2 1
, p p inlet and outlet absolute pressure ( )
2
Nm ,
h inclination of the pipe,
g acceleration of gravity.

Assuming that the pipe is horizontal, ( , the elevation
term is zero and the equation simplifies:
) 0 = h


k
p pressure drop for pipe ; k
The value of is given by: f
i
p absolute pressure at node ; i
( ) ( ) E f
076 . 0
Re 338 . 5 1 =
pj absolute pressure at node ; j
where:

E is equal to 0.8
For medium and high pressure:
Re is the Reynolds number

( ) | | ( )
k j i k
m
n k k n
P P P Q K Q = = =
1
(10)
Assuming T and , the equation (6) can
be rearranged:
K 288 = 589 . 0 = S
where: .
2 2
,
j j i i
p P p P = =


848 . 4 2
848 . 1 2
2
2
1
24 . 27 :
D E
L
K where
KQ p p
n
=
=
(7)
The equations for low pressure, and for medium and high
pressure can be rearranged:

( ) ( )
( ) ( )
1
1
2 1
m
k k k n
k k k n
K P Q
K p Q
=
=
(11)
C. High-pressure (above 7.0 Bar gauge)


For pressures above 7.0 Bar, the Panhandle equation can be
used:
F. Network equations

( )
fSLTZ
D p p
p
T
Qn
n
n
5 2
2
2
1 4
10 57 . 7

=

(8)
Many methods of meshed gas flow simulation may be used,
such as, the Newton-nodal method, Hard-Cross nodal method,
Newton-loop method and Hard-Cross loop method. The
Newton-loop method has a good convergence compared with
the other ones [2], [3],[4] and will be used in this paper.
where: . ( ) bar p
The friction factor is given by: f

( ) ( ) E f
073 . 0
Re 872 . 6 1 =
The Equation (12) represents a set of loop equations that
describe the gas network. In this approach, the non-pipe
elements are not considered.
Assuming , the
equation (8) can be rearranged:
589 . 0 288 , 95 . 0 = = = S and K T Z

( ) | | 0 . = Q B (12)

854 . 4 2
854 . 1 2
2
2
1
43 . 18 :
D E
L
K where
KQ p p
=
=
(9)
where:
B branch-loop incidence matrix;
( ) Q vector of flow equation.

This equation is a mathematical representation of Kirchhoffs
second law which states that the sum of pressure-drops around
any loop is zero. The loop method requires that a set of loops
in the network be defined. An initial approximation made
to the branch flows ensures that a flow balance exists at each
node. Since the branch flows are approximations to their true
values, a loop flow q is introduced. The loop flow is the flow
correction to be added to the branch flow approximations to
yield the true values. A loop error into each loop is also
introduced because the pressure-drop calculated from these
flows will not summate to zero around each loop. This loop
error is a function of all the loop flows.
0
Q
D. General formulae

From above equations, it is possible to derive a general
expression for steady-state gas flow. For any pipe , the pipe
flow equation from node i to node can be expressed as [7]:
k
j
( ) | | ( )
k
m
n k k n
Q K Q
1
=
where:
( ) | |
k n
Q flow function for pipe k ;
k
K pipe constant for pipe k ;
( )
k n
Q flow in pipe k ;

1
m flow exponent: 2 for low-pressure
The loop errors F(q) are incorporated into equation (12):
( ) ( ) | | Q B q F . =
1.848 for medium-pressure
1.854 for high-pressure

But, q B Q Q
T
.
0
+ =
For low pressure:
Then, ( ) ( ) | | q B Q B q F
T
.
0
+ = (13) ( ) | | ( )
k j i k n k k n
p p p Q K Q = = =
2

where:

We now make an initial approximation to the loop flow .
This approximation is then successively corrected until a
solution is reached. The Newton-loop method solves the set of
equations (13) iteratively until the loop errors are less than a
specified tolerance. The iterative scheme for correcting the
approximations to the loop flows is [6]:
q
T(u) =

k all
k
k
k
f
u f
C
) (
(16)
where:
T(u) transmission tariff for the user u
C
k
cost of circuit k
f
k
(u) k-circuit flow caused by user u

k k k
q q q + =
+1

k
f k-circuit capacity
TTC =

total transmission cost

k all
k
C
where,
( ) | |
k
k
k
J
q F
q

=
The flow f
k
(u) may be assessed by a DC load flow model
based on a given set of power injections and power retrieves
that represent the wheeling transaction of user u. In this case,
it is important to identify the from and to points of each
transaction. For instance, if a transaction is defined by an
injection of x MW at bus i and a retrieve of the same x at bus
j, the vector P has zeros at all positions except at i row with
the value x and at j row with the value x. The circuit flows
are calculated using equations (14) and (15).

J is the loop Jacobi matrix given by:

T
B M B J . . =
where:
( ) m i Q K m diag M
M
i i
,...., 1
1
1
1
= =


= m number of branches.


The advantage of this method is its good convergence
characteristics and its little sensitivity to the initial conditions.
Similar approach may be applied to the gas network. For an
agent u represented by an injection of gas at node i and a
retrieve at node j, the variation of gas flow, Q, at each
pipeline may be determined by considering the case with and
without the transaction using equation (12). With such a
procedure, we are assessing the sensibilities of the gas flow of
each pipeline.

III. COMBINED TRANSMISSION AND GAS NETWORK PRICING

Many transmission pricing methods have been proposed since
the popular postage stamp method. The majority of them
starts with the MW-mile method [5], which incorporate the
distance as an important measure to account. This method
allocates the allowed transmission income among the system
users in proportion to their "extent of use" of the transmission
resources [1]. The DC load flow equations identify how each
user contributes to the use of circuit capacities.

The IRTC method also uses the DC load flow but it is not
necessary to identify the counterparties of a transaction. Nodal
prices are determined independently from the injection and
retrieve points. The method uses the sensibility,
ki
, related to
the variation on the load flow in one circuit, k, due to an
injection of 1 MW at bus i. The nodal tariff at node j,
j
, is
determined by:
= P (14)

=
k all
kr kj
k
k
k j
f
C
C ) ( (17)
where:
susceptance matrix
voltage angle vector
The
kr
is the sensibility of k-circuit to the reference bus r. In
most of the cases the tariffs cannot provide the total allowed
revenue (AR) for the transmission grid. Thus, an adjustment
is needed to match the revenue. Equation (18) shows how this
adjustment can be introduced.
P bus power injection vector

The power flow at circuit k, f
k
, may be obtained by:

(15)
k j i k
f ) ( =

+ =
i all
i
i all
i i
j j
L
L AR
(18)
where

i
voltage angle of bus i

k
susceptance of circuit k

where:
Three transmission pricing methods will be proposed for the
gas meshed network: the traditional postage stamp, the MW-
mile and the Invested Related Transmission Cost (IRTC) [9].
The last one has been applied in Brazil.
j
adjusted tariff at bus j
L
i
load at bus i


Rearranging the equation (18) yields:
The MW-mile method calculates the flow at each circuit
caused by the generation/load pattern of each agent based on a
power flow model. Costs are then allocated in proportion to
the ratio of power flow and circuit capacity:

+ =
i all
i
i all
i i j
i all
i
j
L
L
L
AR
) (
(19)
Table 1: Gas wheeling charge under MW-mile
(million US$/Mm3/h)
R. Janeiro S. Paulo P. Alegre
Bolvia 0.0304 0.0290 0.0299
Argentina 0.0288 0.0199 0.0198
Campos 0.0037 0.0092 0.0335 where:

=
k all
ki kj
k
k
i j
f
C
) ( ) (

Table 2 shows the nodal prices for the three cities considering
the IRTP method (injected and consumed gas are in Mm
3
). In
this case, both the production and consumption pay the
wheeling charge, i.e., 50 % of the total cost is paid by the
production and 50 % by the consumption.
Note that

j
does not depend on the reference bus. The first
term of equation (19) is the postage stamp tariff and the
second one is the modulation.

A similar procedure is applied to the gas grid. The
kj
represents, for the gas network, the variation in the gas flow at
pipeline k caused by one variation in the injection of gas at
node j. A reference node is also chosen to build the matrix B
of all sensibilities . It is important to note that those
sensibilities are related to the base case with all transactions
incorporated.
Table 2: Gas wheeling charge under IRTP method
Node Injected
Gas
Generation
Tariff
Consumed
Gas
Load
Tariff
S Paulo 0 0.0410 253 0.0361
R Janeiro 0 0.0386 85 0.0386
P Alegre 0 0.0.424 167 0.0348
Bolivia 94 0.0153 0 0.0618
Argent 73 0.0061 0 0.0710
Campos 338 0.0483 0 0.0288
Based on the B matrix, the same approach related to the
revenue reconciliation can be applied for the gas network.

Table 3 shows the transmission charges under MW-mile for
the three points chosen as the candidate points of electricity
injection. In this case, it was assumed that the natural gas
generation plants will be sited at So Paulo, Rio de Janeiro or
Porto Alegre.
IV. BRAZILIAN GAS AND TRANSMISSION GRIDS

The development of the gas grid in Brazil is recent, but some
pipe loops can be visualized in the future. Figure 1 presents
the planned and the current configuration of the gas network
in the south part of the country. There are basically three
sources of gas: two from the neighboring countries Bolivia
and Argentina and one from the off-shore oil fields at the
north of Rio de Janeiro city. In this paper, it will be analyzed
three options in terms of sitting a thermal plant: So Paulo,
Rio de Janeiro and Porto Alegre cities.

Table 3: Transmission charges under MW-mile
(US$/KWmes)
R. Janeiro S. Paulo P. Alegre
R. Janeiro 0 0.468 1.581
S. Paulo 0.468 0 1.3143
P. Alegre 1.581 1.3143 0

The Table 4 shows the transmission charges under IRTP for
the same points of Table 3. The IRTP method is the current
method used in Brazil where generation pays 50% of the total
cost and load pays the remaining 50%. This is the reason why
there are two different transmission tariffs: one for the
generation and one for the load.


Table 4: Transmission charges under IRTP (US$/KWms)
Generation Load
R. Janeiro -0.291 1.025
S. Paulo 0.0243 0.759
P. Alegre 0.555 0.180

Tables 5, 6 and 7 present the total costs for one combined
cycle thermal plant with capacity of 10 MW being installed at
the three cities respectively. The wheeling charges of both
transmission and gas network vary according to the method.
For instance, at Table 5, if the thermal plant is located at Rio
de Janeiro and the IRTP is used for the gas and the postage
stamp for the electricity, the plant owner will pay US$
249,000.00 per month for the ship-or-pay contracts. The costs
at shaded cells derive from the same wheeling method at gas
and electricity.
Figure 1: Brazilian Gas Network

Table 1 presents the gas network wheeling charges under
MW-mile method considering a transportation of one
thousand m
3
(Mm
3
) from the three sources to the
aforementioned cities. Campos is the city at Rio de Janeiro
state where oil fields are located.




Table 5: Thermal plant at Rio de Janeiro (millionUS$)
Gas
Gas-mile P. Stamp
Campos Argentina Bolvia
IRTP
P. Stamp 0.3219 0.1746 0.2286 0.2321 0.2497
R Janeiro 0.1552 0.0079 0.0619 0.0654 0.0829
So Paulo 0.1598 0.0126 0.0666 0.0700 0.0823
MWm
P.Alegre 0.1710 0.0238 0.0777 0.0812 0.0906












E
l
e
c
t
r
i
c
i
t
y

IRCP 0.1523 0.005 0.0643 0.0709 0.0801

Table 6: Thermal plant at So Paulo (millionUS$)
Gs
Gas-mile P. Stamp
Campos Argentina Bolvia
IRTP
P. Stamp 0.3219 0.1865 0.2095 0.2291 0.2443
R Janeiro 0.1598 0.0245 0.0475 0.0670 0.0835
So Paulo 0.1552 0.0198 0.0428 0.0624 0.0776
MWm
P.Alegre 0.1683 0.0211 0.0559 0.0755 0.0879












E
l
e
c
t
r
i
c
i
t
y

IRCP 0.1554 0.0227 0.0452 0.0679 0.0779

Table 6: Thermal plant at Porto Alegre (millionUS$)
Gs
Gas-mile P. Stamp
Campos Argentina Bolvia
IRTP
P. Stamp 0.3219 0.2387 0.2093 0.2310 0.2415
R Janeiro 0.1710 0.0878 0.0584 0.0801 0.0988
So Paulo 0.1683 0.0852 0.0557 0.0774 0.0907
MWm
P.Alegre 0.1552 0.0720 0.0426 0.0643 0.0748












E
l
e
c
t
r
i
c
i
t
y

IRCP 0.1608 0.0749 0.0450 0.0698 0.0804

Based on the tables, it is possible to identify problems in terms
of economic signals when postage stamp is chosen to be the
wheeling method. The three points, when postage stamp is
applied to both gas and electricity, present high tariffs, which
shows the disadvantage of this method. The tariffs continue to
be high even when only one network is under postage stamp.

The MW-mile and gas-mile method presents better results to
signalize for the agents, that it is better to site the power plant
closer to gas resources and to the electricity load. This is the
case of Rio de Janeiro.

The IRTC method has the advantage of not identifying the
bilateral contracts and, therefore, more suited for market
purposes. However, in terms of location, the mile methods
present a more accurate performance. The IRTC method is
also much better than the PS method.

Comparing the three points, Rio de Janeiro is the better
location to install the thermal plants when gas and electricity
networks are considered.
V. CONCLUSION

Methodologies to charge gas pipeline networks were proposed
in this paper taking into account the transmission pricing
methods. The combined gas and transmission network pricing
is necessary mainly when thermal plants are under
consideration. In Brazil, the development of the gas network
is being justified by the electrical sector that was recently
exposed to a rationing process, i.e., the electrical sector is the
anchor of the natural gas sector. Therefore, finding the best
locations for the thermal plants has become a very important
issue. With the introduction of market mechanisms at these
both sectors, the establishment of reasonable charging
methods for the gas and electricity networks is crucial,
especially because the government has no more control over
the new private investments.

This paper focused on three most used methods to show how
they can change the location of new generation. The results
prove that the influence of gas and transmission grids can not
be neglected by the regulators. Therefore, pricing methods
that incorporate locacional information are better to introduce
a more reasonable policy. Particularly, in Brazil, the IRTP and
postage stamp methods are being used at transmission and gas
respectively.
VI. REFERENCES
Periodicals:
[1] J. W. Marangon Lima, "Allocation of Transmission Fixed Charges: An
Overview", IEEE Trans. On Power Systems, vol. 11, N 3, pp. 1409-
1418, Aug. 1996.
[2] B. Gay, P. Middleton, The Solution of Gas Network Problems, Chem.
Eng. Sci, Vol 28, pp 109-123, 1971.
[3] B. Gay, P. E. Preece, Matrix Methods for the Solution of Fluid
Network Problems, Trans. Inst. Chem. Engers., Vol 53, pp 12-15, 1975.
[4] T. W. Cochrant, Calculate Pipeline Flow of Compressible Fluids,
Chem. Eng. Sci., Vol 103, N2, pp 115-122, 1996.
[5] D Shirmohammadi, P R Gribik, E T K Law, J H Malinowski, R E
O'Donnel, "Evaluation of Transmission Network Capacity Use for
Wheeling Transactions", IEEE Trans on PWRS, Vol. 4, No. 4, October
1989.

Books:
[6] A. J. Osiadacz, Simulation and Analysis of Gas Networks, Gulf
Publishing Company, 1987.

Technical Reports:
[7] Gas Enginnering and Operating Practices GEOP. American Gas
Association, 1996.

Papers from Conference Proceedings (Published):
[8] S. Hecq, Y. Bouffioulx, P. Doulliez, P. Saintes, "The Integrated
Planning of the Natural Gas and Electricity Systems Under Market
Conditions, in Proc. 2001 IEEE Porto Power Tech Conference, Porto
Portugal, 2001.
[9] M C Calviou, RM Dunnet, P H Plumptre, "Charging for the Use of
Transmission System Using Marginal Cost Methods", Proc. 11th PSCC
Conference, pp. 385-391, Avignon France, Aug 30-Sep 4, 1993.

VII. BIOGRAPHIES
M. S. Morais, has B.Sc., M.Sc. in Electrical Engineering, the last one,
from the Federal University at Itajub (UNIFEI), in 1990.From 1987 to 1991
she was with Engenharia Projeto e Consultoria in Belo Horizonte, as engineer
of Transmission Line sector. From 1992 to 2001 she was with University of
Alfenas (UNIFENAS) as a professor and coordinator of Electrical
Engineering. She is currently working on her D.Sc. degree at UNIFEI.

J W Marangon Lima (M1994), has B.Sc., M.Sc. and D.Sc. degrees in
Electrical Engineering, the last one, from the Federal University of Rio de
Janeiro (COPPE-UFRJ), in 1994. From 1980 to 1993 he was with Eletrobrs,
the former Brazilian holding company of the power sector. Since 1993, he has
been with Federal University of Itajub (UNIFEI) as a Professor of Electrical
Engineering. In 1998, he was also with ANEEL, the Brazilian National
Regulatory Agency, as a director advisor. From 2001 to 2002, he joined the
Energy Policy Council in Brazil, as the coordinator of Energy Prices and
Tariffs Technical Committee.