You are on page 1of 11

The development of a natural gas transportation logistics

management system
Sidney Pereira dos Santos
a
, Jose Eugenio Leal
b,n
, Fabrcio Oliveira
b
a
Petroleo Brasileiro S.A.PETROBRAS, Av. Almirante Barroso, 81, 12 andar, Centro, Rio de Janeiro RJ 20031-004, Brazil
b
Pontical Catholic University of Rio de Janeiro (PUC-Rio), Department of Industrial Engineering, R. Marques de S. Vicente 225, Gavea. Rio de Janeiro RJ 22451-900, Brazil
a r t i c l e i n f o
Article history:
Received 30 June 2010
Accepted 21 June 2011
Keywords:
Natural gas industry
Investment analysis
Planning framework
a b s t r a c t
Efcient management of the natural gas business chain based on pipeline transmission networks and
taking into consideration the interaction among the main players (e.g., shippers, suppliers, transmission
companies and local distribution companies) requires the use of decision-making support systems.
These support systems maximise resources and mitigate contingencies due to gas supply shortfalls,
operational contingencies from scheduled and non-scheduled equipment outages and market demand
shortfalls. This study presents a practical use for technologies, such as a thermohydraulic simulation of
gas ow through pipelines, a Monte Carlo simulation for compressor station availability studies, an
economic risk evaluation related to potential revenue losses and contractual penalties and linear
programming for the maximisation of income and the minimisation of contractual penalties. The
proposed system allows the optimum availability level to be dened and maintained by the Transporter
(by installing reserve capacity) to mitigate losses related to revenue and contractual penalties. It also
economically identies, quanties and justies the installation of stand-by compressor units that can
mitigate the Transporters exposure to losses caused by capacity shortfalls as a consequence of
scheduled and non-scheduled outages.
& 2011 Elsevier Ltd. All rights reserved.
1. Introduction
The main objective of an energy distribution system is to
provide efcient and reliable distribution of energy from the
source to the client. The operation of a distribution system must
be robust enough to deal with uctuations in the transportation
capacity, which is mainly the result of supply disruptions. The
reliability of the gas supply relative to the needs of the consumer
is a matter of risk evaluation and mitigation. All energy supply
systems operate at a certain intrinsic level of risk, but it is
important to assess whether the level and type of risks are
acceptable in the operational context. Furthermore, the reliability
of supply in the energy and gas sector is generally more important
(e.g., for political and economic reasons) relative to other indus-
tries due to the lack of alternative options in the short term.
Reliability is also related to the natural gas quality demanded by
the consumers. According to Kabirian and Hemmati (2007), the
risk of poor reliability is signicant because an inadequate supply
of natural gas to an energy consumer (such as a power plant) may
shut down these facilities for relatively long periods.
An important characteristic of the gas supply is its complete
dependence on dedicated pipeline networks. Consequently, there
are generally high costs involved in interruptions to the gas
supply. Complete interruptions are more likely to occur when
the pipelines pass through a transit country. This issue was
considered in recent works (S oderbergh et al., 2010 and
Omonbude, 2007) but is not applicable to the BoliviaBrazil
pipeline. However, recent conicts caused by the nationalisation
policy of Bolivia could cause a similar type of interruption. The
denition of adequate security depends heavily on the consu-
mers willingness to pay for higher reliability, which then has a
direct impact on the cost of the service. Supply reliability can be
improved by constructing sufcient alternatives to potential
supply disruptions in the form of spare capacity, storage capacity,
alternative supplies and contractual guarantees from suppliers.
These options can cushion sudden disruptions in supply and
reduce the impact on energy prices (Lise et al., 2008).
Optimal operating methodology depends on the scale and
complexity of the transmission and distribution system. Nor-
mally, the only economically feasible way to achieve an efcient
and reliable energy supply is to optimise the system operation
(Potocnik et al., 2007). Doing so makes it possible to mitigate the
negative effects of equipment unavailability (due to scheduled
maintenance or unscheduled outages) or failure for both the
transmission and distribution systems.
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/enpol
Energy Policy
0301-4215/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enpol.2011.06.047
n
Corresponding author. Tel./fax: 55 21 3527 1289.
E-mail address: jel@puc-rio.br (J. Eugenio Leal).
Energy Policy 39 (2011) 47744784
By 2000, the Brazilian energy matrix was mainly composed of
16% hydro energy, 1% nuclear energy, 46% petroleum-derived
fuels, 7% coal and 5% natural gas. Sustainability of projects,
environmental issues and new discoveries of gas reserves in San
Alberto and San Antonio in southern Bolvia were all important
factors in the introduction of imported natural gas to the Brazilian
market. In the context of the Brazilian gas industry, the national
Gas and Energy Commission of the MME (Cogas/MME) decided in
1992 to increase the usage of natural gas from 2.5% to 12% of the
Brazilian Energy Matrix by 2010. This decision became an
institutional target of the ministry and was later (in 2000)
endorsed by the National Energy Policy Council (CNPE), the body
responsible for establishing policies and guidelines related to
energy in Brazil (Mathias and Szklo, 2007). The governments
strategic objective already accounted for in the strategic
planning of the state-owned companies Petroleo Brasileiro S.A.
(PETROBRAS) and ELETROBRAS of the petroleum and electricity
sectors, respectively led to important, concrete initiatives, such as
the construction of the BoliviaBrazil gas pipeline (Gasbol), whose
commercial operation started in June, 1999 (Fernandes et al., 2008).
Achieving this goal was a binational effort that involved multi-
national companies, such as Petrobras in Brazil and YPFB in Bolivia,
and multilateral agencies of credit were established. The complete
pipeline project required 1.8 billion US$ as capital investment.
The pipeline project was conceived as a special purpose
company to make use of the many agreements to support the
feasibility of the project and to mitigate the risks associated with
this kind of project. Take-or-pay, delivery-or-pay and ship-or-pay
contracts with a high percentage level of rm commitment (e.g.,
80% take-or-pay and 100% ship-or-pay) set the scenario for having
a management system to guarantee the maximisation of
resources and/or the mitigation of losses to all of the players
associated with the gas chain business, even in the face of the
eventual failure of any of the players. The Gasbol pipeline is the
object of application of the management system that is presented
in Section 6 of this work.
Interstate gas pipelines may be seen as a way to approximate
neighbour states and to establish co-operations that contribute to
peace and stability in the region (SOVACOLL, 2009). Nevertheless,
such concepts must be thoroughly developed to overcome the
mistrust that often appears in response to the commercial and
political conicts that arise during the lifecycle of the project.
Concerns about gas supply security are still bigger in the cases in
which transnational pipelines pass through several countries. An
example is the gas pipeline between Russian and Poland, which
traverses Ukraine and Belarus and allows those countries the
ability to interrupt the gas supply from Russian to Europe
(S oderbergh et al., 2010). The issue of transit pipelines and the
various ways of negotiating them is well presented by Omonbude
(2007), who discusses the Russian Ukraine dispute over gas
transit pipelines.
Pandian (2005) presents a deep analysis of the concerns and
barriers related to a multinational pipeline project between Iran,
India and Pakistan. It is clear that the availability of energy
sources in India is not enough to ensure the friendly environment
that would make the project politically feasible. Political issues
may inuence the decisions of whether to implement transna-
tional pipelines; they can also produce disturbances during the
lifecycle of the project.
In the case of the Gasbol project, with the exception of a
political crisis in 2006 due to the nationalisation policy of the
Bolivian government, no conicts have threatened to disrupt the
gas supply.
This study presents a framework for dening the optimum
availability level that should be maintained by the natural gas
Transporter. The Transporter can maintain this availability level
by installing redundancy equipment to mitigate losses related to
revenue and contractual penalties. The framework also identies,
quanties and economically justies the installation of stand-by
compressor units that can mitigate the Transporters exposure to
losses due to capacity shortfalls as a consequence of scheduled
and non-scheduled outages.
This framework was applied in a real case study of the Gasbol
gas pipeline, illustrating its practical application. Here, we con-
sider strategic aspects of pipeline network planning. An integrated
model for the strategic planning may be found in Kabirian and
Hemmati (2007).
The study is structured as follows. In Section 2, the framework
for the Management System for Natural Gas Transportation
Logistic (MSGTL) is presented. In Section 3, the objectives of the
MSGTL are detailed. Section 4 describes the set of tools used in
the MSGTL. Section 5 presents the MSGTL methodology. Section 6
details the study of transportation availability and Section 7
conducts the economic evaluation. In Section 8, the conclusions
are presented.
2. The MSGTL framework
Based on the natural gas networks process map (in which the
gas begins with the Producer and passes through the Transporter
to be delivered nally by the Loader to the Distributor), one can
optimise its operations using the MSGTL as it is presented in this
study. A functional diagram of the system is presented in Fig. 1.
Fig. 1. Functional diagram of the MSGTL.
Fig. 2. MSGTLs architecture.
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4775
The MSGTL (whose architecture, as proposed by Santos (2008),
appears in Fig. 2) contemplates the three situations presented below:
(A) New pipeline project: uses the results of the Monte Carlo
simulations to determine the availability of compression
stations and the economic risks analysis. This information
allows decisions to be made concerning the contractual rm
capacity allocation and the transport taxes to be levied for
using the gas pipeline.
(B) Operational gas pipeline: uses information about compres-
sion unit failures as detected by the Supervision, Controlling
and Data Acquiring System (SCADA). Those data are com-
pared with failure frequency and transport capacity tables,
which is fed into the Linear Programming (LP). The objective
function of the LP is assessing the total revenue losses and
contractual penalties that must be minimised.
(C) Gas nomination from the Loader to the Distributor: the
optimum allocation of the available gas supply takes into
consideration contractual constraints, the price of gas and the
Distributors required volumes for a given operational day.
There is also an LP problem with an objective function for
revenue maximisation.
The components of the MSGTL used in the project phase are
depicted in the right part of Fig. 1, whereas the components
needed to solve the operational optimisation under the applicable
constraints and the gas nomination optimisation problems are
depicted in the left side of the same gure.
Fig. 2 shows the corporate web business system architecture,
which is based on an Oracle data server with the data model for
gas pipelines APDM and the applications ArcSDE (georeferenced
data manager), ArcIMS (web publisher), gas pipeline viewer,
Thermohydraulic Simulator (a linear optimisation program) and
the scenario builder (which is the users graphical interface).
The mapped processes (Figs. 1 and 2) include the natural gas
nomination, the transport by gas-lines, the management of the
gas Distributor, the operational contingencies, the transport
system maintenance scheduling and a simulation of equipment
failure. The gas nomination process involves the Distributors
solicitation of the desired volumes from the Loader. It also
involves the pairing of those solicitations with the supply and
the available transport capacity, which results in monthly, weekly
and daily scheduled volumes for gas delivery.
The events associated with operational contingencies, scheduled
maintenance and non-scheduled equipment outages of the gas
transport system may reduce the capacity of the whole system
and expose the Loader and the Transporter to contractual penalties
due to insufcient gas delivery (i.e., lower than the previously
nominated volumes). To mitigate the negative effects of equipment
unavailability or failures during operation, a Monte Carlo simulation
study is performed in tandem with a thermohydraulic simulation.
The Monte Carlos simulations, together with the economic evalua-
tion, indicate the possible levels of redundancy based on the capital,
operational and maintenance costs versus the exposure to revenue
losses due to contractual penalties.
The transportation reliability (and thus, consequently, the
equipment availability) becomes a key point in order to achieve
the contractually rmed gas volume. Those volumes are often
subject to ship-or-pay or take-or-pay contractual clauses. The
ship-or-pay clause involves the Loader and the Transporter. The
Loader promises to use the contracted transportation capacity for
a certain period of time. He will pay for those services even if the
demand falls below the contracted capacity. The Transporter
commits to transporting the contracted volume of gas and will
suffer penalties if he does not full these obligations. The take-or-
pay clause involves the Loader and the Distributor. The Loader
promises to offer the agreed upon natural gas volume to the
Distributor at certain delivery points. The Distributor, in turn, is
compelled to receive the agreed upon volume. A failure to meet
these obligations exposes both the Distributor and the Loader to
contractual penalties. Fig. 3 illustrates how such contractual
relationships work among the involved parties in the natural
gas supply chain. The situation illustrated here reveals the
necessity of a methodology that can quantify the availability in
the system. Additionally, the system must be able to recommend
actions to increase the availability level as well as the prot-
ability according to the allowable risk for each agent.
3. The objectives of the MSGTL
The Management System for Natural Gas Transportation Logis-
tics (MSGTL) was conceived to attend to the following objectives:
(1) To maximise the selling of the available gas supply and
fullling the Distributor demand.
Fig. 3. Natural gas business network.
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4776
(2) To mitigate the network agents risk of exposure to potential
contractual penalties that might be incurred under contingent
conditions of reduced capacity due to equipment failures or
gas supply shortfalls. In the case of the Transporter, the risk is
not transporting the total nominated gas volume. In the case
of the Loader, however, the risk is not delivering the nomi-
nated gas volume to the Distributor.
(3) To quantify the availability level and the rm capacity of gas
transportation using a Monte Carlo simulation (which iden-
ties the failure frequencies of the transportation system),
a thermohydraulic simulation of the gas as it travels through
the pipelines (which provides an economic and technical
evaluation of the optimum availability level to be determined
for the transportation system) and mathematical linear pro-
gramming (which manages the gas delivery costs under
certain demand and supply scenarios).
(4) To incorporate the core processes related to the natural gas
business, involving the Producer, the Loader, the Transporter
and the Distributor.
(5) Under a consistent and coherent protability perspective, to
optimise the available natural gas supply allocation in a
manner that better meets the needs of society and the
contractual requirements between the agents.
(6) To manage contingent situations of supply, transportation
and marketing to mitigate negative effects, in accordance
with the previously established procedures.
For new natural gas pipeline projects, it is necessary to
identify the transport capacity that could be compromised (on a
rm contractual basis) and the corresponding transportation
taxes that will pay the investments at some internal rate of
return expected by the investors, considering that such equip-
ments has nearly twenty years of a useful life. Monte Carlo
simulations for equipment failures and thermohydraulic simula-
tions are essential to identify, quantify and then mitigate the
economic risks for investors who embrace an adequate redun-
dancy level. Such redundancy can be achieved using spare
equipment.
Revenue losses and the mitigation of contractual penalties will
be necessary if the system only manages to deliver some fraction
of the accorded total volume. In order to mitigate the risk of these
events, it is necessary to use Monte Carlo and thermohydraulic
simulations. The real gas supply available for a specic operation
day must be allocated to satisfy the contractual demands for that
specic day. This fullment of the contractual demands must be
in accordance with the criteria of constraints that assures that the
optimum resources have been allocated in the face of eventual
reductions in capacity or gas supply. The eventual contingent
reductions in gas delivery to the Distributor must follow consis-
tent rules, and they must be described in reports that support
such decisions.
The optimisation of the nomination process is achieved by a
process of results maximisation over the operation of the avail-
able commercial natural gas supply. The primary objective is to
optimally attend to the Distributor demand, which is usually
greater than the available supply. Contractual prices and volume
conditions are taken into consideration as well.
4. The set of tools
The thermohydraulic simulation of the natural gas pipeline
transportation network requires a model that represents the lines,
equipment and accessories that inuence the gas ow directly, as
well as the equations and system conditions of delivery, supply
and operations. It also requires specialised software that is
capable of simulating thermohydraulic ow in a permanent
(static) or transient (dynamic) regime.
Simulation technology of this kind has developed enormously
over the last few decades. It allows for the simulation of gas-line
network behaviour, under both permanent and transient regime.
According to Santos (1997), this technology is absolutely neces-
sary for the development of gas-line projects. Without the use of
transient simulation during the gas-line project phase, the inves-
tor risks creating an oversized system capacity, which can affect
its viability. There is also a risk of undersizing the systems
transportation capacity, which can require reinvestment to cover
gas failure issues caused by the undercapacity of the system to
cope with the demand patterns of the market. In the case of
undersizing, the project value will deteriorate, reducing the
internal rate of return for the entire project.
Using statistical distribution curves as a method to evaluate
processes and projects containing many variables and volatile
behaviour has become increasingly popular in the last few
decades.
According to Evans and Olson (1998), the Monte Carlo simula-
tion is essentially a sampling experiment that aims to estimate
the distribution of a result variable that depends on other input
variables, each one with its own probabilistic distribution curve.
The Monte Carlo simulation is often used to assess the impact of
strategic changes and the risks behind such decision making. The
risk is usually dened as the probability that an undesired result
occurs.
The Monte Carlo simulation has been used for gas pipeline
analysis in other contexts. Monforti and Szikszai (2010) used a
Monte Carlo based model in a macro-analysis of the EU gas
network. In our case, Monte Carlo is used in a microanalysis of the
gas pipeline operation.
According to Ragsdale (2006), deciding how to best use the
limited resources of an individual or company is a universal
problem. In the current competitive business environment, it is
critical to guarantee that an enterprise uses its limited resources
as efciently as possible.
When applied to the gas distribution problem, linear program-
ming considers two objective functions: rst, to maximise the
prots from natural gas commercialisation and, second, to
minimise losses (via lost income and contractual penalties).
Consequently, two models were developed to attempt to
maximise prots and minimise losses, respectively. Each model
has its own particular set of constraints and an objective function.
Both models are subject to gas supply and gas-line transportation
capacity and minimum and maximum delivery point capacity
constraints. When the demanding point is a thermoelectric
facility, there is also a minimum and maximum gas supply
constraint required to allow the facility to operate.
4.1. Maximising gas commercialisation
To maximise income, the maximisation objective function
considers the commercialisation of a certain available gas supply
subject to the transportation capacity declared by the Transpor-
ter. It also considers the volumes required by the Distributor
and/or the large consumers (such as thermoelectric facilities) in
addition to the individual commercial price that would be
charged once there was a difference between the nal prices
(commoditytransportation) for each type of consumer (indus-
trial and thermoelectric facility). Other parameters that must be
taken into account include the transfer costs incurred when the
gas is used for the internal consumption of reneries, industrial
units and thermoelectric plants that belong to PETROBRAS.
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4777
The linear program is formulated as
max f z
X
n
i 1
n
i
p
i
Subject to
X
n
i 1
n
i
rTS 1
X
N
i 1
n
i
rCAP 2
n
i
rr
i
Y
i
, 8i 1, :::, n 3
n
i
Zqmin
i
Y
i
8i 1, :::, n 4
p
i
,Cp
i
,S
i
Z0, 8i 1, :::, n 5
S
i
n
i
S
i 1
, 8i 1, :::, n1 6
S
i
rCp
i
, 8i 1, :::, n 7
Cp
i
r30, 8i 1, :::, n 8
qmax
i
n
i
Z0, 8i 1, :::, n 9
Y
i
Af0,1g, 8i 1, :::, n 10
where
Parameters
TS Total gas supply
CAP main pipeline transportation capacity
Cp
i
delivery point i transportation capacity
r
i
gas volume required by the Distributor at delivery point i
qmin
i
minimum gas volume nominated to the Distributor at
delivery point i
qmax
i
maximum gas volume nominated to the Distributor at
delivery point i
p
i
gas price at delivery point i
q
i
delivery point i capacity
Auxiliary variables
S
i
amount of gas supplied at delivery point i
Y
i
1, if there is ow for the delivery point i; 0 otherwise
Decision variable
n
i
nominated volume of gas to the Distributor at delivery
point i
For industrial and thermoelectric usage and for PETROBRAS
internal usage, the maximisation objective function aims to
maximise the sum of the product of the nominated gas volume
delivered to the Distributor with the nal gas price at each point
of delivery.
The applicable constraints require the following conditions: the
total nominated volume of gas delivered to each Distributor can
exceed neither the total gas supply offered by the Producer
(constraint 1) nor the transportation capacity declared by the
Transporter (constraint 2); the nominated gas volume at each point
cannot exceed the Distributors required volume (constraint 3) and
cannot be less than the minimum operational volume for that
specic point (constraint 4) (Fig. 4); the non-negativity of some of
the variables are dened in constraint 5; the gas supply before each
delivery point must be sufcient to meet the nominated volume for
that point and all the supply located downstream that same point
(constraint 6); the gas supply at each delivery point cannot exceed
the transportation capacity to that delivery point (constraint 7);
the transportation capacity to each delivery point is limited to
30 MMm
3
/day (constraint 8); the nominated gas volume for each
delivery point cannot exceed the Maximum gas volume nominated
to the Distributor at delivery point i (constraint 9) and the binary
variable Y
i
indicates the existence (1) or not (0) of ow to
delivery point i (constraint 10).
4.2. Minimising losses (via lost income and contractual penalties)
The minimisation objective function mitigates income losses
and the need to pay contractual penalties due to the incapacity to
cope with the total nominated volume accorded for a specic day
as the result of operational contingencies that reduce the gas
transportation capacity of the Transporter or due to gas supply
reductions generated by the Producer. Operational contingencies
that can reduce the transportation capacity include the failure
of compression units, the failure of gas delivery points, or an
improper closing of a gas-line block valve. The normal closure of
block valves (which takes about 0.001% of the total availability of
the line) do not signicantly impact the transportation capacity
and were, therefore, not considered. The same approach was
adopted for failures at the gas delivery points.
min f z
X
n
i 1
n
i
e
i
p
i
m
i

Subject to
X
n
i 1
e
i
rTS 11
X
n
i 1
e
i
rCAP 12
e
i
rn
i
Y
i
, 8i 1, :::, n 13
e
i
Zqmin
i
Y
i
8i 1, :::, n 14
e
i
,S
i
Z0, 8i 1, :::, n 15
S
i
e
i
S
i 1
, 8i 1, :::, n1 16
S
i
rCp
i
, 8i 1, :::, n 17
0rCp
i
r30, 8i 1, :::, n 18
Y
i
Af0,1g, 8i 1, :::, n 19
ri,ni
Cpi, S
i
S
i+1
q
min,
q
max
Detail of a
delivery
point
S
r
1
,n
1
,q
1
,p
1
r
2
,n
2
,q
2
,p
2
r
n
,n
n
,q
n
,p
n
Cp
1
,S
1
Cp
2
,S
2
Cp
n
,S
n
i=1 i=2 i=n
Transportation network
Delivery points
Fig. 4. Illustration of the pipeline process.
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4778
where
Decision variable
e
i
Gas volume to be delivered to the distributor at delivery
point i
The objective function consists of the sum of the product of the
difference between the nominated volumes and the real volumes
delivered to the Distributor on a specic operational day, with the
prices and contractual penalties dened by the solver for the
partial fullment of the Distributor demand.
The applicable constraints 1119 have essentially the same
meaning as the constraints 18 and constraint 10, except with the
decision variable e
i
(gas volume to be delivered to the Distributor
at delivery point i) replacing the decision variable n
i
(nominated
volume of gas to the Distributor at delivery point i) (Fig. 5).
4.3. Geographical information system (GIS)
ArcGIS from ESRI, a world leader in the Geographic Informa-
tion System (GIS) technology, was chosen to develop the system
discussed here. The system is based on georeferenced databases
and is used by large petroleum, gas and energy companies around
the globe. The Arc Pipeline Data Model (APDM) (used for the
development of the system described in this paper) is the
relational georeferenced data model and also the intellectual
property of ESRI. The APDM incorporates activities such as oil
transportation and distribution as well as the transportation and
distribution of oil derivatives, general liquids and natural gas.
5. Methodology
The methodology adopted for the MSGTL conception consid-
ered the following activities:
1. The Gasbol gas pipeline network modelling: the network is
composed of lines, equipment, accessories, terrain elevation
proles, gas ow equations, ambient and soil temperatures
and other relevant characteristics. It constitutes the basis for
the execution of the thermohydraulic simulation.
2. Monte Carlo simulation for equipment failure: it consists of
the statistical model, which considers the main components of
the gas-line network, each with its own failure rates. The
system was created using Microsoft Excel sheets and the risk
simulation software @Risk 4.5.
3. Thermohydraulic simulation: evaluates the operational beha-
viour of the pipeline network under different failures and
redundancy level scenarios. The thermohydraulic simulator
used here was the PipelineStudio, which can simulate both
transient and permanent regimes. The results obtained from
this simulation identied the total available transportation
capacity and the frequencies of failures for each failure scenario.
4. System availability study: compares the equipment failure
frequency results (activity 2) and the gas-line networks
transportation capacity (obtained from activity 3) to identify
the investment required for a certain redundancy level in a
certain scenario. This information is essential to dene the
most appropriate redundancy level that suits the total amount
available for investment.
5. Economic viability study: gathers all the information received
from the previous phases (activities 14) in a structured
analysis. It uses a Microsoft Excel spread sheet and the risk
simulation software @Risk 4.5 to explore economic premises,
such as the internal rate of return, the projects economic life,
the exposure to revenue loss, and the contractual penalties due
to incomplete gas deliveries. It also uses a Monte Carlo
simulation, which incorporates the statistical distribution of
the transportation capacities (activity 3) due to contingent
equipment failure (activity 2). The economic viability study is
the pillar on which the decision will be based in terms of the
optimum level of equipment redundancy.
6. Mathematical linear programming: consists of the mathema-
tical system model that can be solved through optimisation
techniques. The system of equations is led by an objective
function and a certain optimisation direction. The minimisa-
tion function denes the volume of gas that will actually be
delivered to each delivery demand point on a certain operation
day. Those volumes are chosen to minimise the effects of an
unexpected reduction in the gas supply or the constraints of
transportation capacity. The program denes which points will
be fully covered and which will be partially fullled or even
cut from the operational plan, according to previously deter-
mined criteria. The maximisation function allows the optimi-
sation of the available gas supply reallocation from the Loader
to the Distributor. It takes into consideration the contractual
constraints, gas prices and the required volumes from the
Distributor for a specic operation day. The system was
modelled using Microsoft Excel spreadsheets, and its optimisa-
tion was accomplished using the Solver tool.
7. Scenario builder application: consists of the integration mod-
ule of all the different technologies that compose the MSGTL
system (APDM, ArcGIS, PipelineStudio, linear mathematical
programming and the economic evaluation spread sheet).
Using this application, one can directly access the GIS and
APDM systems to execute the scenarios on the PipelineStudio.
6. Transportation system availability study
The BoliviaBrazil gas pipeline (Gasbol) consists of 557 km of
pipelines from Rio Grande (Bolivia) to the border between Bolivia
and Brazil, 1264 km from the border to Campinas (S~ ao Paulo,
Brazil), 1190 km from Campinas to Canoas (Porto Alegre,Brazil)
and a 153 km connection segment from Campinas to Guararema
(S~ ao Paulo, Brazil). The length of the complete line is 3164 km.
The current gas transportation capacity of the pipeline reaches
30 MMm
3
/day (millions of cubic metres per day). The Gasbol
project was implemented by the transportation companies Gas
Transboliviano S.A. (GTB) and Transportadora Brasileira Gasoduto
Bol via Brasil (TBG).
In order to map the availability and the redundancy invest-
ments for each company involved in the transportation
S
e
1
,n
1
,q
1
,p
1
e
2
,n
2
,q
2
,p
2
e
n
,n
n
,q
n
,p
n
Cp
1
,S
1
Cp
2
,S
2
Cp
n
,S
n
i=1 i=2 i=n
Transportation network
Delivery points
n
i
, e
i
Cpi, S
i
S
i+1
qmin, qmax
Detail of a
delivery
point
Fig. 5. Illustration of the pipeline process.
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4779
infrastructure, one has to consider that the business conguration
of the Gasbol project involves various companies. The simulations
were driven by the following parameters:
Software: Microsoft Excel, @RISK 4.5
Number of iterations: 5000
Sampling type: Latin hypercube
Random generator seed: Fixed and equal to 1
Standard recalculation: Expected value
Distributions considered: All
The 5000 iterations for each scenario were converted to
histograms for each type of failure frequency. These frequencies
were obtained from the Monte Carlo simulation process that had
been previously executed. Following the identication of the
failures and their particular frequencies, it was possible to start
the thermohydraulic simulations. These simulations were carried
out by considering each type of failure and its impact on the
transportation capacity of the pipeline network. Finally, with the
data extracted from the previous steps, the economic evaluation
for the adequate redundancy level could be prepared.
For Gasbols gas-line transportation network reliability assess-
ment, the methodology presented by Santos et al. (2006) was
applied to ten Gasbol compression stations (Brazilian side), for
which the authors present a comparative statistical evaluation
based on the binomial discrete probability distribution associated
with the Monte Carlo simulations. Because the binomial evalua-
tion required that every single piece of equipment has the same
availability (i.e., probability of failure), it was necessary to
consider some simplications for the system. In Santos et al.
(2006), it was assumed that the compressing units were identical
and that each station had the same number of units. The Monte
Carlos simulation is not subject to such a constraint and can be
applied to any conguration prole, including proles with differing
numbers of units and different availabilities associated with then.
The authors highlight the applicability of the Monte Carlos simula-
tion method due to its simplicity and greater exibility.
6.1. Availability of compression units
Data from the Electric Power Research Institute, Epri (1999)
and the North American Electric Reliability CouncilNERC (2005)
compression stations were considered. Both sources are inter-
nationally respected and their work is strongly based on statis-
tical data obtained from the compression stations (EPRI) and
thermoelectrical generation (NERC) operations. For this study, the
availability value of 0.9294 (92.94%) was adopted from NERCs
methodology instead of the 0.971 (97.10%) found in the EPRI
study. According to the operations team of the transportation
company Transportadora Brasileira Gasoduto Bol via Brasil (TBG),
NERCs value was slightly closer to the last eight years of Gasbol
operations history than ERPIs value. Additionally, the NERC value
expects more failures between the operational time horizons than
the ERPI value, reecting a more conservative approach. As shown
in Figs. 6 and 7, a simplied approach to arranging the compres-
sion stations was maintained. In these gures, it is possible to
compare how the compression units are arranged in reality to
how they are arranged using the simplied approach. Compres-
sion units 1 through 4 are on the Bolivian side, and units 5 through
14 are on the Brazilian side.
6.2. Studied congurations
The results of the study of stations on the Brazilian side are
presented below. Three alternatives to the conguration of the
spare compression units were considered:
1. Conguration with no spare compression units
2. Conguration with 5 spare compression units
3. Conguration with 10 spare compression units.
Tables 25 present the thermohydraulic and Monte Carlo
simulation results. The availability level for each conguration
is shown in Table 1.
The average availability for each case, considering the installed
redundancy level (i.e., the number of spare compression units) is
shown in Table 2 (0 spare units), Table 3 (5 spare units), Table 4
(10 spare units) and Table 5 (summary of all of the congura-
tions). In Tables 24, column 1 indicates the average capacity
resulting from a failure in the station denoted in column 2.
Column 3 indicates the average number of occurrences of one
failure in one unit in days per year. Column 4 indicates the
average number of occurrences of two failures in one station in
days per year. Column 5 indicates the average capacity corre-
sponding to failures in two adjacent stations, which are indicated
#5
4 x 7000 hp
#1
2 x 19500 hp
3 x 7000
1 x 15000 hp
2 x 15000 hp
#2 #3 #4 #6 #7 #8 #9 #10 #11 #12 #13 #14
4 x 7000 hp
2 x 15000 hp 2 x 15000 hp
Fig. 6. Gasbols compression stations, as installed.
#5
2 x 15000 hp
#1 #2 #3 #4 #6 #7 #8 #9 #10 #11 #12 #13 #14
Fig. 7. Simplied array of Gasbols compression stations.
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4780
in column 6. The number of failures in days in one year for each
combination of stations is presented in column 7. Table 5 sum-
marises, for each spare unit conguration and for each failure
condition (0, 1, 2, or (11) adjacent), the total number of
occurrences in days per year. This table also presents the average
capacity and availability of each spare conguration.
The average available capacity can be obtained by calculating
the sum for each product of the capacity by failure occurrence
and dividing it by the total number of days in a year. Only
compression unit failures were considered in the equation; it does
not take other factors into account.
Average Capacity
P
uAU
Capacity
u
Failure Occurence
u
365
20
U represents the set of compression units.
The systems availability can be obtained by dividing the
average capacity by the gas-line rm contracted capacity.
Systems Availability
Average Capacity
Contracted Capacity
21
Fig. 8, which shows a graph representing the capacity curve
versus the percentage of occurrence frequency, indicates the risk
exposure incurred by failing to full the contracted rm capacity
of 30.08 MMm
3
/day for each conguration of spare compression
units. Fig. 8 clearly indicates that the more spare units installed,
the lesser the probability that the system will operate below
the rm-contracted capacity. The bars in Fig. 8 represent the
Table 1
Availability for each conguration.
Conguration Availability
No spare compression unit (1) 0.9056
With 5 spare compression units (2) 0.9573
With 10 spare compression units (3) 0.9956
Table 2
Monte Carlo and thermohydraulic simulations resultsNo spare compressor units.
1 Failure/station 2 Failures/station (11) Adjacent failures
Capacity MMm
3
/day Station Failure occurrence Failure occurrence Capacity MMm
3
/day Stations Failure occurrence
25.90 #5 36.07 1.79 21.60 5&6 5.93
26.10 #6 31.57 1.86 21.70 6&7 5.07
26.70 #7 29.07 1.79 21.90 7&8 4.93
26.70 #8 24.50 1.71 22.00 8&9 5.43
27.20 #9 19.21 1.79 23.00 9&10 5.86
27.50 #10 19.14 1.71 23.70 10&11 4.29
28.70 #11 17.57 1.71 24.40 11&12 5.21
29.80 #12 14.21 1.64 24.90 12&13 4.14
29.30 #13 12.29 1.71 21.90 13&14 5.29
25.80 #14 11.21 1.86
Table 3
Monte Carlo and thermohydraulic simulation results5 spare compressor units.
1 Failure/station 2 Failures/station (11) Adjacent failures
Capacity MMm
3
/day Station Failure occurrence Failure occurrence Capacity MMm
3
/day Stations Failure occurrence
25.90 #5 1.7 0 21.60 5&6 0
26.10 #6 1.6 0 21.70 6&7 0
26.70 #7 1.5 0 21.90 7&8 0.1
26.70 #8 1.4 0 22.00 8&9 0
27.20 #9 1.4 0 23.00 9&10 0.3
27.50 #10 39.3 1.8 23.70 10&11 5.5
28.70 #11 33.9 1.9 24.40 11&12 5.7
29.80 #12 30.0 1.8 24.90 12&13 5.0
29.30 #13 25.3 1.8 21.90 13&14 5.6
25.80 #14 24.9 1.9
Table 4
Monte Carlo and thermohydraulic simulation results10 spare compressor units.
1 Failure/station 2 Failures/station (11) Adjacent failures
Capacity MMm
3
/day Station Failure occurrence Failure occurrence Capacity MMm
3
/day Stations Failure occurrence
25.90 #5 1.8 0 21.60 5&6 0
26.10 #6 1.9 0 21.70 6&7 0
26.70 #7 1.7 0 21.90 7&8 0.1
26.70 #8 1.7 0 22.00 8&9 0
27.20 #9 1.8 0 23.00 9&10 0
27.50 #10 1.7 0 23.70 10&11 0
28.70 #11 1.7 0 24.40 11&12 0
29.80 #12 1.6 0 24.90 12&13 0
29.30 #13 1.7 0 21.90 13&14 0
25.80 #14 1.9 0
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4781
frequencies of occurrence of each value of capacity resulting from
the spare congurations of compressor equipment. Without any
spare units, the maximum capacity of 30 MMm
3
/d is reached at a
frequency below 60%, whereas values of 26.8727.79 MMm
3
/d
are observed with at a frequency below 10% each. With 5 spare
units, the maximum capacity is reached in almost 80% of the
cases, and values ranging from 27.0527.79 MMm
3
/d are also
observed. With 10 spare units, the maximum capacity is reached
in almost 100% of the cases. As summarised in Table 5, without
any spare units, the availability of the compression system is
0.9055; with 5 spare units, the availability rises to 0.9673 and the
availability reaches a maximum of 0.9956 with 10 spare units at
each compression station.
7. Economic evaluation
According to Santos (2003), economic evaluation for the
purpose of decision making has been a subject of major attention,
both in the academic community and in the corporate world.
When it comes to evaluating projects, one of the most powerful
approaches has been the use of the Discounted Cash Flow (DCF)
method with the adoption of Net Present Value (NPV) associated
with the expected Internal Rate of Return (IRR). Several authors,
such as Ross et al., (2004) and Copeland et al., (2000), have
strongly recommended this approach, and it has been widely
disseminated in the corporate environment where it has gained
many adepts and defenders. However, the use of such methodol-
ogy in isolation is not enough to support efcient decision
making, even if the methodology is associated with the analysis
of what if scenarios. For example, this approach does not
constitute a quantitative risk analysis and assigns the same
probabilistic weight for each scenario, including scenarios in
which all the variables assume their maximum/minimum values
(Vose, 1996).
According to Hertz (1984), a project is always vulnerable to the
uncertainties created by the volatile aspects of production (such
as the market conditions, human resources and deadlines) that
can be associated with high levels of stochasticity. Such uncer-
tainties include the material and service costs, deadlines, the
requirement of environmental licences, workforce management
and other similar aspects whose occurrence are strongly
associated with probabilities and correlations among themselves.
These probabilities and, in particular, these correlations must be
known to clearly identify the risks involved in the project so that
one can properly mitigate, or even eliminate, those risks. Hertz
(1984) warns that combined uncertainties within a project can
generate global risks of critical proportions to that project. Risk
analysis in the context of the gas industry has also been studied
by Pelletier and Wortmann (2009), who used the probability of
negative NPV for the investment as the measure of risk.
The Monte Carlo simulation has been used to probabilistically
evaluate the risk involved in projects that use the DCF method.
Thus, an investor can measure the risk level associated with his
project and then identify actions to mitigate that risk (and/or
discount taxes that can absorb it), reducing or even eliminating
his risk exposure.
As a consequence, if the uncertainties are incorrectly
accounted for during the project evaluation, it can generate
adverse future results in terms of negative NPV. This situa-
tion could turn into an economic disaster for the corporation,
compromising the image of the company among both its share-
holders and society.
The Monte Carlo simulations, together with the thermohy-
draulic simulations conducted for Gasbols pipeline network, had
the basic goal of determining the global systems transportation
availability in terms of capacity. The basic purpose was to dene
the rm capacity available to dene the ship-or-pay contracts
between the Transporter and the Loader. The rm transportation
capacity must be considered when the gas-line transportation tax
is calculated, chiey because the company must pay back the
whole of the capital investment and cover the operational costs
throughout the gas pipelines economic life. Such a procedure had
not been adopted previously for the gas-line network project,
thus exposing the Transporter to economic risks from contractual
penalties due to the failure to full its contractual rm volumes.
The Transporter also exposes himself to the risks associated with
the need for new investments in spare equipment installation,
which can lead to the underestimation of the transportation tax
and, thus, reduce the internal rate of return of the project.
Because the costs of spare equipment were not originally
considered in Gasbols capital costs, it was necessary to begin
an economic evaluation to dene the total number of spare
compression units that must be installed. The required number
of spare compression units was determined using both Monte
Carlo simulations for failure and thermohydraulic simulations.
Because the Gasbol project was constituted by the transporta-
tion companies Gas Transboliviano S.A. (GTB) and Transportadora
Brasileira Gasoduto Bol via Brasil (TBG), economic studies evaluat-
ing spare equipment installation were conducted separately for
the Brazilian and Bolivian sections.
The main objective of the economic evaluation was to identify
the most adequate redundancy level to manage the revenue
losses and the risk of exposure to contractual penalties. The
discounted cash ow methodology was adopted to compare the
results for three different conditions. The aim here was to identity
which condition had the highest present liquid value. The
volumes delivered and the penalties avoided were considered as
income and the installation costs of the spare compression units
Table 5
Monte Carlo and thermodynamic simulation results for each conguration.
Without spare
stations
With 5 spare
stations
With 10 spare
stations
Failure (units by
station)
Occurrences in
one year
Occurrences in
one year
Occurrences in one
year
0 86.43 172.93 347.43
1 214.86 160.79 17.5
2 17.57 9.07 0
(11) Adjacent
stations
46.14 22.21 0.1
Average capacity 27.2403 28.7957 30
Firm contractual
capacity
30.08 30.08 30.08
Availability 0.9055 0.9673 0.9956
Fig. 8. Capacity versus percent frequency versus # of spare compression units.
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4782
were considered as investments in each condition. There were no
additional costs related to the spare units fuel and maintenance
consumption because the operational time of these units is
limited.
The present value of cash ows of contractual penalties and
potential loss of revenue utilises a discount rate of 15% per annum
(pa), which is a conservative value. A rate normally used for
infrastructure projects is approximately 12%. When we consider
the projects nancial leverage (e.g., 30% equity70% debt) and
depending on the funding costs in the nancial market, the
discount rate on equity of the project sponsors may rise signi-
cantly. The use of discount rates for the project of less than 15%
makes the results even more favourable to the adoption of the
proposed solution.
7.1. Economical premises
Firm contractual capacity: 30.08 MMm
3
/d
Superior caloric power: 36,480 BTU/m
3
Transportation tax: 1.20 US$/MMBTU
Revenue loses: 1non-delivered capacity
Contractual penalties: 1non-delivered capacity
Contractual duration: 10 years
Discount rate: 15% yearly
Spare unit installation cost: US$ 12,900,000.00
7.2. Results from the economic evaluation
The results of the economic evaluation are summarised in Table 6.
Columns Min, Average and Max indicate the values of the decision
variables described in column Name for each conguration of spare
compression units. The values in column x1 are the values for each
variable associated with the probability of 5% in column p1. The same
occurs with column x2 for the probability of 95% in column p2. In
column x2x1, the differences between columns x2 and x1 are the
values associated with the probability of 90% occurrence. The results
of the economic evaluation (based on the previous premises) identi-
ed alternative number 3 as the most viable (i.e., the installation of
ten spare compression units) according to the NPV obtained, which is
the highest among the alternatives (i.e., not to install any redundancy
or to install ve spare units). The NPV for the conguration without
spare units is equal to the PV of losses once this alternative implies no
extra investment. Alternative 3, as shown in Table 6, is the only
alternative that does not present a capacity loss (on average) while
showing lower revenue losses and contractual penalties. The combi-
nation of these aspects leads alternative 3 to a more protable NPV.
The NPV for alternative 3 is MMUS$ 305.02, as indicated in Table 6.
8. Conclusions
The implementation and use of the MSGTL are fundamentally
important for the management of a natural gas business due to
the complexity of networks and the contractual relationship
among the agents. Such contractual relationships can be compli-
cated by the many penalty clauses resulting from the failure to
full obligations related to supply, transportation and the receiv-
ing and delivery of natural gas.
Using the methodology presented in this study to assess the
transportation systems optimum reliability level is fundamentally
important for the safekeeping of the transportation processs nancial
equilibrium under a ship-or-pay contractual structure that penalises
the failure to full rm contracted volumes. In the case of Gasbol, an
investment in spare compression units was shown to be economically
viable and benecial to the Transporter, as indicated by the NPV for
the Brazilian section of the gas pipeline. The adoption of the Monte
Carlo method for risk simulation, together with the Discounted Cash
Flow (DCF) method, provides a statistical understanding of the
exposure to revenue loss, contractual penalties and the Net Present
Value (NPV) of the alternatives. Such an understanding allows the
decision maker to identify the expected value trust intervals. This
Table 6
Comparative summary of the spare compression units installation.
Name Min. Average Max. x1 p1 (%) x2 p2 (%) x2x1 p2p1 (%)
Without spare compression unit
Transportation capacity (MMm
3
/d) 21.60 27.24 30.08 21.90 5 30.08 95 8.18 90
Loss of capacity (MMm
3
/d) 8.4800 2.8397 8.1800 5 95 8.1800 90
Loss of revenue (MMUS$) 483.21 227.71 10.19 338.32 5 125.43 95 212.89 90
Contractual penalty (MMUS$) 483.21 227.71 10.19 338.32 5 125.43 95 212.89 90
PV of losses (MMUS$) 966.41 455.43 20.38 676.64 5 250.86 95 425.78 90
NPV (MMUS$) 966.41 455.43 20.38 676.64 250.86 95 425.78 90
With 5 spare compression units
Transportation capacity (MMm
3
/d) 21.9 28.80 30.08 24.9 5 30.08 95 5.18 90
Loss of capacity (MMm
3
/d) 8.1800 1.2843 5.1800 5 95 5.1800 90
Loss of revenue (MMUS$) 333.58 103.00 196.95 5 28.74 95 168.21 90
Contractual penalty (MMUS$) 333.58 103.00 196.95 5 28.74 95 168.21 90
PV of losses (MMUS$) 667.16 205.99 393.90 5 57.48 95 336.42 90
Capacity recovery (MMm
3
/d) 8.1800 1.5554 8.4800 3.9800 5 7.0800 95 11.0600 90
Avoided revenue (MMUS$) 199.87 124.72 401.59 14.48 5 261.29 95 275.77 90
Avoided contractual penalty (MMUS$) 199.87 124.72 401.59 14.48 5 261.29 95 275.77 90
PV of revenues (MMUS$) 399.74 249.43 803.17 28.95 5 522.59 95 551.54 90
NPV (MMUS$) 464.24 184.93 738.67 93.43 5 458.19 95 551.62 90
With 10 spare compression units
Transportation capacity (MMm
3
/d) 21.9 29.95 30.08 30.08 5 30.08 95 0 90
Loss of capacity (MMm
3
/d) 8.1800 0.1334 5 95 90
Loss of revenue (MMUS$) 212.48 10.70 51.35 5 95 51.35 90
Contractual penalty (MMUS$) 212.48 10.70 51.35 5 95 51.35 90
PV of losses (MMUS$) 596.84 21.41 95.28 5 95 95.28 90
Capacity recovery (MMm
3
/d) 4.2800 2.7062 8.4800 5 8.1800 95 8.1800 90
Avoided revenue (MMUS$) 47.34 217.01 483.21 109.68 5 329.57 95 219.89 90
Avoided contractual penalty (MMUS$) 47.34 217.01 483.21 109.68 5 329.57 95 219.89 90
PV of revenues (MMUS$) 94.69 434.02 966.41 219.35 5 659.13 95 439.78 90
NPV (MMUS$) 223.69 305.02 837.41 90.35 5 530.13 95 439.78 90
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4783
methodology also provides the decision maker with better insight
into the risks associated with the investment alternatives, thus
supporting him with a quality statistical basis for decision making.
These ndings highlight the need to incorporate a statistical
approach into the development of the new natural gas-line
project. The thermohydraulic simulations under permanent and
transient regimes, together with the Monte Carlo simulation,
allow us to assess the availability of the gas pipeline networks
components and to measure the effects of failure on the networks
gas transportation capacity.
The proposed management system aims at the improvement
of energy security and may be applied to other gas pipelines in
Brazil, such as the recently opened pipeline UrucuManaus. In
this region, the supply of gas is a critical issue due to the
dependency of the energy matrix of the region on fossil fuels for
the generation of electricity (Frota et al., 2010).
Acknowledgements
The rst author thanks PETROBRAS for supporting this tech-
nological development in association with PUC-Rio. The second
author thanks the National Council of Research of Brazil (CNPq)
for the grant that supported this work. The third author thanks
the CAPES agency for the scholarship that supported his research.
References
Copeland, T., Koller, T., Murrin, J., 2000. Valuation: measuring and managing the
value of companies. John Wiley & Sons, New York.
Electric Power Research Institute, Epri, 1999. Report no. RP 4CH2983. California.
Evans, J.R., Olson, D.L., 1998. Introduction to simulation and risk analysis. Prentice
Hall, Upper Sadle River.
Fernandes, E., De Oliveira, J.C.S., De Oliveira, P.R., Alonso, P.S.R., 2008. Natural-gas-
powered thermoelectricity as a reliability factor in the Brazilian electric sector.
Energy Policy 36, 9991018.
Frota, W.M., Sa , J.A.S., Moraes, S.S.B., Rocha, B.R.P., Ismail, K.A.R., 2010. Natural gas:
The option for a sustainable development and energy in the state of Amazonas.
Energy Policy 38, 38303836.
Hertz, D., 1984. Practical risk analysis: An approach through case histories. John
Wiley & Sons, New York.
Kabirian, A., Hemmati, M.R., 2007. A strategic planning model for natural gas
transmission networks. Energy Policy 35 (11), 56565670.
Lise, W., Hobbs, B.F., Oostvoorn, F.V., 2008. Natural gas corridors between the EU
and its main suppliers: Simulation results with the dynamic GASTALE model.
Energy Policy 36, 18091906.
Mathias, M.C., Szklo, A., 2007. Lessons learned from Brazilian natural gas industry
reform. Energy Policy 35, 64786490.
Monforti, F., Szikszai, A., 2010. A Monte Carlo approach for assessing the adequacy
of the European gas transmission system under supply crisis conditions.
Energy Policy 38, 24862498.
North American Electric Reliability Council, 2005. Report of January 2005. New
Jersey.
Omonbude, E.J., 2007. The transit oil and gas pipeline and the role of bargaining: A
non-technical discussion. Energy Policy 35, 61886194.
Pandian, S., 2005. The political economy of trans-Pakistan gas pipeline project:
assessing the political and economic risks for India. Energy Policy 33, 659670.
Pelletier, C., Wortmann, J.C., 2009. A risk analysis for gas transport network
planning expansion under regulatory uncertainty in Western Europe. Energy
Policy 37, 721732.
Potocnik, P., Thaler, M., Govekar, E., Grabec, I., Poredo s, A., 2007. Forecasting risks
of natural gas consumption in Slovenia. Energy Policy 35, 42714282.
Ragsdale, C.T., 2006. Spreadsheet modeling and decision analysis. South-Western
College Publishing, Cincinnati.
Ross, S.A., Westereld, R.W., Jaffe, J.F., 2004. Corporate Finance. McGraw-Hill,
Boston.
Santos, S.P., 1997. Transient analysis a must in gas pipeline design. In: Pipeline
Simulation Interest Group, 29., 1997, Tucson. ProceedingsyHouston: PSIG.
Santos, S.P., 2003. Viabillidade de projetos sob condic- ~ oes de risco. In: Simpo sio
Brasileiro De Pesquisa Operacional, 35., 2003, Natal. Anais, Rio de Janeiro:
SBPO.
Santos, S.P.,2008. Sistema de gest~ ao logstica de transporte de ga s por gasodutos.
2008.101 f. dissertac- ~ ao (Opc- ~ ao Prossional)Departamento de Engenharia
Industrial, Pontifcia Universidade Cato lica do Rio de Janeiro, Rio de Janeiro.
Santos, S.P.; Bittencourt, M.A.S.; Vasconcellos, L.D.,2006. Compressor station
availabilityManaging its effects on gas pipeline operation. In: International
Pipeline Conference, 6., 2006, Calgary. ProceedingsyNew York: ASME.
S oderbergh, B., Jakobsson, K., Aleklett, K., 2010. European energy security: An
analysis of future Russian natural gas production and exports. Energy Policy
38, 78277843.
Sovacool, B.K., 2009. Energy policy and cooperation in Southeast Asia: The
history,challenges,and implications of the trans-ASEANgaspipeline(TAGP) net-
work. Energy Policy 37, 23562367.
Vose, D., 1996. Quantitative risk analysis: A guide to Monte Carlo Simulation
Modelling. John Wiley & Sons, New York.
S.P. dos Santos et al. / Energy Policy 39 (2011) 47744784 4784

You might also like